Advances in Experimental Social Psychology, Volume 35 (Advances in Experimental Social Psychology)

CONTRIBUTORS Numbers in parentheses indicate the pages on which authors’ contributions begin.

Peter J. Carnevale (235), Department of Psychology, New York University, New York, NY 10003 Carsten K. W. de Dreu (235), Department of Psychology, University of Amsterdam, 1018 WB Amsterdam, THE NETHERLANDS Daniel T. Gilbert (345), Department of Psychology, Harvard University, Cambridge, MA 02138 E. Tory Higgins (293), Department of Psychology, Columbia University, New York, NY 10027 Michael A. Hogg (1), Centre for Research on Group Process, University of Queensland, Brisbane, OLD 4072, AUSTRALIA Olivier Klein (135), Service de Psychologie Sociale, Universite´ Libre de Bruxelles, 1050 Bruxelles, BELGIUM Daan van Knippenberg (1), School of Management, Erasmus University, Rotterdam, 3000 DR, Rotterdam, THE NETHERLANDS Arie W. Kruglanski (293), Department of Psychology, University of Maryland, College Park, MD 20742 Mario Mikulincer (53), Department of Psychology, Bar-Ilan University, Ramat Gan 52900, ISRAEL Antonio Pierro (293), Department of Development and Socialization, Univerita` di Roma ‘‘La Sapienza,’’ Rome, ITALY Phillip R. Shaver (53), Department of Psychology, University of California, Davis, CA 95616 Mark Snyder (135), Department of Psychology, University of Minnesota, Minneapolis, MN 55455 Timothy D. Wilson (345), Department of Psychology, University of Virginia, Charlottesville, VA 22904

ix

SOCIAL IDENTITY AND LEADERSHIP PROCESSES IN GROUPS

Daan van Knippenberg

Gandhi, Mandela, Roosevelt—all great leaders, but without followers would they have been able to lead? Leadership is a relational term—it identifies a relationship in which some people are able to influence others to embrace, as their own, new values, attitudes, and goals, and to exert effort on behalf of and in pursuit of those values, attitudes, and goals. The relationship is almost always configured by and played out within the parameters of a group—a small group like a team, a medium sized group like an organization, or a large group like a nation. Arguably, good leadership inspires others to adopt values, attitudes, and goals, and to behave in ways that serve the group as a collective, and that define membership of the group (e.g., Bass, 1985; Burns, 1978). Thus, effective leaders are able to transform individual action into group action. Leadership is a core feature of social groups—it is very difficult to think about groups without thinking about who leads or manages them, and about how well they are led and managed. This characterization of leadership, which is not uncommon (e.g., Cartwright & Zander, 1968; Chemers, 2001; Conger & Kanungo, 1998), places a premium on the role of group membership and group life in the analysis of leadership. Leadership is quite clearly a social psychological phenomenon that is inextricably grounded in social relations, group life, and the psychology of group membership. Not surprisingly, the study of leadership has long been a central concern of social scientists—the literature is enormous, stretching back to Plato and beyond—and in particular of social psychologists. What is surprising, then, is that in the past 25 years there has been relatively little leadership research conducted within social psychology. Only in the past few 1 ADVANCES IN EXPERIMENTAL SOCIAL PSYCHOLOGY, VOL. 35

Copyright 2003 Elsevier Science (USA). All rights reserved. 0065-2601/03 $35.00

2

HOGG AND VAN KNIPPENBERG

years has this changed—there has been a strong resurgence, framed to a significant extent by a broad social identity perspective in social psychology. Our goal in this chapter is to review and integrate key components of this revival. More specifically, we describe how a focus on group membership, framed by the social identity perspective, can explain important aspects of leadership. We provide a brief critical and historical overview of current leadership research, and an overview of key components of the social identity perspective. We then provide a detailed account of the social identity theory of leadership and a detailed review of empirical support. The final part of the chapter explores implications and extensions of the analysis in the context of the study of leadership.

I. A Brief Historical Overview of Leadership Research A. LEADERSHIP RESEARCH MOVES FROM SOCIAL PSYCHOLOGY TO THE ORGANIZATIONAL SCIENCES Leadership is about social relations, usually within a group, and about changing people’s behaviors and attitudes to conform to the leader’s vision for the group. Not surprisingly, the study of leadership has been an important research topic for social psychology. This was particularly the case during the boom years of small group dynamics (e.g., Cartwright & Zander, 1968; Shaw, 1981). It has also been a part of some of social psychology’s classic research programs (e.g., Bales, 1950; Hollander, 1958; Lippitt & White, 1943; Sherif, 1966; Stogdill, 1974). Moving from a focus on traits that distinguish effective from ineffective leaders (for a review, see Stogdill, 1974), via a focus on broad categories of leader behavior as universal predictors of leadership effectiveness (e.g., Stogdill & Coons, 1957), this tradition of leadership research culminated in Fiedler’s (1964, 1971) contingency theory—the leadership effectiveness of a particular behavioral style is contingent on the favorability of the situation to that behavioral style. During the 1970s and 1980s there was a new excitement in social psychology with attribution processes, and then social cognition (e.g., Devine, Hamilton, & Ostrom, 1994; Fiske & Taylor, 1991), and with large-scale intergroup relations and the social context of human behavior (e.g., Tajfel, 1984). These developments were associated with a decline in interest in small interactive groups (Steiner, 1974, 1986), which impacted the study of leadership. The study of small group processes and of leadership shifted to neighboring disciplines, most notably organizational psychology (Levine & Moreland, 1990, 1995; McGrath, 1997; Sanna & Parks, 1997; Tindale & Anderson,

SOCIAL IDENTITY AND LEADERSHIP PROCESSES IN GROUPS

3

1998). Thus, the body of research on contingency theories of leadership took place largely outside social psychology, in the organizational sciences (e.g., House, 1971; Kerr & Jermier, 1978; Vroom & Yetton, 1973). Here leadership research found a natural home (e.g., Bass, 1990a; Yukl, 2001). 1. Leader Categorization Theory Within social psychology, the emerging interest in attribution processes and social cognition had its impact on the study of leadership. There was a shift from contingency theories of effective leader behavior to a focus on factors leading people to attribute events to leadership (e.g., Meindl, Ehrlich, & Dukerich, 1985) or to view an individual as an effective leader (e.g., Lord & Maher, 1991). In what is probably the most extensive of these research programs, Lord and his colleagues developed leader categorization theory (e.g., Lord, Foti, & DeVader, 1984; Lord & Maher, 1991; Nye & Forsyth, 1991; Palich & Hom, 1992; Rush & Russell, 1988; also see Nye & Simonetta, 1996). Based on implicit leadership theory (Hollander & Julian, 1969) and on contemporary social cognition principles, leader categorization theory states that people have general and task-specific leader schemas that are activated when a person is categorized as a leader. The schema then generates assumptions about the leader’s attributes and behaviors, and if there is a good fit of the person to the schema then the person is considered an effective leader. This perspective treats leader categories as nominal categories, that is, cognitive groupings of instances that share attributes, but do not have any psychological existence as a real human group. Leadership is viewed as a product of individual information processing, not as a structural property of real groups, nor as an intrinsic or emergent property of psychological group membership. What may be missing from this analysis is an analysis of leadership as a relationship between people within a human group. After all, leadership identifies a particular relationship among people, and so a group made up only of leaders is implausible—who would lead and who would follow? 2. Charismatic and Transformational Leadership Although the study of social-cognitive processes provided some linkage with developments in social psychology, the mainstream of leadership research remained focused on leadership effectiveness. When leadership research’s central focus became charismatic and transformational leadership, this refocusing took place outside of social psychology. Starting in the late 1970s (Burns, 1978; House, 1977), and gathering momentum in the mid-1980s (Bass, 1985; Conger & Kanungo, 1987), charismatic and

4

HOGG AND VAN KNIPPENBERG

transformational leadership soon became leadership research’s central focus. Charismatic/transformational leaders motivate followers to work for collective goals that transcend self-interest and transform organizations (Bass, 1990b; Bass & Avolio, 1993; Shamir, House, & Arthur, 1993; see Mowday & Sutton, 1993, for critical comment). This focus on charisma is particularly evident in ‘‘new leadership’’ research (e.g., Bass, 1985, 1990b, 1998; Bryman, 1992; Burns, 1978; Conger & Kanungo, 1987, 1988), which proposes that effective leaders should be proactive, change-oriented, innovative, motivating, and inspiring, and have a vision or mission with which they infuse the group. They should also be interested in others, and be able to create commitment to the group, and extract extra effort from and empower members of the group. 3. Leader–Member Exchange (LMX) Theory Another significant organizational perspective on leadership is represented by leader–member exchange (LMX) theory (Graen & Uhl-Bien, 1995; Schriesheim, Castro, & Cogliser, 1999; Sparrowe & Liden, 1997). LMX theory focuses on dyadic exchange relationships between the leader and specific subordinates. The quality of these relationships can vary widely. Low-quality LMX relationships are ones where subordinates are disfavored by the leader and thus receive fewer valued resources. Leader–subordinate exchanges simply adhere to the terms of the employment contract, with little attempt by the leader to develop or motivate the subordinate. In contrast, high-quality LMX relationships are ones where subordinates are favored by the leader and thus receive many valued resources. Leader–subordinate exchanges go beyond the formal employment contract, with managers showing influence and support, and giving the subordinate greater autonomy and responsibility. LMX theory predicts that effective leaders should develop high LMX relationships with their subordinates, which should enhance subordinates’ well-being and work performance.

B. LEADERSHIP RESEARCH REVISITED Over the past 20 years, social psychology has, through social cognition, become more sophisticated in its methods and theories (Devine, Hamilton, & Ostrom, 1994), and, through the social identity perspective, begun once again to focus on group processes, intergroup phenomena, and the collective self (Abrams & Hogg, 1998; Moreland, Hogg, & Hains, 1994; Sedikides & Brewer, 2001). There has been a reinvigorated focus on leadership (e.g., Chemers, 2001; Lord, Brown, & Harvey, 2001; Messick & Kramer, in press;

SOCIAL IDENTITY AND LEADERSHIP PROCESSES IN GROUPS

5

van Knippenberg & Hogg, in press), an integration of social cognition and social identity approaches within social psychology (Abrams & Hogg, 1999), and a closer relationship between social identity research and organizational psychology (e.g., Haslam, 2001; Haslam, van Knippenberg, Platow, & Ellemers, in press; Hogg & Terry, 2000, 2001; van Knippenberg & Hogg, 2001). The recent social psychological focus on leadership has raised some concerns about contemporary organizational psychology leadership research. Although most research now acknowledges that leadership is a relational property within groups (i.e., leaders exist because of followers, and followers exist because of leaders; e.g., Hollander, 1995), the idea that effective leadership may be grounded in ordinary social-cognitive processes associated with psychologically belonging to a group has not really been elaborated. Instead, the most recent organizational psychology emphases are upon (1) individual cognitive processes that categorize individuals as leaders—the social orientation between individuals is not considered, and thus group processes are not incorporated, (2) dyadic relationships between leaders and individual followers—leadership is a matter of the development of favorable or unfavorable interpersonal leader–member exchange relationship, or (3) whether individuals have the charismatic properties necessary to meet the transformational objectives of leadership—leadership is a matter of situationally attractive individual characteristics rather than group processes. All three of these perspectives have attracted criticism for neglecting the effects of larger social systems (e.g., groups) within which the individual is embedded (e.g., Hall & Lord, 1995; Hogg & Martin, 2003; Lord, Brown & Harvey, 2001; Pawar & Eastman, 1997; also see Chemers, 2001; Haslam & Platow, 2001a). Lord, Brown, & Harvey (2001) explain that leadership cannot be properly understood in terms of a leader’s actions alone or in terms of abstract perceptual categories of types of leader. They advocate a paradigm shift in how we understand leadership. Haslam and Platow (2001a) express the same concern, and warn against any explanation of leadership that rests too heavily, or at all, on invariant properties of individuals and their personalities. These concerns about leadership research are consistent with wider metatheoretical stances concerning the explanation of group processes in general. The warning that a focus on personality, asocial cognition, dyadic relationships, and decontextualized groups does not provide an adequate explanation of group processes and the collective self is one that forms the metatheoretical context for the social identity perspective (see Taylor & Brown 1979; also see Hogg, 2001b; Hogg & Williams, 2000; Turner, 1999). The aim of this chapter is to offer a social identity analysis of leadership, as a group membership-based perspective on leadership. This perspective has attracted growing attention in recent years, and has produced a number of

6

HOGG AND VAN KNIPPENBERG

conceptual and empirical publications. We first briefly overview the social identity perspective and the social identity analysis of social influence. Then we provide a detailed description of the social identity analysis of leadership and an examination of relevant empirical tests.

II. Social Identity and Social Influence A. THE SOCIAL IDENTITY PERSPECTIVE The social identity perspective (Hogg & Abrams, 1988; Tajfel & Turner, 1986; Turner, Hogg, Oakes, Reicher, & Wetherell, 1987; also see Brewer, 1991; Brewer & Gardner, 1996) has become increasingly central to social psychology, and has recently been summarized in detail elsewhere (e.g., Abrams & Hogg, 2001; Hogg, 2001c, 2003). We provide only a brief overview of key features here. From the social identity perspective, a group exists psychologically when people share a self-conception in terms of the defining features of a selfinclusive social category. More specifically, this representation of the group is a prototype—a fuzzy set of features that captures ingroup similarities and intergroup differences regarding beliefs, attitudes, behaviors, and feelings. Prototypes are configured according to the principle of metacontrast, to maximize the ratio of intergroup differences to intragroup differences, or, in other words, to maximize the entitativity of the group (e.g., Hamilton & Sherman, 1996; Hamilton, Sherman, & Castelli, 2002; Hamilton, Sherman, & Lickel, 1998; Yzerbyt, Castano, Leyens, & Paladrino, 2000; Yzerbyt, Corneille, & Estrada, 2001). 1. Social Categorization and Depersonalization A key insight of the social identity perspective is that the basis of perception, attitudes, feelings, behavior, and self-conception is contextually fluid. Self-conception can vary from being based on idiosyncratic personal attributes and the unique properties of a specific interpersonal relationship, to being based on a shared representation of ‘‘us’’ defined in terms of an ingroup prototype. In the latter case, the situation represents a group situation and perceptions, attitudes, feelings, and behavior acquire the familiar characteristics of inter- and intragroup behaviors—conformity, normative behavior, solidarity, stereotyping, ethnocentrism, intergroup discrimination, ingroup favoritism, and so forth. Put another way, the more that an aggregate of people is a salient basis for self-conception as a group

SOCIAL IDENTITY AND LEADERSHIP PROCESSES IN GROUPS

7

member, then the more strongly is self-definition, perception, cognition, affect, and behavior based upon prototypicality. When group membership is the salient basis of self-conception, people, including self, are represented and treated in terms of the relevant ingroup- or outgroup-defining prototype. Self-categorization depersonalizes self in terms of the ingroup prototype (producing self-stereotyping, conformity, normative behavior, social attraction, social identification, and so forth), and it depersonalizes perception of others so that they are seen as reflections of the relevant prototype. 2. Motivation Because groups define self, the social value or status of a group becomes the social value or status of self. Intergroup relations may become, therefore, a struggle for evaluatively positive distinctiveness for one’s own group relative to other groups (Brewer, 1979; Tajfel & Turner, 1986). This, in turn, is underpinned by a self-enhancement motive and a striving for positive self-esteem (e.g., Abrams & Hogg, 1988). The strategies that groups and their members adopt to manage positive distinctiveness and selfenhancement is influenced by people’s beliefs about the nature of relations between groups—beliefs about the legitimacy and stability of status relations, about the permeability of intergroup boundaries, and about the possibility of an alternative social order. Social identity is also motivated by a basic human need to reduce self-conceptual uncertainty (e.g., Hogg, 2000)—people need to reduce uncertainty about their world, their place within it, and how they interact with and relate to other people. This motive causes people to identify more strongly, and with higher entitativity groups, as conditions of uncertainty become more extreme (Hogg, in press b).

B. PROTOTYPICALITY AND SOCIAL INFLUENCE The social identity perspective was originally developed as an analysis of intergroup relations (e.g., Tajfel & Turner, 1986), and intergroup relations remains the main focus for social identity research (e.g., Brewer & Brown, 1998). The approach is, however, perfectly suited to an analysis of processes within groups—a direction that has increasingly been taken in recent years. Of particular relevance to our proposed social identity analysis of leadership as an influence relationship within groups is the social identity analysis of social influence. The social identity analysis of social influence (e.g., Abrams & Hogg, 1990; Hogg & Turner, 1987; Turner, 1991; Turner & Oakes, 1989) centers around the concept of prototypicality. Within any salient group there is a prototypicality

8

HOGG AND VAN KNIPPENBERG

5

Metacontrast ratio

4 3 2 1 0

−

A

B

C

D

E

F

G

−

O

O

O

O

O

O

O

−

17-point intergroup comparative dimension Fig. 1. Metacontrast ratios for each of seven members (A through G) of an ingroup, in an intergroup context (17-point comparative dimension) where outgroup positions are indicated ‘‘O’’ and unoccupied positions ‘‘–.’’ Comparative frame with the outgroup on the right.

gradient, with some members being more prototypical than others (see Hogg, 2001d). The prototypicality of a specific position in the group, and thus the person who occupies that position, can be expressed by a metacontrast ratio— the mean difference between the specific position and each outgroup position, divided by the mean difference between the specific position and each other ingroup position. Figure 1 shows the ingroup metacontrast distribution, on a single 17-point prototypical dimension, for a hypothetical intergroup context where individual ingroup positions are indicated by ‘‘A’’ through ‘‘G,’’ outgroup positions by ‘‘O,’’ and unoccupied positions by ‘‘–.’’There is a marked asymmetry within the group, with more prototypical members (higher metacontrast ratios) located toward the left. Person C is the most prototypical member but is not the average ingroup member, and marginal member A is far more prototypical than marginal member G. Where the social context is in flux, the prototype will likewise be in flux. As the prototype changes so will the person who appears to be most prototypical. Figure 2 illustrates this—it is the same ingroup distribution of people/positions as in Fig. 1, but now the salient comparison outgroup (O) is located to the left, not the right of the ingroup. Person C is no longer the most prototypical member; it is E. In addition, G who was originally the least prototypical ingroup member is now significantly more prototypical. Of course, under conditions of enduring contextual stability the same individual may occupy the most prototypical position over a long period. Where group membership is situationally or enduringly salient, people self-categorize in terms of the ingroup prototype and become

9

SOCIAL IDENTITY AND LEADERSHIP PROCESSES IN GROUPS

5

Metacontrast ratio

4 3 2 1 0

−

O

O

O

− O O O O A B C D 17-point intergroup comparative dimension

E

F

G

−

Fig. 2. Metacontrast ratios for each of seven members (A through G) of an ingroup, in an intergroup context (17-point comparative dimension) where outgroup positions are indicated ‘‘O’’ and unoccupied positions ‘‘–.’’ Comparative frame with the outgroup on the left.

depersonalized—they conform to the ingroup prototype and exhibit groupnormative behavior. This alignment of individual attitudes and behavior with the group prototype may result in attitude and behavior changes typical of group influence processes like conformity and group polarization (e.g., Hogg, Turner & Davidson, 1990; Mackie, 1986; for reviews see, Abrams & Hogg, 1990; Turner, 1991; Turner & Oakes, 1989). The group prototype is a cognitive representation that may not be exemplified by any specific group member. Thus, the social identity analysis of social influence focused on conformity to this cognitive representation (e.g., Hogg & Turner, 1987; Mackie, 1986) rather than on the influence of any specific group member. However, group members differ in the extent to which they are group prototypical, and it is very likely that relative prototypicality may affect group member influence such that more prototypical members are more influential. Indeed, because depersonalization is based on prototypicality, group members are very sensitive to prototypicality. Prototypicality is the basis of perception and evaluation of self and other members, and thus people notice and respond to subtle differences in how prototypical fellow members are—they are very aware not only of the prototype, but also of who is most prototypical (e.g., Haslam, Oakes, McGarty, Turner, & Onorato, 1995; Hogg, 1993). Research within a persuasive communication paradigm has confirmed that people are more likely to attend to communications from prototypical group members than from nonprototypical members, and are more likely to align their attitudes with the attitudes of prototypical members (van

10

HOGG AND VAN KNIPPENBERG

Knippenberg, 2000a; van Knippenberg, Lossie, & Wilke, 1994; cf. van Knippenberg & Wilke, 1992). In view of this evidence that prototypical group members are more influential if and when they attempt to influence others, the question from the perspective of an analysis of group-based leadership is whether prototypical group members are also more likely to emerge as leaders (i.e., actively pursue this influential role, and be endorsed when they assume a leader position), and whether, once they are established as leaders, they are more likely to be endorsed and be effective (e.g., be influential, able to motivate and satisfy followers)?

III. Social Identity and Leadership The effect of social identity processes on leadership is closely tied to the concept of prototypicality, and the core propositions of the social identity analysis of leadership directly follow from the social identity analysis of social influence. As group membership becomes salient, and group members identify more strongly with the group, leadership perceptions, evaluations, and effectiveness become increasingly based on how group prototypical the leader is perceived to be (e.g., Hogg, 2001a, in press a). Two prototypicality based processes are likely to contribute to this effect: depersonalized social attraction and attribution and information processing.

A. DEPERSONALIZED SOCIAL ATTRACTION Social categorization affects not only perceptions, but also feelings, about other people. Social identification transforms the basis of liking for others from idiosyncratic preference and personal relationship history (personal attraction) to prototypicality (social attraction)—ingroup members are liked more than outgroup members and more prototypical ingroup members are liked more than less prototypical ingroup members. Where there is a relatively consensual ingroup prototype, social categorization renders more prototypical members socially popular—there is consensual and unilateral liking for more prototypical members. This depersonalized social attraction hypothesis (Hogg, 1992, 1993) is well supported by a series of laboratory and field studies (e.g., Hogg, Cooper-Shaw, & Holzworth, 1993; Hogg & Hains, 1996; Hogg & Hardie, 1991; Hogg, Hardie, & Reynolds, 1995). From the perspective of leadership, the person occupying the most prototypical position may thus acquire, in new groups, or possess, in established groups, the ability to actively influence. He or she is socially

SOCIAL IDENTITY AND LEADERSHIP PROCESSES IN GROUPS

11

attractive and thus able to secure compliance with his or her suggestions and recommendations. If you like someone you are more likely to agree with them, and to comply with their requests and suggestions, and less likely to disagree with them (e.g., Berscheid & Reis, 1998). In this way, the most prototypical person can actively exercise leadership by having his or her ideas accepted more readily and more widely than ideas suggested by others. Social attraction is prototype-based—it has a unilateral and consensual quality to it that puts tacit pressure on fellow group members to comply with the leader’s suggestions, in order to be acknowledged as fitting into the group. Because depersonalized liking is unilateral and consensual, it also publicly confirms that prototypical members are more popular, more highly valued, and so forth than less prototypical members of the group. It thus imbues prototypical members with prestige and status, which are important influences on emergent leadership and on the effectiveness of established leaders. The consensual liking and status associated with prototypicality positions prototypical members for a leadership role. This may arise actively through stronger feelings of entitlement and self-efficacy as a leader, and passively through prototype-based expectations of their fellow group members (van Knippenberg, van Knippenberg, & van Dijk, 2000; cf. Berger, Wagner, & Zelditch, 1985; Ridgeway, 2001). Once leadership is established, these social attraction and status processes work to bolster the leader’s position and render prototypical leaders more effective in comparison to less prototypical leaders. Social attraction may also be strengthened by the behavior of highly prototypical members. More prototypical members tend to identify more strongly, and thus display more pronounced group behaviors; they are more normative, show greater ingroup loyalty and ethnocentrism, and generally behave in a more group-serving manner. These behaviors further confirm prototypicality and thus enhance social attraction. A leader who acts as ‘‘one of us,’’ by showing ingroup favoritism and intragroup fairness, is not only more socially attractive, but is also furnished with legitimacy (Tyler, 1997; Tyler & Lind, 1992; see Platow, Reid, & Andrew, 1998). These groupnormative or group-oriented behaviors should, however, not be equated with prototypicality (Platow & van Knippenberg, 2001). We address the relationship between prototypicality, group-oriented acts, and leadership below.

B. ATTRIBUTION AND INFORMATION PROCESSING Social attraction works alongside attribution and information processing to translate group prototypicality into active leadership. Attribution processes operate within groups to make sense of others’ behavior. As

12

HOGG AND VAN KNIPPENBERG

elsewhere, attributions for others’ behavior are prone to the fundamental attribution error (Ross, 1977), correspondence bias (Gilbert & Jones, 1986; also see Gilbert & Malone, 1995; Trope & Liberman, 1993), or essentialism (e.g., Haslam, Rothschild, & Ernst, 1998; Medin & Ortony, 1989; Yzerbyt et al., 2001); a tendency to attribute behavior to underlying dispositions that reflect invariant properties, or essences, of the individual’s personality. This effect is more pronounced for individuals who are perceptually distinctive (e.g., figural against a background) or cognitively salient (e.g., Taylor & Fiske, 1978). There is evidence that this tendency to make dispositional attributions is especially strong for attributions about leaders (Meindl et al., 1985). Fiske (1993; Fiske & De´ pret, 1996) shows how followers pay close attention to leaders, and seek dispositional information about leaders because detailed individualized knowledge helps redress the perceived power imbalance between leader and followers. Conger and Kanungo (1987, 1988) describe how followers attributionally construct a charismatic leadership personality for organizational leaders who have a vision that involves substantial change to the group. The social identity analysis of leadership suggests that prototypicality enhances attributions of leadership. When group membership is salient, people are sensitive to prototypicality and attend to differences in prototypicality of fellow members. Highly prototypical members are most informative about what is prototypical of group membership (see Turner, 1991), and so in a group context they attract most attention. They are subjectively important and are distinctive or figural against the background of other less informative members. Research in social cognition shows that people who are subjectively important and distinctive are seen to be disproportionately influential and have their behavior dispositionally attributed (e.g., Erber & Fiske, 1984; Taylor & Fiske, 1975). We have also seen how highly prototypical members may be expected to assume a leadership role due to their relative prototypicality, and may actively exercise influence and gain compliance as a consequence of consensual social attraction. Together, the leadership nature of this behavior and the relative prominence of prototypical members are likely to encourage an internal attribution to intrinsic leadership ability, or charisma. In groups, then, the behavior of highly prototypical members is likely to be attributed, particularly in stable groups over time, to the person’s personality rather than the prototypicality of the position occupied. The consequence is a tendency to construct a charismatic leadership personality for that person, which, to some extent, separates that person from the rest of the group and reinforces the perception of status-based structural differentiation within the group into leader(s) and followers. This may make the leader stand out more starkly against the background of less

SOCIAL IDENTITY AND LEADERSHIP PROCESSES IN GROUPS

13

prototypical followers, as well as draw attention to a potential power imbalance, thus further fueling the attribution effect. It should be noted that this analysis views charisma as a product of socialcognitive processes operating under conditions of self-categorization, and not as an invariant personality attribute that determines leadership effectiveness (see below). In this respect our analysis is consistent with Haslam and Platow’s (2001a) critical appraisal of the role of charisma in contemporary transformational leadership theories.

C. PROTOTYPICALITY AND SOCIAL IDENTITY SALIENCE/ SOCIAL IDENTIFICATION The social identity analysis of leadership assigns a central role to leader prototypicality. However it is important to keep this in perspective. Prototypicality is not the only basis of leadership. People also rely on general and more task-specific schemas of leadership behaviors (what Lord and his colleagues call leader categories or leader schemas—e.g., Lord et al., 1984). In addition, as outlined in LMX theory (Graen & Uhl-Bien, 1995), the quality of the interpersonal relationship between leader and follower may be an important determinant of leadership effectiveness. The influence of these more personalized leader attributes and more interpersonal processes is not incompatible with the social identity analysis of leadership. The social identity analysis of leadership does, however, suggest that the relative importance of group-based prototypicality as compared with these more personalized or interpersonal processes is contingent on social identity salience and group identification. Prototypicality becomes an increasingly powerful determinant of effective social influence and of effective leadership as people more strongly define themselves (self-categorize) in terms of group membership (social identity), rather than in terms of individuality, idiosyncracy, or interpersonal relationships (personal identity). The principles governing contextual salience are described by self-categorization theory (Turner et al., 1987; also see van Knippenberg, 2000b). People are more likely to self-categorize (strongly) in terms of a specific group membership in a particular context if that categorization (1) is chronically accessible (e.g., it is an important aspect of one’s social identity), (2) is contextually accessible (e.g., there are immediate cues that draw attention to it), (3) accounts for relevant similarities and differences among people (i.e., comparative fit), and (4) makes sense of how people are behaving (i.e., normative fit). When personal rather than social identity is salient, prototypicality will not be a salient attribute and, accordingly, leadership effectiveness should be

14

HOGG AND VAN KNIPPENBERG

less contingent on prototypicality. Relationships will be governed by personal rather than social attraction (Hogg, 1992, 1993), and leadership effectiveness should be more contingent on interpersonal leader-follower relationships. Although the importance of leader schemas as outlined in leadership categorization theory (Lord et al., 1984) is not necessarily affected by self-categorization, leadership schemas should become less influential relative to group prototypicality as a determinant of leadership perceptions as group membership becomes psychologically more salient. To summarize, the core proposition of the social identity analysis of leadership is that (1) more prototypical group members are more likely to emerge as leaders, and more likely to be endorsed and effective as leaders, and (2) that this relationship between prototypicality and leadership is stronger the more salient the relevant group membership is and the more followers identify with the group. In addition, (3) more personalized or interpersonal aspects of leadership become less important relative to leader prototypicality as determinants of leadership effectiveness. In the following section, we survey empirical evidence for the social identity analysis of leadership.

IV. Empirical Support for the Social Identity Analysis of Leadership A. LEADER PROTOTYPICALITY, LEADER SCHEMA CONGRUENCE, AND GROUP MEMBERSHIP SALIENCE Core predictions of the social identity analysis of leadership were tested in a series of studies by Hogg and associates. Hains, Hogg, and Duck (1997) conducted a laboratory study of leadership perceptions and evaluations in ad hoc and relatively minimal groups. The methodology was closely based on standard procedures and methods used in social identity research. Three independent variables (group salience, group prototypicality, and leader schema congruence) were manipulated in a 2
2
2 design. Under conditions of high or low group salience, student participants (N ¼ 184) anticipated joining a small discussion group ostensibly formed on the basis of attitude congruence among members. Salience was manipulated by always referring in instructions and explanations to groups or to loose aggregates of individuals, and by having participants consider commonalities within the group or differences among members, and by referring to themselves in group terms or only in individual terms. Participants were informed that a group leader had been randomly appointed from among the members. Information was given that revealed the leader to be group prototypical or nonprototypical (group

SOCIAL IDENTITY AND LEADERSHIP PROCESSES IN GROUPS

4.20

15

4.14

Leader effectiveness

4.03 4.00

3.80

3.80

3.60

3.53

3.40 Low

High Salience

Group prototypical Group non-prototypical Fig. 3. Hains, Hogg, and Duck (1997): Leader effectiveness (1–9 scale, 10 items,
= .88) as a function of group salience and group prototypicality of the leader (p < .001).

prototypicality) in terms of the attitude dimension, and to have a behavioral style (on the basis of a pretest) that was congruent or incongruent with a very general schema of effective leadership (leader schema congruence). Dependent measures were taken ostensibly in anticipation of the upcoming discussion. In addition to checks on each of the three manipulations, group identification (11-item scale,
¼ .87, for example ‘‘how much do you feel you identify with your group’’) and anticipated leader effectiveness (10-item scale,
¼ .88, for example ‘‘how effective do you feel your leader will be’’) were also measured. As predicted, when group membership was salient people identified more strongly with the group, and endorsed the prototypical leader as likely to be much more effective than the nonprototypical leader; low salience participants did not differentiate between prototypical and nonprototypical leaders (Fig. 3). Although leader schema-congruent leaders were perceived overall to be more effective than schema-incongruent leaders, Hains et al. (1997) found that this effect disappeared for high salience participants on one leadership effectiveness item measuring the extent to which the leader was anticipated to exhibit leadership behavior (Fig. 4). Although social attraction for the leader was not explicitly tested, the 10-item leadership effectiveness scale contained an item measuring liking for the leader. Thus, as predicted from the social identity analysis of leadership, perceived leadership effectiveness was associated with group membership-based liking for the leader.

16

HOGG AND VAN KNIPPENBERG

Leadership behavior

6.50

6.47

6.00

5.91 5.59

5.50

5.00

4.93

4.50 Low

High Salience

Leader schema congruent Leader schema incongruent Fig. 4. Hains, Hogg, and Duck (1997): Leader behavior (1–9 scale, one item) as a function of group salience and leader schema congruence of the leader (p < .01).

To complement this controlled laboratory experiment, Fielding and Hogg (1997) conducted a naturalistic field study of leadership in small interactive ‘‘Outward Bound’’ groups where real leaders emerged to lead groups in wilderness and outdoor experiences. There were 13 mixed-sex, approximately 11-person groups of people mainly in their 20s from around Australia (N ¼ 143). The groups stayed together for 3 weeks. The laboratory experiment was replicated as closely as possible, but in a measurement-based regression format. Leadership schemas, group membership variables, and leadership effectiveness perceptions were measured a week to 10 days apart. In this study it was also possible to measure social attraction. As predicted, (1) group identification, perceived leadership effectiveness, and social attraction for the leader increased over time as the group became a more cohesive entity, and (2) perceived leadership effectiveness was a positive function of social attraction for and group prototypicality of the leader, and this was amplified among high identifying participants. Perceived schema congruence of the leader was a predictor of perceived leadership effectiveness, but was uninfluenced by identification. Hogg, Hains, and Mason (1998) returned to the laboratory to conduct two minimal group studies based closely on the methodology of Hains et al. (1997). The main aim of these somewhat complex studies was to treat prototypicality and leadership as relativistic properties of a comparative frame of reference in which individuals are perceived and evaluated in

SOCIAL IDENTITY AND LEADERSHIP PROCESSES IN GROUPS

17

relation to other individuals who are ingroup or outgroup members. This aim recognizes that perceptions or evaluations of whether someone is prototypical or a leader are actually perceptions of how prototypical or how much of a leader someone is relative to other people. The first study had student participants (N ¼ 82) nominate a leader for a four- or five-person high salience discussion group they were ostensibly going to join. They were provided with a carefully constructed transcript of an earlier meeting of the group, which provided leadership schema congruence and group prototypicality information for each of the other members. They rated the group, their nominated leader, and all other group members, and their identification with the group. The second study was a 2
2
2 experiment (N ¼ 164) in which group salience, leadership schema congruence of the leader, and group prototypicality of the leader were orthogonally manipulated. The key feature of this second experiment was that the prototypicality of the leader was indirectly manipulated by constructing an intergroup comparative context that, on the basis of metacontrast, influenced the location of the prototype in precisely calculable ways. Across these two studies Hogg et al. (1998) found that leadership schema congruence became a less influential, and group prototypicality a more influential, determinant of leadership endorsement in more cohesive groups with which people identified more strongly. Also, as expected, identification accentuated perceived prototypical similarities between ingroup members and thus between leader and nonleaders, and this weakened the prototypeleadership relationship when leadership and prototypicality were measured relativistically (leaders relative to nonleaders). Hogg et al. (1998) argued, however, that this effect was due to the ad hoc and emergent nature of the group, and that over time the perceptual gulf between leader and followers, described by the theory, would gradually develop. This pattern of results was replicated by Platow and van Knippenberg (2001). In addition to the experimental tests of their hypotheses that are discussed in more detail later, Platow and van Knippenberg conducted a measurement-based test of the core predictions of the social identity theory of leadership. Multiitem scales were used to measure identification, leader prototypicality, and leader schema congruence. Leader endorsement was measured as participants’ willingness to vote for the leader to remain as leader. A regression analysis replicated the findings of Hains et al. (1997). Leader prototypicality was more strongly related to leadership endorsement as members identified more strongly with the group, whereas the relationship between leader schema congruence and leader endorsement became weaker as members identified more strongly with the group.

18

HOGG AND VAN KNIPPENBERG

1. Summary The studies described above yield consistent evidence for the basic predictions of the social identity analysis of leadership. When social identity is salient and followers identify with their group, prototypical leaders are more strongly endorsed and more effective than nonprototypical leaders. Moreover, when social identity is salient, and leader prototypicality is relatively important, more personalized leader characteristics that affect the match between leader attributes and leader stereotypes (i.e., leader schema congruence; Lord et al., 1984) affect perceptions of leadership effectiveness to a lesser degree. These findings are consistent across research setting (i.e., field vs. laboratory) and research paradigm (i.e., the Hains et al. paradigm vs. the Platow and van Knippenberg paradigm), and hold for both emergent leadership (Fielding & Hogg, 1997) and endorsement of appointed leaders. Additional evidence for the importance of prototypicality as a determinant of emergent leadership comes from a study by van Knippenberg et al. (2000). van Knippenberg and associates manipulated whether the group was faced by an ambiguous decision-making task, and thus the group was in need of leadership (Hemphill, 1961), or by a clear-cut decision task where the decision was self-evident. In all conditions there was a salient comparative outgroup, so it can be assumed that social identity was relatively highly salient. van Knippenberg et al. (2000) predicted and found that when the decision task was ambiguous prototypical members were more likely to take the lead and nonprototypical members were less likely to take the lead than when the task was unambiguous.

B. INDIRECT EVIDENCE: PROXIES FOR PROTOTYPICALITY 1. Ingroup versus Outgroup Leaders The studies above involve direct measures or manipulations of the prototypicality of the leader. In this section we discuss studies that operationalize prototypicality indirectly. For example, leaders who are ingroup members are, by definition, more ingroup prototypical than are leaders who are outgroup members. From our social identity perspective, we would predict that as group membership becomes more salient, ingroup leaders will become increasingly effective relative to outgroup leaders. So, for example, where groups, organizations, or even nations merge, leadership processes in the merged entity will be significantly affected by perceptions of whether the superordinate leader is a member of one’s former ingroup or outgroup—is the CEO a member of premerger organization A or premerger organization B; is the national president from East or West Germany?

SOCIAL IDENTITY AND LEADERSHIP PROCESSES IN GROUPS

19

Support for the social identity analysis of the group salience-mediated impact of the leader’s group membership on leadership effectiveness can be found in studies by Duck and Fielding (1999) and Van Vugt and De Cremer (1999). Duck and Fielding (1999) conducted an experiment grounded in recent social identity analyses of organizational mergers and acquisitions (e.g., Terry, Carey, & Callan, 2001; van Knippenberg & van Leeuwen, 2001). Mergers are notoriously unsuccessful (Cartwright & Cooper, 1992) because of problems associated with forging a single identity from two erstwhile separate groups (e.g., Hornsey & Hogg, 2000). Associated with this is a leadership problem hinging on whether the new group leader is a member of one’s former ingroup or one’s former outgroup. Duck and Fielding (1999) conducted two laboratory experiments that, in a relatively minimal way, simulated equal status subgroups nested within a larger organization (total of 328 participants). They measured group identification and evaluations of leaders who were randomly appointed from participants’ own or the other subgroup. Ingroup, thus prototypical, leaders were more strongly supported than outgroup, thus non prototypical, leaders, and this effect was more pronounced to the extent that participants identified strongly with their own subgroup. Van Vugt and De Cremer (1999) conducted an experimental study of leadership preferences in social dilemmas. Social dilemmas are typically difficult to resolve, and often require structural solutions, such as appointing a leader to manage the scarce resources for all parties who have access to or are competing over the resources (Messick & Brewer, 1983; Foddy, Smithson, Schneider, & Hogg, 1999). To study the relationship between group identification and leader preferences in social dilemmas, Van Vugt and De Cremer (1999; Experiment 1, N ¼ 96) manipulated group identification by making an intergroup comparison salient or not. They then assessed preferences for different types of leaders, including an ingroup versus an outgroup leader. Consistent with the social identity analysis of leadership, they found that participants generally preferred ingroup (i.e., prototypical) over outgroup (i.e., nonprototypical) leaders, and that this differential preference was more pronounced among high than low identifiers. 2. Appointed versus Elected Leaders Another proxy for prototypicality may hinge on the process by which leaders are selected. Some leaders are appointed by external authorities (e.g., business leaders), whereas others are elected by the group (e.g., political leaders, leaders of social movements). It is plausible to assume that the fact

20

HOGG AND VAN KNIPPENBERG

of being elected by the group imbues the leader with a degree of group representativeness (i.e., prototypicality) that externally appointed leaders lack. From the social identity analysis of leadership, we would expect elected leaders to be more strongly endorsed than externally appointed leaders, especially by group members who strongly identify with the group. Support for this hypothesis can be found in Van Vugt and De Cremer’s (1999) first experiment. In addition to preferences for ingroup versus outgroup leaders, Van Vugt and De Cremer assessed preferences for leaders elected by the group versus externally appointed leaders. They found that elected leaders were generally preferred over appointed leaders, especially by participants in the high, as opposed to low, identification condition (see also Haslam, McGarty, Brown, Eggins, Morrison, & Reynolds, 1998). Extending these results from effects on leader preference to effects on cooperation with established leaders, De Cremer and Van Vugt (2002) showed in another social dilemma experiment that the preference for elected vs. appointed leaders generalizes to greater cooperation with elected leaders. 3. Summary These studies consistently show that factors that can be considered proxies for prototypicality enhance leadership endorsement and effectiveness when followers identify with the group. This confirms the core prediction from the social identity analysis of leadership that protypicality becomes increasingly influential in leadership effectiveness as people identify more strongly with the group. These studies also show how the social identity analysis of leadership and its emphasis on leader prototypically may be used to understand leadership effectiveness in intergroup contexts (e.g., mergers) and in the context of different leader selection methods.

C. LEADERS’ GROUP-ORIENTED ATTITUDES AND BEHAVIOR The social identity analysis of leadership is not limited to the direct effects of prototypicality. From the proposition that group members are more likely to endorse leaders who represent the group’s identity (i.e., prototypical leaders), follows the implication that members should be more likely to endorse leaders who exert themselves on behalf of the group or are in other ways group oriented in their attitudes and behavior. As Haslam and Platow (2001a) phrase it, social identity-based leadership endorsement may not only derive from being ‘‘one of us’’ (i.e., being prototypical), but also from being seen to be ‘‘doing it for us.’’ Relevant group-oriented leadership attitudes and behaviors include (1) the leader’s commitment to

SOCIAL IDENTITY AND LEADERSHIP PROCESSES IN GROUPS

21

and sacrifices for the group, (2) how much the leader favors the ingroup over relevant outgroups, and (3) how procedurally fair and relationship oriented the leader is. Leader commitment to the group reflects the leader’s willingness to exert effort on behalf of the group, and thus should be associated with enhanced leadership effectiveness, especially when social identity is salient. Commitment can be revealed through the leader’s behavior or his or her expressed attitudes. A particularly powerful indicator of leader commitment to the group is leader self-sacrifice for the group. Indeed, theories of charismatic leadership identify self-sacrifice for the collective as a constituent of leader charisma (Yorges, Weiss, & Strickland, 1999). When leaders incur personal costs (e.g., time, effort, money) to achieve group goals, this conveys the leader’s commitment to the group, and more generally that the leader places the group’s interest before personal interests. It shows that the leader is ‘‘doing it for us.’’ On the basis of the social identity analysis of leadership we expect that self-sacrificing leader behavior would improve leader effectiveness. Ingroup favoritism is a distinctly group-oriented behavior. Thus, when a leader behaves in ways that favor the ingroup over relevant outgroups, he or she is acting in a highly ingroup prototypical manner. Ingroup favoritism not only benefits the group materially, when a valued resource such as money is being allocated, but also benefits the group symbolically because it creates evaluatively positive distinctiveness for the ingroup—it imbues the ingroup and its members with a relatively more favorable social identity (e.g., Tajfel & Turner, 1986). A leader who favors the ingroup over relevant outgroups may not only be seen as ‘‘doing it for us,’’ but also as exemplifying ‘‘what makes us better than them’’ (Haslam & Platow, 2001a). Therefore, on the basis of the social identity analysis of leadership, we expect leaders who favor the ingroup in intergroup contexts to be more strongly endorsed. Procedural fairness is another important feature of group life. People in groups like to be treated fairly by authorities, in particular as regards fair procedures. Procedural fairness conveys to members respect, and recognition of membership in the group (Koper, van Knippenberg, Bouhuijs, Vermunt, & Wilke, 1993; Tyler & Lind, 1992). Procedurally fair treatment fulfills an important social identity function in groups (Tyler, 1999), and procedurally fair behavior should therefore be associated with greater leader effectiveness, especially when people identify with the group (cf. Smith, Tyler, Huo, Ortiz, & Lind, 1998; Tyler & Degoey, 1995). Similarly, other relationship-oriented behaviors should have a positive effect on leader effectiveness, again especially when people identify with the group.

22

HOGG AND VAN KNIPPENBERG

1. Studies of Leader Commitment and Leader Procedural Fairness Evidence for the effects of leader commitment comes from a study by De Cremer and Van Vugt (2002). In a social dilemma study (N ¼ 94), De Cremer and Van Vugt manipulated group identification and a randomly appointed leader’s ostensible commitment to the group. The latter was manipulated via controlled feedback about the leader’s earlier responses on a group commitment questionnaire. The main dependent measure was member cooperation. De Cremer and Van Vugt found that members cooperated more with a leader who was highly committed to the group than one who was less committed, and that this was particularly so for members who identified strongly with the group. In addition, De Cremer and Van Vugt manipulated whether the leader was procedurally fair or not (operationalized as giving or withholding voice from group members). Arguing that procedural fairness fulfills a social identity function (cf. Tyler, 1999), De Cremer and Van Vugt predicted that fair leaders would be more effective, especially when group members identified with the group. This was exactly what they found. In a second experiment (N ¼ 72), De Cremer and Van Vugt (2002) measured high and low identifying group members’ reactions toward highly committed leaders or toward leaders who scored high on leadership skills. Commitment and leadership skills were manipulated via controlled feedback of leaders’ responses on a prior commitment questionnaire (as in De Cremer and Van Vugt, 2002, Experiment 1), and on a prior leadership skills questionnaire. There is a clear correspondence between this manipulation, of the leader being committed to the group versus being skilled as a leader, and the group prototypicality and leader schema congruence manipulations in the experiment of Hains et al. (1997; Hogg et al., 1998). On the basis of the social identity analysis of leadership, we would expect that committed leaders would be more strongly supported (i.e., would elicit greater cooperation) by high identifiers, and skilled leaders would be more strongly supported by low identifiers. De Cremer and Van Vugt found support for this prediction. The social identity analysis of leadership also predicts, as described above, that procedurally fair leaders should attract more support than procedurally unfair leaders (because procedural fairness conveys a relationship orientation to members), and that this is particularly the case where members identify strongly with the group. Van Vugt and De Cremer (1999, Experiment 2) found support for this prediction from a social dilemmas experiment (N ¼ 93). They manipulated how strongly members identified with the group (manipulated as in Experiment 1), and whether the leader adopted punishment-oriented or relationship-oriented behavior toward

SOCIAL IDENTITY AND LEADERSHIP PROCESSES IN GROUPS

23

them (manipulated via a controlled communication from the leader about how he or she planned to treat followers). They found that relationshiporiented leaders were better at eliciting group member cooperation when group members identified strongly with the group than when members identified less strongly. Identification did not influence the impact of punishment-oriented behavior on cooperation. As noted earlier, leader commitment to the group may also express itself in the extent to which the leader makes personal sacrifices for the benefit of the group. De Cremer and van Knippenberg (in press, Experiments 1 and 3, N ¼ 62 and 99) studied the effects of leader self-sacrifice and procedural fairness on leadership effectiveness in the context of a social dilemma. They manipulated whether the leader exhibited self-sacrifice for the benefit of the group or displayed self-benefiting behavior (the leader informed members that he or she would invest extra time in the group at personal cost, or would personally benefit from the group), and whether or not the leader allowed members voice in decisions about the allocation of resources (i.e., procedurally fair versus unfair behavior). Self-sacrificial behavior (De Cremer, 2002) and procedural fairness (Tyler & Lind, 1992) communicate that the leader is group oriented, and thus both instill a sense of identification and group belongingness among followers. Where either selfsacrificial behavior or procedural fairness is high, a threshold is reached and followers’ sense of belonging is ensured. Information provided by the other variable does not add anything new. Identification and group belongingness were predicted to mediate the effects of the leader’s behavior on group member cooperation. Results indicated that self-sacrificing leaders and procedurally fair leaders elicited stronger identification and more cooperation. Moreover, the effects of leaders’ self-sacrificing on identification and cooperation were more pronounced when the leader was procedurally unfair, and leader fairness had a stronger impact when the leader displayed self-benefiting behavior. As predicted, identification mediated the interactive effect of self-sacrifice and procedural fairness on cooperative behavior. Because identification was treated as a dependent variable rather than a moderator variable, these findings do not speak to the moderating role of social identity salience proposed in the social identity analysis of leadership. However, the fact that group-oriented behavior (i.e., self-sacrificing, procedural fairness) was more effective than behavior that was not group oriented (i.e., self-benefiting, procedural unfairness), and that these effects were mediated by the social identity variable of identification/group belongingness, is clearly in line with the social identity analysis. Moreover, these results suggest that prototypical leaders may chose to behave in ways aimed at strengthening members’ feelings of identification and

24

HOGG AND VAN KNIPPENBERG

belongingness, in order to render prototypicality a stronger basis for effective leadership. 2. Studies of Ingroup Favoritism The hypothesis that leaders may gain support through ingroup favoritism has been addressed in a program of studies by Platow and colleagues. Platow, Hoar, Reid, Harley, and Morrison (1997) argued that group members would show greater support for leaders who allocated resources fairly than unfairly among ingroup members, whereas in intergroup contexts they would prefer leaders who favored the ingroup over the outgroup than leaders who treated ingroup and outgroup members equally. Results of a series of three experiments in which leaders allocated resources fairly or unfairly in intragroup or intergroup contexts supported these predictions. In an intragroup context, fair leaders were more strongly endorsed (operationalized as participants’ willingness to vote for the leader to be leader again), whereas in intergroup contexts ingroup favoring leaders were endorsed equally or more strongly than fair leaders. In Experiments 2 and 3, which incorporated a high identification and a low identification intergroup condition (the former tapped into groups that participants belonged to, whereas the latter used groups that participants could only imagine belonging to), this pattern of results emerged only in the high identification condition, not the low identification condition. Experiment 3 also showed that this pattern of results generalized to leader social influence—in this case, the extent to which followers agreed with a statement made by the leader about the utility of a particular practice within the group (see also Platow, Mills, & Morrison, 2000). Platow, Reid, and Andrew (1998) replicated the leadership endorsement findings for distributively fair and biased leaders, but also showed that these effects generalize to procedurally fair and unfair behavior. Leaders who are procedurally unfair within their group are less strongly endorsed than those who are procedurally fair, but the effect is reversed in intergroup contexts. Leaders who procedurally favor their own group over an outgroup (i.e., give voice to ingroup but not to outgroup members) are endorsed as strongly as those who are procedurally fair between groups. Haslam and Platow (2001b) used a different manipulation of ingroup favoritism, in which committees were composed to favor the ingroup point of view, the outgroup’s point of view, or in an unbiased manner. The results showed that in addition to receiving as much support as fair leaders and more support than outgroup-favoring leaders, ingroup-favoring leaders elicit higher productivity from group members than either fair or outgroupfavoring leaders.

SOCIAL IDENTITY AND LEADERSHIP PROCESSES IN GROUPS

25

3. Summary The studies discussed in this section support the proposition that when group members identify with a group and group membership is salient, leaders will be more strongly supported when they are group oriented in their attitudes (i.e., group commitment) and their actions (i.e., selfsacrificing, procedurally and distributively fair within the group, relationship oriented, ingroup favoring). Given the diversity of these studies, both in terms of operationalizations of leader attitudes and behavior, and measures of leader effectiveness, and in terms of the research paradigms used, there seems to be robust support for the proposition from the social identity analysis of leadership that under conditions of high social identity salience group-oriented leaders will be more effective and better supported. In view of the strong emphasis on prototypicality in the social identity analysis of leadership, an obvious question is how these findings for group-oriented attitudes and behavior relate to findings from studies of prototypicality and of proxies for prototypicality.

D. PROTOTYPICALITY AND GROUP-ORIENTED BEHAVIOR Although prototypicality and group-oriented behavior generally go together, they should not be equated. People who are highly prototypical may sometimes fail to act in the group’s best interest, and people who act in the group’s best interest may not always be highly prototypical. Likewise, leaders may engage in acts that are group oriented, yet not be very representative of the group, or, vice versa, they may be very prototypical yet act in ways that are clearly not in the group’s best interest. Generally speaking a person’s perceived prototypicality is unlikely to be changed dramatically by the commission of a small number of behaviors that are inconsistent with their level of prototypicality. However, widespread and frequent inconsistency would be expected to change perceived prototypicality. A key question is whether, and how, leader prototypicality and the group-orientation of the leader’s behavior jointly determine follower responses and leadership effectiveness. This was addressed in a study by Platow and van Knippenberg (2001). Echoing and developing on Hollander’s (1958) notion of idiosyncracy credit, Platow and van Knippenberg (2001) argued that prototypical leaders have solid membership credentials, and therefore have considerable leeway in their behavior, with little detrimental impact on their perceived prototypicality or on their endorsement as leaders. Less prototypical leaders have insecure membership credentials and so the group would be less

26

HOGG AND VAN KNIPPENBERG

tolerant of non-group-oriented behavior—less prototypical leaders would be endorsed only when they overtly engaged in group-oriented behavior. Prototypical leaders would be endorsed anyway because they clearly are ‘‘one of us,’’ but nonprototypical leaders need to be seen to be ‘‘doing it for us’’ to be endorsed. Platow and van Knippenberg hypothesized that this pattern would arise only when followers identified strongly with the group. 1. Studies of the Leader’s Allocative Behavior Platow and van Knippenberg (2001) conducted an experiment (N ¼ 216) modeled on the experiments of Platow et al. (1997) discussed in the previous section. They manipulated the leader’s prototypicality (three levels) and the leader’s intergroup allocative behavior (three levels), and added a measurement-based group member identification factor (median-split), to produce a 3
3
2 design. To manipulate leader prototypicality, participants were given feedback from a prior personality test that ostensibly indicated how much the leader had in common with the ingroup (the participants’ university) and with the outgroup (another university). The feedback was graphically displayed, as two partially overlapping personality distributions, in such a way that the leader occupied the middle of the ingroup personality distribution (prototypical leader), the overlapping area of the ingroup and outgroup distribution (marginal leader), or the tail of the ingroup distribution that was remote from the outgroup (extreme leader). The leader then made an intergroup allocation, which was either ingroup favoring, fair, or outgroup favoring. The main dependent variable, leadership endorsement, was operationalized as willingness to vote for the leader to be group leader again. Results showed that, as predicted, the three-way interaction between leader prototypicality, leader allocation behavior, and group member identification was significant. Among low identifiers, fair leaders received more support than ingroup-favoring or outgroup-favoring leaders, and prototypicality was unrelated to leadership endorsement. Among high identifiers prototypicality and leader behavior interacted. This interaction is displayed in Fig. 5. Prototypical leaders were endorsed irrespective of their allocation behavior. Marginal leaders, in contrast, received relatively high levels of endorsement (i.e., comparable to those of the prototypical leader) only if they were ingroup serving, and not if they made fair or outgroupfavoring allocations. Endorsement of the extreme leader was relatively low irrespective of the leader’s allocation behavior. These data are consistent with the core proposition of the social identity analysis of leadership. Relative to less prototypical leaders, prototypical

SOCIAL IDENTITY AND LEADERSHIP PROCESSES IN GROUPS

27

7

Leadership endorsement

6.31 6

5.82

5.93

5.82

5.00

5 4.50 4.31 4

3.91

3.87

3 Extreme

Prototypical

Marginal

Leader Prototypicality Ingroup-favoring Fair Outgroup-favoring Fig. 5. Platow and van Knippenberg (2001): Leadership endorsement by high identifiers as a function of leader prototypicality and the leader’s allocation behavior.

leaders are more strongly endorsed when group members identify with the group. They also show how leader prototypicality and the group orientation of leader behavior interact. Prototypical leaders need not engage in grouporiented behavior to be endorsed, whereas nonprototypical leaders are endorsed only when they engage in group-oriented behavior. Finally, these data suggest that endorsement of nonprototypical leaders may be influenced by the function of the leader’s behavior in an intergroup comparative context. Nonprototypical leaders who were on the margin between ingroup and outgroup were endorsed if they showed ingroup favoritism (in this context a behavior that consolidates ingroup credentials), whereas favoritism expressed by nonprototypical leaders who were at the other extreme of the ingroup distribution (in this context a behavior that contributes little to confirming the leaders distinctiveness from the outgroup) did not strengthen endorsement. 2. Studies of Leader Influence Tactics Two studies by van Knippenberg and van Knippenberg (2000, 2001; see also van Knippenberg & van Knippenberg, in press) extend the finding that

28

HOGG AND VAN KNIPPENBERG

prototypical leaders are permitted more behavioral leeway than nonprototypical leaders, to behaviors other than intergroup allocation. van Knippenberg and van Knippenberg studied leader effectiveness (operationalized as followers’ perceptions of the leader) as a function of the leader’s use of hard versus soft influence tactics. Hard tactics involve coercion and are compliance based, whereas soft tactics involve persuasion and are conformity based—thus the former are seen as more unfriendly and less socially desirable, and as placing a strain on leader–follower relations [e.g., van Knippenberg & van Knippenberg, in press; van Knippenberg, van Knippenberg, Blaauw, & Vermunt, 1999; Yukl, 2001—also see Moscovici’s (1976) distinction between power and influence]. Based on the social identity analysis of leadership, and the reasoning underlying Platow and van Knippenberg’s (2001) predictions, van Knippenberg and van Knippenberg (2000, 2001) predicted that prototypical leaders would maintain good relationships with their followers (as rated by the followers) irrespective of their use of hard or soft tactics, whereas less prototypical leaders would be more effective when they used soft rather than hard tactics. To examine these predictions, members of 69 different teams in a large organization were asked about their supervisors’ influence tactics, the quality of their relationship with their supervisor, and the prototypicality of the team leader. The data showed that leader–subordinate relations were better when the leader was more prototypical, that where the leader was prototypical leader– follower relations improved with increasing use of soft influence tactics, and that increasing use of soft influence tactics improved leader–follower relations for prototypical leaders much more than for less prototypical leaders. Increasing use of hard influence tactics impoverished leader– subordinate relations where the leader was less prototypical, but improved it where the leader was more prototypical. These findings confirm the prediction that nonprototypical leaders are more effective in maintaining good relationships with their subordinates if they use soft rather than hard tactics, whereas prototypical leaders are able to maintain good relationships with their subordinates regardless of whether they use hard or soft tactics. In a second study, van Knippenberg and van Knippenberg (2001) replicated these findings experimentally, and tested the idea that the effects of prototypicality should mainly be obtained when followers identify more strongly with the group. They manipulated leader prototypicality and the leader’s use of a hard versus a soft influence tactic, and classified participants as high or low identifiers on the basis of a median-split. The results confirmed that leader–follower relations were perceived to be worse with hard than soft influence tactics, where the leader was nonprototypical, and that this effect was significantly weakened where the leader was

SOCIAL IDENTITY AND LEADERSHIP PROCESSES IN GROUPS

29

prototypical. As predicted, these effects emerged only among high identifying participants. 3. Summary Taken together, the studies discussed in this section add further support for the role of leader prototypicality and group/relationship-oriented behavior in leader effectiveness. Prototypical leaders are not as tightly constrained as less prototypical leaders regarding the extent to which followers expect them to behave in group-oriented ways.

V. The Social Identity Analysis and Other Contemporary Theories of Leadership Thus far we have primarily described the social identity analysis of leadership and examined the role of social identity variables such as prototypicality and group-oriented behavior and attitudes. In this section we explore the relationship between the social identity theory of leadership and some other major contemporary psychological theories of leadership. We have already discussed leader categorization theory (e.g., Lord et al., 1984)—the theory that leadership endorsement and effectiveness rest on the match between the leader’s attributes, and people’s general or more task specific schemas of effective leadership. The social identity studies we discussed showed that as social identity becomes more salient and group members identify more with the group, group prototypicality becomes more important as a determinant of leadership effectiveness, and the influence of leader schema congruence is either unaffected (Fielding & Hogg, 1997; Hains et al., 1997) or diminished (De Cremer & Van Vugt, 2002; Hogg et al., 1998; Platow & van Knippenberg, 2001). However, it should be noted that Lord and his colleagues have recently modified their leadership categorization theory in a way that is consistent with our social identity analysis (Lord & Hall, in press; see also Lord, Brown, & Harvey, 2001; Lord, Brown, Harvey, & Hall, 2001). Lord and his colleagues now view leadership categorization as a highly dynamic and context-dependent process, and propose that under conditions of high social identity salience the leadership schema may be construed in terms of the group prototype. In this section we assess how the influence of prototypicality on leadership effectiveness relates to what are perhaps the two most closely studied perspectives on effective leadership in contemporary psychological leadership research: charismatic/transformational leadership and leader–member exchange (LMX).

30

HOGG AND VAN KNIPPENBERG

A. PROTOTYPICALITY AND CHARISMATIC/ TRANSFORMATIONAL LEADERSHIP Since the mid-1980s, a major focus in leadership research has been on charismatic and transformational leadership. Although there are differences between theories of transformational and charismatic leadership (see Yukl, 2001), there is enough common ground, we feel, for them to be treated together here. The central idea is that charismatic leadership is more effective than noncharismatic leadership (e.g., transactional leadership; Bass, 1985), because charismatic leadership makes collective identity salient and mobilizes the individual by linking group goals to the individual’s self-concept. Charismatic leadership motivates people to move beyond self-interest and toward the group’s goals and mission (Bass, 1985; Conger & Kanungo, 1988; De Cremer & van Knippenberg, 2002; Lord, Brown, & Freiberg, 1999; Shamir et al., 1993). A limitation (see Haslam & Platow, 2001a; Mowday & Sutton, 1993) of the charismatic leadership perspective is that it has a tendency to conceptualize charismatic leadership in terms of charisma, and to conceptualize charisma as a set of invariant personality traits, such as need for power and self-confidence, possessed by certain people but not others (House, 1977; Shamir et al., 1993). However, theories of charismatic leadership also identify a number of key behaviors that are quite clearly related, from our perspective, to social identity processes, such as articulating an inspiring vision for the collective, emphasizing collective goals, and making selfsacrifices for the collective good (e.g., Bass, 1985; Conger & Kanungo, 1987; Shamir et al., 1993). The social identity theory of leadership allows us to propose a more dynamic relationship between charisma and leadership. The background premise is that charisma is not a static property of individuals, but a social-cognitive attributional consequence of the interaction of group salience and leader prototypicality. As was described in our earlier theory section (also see Hogg, 2001a), in salient groups with emergent leaders, members internally attribute the leader’s prototype-based leadership behavior (e.g., the ability to influence the group). In this way, members construct a charismatic personality for the leader, which in turn strengthens the leader’s leadership effectiveness. However, building on van Knippenberg and colleagues’ work on group-oriented behavior, we can also predict that in a group where a leader happens to be less prototypical, the attribution of charisma rests more heavily on commission of group-oriented behavior than it does for more prototypical leaders (Platow & van Knippenberg, 2001; van Knippenberg & van Knippenberg, 2000, 2001).

SOCIAL IDENTITY AND LEADERSHIP PROCESSES IN GROUPS

31

Initial support for these ideas comes from Platow, van Knippenberg, Haslam, van Knippenberg, and Spears (2002). Across two experiments, Platow et al. (2002) had prototypical or nonprototypical leaders request help from the rest of the group in such a way as to emphasize how such help to the leader would serve the followers’ self-interest or would serve the collective good of the group. The key dependent measure was attributions of charisma to the leader. It was predicted that nonprototypical leaders would have to emphasize the interest of the collective in order to be seen as charismatic. In Experiment 1 there was only a main effect for prototypicality. Prototypical leaders were rated as more charismatic than nonprototypical leaders, but type of message had no effects on attributed charisma. Experiment 2 used the same design but with two key modifications. There was an explicit outgroup, and the leader was described quite fully and concretely in terms of group prototypical (or nonprototypical) characteristics. In Experiment 2, leader prototypicality and type of message interacted significantly as predicted. Nonprototypical leaders were perceived to be more charismatic when they invoked collective interest rather than followers’ personal interest, whereas prototypical leaders were seen as relatively charismatic regardless of the type of message sent.

B. PROTOTYPICALITY AND LEADER-MEMBER EXCHANGE Another major perspective in contemporary leadership research is the social exchange perspective on leader–follower relations, which is primarily represented by leader–member exchange (LMX) theory of leadership (Graen & Uhl-Bien, 1995; cf. Hollander & Offerman, 1990). Testifying to its impact, a recent review of the LMX literature lists 147 studies since the concept was originally introduced in the early 1970s (Schriesheim et al., 1999). LMX theory attributes effective leadership to the development of high-quality dyadic exchange relationships between the leader and specific subordinates— exchange relationships that treat specific members as unique individuals within the group. From a social identity leadership perspective we would predict that this kind of interpersonal relationship might be effective in low salience groups, but not in high salience groups (cf. the effects of leader schema congruence discussed above). In high salience groups, it would be more effective to treat members in a more depersonalized manner that recognizes them as group members rather than individuals (cf. Tyler, 1999). This analysis has recently been tested by Hogg, Martin, Weeden, and Epitropaki (2001 Study 1; Hogg & Martin, 2003). Hogg et al. (2001) conducted a naturalistic correlational study within organizations. They administered a questionnaire to 439 respondents

32

HOGG AND VAN KNIPPENBERG

performing a range of jobs in seven UK service and manufacturing companies. There were two dependent variables: a three-item scale measuring perceived leader effectiveness,
¼ .80, and a two-item scale measuring leader satisfaction, r ¼ .76, p < .001). There were two predictor variables: a five-item scale measuring the extent to which respondents felt that the company encouraged people to work in teams or not,
¼ .85 (high scores approximate a high salience group and low scores a low salience group), and a two-item scale measuring the extent to which the leader’s leadership style treated subordinates as unique individuals or as one of the team, r ¼ .44, p < .001. Stepwise regression revealed that after demographic variables had been removed at Step 1, leader effectiveness and leader satisfaction were each predicted by salience and by leader style (betas all significant at p < .001), and by the interaction of salience and leader style (beta ¼ .078 on effectiveness, and .090 on style—both p < .05). The interaction generally supports predictions derived from the social identity analysis. To illustrate the effect we have only presented the interaction on leader effectiveness. Figure 6 displays cell means for the 2
2 interaction on leader effectiveness, based, for ease of display, on median splits on the predictor variables. It shows that in low salience groups leaders who adopt an interpersonal leadership style are perceived to be significantly more effective than leaders who adopt a depersonalized leadership style, and that this difference disappears in high salience groups. Put differently, the leadership effectiveness of a depersonalized leadership style is significantly increased in high salience groups—the effectiveness of an interpersonal style is not significantly changed. These results provide some support for the hypothesis that a depersonalized leadership style becomes more important relative to an interpersonal leadership style as group membership becomes more salient. The study represents a step toward integration of the social identity analysis of leadership and the LMX theory of leadership. The processes described in LMX theory are more important under low rather than high group salience, whereas the processes described in the social identity analysis of leadership have a greater impact on leadership effectiveness under conditions of high social identity salience.

C. SUMMARY The social identity analysis of leadership provides significant qualifications and modifications of two major (organizational) psychological perspectives on effective leadership. For LMX theory, it complements the focus on interpersonal relationships that may be most effective in low salience groups, with a focus on depersonalized, social identity-based

SOCIAL IDENTITY AND LEADERSHIP PROCESSES IN GROUPS

33

5

Leadership effectiveness

4.5

4

3.5

3.82 3.45

3.42

3 2.65 2.5

2 Low

High Salience

Interpersonal Group Fig. 6. Hogg, Martin, Weeden, and Epitropaki (2001, Study 1): Perceived leader effectiveness as a function of group salience and leader–subordinate leadership style (p < .05).

leadership that may be more appropriate for high salience groups. For theories of charismatic and transformational leadership, it complements the focus on invariant traits as determinants of charisma, with contextdependent social identity-based attributes (i.e., prototypicality) as determinants of charisma. Moreover, it suggests that group-oriented acts that are associated with charismatic leadership may serve as ‘‘substitutes’’ for prototypicality as a basis of charisma. Further studies are needed to explore these ideas more fully.

VI. Implications and Future Directions for Leadership Research So far, our concern has been with assessing empirical support for the social identity analysis of leadership and with its impact on contemporary psychological theories of leadership effectiveness. In this section, we explore some broader implications of the social identity analysis for theory and

34

HOGG AND VAN KNIPPENBERG

research in leadership effectiveness. This discussion is by no means exhaustive—the range of possible applications and extensions of the social identity analysis is very large. Rather, we focus on three implications that we believe to be among the more obvious and important for leadership research: demographic differences and the ‘‘glass ceiling,’’ the dynamics of maintaining effective leadership over time, and ‘‘leading change.’’

A. DEMOGRAPHIC DIFFERENCES AND THE GLASS CEILING Research, primarily in western organizational contexts, has discovered evidence for a ‘‘glass ceiling,’’ an invisible barrier, to advancement of social minorities from managerial positions to top leadership positions (e.g., Eagly, Karau, & Makhijani, 1995; Rudman & Glick, 1999). The research primarily focuses on women, but the analysis is intended to apply to other societal minorities, such as those based on race, ethnicity, disability, and so forth. The effect may be produced by a number of different social psychological processes. The role of social identity-related processes has only recently been proposed by Hogg and his associates (Hogg, Fielding, Johnston, Masser, Russell, & Svensson, 2001; also see Hogg, 2001a,d,e). The key idea is that as group membership becomes more salient, leadership endorsement, and effective leadership itself, rests increasingly on a good perceived match between properties of the leader and properties of the group prototype. Properties of the leader and properties of the prototype can include attitudes and behavior, but also demographic properties such as gender, ethnicity, socioeconomic status, and so forth. Thus, if there is a demographic mismatch between leader and group, then this will become an insurmountable obstacle to effective leadership as the group becomes an increasingly strong basis for self-definition. Hogg (e.g., 2000, in press b) suggests that group identification is often very strong when a group is under status or entitativity threat. These are circumstances where real leadership, as opposed to middle-level management, is required. The glass ceiling, for women, may, therefore, be partly a result of the fact that many western organizational prototypes are more masculine than feminine (e.g., Cejka & Eagly, 1999; Glick, Wilk, & Perrault, 1995), and therefore women are less group prototypical than men when organizations really need leadership rather than management. The idea that demographic minorities may find leadership difficult in salient groups whose prototypes represent the demographic majority has been tested by Hogg, Fielding, Johnston, Masser, Russell, and Svensson (2001). Hogg et al. (2001) conducted two very closely related experiments employing a computerized modification of the paradigm of Hains et al.

SOCIAL IDENTITY AND LEADERSHIP PROCESSES IN GROUPS

35

(1997). These experiments investigated leadership effectiveness as a function of identity salience and the match between the leader’s gender and the gender stereotypicality of the group norm. The two experiments differed only in that Experiment 2 (N ¼ 260) employed slightly stronger and cleaner manipulations than Experiment 1 (N ¼ 147). In these experiments, students anticipated joining a group to discuss university resource allocations for undergraduate classes. Group salience was manipulated and participants were informed that their group had an agentic/instrumental (i.e., male stereotypical) or a communal/expressive (i.e., female stereotypical) norm for how the discussion was to be conducted. Participants were also told that a leader had been randomly appointed— they discovered that the leader was either male or female. The three manipulated variables were thus: 2 (group salience)
2 (group norm)
2 (sex of leader). There was a fourth variable formed by median split of participants, on the basis of their responses on the Hostile Sexism subscale of Glick and Fiske’s (1996) Ambivalent Sexism Inventory, into those with traditional and those with progressive sex role orientations. In addition to manipulation checks, there were three dependent variables: a four-item measure of anticipated group effectiveness (
¼ .84 in both experiments), a 12-item measure of anticipated leader effectiveness (
¼ .89 and .91 in Experiments 1 and 2, respectively), and a six-item measure of the actual quality of the leader’s decisions (
¼ .92 in both experiments). Because the two experiments were so similar and were conducted by the same research team with participants from the same population it was possible to pool the two data sets (N ¼ 407) and perform five-way analyses with Experiment (1 vs. 2) as the fifth variable. Across the three dependent variables, the predicted four-way interaction of salience, norm, leader, and sex role orientation was significant [F(3, 373) ¼ 2.84, p < .05]. This interaction was also significant on the constituent variable of anticipated leader effectiveness. At the individual experiment level, the multivariate four-way effect was significant in Experiment 1, and the univariate four-way effect was significant on decision quality in Experiment 1, and on group performance and anticipated leader effectiveness in Experiment 2. Figure 7 illustrates what was found. It shows the interaction of salience, norm, and leader gender on anticipated group performance in Experiment 2 among traditional participants only. As predicted, traditional participants (those who believe that men are agentic/instrumental and that women are communal/expressive) under high salience felt instrumental groups would perform better under male than female leadership, and expressive groups would perform better under female than male leadership (although the means are in the correct direction, simple main effects were not significant). Another way to frame this is that relative to low salience groups, group

36

HOGG AND VAN KNIPPENBERG

7.5

Group performance

7.15 6.97

7 6.88 6.70

6.75

6.68

6.5

6.38 6.28

6 Male Instrum

Female Instrum

Male Express

Female Express

Norm x Leader gender Low salience High salience Fig. 7. Hogg, Fielding, Johnston, Masser, Russell, and Svensson (2001; Experiment 2): The effect of salience by norm by leader gender on anticipated group performance, among traditional participants (p < .05).

performance was upgraded under male leadership in instrumental groups and upgraded under female leadership in expressive groups. Overall the interaction of salience, norm, and leader gender on the three leadership perception variables was consistent with the social identity analysis of leadership—as group salience increased, leadership effectiveness became more dependent on the group prototypicality of the leader (in this case the gender match between leader and norm). Progressive participants behaved in almost exactly the opposite manner to the traditionals. Under high salience they felt instrumental groups would do better under female than male leadership, and expressive groups would do better under male than female leadership. Relative to low salience groups, group performance was upgraded under female leadership in instrumental groups and under male leadership in expressive groups, and downgraded under male leadership in instrumental groups and under female leadership in expressive group. This is as predicted, based on Swim, Aikin, Hall, and Hunter’s (1995) finding that ‘‘progressives’’ are more aware that occupational segregation may be a result of prejudice, and Masser and Abrams’s (2001) argument that progressives might exhibit ‘‘reverse discrimination’’ in favor of women in order to combat gender stereotypes.

SOCIAL IDENTITY AND LEADERSHIP PROCESSES IN GROUPS

37

B. MAINTAINING LEADERSHIP: PROTOTYPE MANAGEMENT AND THE PROTOTYPICALITY PARADOX Thus far we have seen how prototype-based depersonalization fairly automatically imbues the most prototypical member of a group with many attributes of leadership—status, charisma, popular support, and the ability to influence. These attributes also allow the leader to actively maintain his or her leadership position. The longer an individual remains in a leadership position the more they will be socially liked, the more consensual will social attraction be, and the more entrenched will be the fundamental attribution effect. There may, however, be two threats to prototype-based leadership effectiveness that have as yet received little attention. First, the group prototype may change over time, rendering the leader less prototypical. Second, as the influence differential between leader and followers becomes more entrenched in group life, followers may start to perceive the leader as different from rather than as prototypical of the group. Prototypicality may be a basis for effective leadership, but that does not mean it lasts forever. 1. Prototype Management Changes in the social comparative context may change the salient outgroup or change the salient dimensions of intergroup comparisons—both of which impact prototypicality within the group, and thus (part of ) the basis for the leader’s position. Therefore, over time and across contexts, the leader may decline in prototypicality whereas other members become more prototypical, opening the door, particularly under high salience conditions, to a redistribution of influence within the group. An established leader is, however, well placed, in terms of material and social psychological resources, to combat this by redefining the prototype in a self-serving manner to marginalize contenders and centralize self. This can be done by accentuating the existing ingroup prototype, by pillorying ingroup deviants, or by demonizing an appropriate outgroup. Work by Reicher, Hopkins, and associates (e.g., Reicher, Drury, Hopkins, & Stott, in press; Reicher & Hopkins, 1996) on the rhetoric used by political leaders shows that generally all three tactics are used, and that the very act of engaging in these powerful rhetorical devices is often viewed as convincing evidence of effective leadership. Indeed, Reicher and Hopkins proposed that leaders are in this sense ‘‘entrepreneurs of identity’’—as we might put it in the context of the present analysis, ‘‘prototypicality managers.’’ In a similar vein, Rabbie and Bekkers (1978) show that leaders whose position is insecure are more likely to seek conflict with other groups. They may do this in order to be seen by the group to be behaving in a grouporiented manner. More generally, leaders who feel they are not, or are no

38

HOGG AND VAN KNIPPENBERG

longer, prototypical, may strategically engage in a range of group-oriented behaviors in order to strengthen their membership credentials (e.g., Platow & van Knippenberg, 2001; Platow et al., 2002). Leadership maintenance also benefits from consensual prototypicality, because of the latter’s effect on social attraction. In groups with less consensual prototypes, there is less consensus of perceptions of and feelings for the leader and thus the leader may have less power and may occupy a less stable position. It is in the leader’s interest to maintain a clearly defined and consensual prototype. Simple and more clearly focused prototypes are less open to ambiguity and alternative interpretations and are thus better suited to consensuality. One way in which the leader can do this is to identify ingroup deviates and orchestrate a campaign of derogation and rejection—this is a process that clarifies the prototype that the leader best represents (see Marques, Abrams, Pa´ ez, & Hogg, 2001). Another strategy is to focus on a carefully chosen threatening or ‘‘wicked’’ outgroup as the target of vilification, in order to polarize or extremitize the ingroup prototype away from the outgroup. These processes frequently operate in totalist or extremist societies or groups that have all-powerful leaders (e.g., Hogg, 2001e; Hogg & Reid, 2001). However, they regularly occur in much more ordinary group contexts. For example, the committee set up by the United States Senator Joe McCarthy during the height of the Cold War in the 1950s to purge the entertainment industry of communist sympathizers (i.e., political deviants) was formally called the House Committee on UnAmerican Activities. By pillorying individuals and groups that were unAmerican, this clarified what was American. There is, thus, some support for the proposition that leaders may actively manage the group prototype to maintain, or bolster, their leadership position. More research is needed to better understand the dynamics of the maintenance of prototype-based leadership, and to develop the temporal dimension in the social identity analysis of leadership. 2. From Influence to Power: The Prototypicality Paradox Even if the comparative context remains stable and the group prototype is highly consensual—probably especially under these conditions—there is another threat to prototype-based leadership effectiveness that, paradoxically, lies in the processes that underlie prototypicality based leadership effectiveness (see Hogg, 2001e; Hogg & Reid, 2001). Prototypicality imbues leaders with status, charisma, and influence and may thus set leaders apart from the rest of the group. Over time, as this intragroup differentiation becomes more entrenched in group life, there may arise a gradual perceptual separation of the leader from the rest of the group, through structural role

SOCIAL IDENTITY AND LEADERSHIP PROCESSES IN GROUPS

39

differentiation grounded in social attraction and attribution processes. The leader may gradually be perceived as ‘‘other’’ rather than ‘‘one of us.’’ In other words, the person who embodied the essence of the group by being most prototypical may become effectively an outgroup member within the group (i.e., become nonprototypical). As a result, the leader will lose part of his or her influence base, and in order to gain compliance may feel the need to resort to the use of power (i.e., coercive forms of influence) instead. Such attempts to force compliance from the group will set the leader further apart from the group, and will work to the further detriment of the leader’s influence base (e.g., van Knippenberg & van Knippenberg, in press). Once this process of separation between leader and follower has been set in motion, a downward spiral may occur. Prototypicality based leadership, which converts followers, is gradually replaced by the use of position-based power that elicits only compliance. The important point here is that an intragroup relationship is effectively transformed into a highly status-marked intergroup relationship. However, prototypicality based leadership does not inevitably end in a separation of leader and followers. Paradoxically, it is more likely to occur in groups where the leader is highly prototypical, the group is very cohesive, and members have a consensual prototype. In these groups prototype-based status, charisma, and influence will be highest, and therefore the differentiation between leader and group, and the instantiation of an intergroup relationship, will occur most readily and most completely. This is unlikely to occur so readily in groups with less markedly prototype-based leadership. Leader prototypicality may actually protect against the slide into power tactics based on intergroup schism. The leader’s prototypicality encourages followers to see the leader as ‘‘one of us’’ and may thus blind followers to the leader’s use of power over them. Prototypicality may legitimate the entire range of leadership tactics, including those based on schism and power. Furthermore, leaders may maintain, or regain, their prototypicality by strategic management of the group’s prototype, as discussed above. These ideas suggest a development of the social identity analysis of leadership to explore the development of leader prototypicality and leader– group relations over time, to better understand the tension between being ‘‘one of us’’ and being set apart from the group.

C. ‘‘LEADING CHANGE’’ For some theorists, the essence of effective leadership is to bring about, or lead, change (e.g., Yukl, 2001). Indeed, the emphasis on charismatic and transformational leadership represents precisely a focus on how leaders bring about social or organizational changes (e.g., Bass, 1985; Conger &

40

HOGG AND VAN KNIPPENBERG

Kanungo, 1987; Shamir et al., 1993). Charismatic leaders are considered more innovative than noncharismatic leaders, and are therefore better at initiating and implementing change. The organizational change literature—there is no reason to assume that this should not hold for societal change as well—cites resistance to change as one of the principal obstacles to effective change (e.g., Connor, 1995). A recent social identity analysis of mergers of groups and organizations has identified social identity concerns as a significant source of resistance to change (van Knippenberg, van Knippenberg, Monden, & De Lima, 2002; van Knippenberg & van Leeuwen, 2001). Mergers often raise the specter of a loss of identity and the construction or acquisition of a new identity. This can pose a significant threat to self-conceptualization, in which case a leader’s attempts to bring about such change will be vigorously resisted. However, resistance to change is not inevitable. Group members may be quite accepting of change, including identity change, as long as a sense of continuity of identity is ensured, that is, as long as group members may feel that, despite the changes, it is still ‘‘their group’’ (Rousseau, 1998). Because they represent the collective identity, prototypical leaders may be better than nonprototypical leaders at ensuring such a sense of continuity. Changes promoted by prototypical leaders will be viewed as more identity consistent than the same changes promoted by less prototypical leaders. As discussed above, prototypical leaders are better placed than less prototypical leaders to act as ‘‘entrepreneurs of identity’’ or ‘‘prototypicality managers’’— that is, to not just make changes to the group, organization, or society, but to actually change the group prototype (i.e., the group identity) in a consensually accepted way. Tentative support for this proposition may be derived from the studies described above that show that prototypical leaders have more leeway in their behavior than less prototypical leaders.

VII. Summary and Conclusions After many years of decline, leadership has once again attracted the attention of social psychologists. This new interest has been spurred by conceptual advances in social cognition and social identity, and by growing synergies between social cognition, social identity, and organizational psychology. Scholars have become concerned that current leadership theories, largely in organizational psychology, are inadequately grounded in an analysis of the generative role of psychological group membership. The social identity analysis of leadership described in this chapter goes some way toward addressing this concern.

SOCIAL IDENTITY AND LEADERSHIP PROCESSES IN GROUPS

41

The social identity perspective views group and intergroup phenomena (perceptions, attitudes, feelings, and behaviors) as being generated by a process in which people categorize and depersonalize themselves and others in terms of the relevant ingroup or outgroup prototype. Prototypes describe and prescribe all group-defining attributes (perceptions, attitudes, feelings, and behaviors) and thus provide people with social identities. They are formed and modified according to the metacontrast principle to maximize the subjective entitativity of groups. Social identity processes are motivated by a need to reduce uncertainty about self, others, and one’s relations with others. They are also motivated by self-enhancement that underpins a struggle to protect or enhance the evaluatively positive distinctiveness of one’s own group relative to other groups. Both these motives are played out against a backdrop of beliefs about the nature of intergroup relations, and thus about the sorts of strategies that can realistically be adopted to reduce uncertainty and to protect/enhance the evaluative self-concept. Applied to leadership, the social identity perspective focuses on the critical role of prototypicality in salient groups. Within low salience groups— loose aggregates of people who do not identify very strongly—leadership effectiveness probably rests on matching the appropriate leadership schema and on developing favorable interpersonal leader–member exchange relations. Things are different in more salient groups—groups that are subjectively important and with which members identify strongly. Attention and information processing are focused on prototypicality and on how prototypical members are. Leadership effectiveness rests firmly on the extent to which the leader appears to match the group prototype and thus to embody the group attributes to which followers aspire and conform. Social attraction processes ensure that prototypical leaders are unilaterally and consensually socially liked—they are popular, in group terms, and thus able to actively influence others and to be innovative within the broad parameters of the group’s prototype. Social attraction also instantiates a status and prestige differential between leader(s) and followers. Attribution processes, specifically the fundamental attribution error, correspondence bias, or essentialism, operate to construct a charismatic and leadership persona for the prototypical leader. This happens because followers pay close attention to prototypicality, and thus prototypical leaders are figural against the background of less prototypical members. One consequence is that the leader acquires greater and more secure leadership ability. Another consequence is that the status and prestige differential between leader(s) and followers is entrenched. In this chapter we described the social identity analysis of leadership in some detail, including a description of the ways that prototypical leaders can protect their tenure through manipulation and control of the group’s

42

HOGG AND VAN KNIPPENBERG

prototype. We also occupied considerable space describing key empirical tests of the analysis—these tests necessarily hinge on a demonstration that leadership processes become more prototype based with increasing group salience. There are now many direct and indirect tests from a number of laboratories and research groups around the world that provide support for the social identity analysis described here. We described the results of about 25 independent samples from 16 different studies in some detail. Overall, these studies show that contingent on social identity salience and group identification, leader prototypicality is positively related to leadership effectiveness, as are proxies of prototypicality such as the group membership of the leader and the leader selection procedure. Similarly, group-oriented attitudes and behaviors, like leader commitment to the group and leader self-sacrifice for the group, are positively related to leader effectiveness in salient groups. This latter relationship, however, primarily occurs for less prototypical leaders—more prototypical leaders tend to be relatively effective irrespective of the group orientation of their behavior. In contrast to leader prototypicality, which grows in importance as group membership becomes more salient, more personalized attributes of the leader and interpersonal aspects of the leader’s behavior, such as leader schema congruence and the quality of the interpersonal leader–follower relationship, diminish in importance as group membership becomes more salient. There is, therefore, good support for the core propositions of the social identity analysis of leadership as outlined in this chapter. What remains to be tested more extensively is the role of attribution processes, and to a lesser extent the role of social attraction. The social identity analysis of leadership may provide important qualifications and extensions of some major contemporary perspectives on leadership. For example, there is some support for the social identity proposition that prototypicality is a basis of charisma, and that leader behaviors that are typically associated with charismatic leadership, such as invoking collective interest, are less important determinants of charisma for prototypical than for less prototypical leaders. There is also some evidence that interpersonal leader–member exchange (LMX) relationships may diminish in importance whereas depersonalized prototype-based relationships increase in importance as a function of increasing group salience. These findings are a significant step toward integration of the social identity analysis of leadership with contemporary organizational psychological theories of charismatic and transformational leadership, and of leader– member exchange processes (LMX theory). The social identity analysis of leadership has great potential to reconceptualize a range of leadership phenomena and leadership contexts. To illustrate this, we identified three possible directions: the glass ceiling

SOCIAL IDENTITY AND LEADERSHIP PROCESSES IN GROUPS

43

effect, leadership maintenance processes in relationship to threats to the prototypicality of leaders (i.e., changes in group prototypes, and the differentiation between leader and group), and the role of leadership in social and organizational change. There is much work yet to be done, but this work can be grounded in strong empirical support for the core propositions of the social identity analysis of leadership. In conclusion, the social identity analysis of leadership views leadership as a group process that arises from the social categorization and depersonalization processes associated with social identity. Prototype-based depersonalization and the behavior of followers in salient groups play a critical role. They empower individuals as leaders, imbue them with charisma, create a status differential between leader(s) and followers that has some of the typical characteristics of uneven status intergroup relations, and set up conditions that are conducive to coercion and the exercise of power. These ideas are a rich source of conceptual explorations, and basic and applied empirical research into social identity and leadership. They help regain the study of leadership for social psychology.

Acknowledgments Michael Hogg would like to acknowledge grant funding from the Australian Research Council that has enabled his research program on leadership, facilitated the writing of this chapter, and allowed him and Daan van Knippenberg to meet to discuss and develop their leadership research program. We would also like to thank Barbara van Knippenberg for her comments on an earlier draft of this article.

References Abrams, D., & Hogg, M. A. (1988). Comments on the motivational status of self-esteem in social identity and intergroup discrimination. European Journal of Social Psychology, 18, 317–334. Abrams, D., & Hogg, M. A. (1990). Social identification, self-categorization and social influence. European Review of Social Psychology, 1, 195–228. Abrams, D., & Hogg, M. A. (1998). Prospects for research in group processes and intergroup relations. Group Processes and Intergroup Relations, 1, 7–20. Abrams, D. & Hogg, M. A. (Eds.) (1999). Social identity and social cognition. Oxford, UK: Blackwell. Abrams, D., & Hogg, M. A. (2001). Collective identity: Group membership and selfconception. In M. A. Hogg & R. S. Tindale (Eds.), Blackwell handbook of social psychology: Group processes (pp. 425–460). Oxford, UK: Blackwell. Bales, R. F. (1950). Interaction process analysis: A method for the study of small groups. Reading, MA: Addison-Wesley.

44

HOGG AND VAN KNIPPENBERG

Bass, B. M. (1985). Leadership and performance beyond expectations. New York: Free Press. Bass, B. M. (1990a). Bass and Stogdill’s handbook of leadership: Theory, research and managerial applications. New York: Free Press. Bass, B. M. (1990b). From transactional to transformational leadership: Learning to share the vision. Organizational Dynamics, 18, 19–31. Bass, B. M. (1998). Transformational leadership: Industrial, military, and educational impact. Mahwah, NJ: Erlbaum. Bass, B. M., & Avolio, B. J. (1993). Transformational leadership: A response to critiques. In M. M. Chemers & R. A. Ayman (Eds.), Leadership theory and research: Perspectives and directions (pp. 49–80). London: Academic Press. Berger, J., Wagner, D., & Zelditch, M. (1985). Expectation states theory: Review and assessment. In J. Berger & M. Zelditch (Eds.), Status, rewards, and influence (pp. 1–72). San Francisco: Jossey-Bass. Berscheid, E., & Reis, H. T. (1998). Attraction and close relationships. In D. T. Gilbert, S. T. Fiske, & G. Lindzey (Eds.). The handbook of social psychology (4th ed., Vol. 2, pp. 193–281). New York: McGraw-Hill. Brewer, M. B. (1979). Ingroup bias in the minimal intergroup situations: A cognitivemotivational analysis. Psychological Bulletin, 86, 307–324. Brewer, M. B. (1991). The social self: On being the same and different at the same time. Personality and Social Psychology Bulletin, 17, 475–482. Brewer, M. B., & Brown, R. J. (1998). Intergroup relations. In D. T. Gilbert, S. T. Fiske, & G. Lindzey (Eds.). The handbook of social psychology (4th ed., pp. 554–594). New York: McGraw-Hill. Brewer, M. B., & Gardner, W. (1996). Who is this ‘‘we’’? Levels of collective identity and self representations. Journal of Personality and Social Psychology, 71, 83–93. Bryman, A. (1992). Charisma and leadership in organization. London: Sage. Burns, J. M. (1978). Leadership. New York: Harper & Row. Cartwright, D. & Zander, A. (Eds.) (1968). Group dynamics: Research and theory (3rd ed.) London: Tavistock. Cartwright, S., & Cooper, C. L. (1992). The impact of mergers and acquisitions on people at work: Existing research and issues. British Journal of Management, 1, 65–76. Cejka, M. A., & Eagly, A. H. (1999). Gender-stereotypic images of occupations correspond to the sex segregation of employment. Personality and Social Psychology Bulletin, 25, 413–423. Chemers, M. M. (2001). Leadership effectiveness: An integrative review. In M. A. Hogg & R. S. Tindale (Eds.), Blackwell handbook of social psychology: Group processes (pp. 376–399). Oxford, UK: Blackwell. Conger, J. A., & Kanungo, R. N. (1987). Towards a behavioral theory of charismatic leadership in organizational settings. Academy of Management Review, 12, 637–647. Conger, J. A., & Kanungo, R. N. (1988). Behavioral dimensions of charismatic leadership. In J. A. Conger & R. N. Kanungo (Eds.). Charismatic leadership: The elusive factor on organizational effectiveness. San Francisco: Jossey-Bass. Conger, J. A., & Kanungo, R. N. (1998). Charismatic leadership in organizations. Thousand Oaks, CA: Sage. Conner, D. R. (1995). Managing at the speed of change: How resilient managers succeed and prosper where others fail. New York: Villard Books. De Cremer, D. (2002). Charismatic leadership and cooperation in social dilemmas: A matter of transforming motives? Journal of Applied Social Psychology, 32, 997–1016. De Cremer, D, & van Knippenberg, D (2002). How do leaders promote cooperation? The effects of charisma and procedural fairness. Journal of Applied Psychology, 87, 858–866.

SOCIAL IDENTITY AND LEADERSHIP PROCESSES IN GROUPS

45

De Cremer, D., & Van Vugt, M. (2002). Intergroup and intragroup aspects of leadership in social dilemmas: A relational model of cooperation. Journal of Experimental Social Psychology, 38, 126–136. Devine, P. G., Hamilton, D. L., & Ostrom, T. M. (Eds.) (1994). Social cognition: Impact on social psychology. San Diego, CA: Academic Press. Duck, J. M., & Fielding, K. S. (1999). Leaders and sub-groups: One of us or one of them? Group Processes and Intergroup Relations, 2, 203–230. Eagly, A. H., Karau, S. J., & Makhijani, M. G. (1995). Gender and the effectiveness of leaders: A meta-analysis. Psychological Bulletin, 117, 125–145. Erber, R., & Fiske, S. T. (1984). Outcome dependency and attention to inconsistent information. Journal of Personality and Social Psychology, 47, 709–726. Fiedler, F. E. (1964). A contingency model of leadership effectiveness. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 1, pp. 149–190). New York: Academic Press. Fiedler, F. E. (1971). Leadership. Morristown, NJ: General Learning Press. Fielding, K. S., & Hogg, M. A. (1997). Social identity, self-categorization, and leadership: A field study of small interactive groups. Group Dynamics: Theory, Research, and Practice, 1, 39–51. Fiske, S. T. (1993). Controlling other people: The impact of power on stereotyping. American Psychologist, 48, 621–628. Fiske, S. T., & De´ pret, E. (1996). Control, interdependence and power: Understanding social cognition in its social context. European Review of Social Psychology, 7, 31–61. Fiske, S. T., & Taylor, S. E. (1991). Social cognition (2nd ed.) New York: McGraw-Hill. Foddy, M., Smithson, M., Schneider, S., & Hogg, M. A. (Eds.) (1999). Resolving social dilemmas: Dynamic, structural, and intergroup aspects. Philadelphia, PA: Psychology Press. Gilbert, D. T., & Jones, E. E. (1986). Perceiver-induced constraint: Interpretations of selfgenerated reality. Journal of Personality and Social Psychology, 50, 269–280. Gilbert, D. T., & Malone, P. S. (1995). The correspondence bias. Psychological Bulletin, 117, 21–38. Glick, P., & Fiske, S. T. (1996). The Ambivalent Sexism Inventory: Differentiating hostile and benevolent sexism. Journal of Personality and Social Psychology, 70, 491–512. Glick, P., Wilk, K., & Perrault, M. (1995). Images of occupations: Components of gender and status in occupational stereotypes. Sex Roles, 32, 564–582. Graen, G. B., & Uhl-Bien, M. (1995). Relationship-based approach to leadership: Development of leader-member exchange (LMX) theory of leadership over 25 years: Applying a multilevel multi-domain approach. Leadership Quarterly, 6, 219–247. Hains, S. C., Hogg, M. A., & Duck, J. M. (1997). Self-categorization and leadership: Effects of group prototypicality and leader stereotypicality. Personality and Social Psychology Bulletin, 23, 1087–1100. Hall, R. J., & Lord, R. G. (1995). Multi-level information processing explanations of followers’ leadership perceptions. Leadership Quarterly, 6, 265–287. Hamilton, D. L., & Sherman, S. J. (1996). Perceiving persons and groups. Psychological Review, 103, 336–355. Hamilton, D. L., Sherman, S. J., & Castelli, L. (2002). A group by any other name: The role of entitativity in group perception. In W. Stroebe & M. Hewstone (Eds.), European Review of Social Psychology (Vol. 12, pp. 136–166). Chichester, U.K.: Wiley. Hamilton, D. L., Sherman, S. J., & Lickel, B. (1998). Perceiving social groups: The importance of the entitativity continuum. In C. Sedikides, J. Schopler, & C. A. Insko (Eds.), Intergroup cognition and intergroup behavior (pp. 47–74). Mahwah, NJ: Erlbaum. Haslam, N., Rothschild, L., & Ernst, D. (1998). Essentialist beliefs about social categories. British Journal of Social Psychology, 39, 113–127. Haslam, S. A. (2001). Psychology in organisations: The social identity approach. London: Sage.

46

HOGG AND VAN KNIPPENBERG

Haslam, S. A., McGarty, C., Brown, P. M., Eggins, R. A., Morrison, B. E., & Reynolds, K. J. (1998). Inspecting the emperor’s clothes: Evidence that random selection of leaders can enhance group performance. Group Dynamics: Theory, Research, and Practice, 2, 168–184. Haslam, S. A., Oakes, P. J., McGarty, C., Turner, J. C., & Onorato, S. (1995). Contextual changes in the prototypicality of extreme and moderate outgroup members. European Journal of Social Psychology, 25, 509–530. Haslam, S. A., & Platow, M. J. (2001a). Your wish is our command: The role of shared social identity in translating a leader’s vision into followers’ action. In M. A. Hogg & D. J. Terry (Eds.), Social identity processes in organizational contexts (pp. 213–228). Philadelphia, PA: Psychology Press. Haslam, S. A., & Platow, M. J. (2001b). The link between leadership and followership: How affirming social identity translates vision into action. Personality and Social Psychology Bulletin, 27, 1469–1479. Haslam, S. A., van Knippenberg, D., Platow, M., & Ellemers, N. (Eds.) (2003). Social identity at work: Developing theory for organizational practice. Philadelphia, PA: Psychology Press. Hemphill, J. K. (1961). Why people attempt to lead. In L. Petrullo & B. M. Bass (Eds.), Leadership and interpersonal behavior (pp. 201–215). New York: Holt. Hogg, M. A. (1992). The social psychology of group cohesiveness: From attraction to social identity. New York: New York University Press. Hogg, M. A. (1993). Group cohesiveness: A critical review and some new directions. European Review of Social Psychology, 4, 85–111. Hogg, M. A. (2000). Subjective uncertainty reduction through self-categorization: A motivational theory of social identity processes. European Review of Social Psychology, 11, 223–255. Hogg, M. A. (2001a). A social identity theory of leadership. Personality and Social Psychology Review, 5, 184–200. Hogg, M. A. (2001b). Social identity and the sovereignty of the group: A psychology of belonging. In C. Sedikides & M. B. Brewer (Eds.), Individual self, relational self, collective self (pp. 123–143). Philadelphia, PA: Psychology Press. Hogg, M. A. (2001c). Social categorization, depersonalization, and group behavior. In M. A. Hogg & R. S. Tindale (Eds.), Blackwell handbook of social psychology: Group processes (pp. 56–85). Oxford, UK: Blackwell. Hogg, M. A. (2001d). Social identification, group prototypicality, and emergent leadership. In M. A. Hogg & D. J. Terry (Eds.), Social identity processes in organizational contexts (pp. 197–212). Philadelphia, PA: Psychology Press. Hogg, M. A. (2001e). From prototypicality to power: A social identity analysis of leadership. In S. R. Thye, E. J. Lawler, M. W. Macy, & H. A. Walker (Eds.), Advances in group processes, (Vol. 18, pp. 1–30). Oxford, UK: Elsevier. Hogg, M. A. (2002). Social identity. In M. R. Leary & J. P. Tangney (Eds.), Handbook of self and identity (pp. 462–479). New York: Guilford. Hogg, M. A. (in press a). Social identity and leadership. In D. M. Messick & R. Kramer (Eds.), The psychology of leadership: Some new approaches. Mahwah, NJ: Erlbaum. Hogg, M. A. (in press b). Uncertainty and extremism: Identification with high entitativity groups under conditions of uncertainty. In V. Yzerbyt, C. M. Judd, & O. Corneille (Eds.), The psychology of group perception: Contributions to the study of homogeneity, entitativity, and essentialism. New York: Psychology Press. Hogg, M. A., & Abrams, D. (1988). Social identifications: A social psychology of intergroup relations and group processes. London: Routledge. Hogg, M. A., Cooper-Shaw, L., & Holzworth, D. W. (1993). Group prototypicality and depersonalized attraction in small interactive groups. Personality and Social Psychology Bulletin, 19, 452–465.

SOCIAL IDENTITY AND LEADERSHIP PROCESSES IN GROUPS

47

Hogg, M. A., Fielding, K. S., Johnson, D., Masser, B., Russell, E., & Svensson, A. (2001). Demographic category membership and leadership in small groups: A social identity analysis. Manuscript submitted for publication, University of Queensland. Hogg, M. A., & Hains, S. C. (1996). Intergroup relations and group solidarity: Effects of group identification and social beliefs on depersonalized attraction. Journal of Personality and Social Psychology, 70, 295–309. Hogg, M. A., Hains, S. C., & Mason, I. (1998). Identification and leadership in small groups: Salience, frame of reference, and leader stereotypicality effects on leader evaluations. Journal of Personality and Social Psychology, 75, 1248–1263. Hogg, M. A., & Hardie, E. A. (1991). Social attraction, personal attraction and selfcategorization: A field study. Personality and Social Psychology Bulletin, 17, 175–180. Hogg, M. A., Hardie, E. A., & Reynolds, K. (1995). Prototypical similarity, self-categorization, and depersonalized attraction: A perspective on group cohesiveness. European Journal of Social Psychology, 25, 159–177. Hogg, M. A., Martin, R. (2003). Social identity analysis of leader-member relations: Reconciling self-categorization and leader-member exchange theories of leadership. In S. A. Haslam, D. van Knippenberg, M. Platow, & N. Ellemers (Eds.), Social identity at work: Developing theory for organizational practice. New York: Psychology Press. Hogg, M. A., Martin, R., Weeden, K., & Epitropaki, O. (2001). Effective leadership in salient groups: Revisiting leader-member exchange theory from the perspective of the social identity theory of leadership. Manuscript submitted for publication, University of Queensland. Hogg, M. A., & Reid, S. A. (2001). Social identity, leadership, and power. In A. Y. Lee-Chai & J. A. Bargh (Eds.), The use and abuse of power: Multiple perspectives on the causes of corruption (pp. 159–180). Philadelphia, PA: Psychology Press. Hogg, M. A., & Terry, D. J. (2000). Social identity and self-categorization processes in organizational contexts. Academy of Management Review, 25, 121–140. Hogg, M. A. & Terry, D. J. (Eds.) (2001). Social identity processes in organizational contexts. Philadelphia, PA: Psychology Press. Hogg, M. A., & Turner, J. C. (1987). Social identity and conformity: A theory of referent informational influence. In W. Doise & S. Moscovici (Eds.), Current issues in European social psychology (Vol. 2, pp. 139–182). Cambridge, UK: Cambridge University Press. Hogg, M. A., Turner, J. C., & Davidson, B. (1990). Polarized norms and social frames of reference: A test of the self-categorization theory of group polarization. Basic and Applied Social Psychology, 11, 77–100. Hogg, M. A., & Williams, K. D. (2000). From I to we: Social identity and the collective self. Group Dynamics: Theory, Research, and Practice, 4, 81–97. Hollander, E. P. (1958). Conformity, status, and idiosyncracy credit. Psychological Review, 65, 117–127. Hollander, E. P. (1995). Organizational leadership and followership. In P. Collett & A. Furnham (Eds.), Social psychology at work: Essays in the honour of Michael Argyle (pp. 69–87). London: Routledge. Hollander, E. P., & Julian, J. W. (1969). Contemporary trends in the analysis of leadership processes. Psychological Bulletin, 71, 387–391. Hollander, E. P., & Offerman, L. R. (1990). Power and leadership in organizations. American Psychologist, 45, 179–189. Hornsey, M. J., & Hogg, M. A. (2000). Assimilation and diversity: An integrative model of subgroup relations. Personality and Social Psychology Review, 4, 143–156. House, R. J. (1971). A path—goal theory of leader effectiveness. Administrative Science Quarterly, 16, 321–339.

48

HOGG AND VAN KNIPPENBERG

House, R. J. (1977). A 1976 theory of charismatic leadership. In J. G. Hunt & L. L. Larson (Eds.), Leadership: The cutting edge (pp. 189–207). Carbondale, IL: Southern Illinois University Press. Kerr, S., & Jermier, J. M. (1978). Substitutes for leadership: Their meaning and measurement. Organizational Behavior and Human Performance, 22, 375–403. Koper, G., van Knippenberg, D., Bouhuijs, F., Vermunt, R., & Wilke, H. (1993). Procedural fairness and self-esteem. European Journal of Social Psychology, 23, 313–325. Levine, J. M., & Moreland, R. L. (1990). Progress in small group research. Annual Review of Psychology, 41, 585–634. Levine, J. M., & Moreland, R. L. (1995). Group processes. In A. Tesser (Ed.), Advanced social psychology (pp. 419–465). New York: McGraw-Hill. Lippitt, R., & White, R. (1943). The ‘social climate’ of children’s groups. In R. G. Barker, J. Kounin, & H. Wright (Eds.), Child behavior and development (pp. 485–508). New York: McGraw-Hill. Lord, R. G., Brown, D. J., & Freiberg, S. J. (1999). Understanding the dynamcs of leadership: The role of follower self-concepts in the leader/follower relationship. Organizational Behavior and Human Decision Processes, 78, 1–37. Lord, R. G., Brown, D. J., & Harvey, J. L. (2001). System constraints on leadership perceptions, behavior and influence: An example of connectionist level processes. In M. A. Hogg & R. S. Tindale (Eds.), Blackwell handbook of social psychology: Group processes (pp. 283–310). Oxford, UK: Blackwell. Lord, R. G., Brown, D. J., Harvey, J. L., & Hall, R. J. (2001). Contextual constraints on prototype generation and their multilevel consequences for leadership perceptions. Leadership Quarterly, 12, 311–338. Lord, R. G., Foti, R. J., & DeVader, C. L. (1984). A test of leadership categorization theory: Internal structure, information processing, and leadership perceptions. Organizational Behavior and Human Performance, 34, 343–378. Lord, R. G., Hall, R. J. (in press). Identity, leadership categorization, and leadership schema. In D. van Knippenberg & M. A. Hogg (Eds.), Leadership and power: Identity processes in groups and organizations. London: Sage. Lord, R. G., & Maher, K. J. (1991). Leadership and information processing: Linking perceptions and performance. Boston: Unwin Hyman. Mackie, D. M. (1986). Social identification effects in group polarization. Journal of Personality and Social Psychology, 50, 720–728. Marques, J. M., Abrams, D., Pa´ ez, D., & Hogg, M. A. (2001). Social categorization, social identification, and rejection of deviant group members. In M. A. Hogg & R. S. Tindale (Eds.), Blackwell handbook of social psychology: Group processes (pp. 400–424). Oxford, UK: Blackwell. Masser, B., & Abrams, D. (2001). When women still lose: The consequence of hostile and benevolent sexism for prejudice and discrimination against individual targets. Manuscript submitted for publication, University of Queensland. McGrath, J. E. (1997). Small group research, that once and future field: An interpretation of the past with an eye to the future. Group Dynamics: Theory, Research, and Practice, 1, 7–27. Medin, D. L., & Ortony, A. (1989). Psychological essentialism. In S. Vosnaidou & A. Ortony (Eds.), Similarity and analogical reasoning (pp. 179–195). Cambridge, UK: Cambridge University Press. Meindl, J. R., Ehrlich, S. B., & Dukerich, J. M. (1985). The romance of leadership. Administrative Science Quarterly, 30, 78–102. Messick, D. M., & Brewer, M. B. (1983). Solving social dilemmas. In L. Wheeler & P. R. Shaver (Eds.), Review of personality and social psychology (Vol. 4, pp. 11–44). Beverly Hills, CA: Sage.

SOCIAL IDENTITY AND LEADERSHIP PROCESSES IN GROUPS

49

Messick, D. M., Kramer, R. (Eds.) (in press). The psychology of leadership: Some new approaches. Mahwah, NJ: Erlbaum. Moreland, R. L., Hogg, M. A., & Hains, S. C. (1994). Back to the future: Social psychological research on groups. Journal of Experimental Social Psychology, 30, 527–555. Moscovici, S. (1976). Social influence and social change. London: Academic Press. Mowday, R. T., & Sutton, R. I. (1993). Organizational behavior: Linking individuals and groups to organizational contexts. Annual Review of Psychology, 44, 195–229. Nye, J. L., & Forsyth, D. R. (1991). The effects of prototype-based biases on leadership appraisals: A test of leadership categorization theory. Small Group Research, 22, 360–379. Nye, J. L., & Simonetta, L. G. (1996). Followers’ perceptions of group leaders: The impact of recognition-based and inference-based processes. In J. L. Nye & A. M. Bower (Eds.), What’s social about social cognition: Research on socially shared cognition in small groups (pp. 124–153). Thousand Oaks, CA: Sage. Palich, L. E., & Hom, P. W. (1992). The impact of leader power and behavior on leadership perceptions: A lisrel test of an expanded categorization theory of leadership model. Group and Organization Management, 17, 279–296. Pawar, B. S., & Eastman, K. (1997). The nature and implications of contextual influences on transformational leadership. Academy of Management Review, 22, 80–109. Platow, M. J., Hoar, S., Reid, S. A., Harley, K., & Morrison, D. (1997). Endorsement of distributively fair and unfair leaders in interpersonal and intergroup situations. European Journal of Social Psychology, 27, 465–494. Platow, M. J., Mills, D., & Morrison, D. (2000). The effects of social context, source fairness, and perceived self-source similarity on social influence: A self-categorisation analysis. European Journal of Social Psychology, 30, 69–81. Platow, M. J., Reid, S. A., & Andrew, S. (1998). Leadership endorsement: The role of distributive and procedural behavior in interpersonal and intergroup contexts. Group Processes and Intergroup Relations, 1, 35–47. Platow, M. J., & van Knippenberg, D. (2001). A social identity analysis of leadership endorsement: The effects of leader ingroup prototypicality and distributive intergroup fairness. Personality and Social Psychology Bulletin, 27, 1508–1519. Platow, M. J., van Knippenberg, D., Haslam, S. A., van Knippenberg, B., & Spears, R. (2002). A special gift we bestow on you for being representative of us: Considering leader charisma from a self-categorization perspective. Paper presented at the 13th General Meeting of the European Association of Experimental Social Psychology. San Sebastia´ n, Spain. June 26–29, 2002. Rabbie, J. M., & Bekkers, F. (1978). Threatened leadership and intergroup competition. European Journal of Social Psychology, 8, 9–20. Reicher, S. D., Drury, J., Hopkins, N., Stott, C. (in press). A model of crowd prototypes and crowd leaders. In C. Barker (Ed.), Leadership and social movements. Manchester, UK: Manchester University Press. Reicher, S. D., & Hopkins, N. (1996). Self-category constructions in political rhetoric; An analysis of Thatcher’s and Kinnock’s speeches concerning the British miners’ strike (1984–5). European Journal of Social Psychology, 26, 353–371. Ridgeway, C. L. (2001). Social status and group structure. In M. A. Hogg & R. S. Tindale (Eds.), Blackwell handbook of social psychology: Group processes (pp. 352–375). Oxford, UK: Blackwell. Ross, L. (1977). The intuitive psychologist and his shortcomings. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 10, pp. 174–220). New York: Academic Press. Rousseau, D. M. (1998). Why workers still identify with organizations. Journal of Organizational Behavior, 19, 217–233.

50

HOGG AND VAN KNIPPENBERG

Rudman, L. A., & Glick, P. (1999). Feminized management and backlash toward agentic women: The hidden costs to women of a kinder, gentler image of middle-managers. Journal of Personality and Social Psychology, 77, 1004–1010. Rush, M. C., & Russell, J. E. A. (1988). Leader prototypes and prototype-contingent consensus in leader behavior descriptions. Journal of Experimental Social Psychology, 24, 88–104. Sanna, L. J., & Parks, C. D. (1997). Group research trends in social and organizational psychology: Whatever happened to intragroup research? Psychological Science, 8, 261–267. Schriesheim, C. A., Castro, S. L., & Cogliser, C. C. (1999). Leader-member exchange (LMX) research: A comprehensive review of theory, measurement, and data-analytic practices. Leadership Quarterly, 10, 63–113. Sedikides, C. & Brewer, M. B., (Eds.) (2001). Individual self, relational self, collective self, pp. 123–143. Philadelphia, PA: Psychology Press. Shamir, B., House, R., & Arthur, M. B. (1993). The motivational effects of charismatic leadership: A self-concept based theory. Organization Science, 4, 577–594. Shaw, M. E. (1981). Group dynamics: The psychology of small group behavior (2nd ed.) New York: McGraw-Hill. Sherif, M. (1966). In common predicament: Social psychology of intergroup conflict and cooperation. Boston: Houghton-Mifflin. Smith, H. J., Tyler, T. R., Huo, Y. J., Ortiz, D., & Lind, E. A. (1998). The self-relevant implications of the group-value model: Group membership, self-worth, and procedural justice. Journal of Experimental and Social Psychology, 34, 470–493. Sparrowe, R. T., & Liden, R. C. (1997). Process and structure in leader-member exchange. Academy of Management Review, 22, 522–552. Steiner, I. D. (1974). Whatever happened to the group in social psychology?. Journal of Experimental Social Psychology, 10, 94–108. Steiner, I. D. (1986). Paradigms and groups. Advances in Experimental Social Psychology, 19, 251–289. Stogdill, R. (1974). Handbook of leadership. New York: Free Press. Stogdill, R. M. & Coons, A. E., (Eds.) (1957). Leader behavior: Its description and measurement. Columbus, OH: Bureau of Business Research. Swim, J. K., Aikin, K. J., Hall, W. S., & Hunter, B. A. (1995). Sexism and racism: Old fashioned and modern prejudices. Journal of Personality and Social Psychology, 68, 199–214. Tajfel, H. (Ed.) (1984). The social dimension: European developments in social psychology. Cambridge, UK: Cambridge University Press. Tajfel, H., & Turner, J. C. (1986). The social identity theory of intergroup behavior. In S. Worchel & W. Austin (Eds.), Psychology of intergroup relations (pp. 7–24). Chicago: Nelson-Hall. Taylor, D. M., & Brown, R. J. (1979). Towards a more social social psychology? British Journal of Social and Clinical Psychology, 18, 173–179. Taylor, S. E., & Fiske, S. T. (1975). Point-of-view and perceptions of causality. Journal of Personality and Social Psychology, 32, 439–445. Taylor, S. E., & Fiske, S. T. (1978). Salience, attention, and attribution: Top of the head phenomena. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 11, pp. 249–288). New York: Academic Press. Terry, D. J., Carey, C. J., & Callan, V. J. (2001). Employee adjustment to an organizational merger: An intergroup perspective. Personality and Social Psychology Bulletin, 27, 267–280. Tindale, R. S., & Anderson, E. M. (1998). Small group research and applied social psychology: An introduction. In R. S. Tindale, L. Heath, J. Edwards, E. J. Posavac, F. B. Bryant, Y. Suarez-Balcazar, E. Henderson-King, & J. Myer (Eds.), Social psychological applications to social issues: Theory and research on small groups (Vol. 4, pp. 1–8). New York: Plenum Press.

SOCIAL IDENTITY AND LEADERSHIP PROCESSES IN GROUPS

51

Trope, Y., & Liberman, A. (1993). The use of trait conceptions to identify other people’s behavior and to draw inferences about their personalities. Personality and Social Psychology Bulletin, 19, 553–562. Turner, J. C. (1991). Social influence. Buckingham, UK: Open University Press. Turner, J. C. (1999). Some current issues in research on social identity and self-categorization theories. In N. Ellemers, R. Spears, & B. Doosje (Eds.), Social identity (pp. 6–34). Oxford, UK: Blackwell. Turner, J. C., Hogg, M. A., Oakes, P. J., Reicher, S. D., & Wetherell, M. S. (1987). Rediscovering the social group: A self-categorization theory. Oxford, UK: Blackwell. Turner, J. C., & Oakes, P. J. (1989). Self-categorization and social influence. In P. B. Paulus (Ed.). The psychology of group influence (2nd ed., pp. 233–275). Hillsdale, NJ: Erlbaum. Tyler, T. R. (1997). The psychology of legitimacy: A relational perspective on voluntary deference to authorities. Personality and Social Psychology Review, 1, 323–345. Tyler, T. R. (1999). Why people cooperate with organizations: An identity-based perspective. Research in Organizational Behavior, 21, 201–246. Tyler, T. R., & Degoey, P. (1995). Collective restraint in social dilemmas: Procedural justice and social identification effects on support for authorities. Journal of Personality and Social Psychology, 69, 482–497. Tyler, T. R., & Lind, E. A. (1992). A relational model of authority in groups. In M. P. Zanna (Ed.), Advances in experimental social psychology (Vol. 25, pp. 115–191). New York: Academic Press. van Knippenberg, B., & van Knippenberg, D. (2000). Leaders’ use of influence tactics, prototypicality, and their effects on subordinates’ perceptions. Paper presented at the EAESP Small Group Meeting on Social Identity Processes in Organizations, Amsterdam, July 2000. van Knippenberg, B., & van Knippenberg, D. (2001). Leader prototypicality, leader influence tactics, and leadership effectiveness. Unpublished manuscript, Free University, Amsterdam. van Knippenberg, B., & van Knippenberg, D. (in press). Leadership, identity, and influence: Relational concerns in the use of influence tactics. In D. van Knippenberg & M. A. Hogg (Eds.), Leadership and power: Identity processes in groups and organizations. London: Sage. van Knippenberg, B., van Knippenberg, D., Blaauw, E., & Vermunt, R. (1999). Relational considerations in the use of influence tactics. Journal of Applied Social Psychology, 29, 806–819. van Knippenberg, D. (2000a). Group norms, prototypicality, and persuasion. In D. J. Terry & M. A. Hogg (Eds.), Attitudes, behavior, and social context: The role of norms and group membership (pp. 157–170). Mahwah, NJ: Erlbaum. van Knippenberg, D. (2000b). Work motivation and performance: A social identity perspective. Applied Psychology: An International Review, 49, 357–371. van Knippenberg, D. & Hogg, M. A. (Eds.) (2001). Social identity processes in organizations. (Special issue of Group processes and Intergroup Relations), London: Sage. van Knippenberg, D., & Hogg, M. A. (Eds.) (in press). Leadership and power: Identity processes in groups and organizations. London: Sage. van Knippenberg, D., Lossie, N., & Wilke, H. (1994). In-group prototypicality and persuasion: Determinants of heuristic and systematic message processing. British Journal of Social Psychology, 33, 289–300. van Knippenberg, D., van Knippenberg, B., Monden, L., & De Lima, F. (2002). Organizational identification after a merger: A social identity perspective. British Journal of Social Psychology, 41, 233–252. van Knippenberg, D., van Knippenberg, B., & van Dijk, E. (2000). Who takes the lead in risky decision making? Effects of group members’ individual riskiness and prototypicality. Organizational Behavior and Human Decision Processes, 83, 213–234.

52

HOGG AND VAN KNIPPENBERG

van Knippenberg, D., & van Leeuwen, E. (2001). Organizational identity after a merger: Sense of continuity as the key to post-merger identification. In M. A. Hogg & D. J. Terry (Eds.), Social identity processes in organizational contexts (pp. 249–264). Philadelphia, PA: Psychology Press. van Knippenberg, D., & Wilke, H. (1992). Prototypicality of arguments and conformity to ingroup norms. European Journal of Social Psychology, 22, 141–155. Van Vugt, M., & De Cremer, D. (1999). Leadership in social dilemmas: The effects of group identification on collective actions to provide public goods. Journal of Personality and Social Psychology, 76, 587–599. Vroom, V. H., & Yeton, P. W. (1973). Leadership and decision making. Pittsburgh, PA: University of Pittsburgh Press. Yorges, S. L., Weiss, H. M., & Strickland, O. J. (1999). The effect of leader outcomes on influence, attributions, and perceptions of charisma. Journal of Applied Psychology, 84, 428–436. Yukl, G. (2001). Leadership in organizations (5th ed.). New York: Prentice Hall. Yzerbyt, V., Castano, E., Leyens, J.-P., & Paladrino, M.-P. (2000). The primacy of the ingroup: The interplay of entitativity and identification. European Review of Social Psychology, 11, 257–295. Yzerbyt, V., Corneille, O., & Estrada, C. (2001). The interplay of subjective essentialism and entitativity in the formation of stereotypes. Personality and Social Psychology Review, 5, 141–155.

THE ATTACHMENT BEHAVIORAL SYSTEM IN ADULTHOOD: ACTIVATION, PSYCHODYNAMICS, AND INTERPERSONAL PROCESSES

Mario Mikulincer Phillip R. Shaver

My life’s work has been directed to a single aim. I have observed the more subtle disturbances of mental function in healthy and sick people and have sought to infer—or if, you prefer it, to guess—from signs of this kind how the apparatus which serves these functions is constructed and what concurrent and mutually opposing forces are at work in it. —Sigmund Freud (1961/1930, p. 208) As my study of theory progressed it was gradually borne in upon me that the field I had set out to plough so lightheartedly was no less than the one that Freud had started tilling sixty years earlier, and that it contained all those same rocky excrescences and thorny entanglements that he had encountered and grappled with—love and hate, anxiety and defense, attachment and loss. —John Bowlby (1982, p. xxvii)

Psychology has changed its focus several times during its relatively short history—psychoanalysis, behaviorism, cognitivism, and neuroscience being prominent examples. Within such conceptual traditions, an emphasis on the individual and the individual’s selfish motives often gives way, over time, to an emphasis on social interrelatedness. Freud’s psychoanalytic theory of drives or instincts, for example, evolved into later theories of ‘‘object relations’’ (a misleading name for social relations) and ‘‘intersubjectivity.’’ Behaviorism focused first on the individual’s drives and the role of drivereduction in learning, but then gradually made room for social learning. The ‘‘cognitive revolution,’’ aimed originally at such topics as perception 53 ADVANCES IN EXPERIMENTAL SOCIAL PSYCHOLOGY. VOL. 35

Copyright 2003 Elsevier Science (USA). All rights reserved. 0065-2601/03 $35.00

54

MIKULINCER AND SHAVER

of letters and memory for words, generated the field of social cognition that deals with person perception and social memory. More recently, neuroscience has bred cognitive neuroscience, which in turn is generating social cognitive neuroscience. These days, no one doubts that the human mind/brain is a social machine. By evolutionary standards, the large human cerebral cortex (including the very large frontal lobes and special modules for language) developed surprisingly rapidly, primarily to serve increasingly complex interpersonal functions (cooperative hunting and child-rearing, for example; identification of reliable mates and allies as well as detection of cheaters and scoundrels). Assuming, as we do, that major contributors to the different traditions in psychology all had valuable insights and solid reasons for pursuing their particular angles on human experience and behavior, we wish to combine them to form a powerful, integrative theory of the social mind. In so doing, we take as primary guides the British psychoanalyst John Bowlby and his North American research collaborator Mary Ainsworth. These investigators combined key insights of psychoanalysis, ethology, developmental psychology, and cognitive psychology to create a theory of ‘‘attachment,’’ or emotional bonding, that has been tested and elaborated in literally thousands of studies over the past 30 years. Both Bowlby and Ainsworth were unusually talented analytic thinkers and clear writers who respected the scientific method. Bowlby was notable for the breadth and openness of his mind, Ainsworth for her keen observational skills and methodological creativity. Together they laid a foundation for studying close interpersonal relationships across the lifespan. Our work can be better understood if we explain briefly how we came, individually, to attachment theory and then began to influence each other and, eventually, work together. Both of us were attracted to psychoanalytic theory as undergraduates, despite the hard knocks it has always taken from diverse critics. Anyone who opens his mind to what goes on in real people’s lives, and anyone who reads novels or poems or watches artful films, realizes that the issues raised by psychoanalysts, beginning with Freud, are important: sexual hunger and attraction; romantic love; the development of personality, beginning in the context of infant–caregiver relationships; negative emotions such as anxiety, sadness, jealousy, guilt, shame, and anger and their role in intrapsychic conflict and psychopathology; intergroup hostility and war. When we first began studying experimental social psychology, it seemed rather tame and superficial compared to psychoanalysis. But its strong point, and the weak point of psychoanalysis, was a collection of powerful and creative research methods. Psychoanalytic theorists seemed capable of endlessly proliferating and debating hypothetical constructs and

THE ATTACHMENT BEHAVIORAL SYSTEM

55

processes without being constrained by operational definitions, sound psychometrics, or replicable research evidence. Both of us began our careers as experimental researchers pursuing existing topic areas (stress and learned helplessness in Mikulincer’s case, self-awareness theory and fear of success in the case of Shaver), but our interest in psychoanalytic ideas never abated. When Bowlby’s books began to appear, we realized it was possible for a psychoanalytic thinker to amass empirical evidence for important psychoanalytic propositions, and that psychoanalytic theory could be altered and elaborated based on empirical research. Ainsworth’s development of a laboratory ‘‘strange situation’’ assessment procedure, which allowed her to classify infants’ attachment patterns and relate them systematically to home observations of infant–parent interactions, added to our confidence that research on an appropriately social and evolutionary version of psychoanalytic theory was possible. In the mid-1980s, Shaver was studying adolescent and adult loneliness (e.g., Rubenstein & Shaver, 1982; Shaver & Hazan, 1984) and noticing both that attachment theory could be useful in understanding loneliness (e.g., Weiss, 1973) and that patterns of chronic loneliness were similar in certain respects to the insecure infant attachment patterns identified by Ainsworth and her colleagues (Ainsworth, Blehar, Waters, & Wall, 1978). Building on this insight, one of Shaver’s doctoral students, Cindy Hazan, wrote a seminar paper suggesting that attachment theory could be used as a broad framework for studying romantic love or ‘‘romantic attachment,’’ as they called it in their initial article on the topic (Hazan & Shaver, 1987). That paper caught the eye of Mikulincer, who had become interested in attachment theory while studying affect-regulation processes related to learned helplessness, depression, combat stress reactions, and posttraumatic stress reactions in Israel. He noticed similarities between (1) certain forms of helplessness in adulthood and attachment-figure (e.g., parental) unavailability in infancy; (2) intrusive images and anxiety in the case of posttraumatic stress disorder and the anxious attachment pattern described by Ainsworth et al. (1978) and Hazan and Shaver (1987); and (3) avoidant strategies of coping with stress and the avoidant attachment pattern described by these same authors. In 1990, Mikulincer, Florian, and Tolmacz published a study of attachment patterns and conscious and unconscious death anxiety, one of the first studies to use the preliminary self-report measure of attachment style devised by Hazan and Shaver (1987). From that point on, both of us continued to pursue the application of attachment theory to the study of adults’ emotions, emotion-regulation strategies, and close relationships, noticing that we were both interested in the experimental study of what might be called attachment-related

56

MIKULINCER AND SHAVER

psychodynamics: the kinds of mental processes, including intense needs, powerful emotions, and defensive strategies, that had captivated both Freud and Bowlby. In recent years we have pooled our efforts to work out a more carefully articulated statement of our theoretical ideas. In the present chapter we summarize what we have learned to date, placing primary emphasis on our own ideas and research (in line with the policy of the Advances series) rather than attempting to provide a comprehensive review of the entire adult attachment literature. The chapter is organized as follows: We begin with an overview of basic theoretical concepts, which will be essential for understanding the remainder of the chapter. We then present an integrative model of attachment-system dynamics in adulthood, a model that has both guided our research and been affected by it. These two theoretical sections of the chapter constitute roughly a third of its length, which is perhaps unusual in a chapter designed to summarize research. In the present case, the research makes better and deeper sense if the reader understands the theory. The subsequent sections of the chapter summarize the diverse empirical studies motivated by the theory and review a variety of unresolved conceptual and empirical issues that provide an agenda for future research.

I. Basic Theoretical Concepts We begin by introducing basic concepts in Bowlby’s (1973, 1980, 1982/ 1969) attachment theory, briefly covering both the normative (general) aspects of what he called the attachment behavioral system and individual differences in what social psychologists call ‘‘attachment style’’ (Levy & Davis, 1988). These styles, first noticed and delineated by (Ainsworth, 1967; Ainsworth et al., 1978), are thought to emerge over the course of development from countless interactions between the attachment system and its interpersonal environment.

A. NORMATIVE COMPONENTS OF THE ATTACHMENT BEHAVIORAL SYSTEM 1. What Is a Behavioral System? Bowlby (1982/1969) defined a behavioral system as a species-universal program that organizes an individual’s behavior in functional ways—i.e., in ways that increase the likelihood of survival in the face of particular

THE ATTACHMENT BEHAVIORAL SYSTEM

57

environmental demands. A behavioral system is an inborn, preset program of the central nervous system that was ‘‘designed’’ by evolution via natural selection. It governs the choice, activation, and termination of behavioral sequences that produce a predictable and generally functional change in the person–environment relationship. (The term ‘‘behavioral system’’ therefore implies the involvement of actual behavior and behavior generation, although it points theoretically to the central neural program we are emphasizing.) Each behavioral system (e.g., attachment, caregiving, exploration, affiliation) follows a particular predictable pattern of activation and termination in almost all members of a species, a pattern that does not depend on learning opportunities. Conceptually, a behavioral system has six components or aspects: (1) a specific biological function that increases the likelihood of an individual’s survival and reproductive success; (2) a set of contextual activating triggers; (3) a set of interchangeable, functionally equivalent behaviors that constitutes the primary strategy of the system for attaining a particular goal state; (4) a specific set-goal—the change in the person–environment relationship that terminates the activation of the system; (5) the cognitive operations involved in the functioning of the system; and (6) specific excitatory or inhibitory neural links with other behavioral systems. 2. Biological Function of the Attachment System According to Bowlby (1982/1969), the biological function of the attachment behavioral system is to protect a person (especially during infancy) from danger by ensuring that he or she maintains proximity to caring and supportive others (attachment figures). In his view, natural selection favored maintenance of proximity to attachment figures [what Bowlby (1973) called ‘‘stronger and wiser’’ caregivers], because it increases the likelihood of survival and eventual reproduction on the part of members of a species born with immature capacities for defense from predators and other dangers. Because infants require a long period of protection, they are born with a tendency to seek proximity to others who can provide care and support. Although the biological function of the attachment behavioral system is most critical during the early phases of life, Bowlby (1988) assumed the system is active over the entire life span and is manifested in thoughts and behaviors related to seeking proximity to attachment figures in times of need. In his view, proximity seeking is a behavioral adaptation to evolutionary pressures that can contribute to adjustment and health throughout the lifespan. This function of the attachment system is most relevant during stressful periods or traumatic experiences, wherein the

58

MIKULINCER AND SHAVER

support and comfort given by attachment figures enhance coping and adjustment. During infancy, primary caregivers (usually one or both parents) are the main attachment figures. In adulthood, however, a wide variety of relationship partners can act as attachment figures, including parents, friends, and romantic partners. Moreover, groups, institutions, and abstract or symbolic figures (e.g., God) can become targets of proximity seeking in times of need. There are also context-tailored attachment figures, who are sources of support only in specific milieus: teachers and supervisors in academic settings; therapists in therapeutic settings; and managers in organizational settings. Attachment theory conceptualizes attachment figures as forming a hierarchical network, with the primary attachment figures being those with whom the individual maintains long-term and strong affectional bonds (parents, close friends, spouses). 3. The Attachment System’s Activating Triggers Originally, Bowlby (1982/1969) claimed that the attachment system is activated by environmental threats that endanger a person’s survival. Encounters with such threats create a need for protection from other people and automatically activate the attachment system. When no threat is present, there is no need to seek care from others and no proximity-seeking tendency is activated, at least not for the purposes of protection. (A person may seek proximity to others for the purpose of some other behavioral system such as affiliation or sexual mating.) In subsequent writings, Bowlby (1973) extended this reasoning and proposed that the attachment system is also activated by ‘‘natural clues of danger’’—stimuli that are not inherently dangerous but that increase the likelihood of danger (e.g., darkness, loud noises, isolation)—as well as by attachment-related threats such as impending or actual separation from, or loss of, attachment figures. In his view, a combination of attachment-unrelated sources of threat and lack of access to an attachment figure compounds distress and triggers the highest level of attachment-system activation. 4. The Primary Attachment Strategy According to Bowlby (1982/1969), proximity seeking is the natural and primary strategy of the attachment behavioral system when a person has a perceived need for protection or support. This strategy consists of a wide variety of behaviors that have a similar meaning (the seeking of proximity) and serve similar adaptive functions (protection from threats). Among these behaviors, one can find signals (interaction bids) that tell a relationship

THE ATTACHMENT BEHAVIORAL SYSTEM

59

partner an individual is interested in restoring or maintaining proximity; overt displays of negative emotion (e.g., anger, anxiety, sadness) that drive the relationship partner to provide support and comfort; active approach behaviors that result in greater physical or psychological contact, including what Harlow (1959) called ‘‘contact comfort’’; and explicit requests for emotional or instrumental support. According to Bowlby (1982/1969), not all of these behaviors are likely to be manifested in every threatening situation. Rather, they are part of a repertoire of behaviors from which an individual can ‘‘choose’’ (consciously or unconsciously) the most adequate means for attaining protection in a given situation. In adulthood, the primary attachment strategy does not necessarily lead to actual proximity-seeking behavior. In fact, this strategy can be manifested in the activation of mental representations of relationship partners who regularly provide care and protection. These cognitions can create a sense of safety and security, which helps a person deal successfully with threats. That is, mental representations of attachment figures can become symbolic sources of protection, and their activation can establish what might be called symbolic proximity to supportive others. Of course there are times when these strategies are insufficient and even adults seek actual proximity to attachment figures. 5. Set-Goal of the Attachment System Bowlby (1982/1969) viewed the attainment of actual or perceived protection and security as the set-goal of the attachment system, which normally terminates the system’s activation [see also Sroufe & Waters (1977), who introduced the term ‘‘felt security’’]. Bowlby also enumerated the provisions attachment figures should supply in order to facilitate the attainment of this set-goal (see also Hazan & Shaver, 1994). First, attachment figures should be responsive to the individual’s proximityseeking attempts in times of need. Second, these figures should provide a physical and emotional safe haven—i.e., they should facilitate distress alleviation and be a source of support and comfort. Third, attachment figures should provide a secure base from which the individual can explore and learn about the world and develop his or her own capacities and personality while feeling confident that care and support will be available if needed. When these provisions are supplied, a person feels secure and safe, and proximity seeking for the purpose of protection and care is terminated. 6. Cognitive Substrate of the Attachment System According to Bowlby (1982/1969), the attachment system operates in a complex goal-corrected manner; that is, the individual evaluates the

60

MIKULINCER AND SHAVER

progress his or her behaviors are making toward achieving the set-goal and then corrects these behaviors to produce the most effective action sequence. In our view, this flexible, goal-directed and goal-corrected adjustment of attachment behavior requires at least three cognitive operations: (1) processing of information about the person–environment relationship, which involves the monitoring and appraisal of threatening events and one’s own inner state (e.g., distress, security); (2) monitoring and appraisal of the attachment figure’s responses to one’s proximity-seeking attempts; and (3) monitoring and appraisal of the viability of the chosen behaviors in a given context, so that an effective adjustment of these behaviors can be made, if necessary, in accordance with contextual constraints. Bowlby (1973, 1982/1969) also stressed that the goal-corrected nature of attachment behavior requires the storage of relevant data in the form of mental representations of person–environment transactions. He called these representations working models and seemed to intend the word ‘‘working’’ to carry two senses: (1) the models allow for mental simulation and prediction of likely outcomes of various attachment behaviors (that is, they can provide dynamic, adjustable, context-sensitive representations of complex social situations); and (2) the models are provisional (in the sense of ‘‘working’’ drafts—changeable plans). Bowlby (1982/1969) distinguished between two kinds or components of working models: ‘‘If an individual is to draw up a plan to achieve a set-goal not only does he have some sort of working model of his environment, but he must have also some working knowledge of his own behavioral skills and potentialities’’ (p. 112). That is, the attachment system, once it has been used repeatedly in relational contexts, includes representations of attachment figures’ responses (working models of others) as well as representations of the self ’s efficacy and value (working models of self ). These working models organize a person’s memory about an attachment figure and him- or herself during attempts to gain protection in times of need (Main, Kaplan, & Cassidy, 1985). 7. Interplay between the Attachment Behavioral System and Other Behavioral Systems Encountering threats and experiencing disruptions in felt security activate the attachment behavioral system, which in turn inhibits the activation of other behavioral systems and prevents or hinders engagement in nonattachment activities (Bowlby, 1982/1969). Under conditions of threat, people turn to others as providers of support and comfort rather than as partners for exploratory, affiliative, or sexual activities. Moreover, at such times they are likely to be so self-focused (so focused on their need for protection) that they

THE ATTACHMENT BEHAVIORAL SYSTEM

61

lack the mental resources necessary to attend empathically and altruistically to others’ needs and engage in caring behavior. Only when relief is attained and a sense of attachment security is restored can the individual deploy attention and energy to other behavioral systems and engage in nonattachment activities. Because of this reciprocal relation between the attachment system and other behavioral systems, the attainment of attachment security fosters engagement in nonattachment activities such as exploration, sex, caregiving, and affiliation, and allows the individual to distance from an attachment figure with the belief that this figure will be available if needed. 8. Summary According to Bowlby (1973, 1980, 1982/1969), the attachment system is an inborn regulatory device that has important implications for personality and interpersonal behavior. The system is activated by perceived threats and dangers, which cause the threatened individual to seek proximity to protective others. The attainment of proximity and protection results in feelings of relief and security as well as positive mental representations of relationship partners and the self. Bowlby (1988) viewed the optimal functioning of this behavioral system as extremely important for the maintenance of emotional stability and mental health, development of a positive self-image, and formation of positive attitudes toward relationship partners and close relationships in general. Moreover, because optimal functioning of the attachment system facilitates relaxed and confident engagement in nonattachment activities, it contributes to the broadening of a person’s perspectives and skills as well as the actualization of his or her unique potentialities. To borrow a term from another theoretical tradition, humanistic psychology (e.g., Maslow, Rogers), attachment security is not only an important prerequisite for healthy love but also a major stepping stone to self-actualization.

B. INDIVIDUAL DIFFERENCES IN THE OPERATION OF THE ATTACHMENT SYSTEM 1. The Role of Actual Interactions with Attachment Figures According to Bowlby (1973), optimal functioning of the attachment system depends on the availability of one or more attachment figures in times of need and on their sensitivity and responsiveness to the individual’s proximity-seeking behaviors. As Cassidy (1999) noted, ‘‘whereas nearly all

62

MIKULINCER AND SHAVER

children become attached (even to mothers who abuse them, Bowlby, 1956), not all are securely attached’’ (p. 7). The quality of interactions with attachment figures in times of need is, according to attachment theory, the major source of individual differences in attachment-system functioning. On the one hand, when an attachment figure is available, sensitive, and responsive to an individual’s proximity-seeking attempts, the attachment figure facilitates the restoration of emotional equanimity and the attainment of a sense of attachment security. As a result, the individual leaves these satisfying interactions with increased confidence in proximity seeking as an effective regulatory strategy, heightened trust in others’ availability and responsiveness, and increased confidence in his or her own resources for dealing with stress. Moreover, these interactions reinforce the individual’s sense that the world is a safe place populated by people of good will, which allows him or her to form rewarding relationships with others and engage freely in nonattachment activities. On the other hand, when the attachment figure is not physically or emotionally available in times of need or is not sensitive or responsive to the individual’s proximity-seeking attempts, the functioning of the attachment system is disrupted and its set-goal is not attained. In these cases, the individual does not experience distress alleviation. Rather, the distress originally elicited by the encounter with threats is now compounded by serious doubts that safety can be attained, that the world is a safe place, that others can be trusted, and that the self has the resources necessary to manage stress. Needless to say, the resulting sense of vulnerability and uncertainty can interfere drastically with a whole range of life activities. 2. Secondary Attachment Strategies Negative interactions with an unavailable and unresponsive attachment figure have an additional immediate effect on attachment-system functioning. These interactions signal that the primary attachment strategy is failing to accomplish its regulatory goal and that pursuing this strategy is exacerbating rather than diminishing distress and insecurity. As a result, the individual is forced to continue to deal with unregulated distress while searching for alternative strategies of affect regulation to replace the primary attachment strategy. Main (1990) labeled these alternatives secondary attachment strategies. They are, one might say, the best a person can do under dreadful circumstances. Attachment theorists (e.g., Cassidy & Kobak, 1988; Main, 1990) have delineated two major secondary attachment strategies that involve either hyperactivation or deactivation of the attachment system. Hyperactivating

THE ATTACHMENT BEHAVIORAL SYSTEM

63

strategies are a kind of ‘‘fight’’ (or as Bowlby, 1982/1969, said, ‘‘protest’’) response to the frustration of attachment needs. A hyperactivating individual does not give up his or her proximity-seeking attempts, and in fact intensifies them in order to coerce an attachment figure’s love and support. This requires that the vigilance, preoccupation, and energization of behavior typical of short-term bids for proximity on the part of a less anxious individual be maintained more or less constantly, even in the face of repeated evidence of attachment-figure unavailability. In contrast, deactivating strategies are a kind of ‘‘flight’’ reaction to the unavailability of an attachment figure. The deactivating individual gives up his or her proximity-seeking efforts, deactivates the attachment system without restoring a sense of attachment security, and attempts to deal with dangers alone. Bowlby (1982/1969) called this strategy ‘‘compulsive self-reliance.’’ Individuals who rely on deactivating strategies of affect-regulation try to keep their attachment system down-regulated to avoid the pain and distress caused by attachment-figure unavailability. The two strategies have a variety of psychological and interpersonal costs, which we will address later in this chapter. 3. From Strategies to Mental Representations: Generalized Individual Differences Beyond producing immediate individual differences in reliance on a specific attachment strategy during a given interaction, variations in the quality of this interaction can also produce more enduring and generalized individual differences in the functioning of the attachment system. According to Bowlby (1973), this passage from context-tailored variations to person-tailored variations is mediated by the cognitive substrate of the attachment system—working models. Every interaction with an attachment figure can be incorporated symbolically into working models of the self and partner, models that allow a person to predict future interactions with the partner and design new proximity-seeking attempts without rethinking each one from the beginning. In this way, the internalization of a specific interaction can affect the functioning of the attachment system in future interactions with a particular attachment figure, and a strategy choice that was originally manifested in a specific interaction can be generalized to other relational episodes. According to Main et al. (1985), early working models organize a child’s memory about the self and the partner during attempts to gain security, as well as the typical outcomes of those attempts: successful attainment of security or failure of the primary attachment strategy. In this way, a child can develop working models for successful proximity-seeking episodes, for

64

MIKULINCER AND SHAVER

episodes in which the attachment system had to be hyperactivated, and for episodes in which the system needed to be defensively deactivated. Each of these models consists of episodic memories of the interaction sequence, declarative knowledge about the partner’s responses and the efficacy of the individual’s responses, and procedural knowledge about the ways in which one responds to such situations and deals with different sources of distress. In our view, the central organizing factor in a working model is the attachment strategy used by the individual in a particular relational episode. These strategies not only shape the procedural knowledge included in the representations but also bias declarative knowledge about the self and the attachment figure according to the goal these strategies are intended to attain (e.g., getting hold of an insufficiently available attachment figure by hyperactivating the attachment system or avoiding punishment or perpetual frustration by deactivating the system). Because of such biases, working models of the self and others reflect only in part the ways the person and the partner actually behaved in a given interaction. They also reflect the underlying regulatory action of attachment strategies that shape cognition, affect, and behavior. That is, working models of self and others are always blended reflections of what actually happened in a social encounter and subjective biases resulting from attachment strategies. Like other mental representations, which are presumably the psychological manifestations of underlying neural networks, working models form excitatory and inhibitory associations with one other, and the activation of one model primes congruent models and inhibits incongruent models. That is, experiencing or thinking about an episode of security attainment activates memories of other congruent episodes of successful proximityseeking attempts and renders memories of hyperactivation and deactivation less accessible. With the passage of time and the recurrent retrieval of related memories, these associative links are strengthened and favor the formation of more abstract and generalized representations of attachment-system functioning with a specific partner. In this way, models of securityattainment, hyperactivation, and deactivation with a specific attachment figure (relationship-specific models) are created, and they form excitatory and inhibitory links with models representing interactions with other attachment figures. With the consolidation of these links, even more generic working models are formed, global representations of the self and others across different relationships. The end product of this cognitive generalization and consolidation process is the formation of a hierarchical associative network, in which episodic memories become exemplars of relationshipspecific models, which in turn become exemplars of generic relational schemas. As a result, with respect to a particular relationship and across different relationships, everyone has models of security-attainment,

THE ATTACHMENT BEHAVIORAL SYSTEM

65

hyperactivation, and deactivation and so can sometimes think about relationships in secure terms and at other times think about them in more hyperactivating or deactivating terms. The idea of multiple working models raises questions about the personal and contextual factors that determine which kind of model is activated and used to guide expectations, concerns, and behaviors in a particular interaction with an attachment figure. In our view, this activation depends mainly on the relative strength of a model, which is determined by the amount of experience on which it is based, the number of times it has been applied in the past, and the density of its connections with other cognitive representations (see also Collins & Read, 1994; Shaver, Collins, & Clark, 1996). At a relationship-specific level, this idea implies that the model representing the typical interaction with an attachment figure has the highest chance of being activated in subsequent interactions with the same person. At a generic level, the model that represents interactions with major attachment figures (e.g., parents, spouse) becomes the most easily activated and chronically accessible attachment-related representation. In addition to the history of attachment interactions, features of the current situation also contribute to the activation of a particular working model. For example, clear-cut contextual cues concerning a partner’s love, availability, and supportiveness can activate models of security attainment. In addition, working models can be invoked by a person’s current motives (e.g., wishing to distance from a partner) or by his or her current mood (Shaver et al., 1996). In our view, the chronically accessible model coexists with less typical working models in the memory network, and these models can be activated by contextual factors or a person’s inner state. The consolidation of a chronically accessible working model is the most important psychological process that explains the passage from contexttailored variations in attachment-system functioning to person-tailored variations. Given a fairly consistent pattern of interactions with attachment figures during childhood and adolescence, the most representative models of these interactions are solidified through thousands of repeated experiences and increasingly become part of the individual’s developing personality. Like other mental schemas, these chronically accessible models tend to operate automatically and unconsciously and are resistant to change. Thus, what began as representations of specific interactions with a specific partner become core personality characteristics and tend to be applied in new situations and relationships where they continue to guide the functioning of the attachment system. Bowlby (1979) aptly illustrated how these chronically accessible models shape a person’s experience. In his words, a person ‘‘tends to assimilate any new person with whom he may form a bond, such as a spouse, or child, or

66

MIKULINCER AND SHAVER

employer, or therapist, to an existing model (either of one or other parent or of self ), and often to continue to do so despite repeated evidence that the model is inappropriate. Similarly he expects to be perceived and treated by them in ways that would be appropriate to his self-model, and to continue with such expectations despite contrary evidence’’ (pp. 141–142). 4. Measurement of Individual Differences: Attachment Style Most of the research examining individual differences in attachmentsystem functioning has focused on attachment styles—patterns of expectations, needs, emotions, and social behavior that result from a particular history of attachment experiences, usually beginning in relationships with parents (Fraley & Shaver, 2000). Attachment style reflects the individual’s most chronically accessible working models and the typical functioning of his or her attachment system in a specific relationship (relationship-specific style) or across relationships (global or general attachment style). As such, each attachment style is closely tied to working models and reflects the underlying, organizing action of a particular attachment strategy (primary or secondary, hyperactivating or deactivating). As mentioned earlier, these styles were first described by Ainsworth (1967; Ainsworth et al., 1978) based on her observations of infants’ responses to separations from and reunions with mother, both at home and in a laboratory strange situation. Using this assessment procedure, infants are classified into one of three attachment categories: secure, avoidant, or anxious/ambivalent. Recently, a fourth category, ‘‘disorganized/disoriented,’’ has been added. It is characterized by odd, awkward behavior during separation and reunion episodes and random fluctuations between signs of anxiety and avoidance (Main et al., 1985). Infants classified as secure seem to hold accessible working models of successful proximity-seeking attempts and security-attainment. In the strange situation, they tend to exhibit distress during separation but then recover quickly and continue to explore the environment with interest. When reunited with mother, they greet her with joy and affection, respond positively to being held, and initiate contact with her. During home observations, mothers of these infants are emotionally available in times of need and responsive to their infants’ proximity-seeking behavior (Ainsworth et al., 1978). It seems reasonable to characterize these mothers as a source of attachment security and as reinforcing reliance on the primary attachment strategy. Avoidant infants seem to hold accessible working models of attachmentsystem deactivation. In the strange situation they show little distress when separated from their mother and avoid her upon reunion. In home-based

THE ATTACHMENT BEHAVIORAL SYSTEM

67

observations, their mothers tend to be emotionally rigid as well as angry at and rejecting of their infants’ proximity-seeking attempts (Ainsworth et al., 1978). Anxious infants seem to hold accessible working models of attachment-system hyperactivation. In the strange situation, they are extremely distressed during separation and exhibit conflictual responses toward their mother at reunion (e.g., they may cling one moment and angrily resist comforting the next, which was Ainsworth’s reason for calling them ‘‘ambivalent’’). During home-based observations, interactions between anxious infants and their mothers were characterized by lack of harmony and lack of caregivers’ consistent responsiveness (Ainsworth et al., 1978). Mothers of both avoidant and anxious infants seem to thwart security attainment, thereby fostering the adoption of secondary strategies. However, whereas avoidant infants deactivate the attachment system in response to attachment-figure unavailability, anxious infants tend to hyperactivate the system (Main, 1990; Main et al., 1985). Disorganized/disoriented infants seem to suffer from a breakdown of organized attachment strategies (primary, hyperactivating, or deactivating). They either oscillate between strategies or do something bizarre like lie face down on the floor without moving or sit under a table, evincing no strategy at all. This pattern of behavior seems to be due to disorganized, unpredictable, and discomfiting behavior on the part of attachment figures who, research shows, are likely to be suffering from unresolved losses or attachment-related traumas (Hesse, 1999; Lyons-Ruth & Jacobvitz, 1999). With the extension of attachment research into adolescence and adulthood, new measures of attachment style were created for use with those age groups. Based on a narrative approach, Main and her colleagues (George, Kaplan, & Main, 1985; Main et al., 1985; see Hesse, 1999, for a review) devised the Adult Attachment Interview (AAI), in which participants answer open-ended questions about their childhood relationships with parents. In the AAI, interviewees are classified into three primary categories that parallel Ainsworth’s infant–mother attachment typology: ‘‘secure’’ (or free and autonomous with respect to attachment), ‘‘dismissing’’ (of attachment), or ‘‘preoccupied’’ (with attachment). If a narrative contains indications of unresolved traumas or losses, it gets a secondary classification of ‘‘unresolved.’’ AAI narratives are coded as indicative of secure working models if an interviewee describes positive relationships with parents in a clear, convincing, and coherent manner or if negative relationships are described coherently with some degree of perspective. The narratives are coded as indicative of dismissing attachment if the interviewee dismisses the importance of attachment relationships or idealizes them and provides no clear examples to support his or her characterizations. This category parallels

68

MIKULINCER AND SHAVER Table I Original Three-Category Measure of Romantic Attachment Style a

I am somewhat uncomfortable being close to others; I find it difficult to trust them completely, difficult to allow myself to depend on them. I am nervous when anyone gets too close and often, others want me to be more intimate than I feel comfortable being. (Avoidant) I find that others are reluctant to get as close as I would like. I often worry that my partner doesn’t really love me or won’t want to stay with me. I want to get very close to my partner and this sometimes scares people away. (Anxious) I find it relatively easy to get close to others and am comfortable depending on them and having them depend on me. I don’t worry about being abandoned or about someone getting too close to me. (Secure) a

Hazan and Shaver (1987, 1990)

Ainsworth’s avoidant style and reflects the underlying action of deactivating strategies. The narratives are coded as preoccupied if the interviewee is entangled in still-intense worries and conflicted feelings about parents, can easily retrieve attachment-related memories, but has trouble coherently discussing them without anger or anxiety. This category parallels Ainsworth’s anxious style and reflects the underlying action of hyperactivating strategies. Despite the richness of AAI narratives, the interview is expensive to administer and score and it deals almost exclusively with memories of child– parent relationships. Given that we are primarily interested in attachment behavior in the context of adult close relationships and would like to be able to run large questionnaire and experimental studies in a reasonably short period of time, our research program relies on simple self-report measures of attachment style. The first such measure was constructed by Hazan and Shaver (1987, 1990) for their studies of romantic attachment. In its original form, the measure consisted of three brief descriptions of feelings and behaviors in close relationships that were intended to parallel Ainsworth’s three types of infants. Participants were asked which of the three descriptions best characterizes them in close relationships (see Table I). However, subsequent studies (e.g., Bartholomew & Horowitz, 1991; Brennan, Clark, & Shaver, 1998; Fraley & Waller, 1998) revealed that attachment styles are best conceptualized as regions in a two-dimensional space, which conceptually parallels the space defined by two discriminant functions in the summary of research on infant–mother attachment by Ainsworth et al. (1978, Fig. 10, p. 102). The two dimensions defining the space are called attachment anxiety and attachment avoidance (Brennan et al., 1998). (See Fraley & Spieker, 2003, for a similar analysis of individual differences among infants in the strange situation.)

69

THE ATTACHMENT BEHAVIORAL SYSTEM

LOW AVOIDANCE

SECURE

PREOCCUPIED

LOW ANXIETY

HIGH ANXIETY

DISMISSINGAVOIDANT

FEARFULAVOIDANT

HIGH AVOIDANCE Fig. 1. Diagram of the two-dimensional space defined by attachment anxiety and attachment avoidance. The terms in the four quadrants are Bartholomew’s (1990) names for the four major attachment styles.

In the two-dimensional attachment-style space (see Fig. 1), what was formerly called the ‘‘secure style’’ is a region in which both anxiety and avoidance are low. This region is defined by a sense of attachment security, comfort with closeness and interdependence, and reliance on the primary attachment strategy in times of need. What was formerly called the ‘‘anxious style’’ refers to the region in which anxiety is high and avoidance is low. This region is defined by a lack of attachment security, a strong need for closeness, worries about relationships, and reliance on hyperactivating strategies. What was called the ‘‘avoidant style’’ refers to a region in which avoidance is high. This region is defined by a lack of attachment security, compulsive self-reliance, preference for emotional distance from others, and reliance on deactivating strategies. In the research of Ainsworth et al. (1978), avoidant infants occupied mainly the region where avoidance is high and anxiety is low. In adult attachment research, Bartholomew and Horowitz (1991) drew a distinction between ‘‘dismissing avoidants’’ (high on

70

MIKULINCER AND SHAVER

avoidance, low on anxiety) and ‘‘fearful avoidants’’ (high on both avoidance and anxiety). Because several different measures have been used to study adult attachment, some based on three categories, some on four categories, and some on two dimensions, it is somewhat difficult to summarize results of studies on a particular topic without constantly raising the issue of measurement. Here, we will generally summarize results in terms of ‘‘secure,’’ ‘‘anxious,’’ and ‘‘avoidant’’ attachment patterns, unless some feature of the results requires reference to the distinction between fearful and dismissing avoidance or to simultaneously high scores on both the attachment anxiety and attachment avoidance dimensions. The issue of ‘‘fearful avoidance’’ is, in any case, less likely to arise in normal samples of college students and community adults. Extremely high scores on both the anxiety and avoidance dimensions are more common in samples of abused or clinical samples (see Shaver & Clark, 1994, for a review), where ‘‘disorganized’’ attachment behavior is a consequence of maltreatment or other contributors to psychopathology. In most of our studies, the results can be adequately described in terms of either anxiety or avoidance. At present, a person’s locations on the two dimensions are measured with two reliable 18-item Likert scales created by Brennan et al. (1998) based on factor analyses of previous measures. The two scales are reliable in both the internal-consistency and test–retest senses and have high construct, predictive, and discriminant validity (Crowell, Fraley, & Shaver, 1999). The following items are representative of the avoidance scale: ‘‘I try to avoid getting too close to my partner,’’ ‘‘I prefer not to show a partner how I feel deep down,’’ and ‘‘I turn to my partner for many things, including comfort and reassurance’’ (reverse scored). The following items are taken from the anxiety scale: ‘‘I need a lot of reassurance that I am loved by my partner,’’ ‘‘I do not often worry about being abandoned’’ (reverse scored), and ‘‘I get frustrated if romantic partners are not available when I need them.’’ The two scales were conceptualized as independent and, like the two discriminant functions reported by Ainsworth et al. (1978), have been found to be uncorrelated in most studies. It is important to note that despite the substantial differences in focus (parent–child vs. adult–adult relationships) and method (brief self-reports vs. extensive interview transcripts), self-report measures of attachment in close relationships are related to AAI coding scales (Bartholomew & Shaver, 1998). For example, in a study of over 100 married women, Shaver, Belsky, and Brennan (2000) found that both avoidance and anxiety scores based on self-report scales could be predicted from AAI coding scales. Analyses running in the other direction revealed that every AAI coding scale except one was predictable from self-report items. These findings imply that scores

THE ATTACHMENT BEHAVIORAL SYSTEM

71

on the two kinds of measures are related to each other in sensible ways, and that both are reflections of an underlying sense of attachment security, even though their differences in focus, method, and interpretation make them far from identical. 5. Summary In this section we introduced major individual differences in the functioning of the attachment system. Although the system is an inborn, prewired device that presumably operates mainly at a subcortical level and in a reflexive, mechanistic, and unintentional manner, its activation can lead to different response strategies (primary, hyperactivating, or deactivating) depending on both the quality of the current interaction with an actual attachment figure and internal representations of past interactions with the same or other attachment figures. These representations (working models of self and others) presumably operate in a more cortical, reflective, constructive, and intentional manner (although they can become habitual and automatic, and can be held out of consciousness by defensive maneuvers), and can transform context-tailored variations in the activation of specific attachment strategies to person-tailored variations. That is, they are a source of within-person continuity in attachment-system functioning. In the next section, we present a theoretical framework that makes sense of the cognitive operations and dynamic processes involved in the activation and functioning of the attachment system in adulthood.

II. An Integrative Model of Attachment-System Dynamics in Adulthood In attempting to characterize the dynamics of the activation and functioning of the attachment system in adulthood, we (Mikulincer, Shaver, & Pereg, in press; Shaver & Mikulincer, 2002a, 2002b) have proposed a control-system model that integrates recent findings with earlier theoretical proposals by Bowlby (1973, 1980, 1982/1969), Ainsworth (1991), Cassidy and Kobak (1988), and Main (1995). This model is a conceptual extension and refinement of previous control-systems representations of attachmentsystem dynamics presented by Shaver, Hazan, and Bradshaw (1988) and Fraley and Shaver (2000). It includes three components (Fig. 2): the primary strategy associated with attachment-system activation, the strategy related to the attainment of a sense of attachment security, and the strategies triggered by attachment-figure unavailability, which lead to a regulatory failure of the attachment system. The model also delineates (1) the main goals of each of

72

MIKULINCER AND SHAVER

−

FIRST MODULE

+

Signs of threat?

No

Continue with ongoing activities

Yes Activation of the attachment system

Seeking proximity to external or internalized attachment figure

+ −

SECOND MODULE Is attachment figure Attachment security, available, attentive, security-based strategies Yes responsive, etc?

Engagement in nonattachment activities (e.g., exploration, caregiving)

No Attachment insecurity (compounding of distress)

THIRD MODULE Is proximity-seeking a viable option?

No

Deactivating strategies

Distancing of threat- and attachment-related cues

Yes Hyperactivating strategies

Hypervigilance regarding threatand attachment-related cues

Fig. 2. Shaver and Mikulincer’s (2002a) integrative model of the activation and dynamics of the attachment system.

THE ATTACHMENT BEHAVIORAL SYSTEM

73

these strategies; (2) their declarative knowledge base and procedural rules for managing interpersonal behavior, coping with stress, and processing information; and (3) their implications for self-image, social judgments, mental health, relationship quality, and other behavioral systems. The first component involves the monitoring and appraisal of threatening events; it is responsible for activation of the attachment system. The second component involves the monitoring and appraisal of attachment-figure availability and is responsible for individual differences in the sense of attachment security and the psychological correlates and consequences of the secure attachment style. The third component involves monitoring and appraisal of the viability of proximity seeking as a way of dealing with attachment insecurity. This component is responsible for individual variations in the use of hyperactivating or deactivating strategies and accounts for the psychological manifestations of attachment anxiety and avoidance. The model also includes hypothetical excitatory and inhibitory ‘‘neural circuits’’ (shown as arrows on the left-hand side of Fig. 2), resulting from the recurrent use of hyperactivating or deactivating strategies. These circuits affect the monitoring of threats and the monitoring of attachment figures’ availability. All components and circuits of the model can operate either consciously or unconsciously. Moreover, these components and circuits can operate either in parallel or in opposite ways at conscious and unconscious levels. (We will provide concrete examples later in the chapter.) Our model is sensitive to both context- and person-tailored individual differences. On the one hand, each component of the model can be affected by specific contextual factors (e.g., actual threats, information about attachment-figure availability or proximity-seeking viability), which initiate a bottom-up process in a person’s working models, activating congruent attachment strategies and producing immediate changes in the functioning of the attachment system. On the other hand, each component of the model is affected by the chronically accessible working model, which biases the appraisal of threats, attachment-figure availability, and proximity-seeking viability. These biases are part of a top-down process by which the attachment system functions in accordance with a person’s chronic attachment style. Overall, the model gives place to both reality—the current context in which the attachment system is activated—and fantasy, the cognitive biases resulting from attachment strategies.

A. ACTIVATION OF THE ATTACHMENT SYSTEM Following Bowlby’s (1982/1969) ideas, we assume that the monitoring of unfolding internal or external events results in activation of the attachment

74

MIKULINCER AND SHAVER

system when a potential or actual threat is perceived. These triggers include both physical and psychological threats. They also include both attachmentunrelated and attachment-related sources of threat. In our view, every unfolding event perceived by a person as threatening his or her sense of security, and consequently his or her adjustment and survival, can activate the attachment system. This activation automatically heightens the accessibility of attachment-related cognitions, which in turn favor the seeking of proximity to supportive others. Although this component of the model represents the normative working of the system, which occurs regardless of individual differences in attachment history and orientation, it is still affected by the excitatory circuits of hyperactivating strategies and the inhibitory circuits of deactivating strategies. In our model, attachment-system activation depends on the subjective appraisal of threats rather than their actual occurrence. Although the actual presence of threats is obviously a very important factor, the individual’s appraisal of internal or external events as potential or actual threats is the critical factor in activating the attachment system. This idea fits with Lazarus and Folkman’s (1984) model of stress and coping, which emphasizes the critical role played by appraisal processes in the regulation of distress and the energization of coping efforts. In fact, the attachment system is designed to cope with threats and regulate distress, and it, like other regulatory devices, depends on a person’s perception that unfolding events are threats that require the mobilization of coping efforts. We view the appraisal of threats as a product of both the actual presence of threat-related cues and the individual’s forecast or expectations that unfolding events may have negative effects on well-being, adjustment, and survival. Although these expectations can follow a rational, conscious riskanalysis of the unfolding events, excitatory and inhibitory circuits resulting from secondary attachment strategies can automatically bias these expectations. That is, the appraisal process is not necessarily conducted at a conscious, rational level and the person may not even be aware of the monitoring and appraisal of threats. These cognitive processes can occur at preconscious levels and be manifested in physiological reactions and the measurable accessibility—readiness to influence information processing—of threat-related thoughts. Another important feature of our model is that inner sources of threat (e.g., thoughts, imagery) can activate the attachment system. For example, thoughts about personal death are usually appraised as extremely threatening (Greenberg, Pyszczynski, & Solomon, 1997) and so can activate the attachment system. In such cases, the person does not need to actually encounter the threat: Merely being reminded of death is enough to activate the system. Again, these thoughts need not be conscious; they can be

THE ATTACHMENT BEHAVIORAL SYSTEM

75

activated by a stimulus as minimal as the word ‘‘death’’ presented subliminally (e.g., Mikulincer, Birnbaum, Woodis, & Nachmias, 2000). These two features—subjective appraisal and the activating power of threat-related thoughts—are critical for delineating the effects of the neural circuits associated with secondary attachment strategies on attachmentsystem activation. On the one hand, hyperactivating strategies, which maintain the attachment system in a chronically activated state, include vigilance with respect to triggers of attachment-system activation and exaggeration of the threatening aspects of person–environment transactions; they also include rumination on potential threats, which occupies working memory. This means anxiously attached individuals can experience attachment-system activation in the absence of any consensual, external sign of danger. On the other hand, deactivating strategies, which maintain the attachment system in a down-regulated or inactive state, foster the ignoring or dismissal of threatening aspects of person–environment transactions and the suppression of any threat-related thought that might activate the attachment system. This means that avoidant individuals can distance themselves from various sources of threat and keep themselves from thinking about either their need for protection or the relief that might be found in the presence of a loving and protective partner. Once a threat is appraised, attachment-system activation is automatically manifested in heightened accessibility of attachment-related nodes within the associative memory network. These nodes include internal representations of security-enhancing attachment figures; episodic memories of supportive and comforting interactions with these figures; thoughts related to proximity, love, and support; and proximity-seeking goals. These preconsciously activated nodes become ready for use in subsequent information processing and can shape a person’s state of mind and influence his or her behavioral plans even before they are consciously formulated. This idea fits with recent findings from social cognition research showing that accessible cognitive material can shape a person’s state of mind before he or she recognizes the material in the stream of consciousness (Wegner & Smart, 1997). It is also congruent with the ‘‘auto-motive’’ model (Bargh, 1990), which suggests that goals can be preconsciously activated and this preconscious activation can automatically guide a person’s behavior without the mediation of conscious behavioral planning. This automatic, preconscious accessibility of attachment-related memory nodes is the normative manifestation of attachment-system-activation in adulthood. We assume that age and development result in an increased ability to gain security from internalized representations of securityenhancing attachment figures without necessarily causing a person to actually seek proximity to these figures. This cognitive activation maintains

76

MIKULINCER AND SHAVER

symbolic proximity to figures who in the past provided security and safety. That is, the regulatory action of the adult attachment system can occur intrapsychically without a person’s awareness. However, like Bowlby (1982/ 1969), we assume that no one of any age is completely free of dependence on others. There are situations, such as physical and psychological traumas, illnesses, or losses, in which symbolic proximity to internalized figures is not sufficient to provide a sense of security, and in such situations attachmentsystem activation leads to proximity-seeking behavior. There are also developmental stages, such as old age, that tax people’s physical and psychological resources to the point where it becomes necessary to seek proximity to and support from others. Overt manifestations of attachment-system activation in conscious thoughts, behavioral intentions, and actual behaviors are dependent on contextual, cultural, and personal factors and do not necessarily follow directly from the accessibility of attachment-related nodes. Furthermore, variations in proximity-seeking behavior are more a reflection of the underlying action of attachment strategies than of attachment-system activation itself. In fact, people differing in attachment style differ in their tendency to seek actual proximity in times of need. These individual differences are also manifested in the content of the attachment-related nodes that are automatically activated by threat appraisal. For people with a history of security attainment and a secure attachment style, threat appraisal makes accessible thoughts about typical, positive interactions with attachment figures—i.e., thoughts of proximity, support, love, comfort, and relief. However, for persons with a painful history of attachment interactions and an insecure attachment style, threat appraisal can make accessible many negative attachment-related thoughts (e.g., thoughts about separation and rejection). These people’s frustrating attachment experiences may create an associative link in their memory networks between activation of the attachment system and worries about separation or rejection, so that the accessibility of these worries increases every time attachment needs are activated. In sum, the appraisal of unfolding events as threatening automatically activates attachment-related nodes in the associative memory network and preconsciously activates internalized representations of attachment figures and proximity-seeking goals. This activation can remain at the preconscious level and be a source of comfort. However, it can also be manifested in attachment-related conscious thoughts and actual proximity-seeking behaviors, depending on contextual factors and the person’s attachment style. Attachment style is also a source of individual differences in threat appraisal and the specific content of the attachment-related nodes activated by the perceived threat.

THE ATTACHMENT BEHAVIORAL SYSTEM

77

B. ATTACHMENT-FIGURE AVAILABILITY AND SECURITY-BASED STRATEGIES Once the attachment system is activated, an affirmative answer to the question, ‘‘Is the attachment figure available?’’ results in a sense of attachment security and positive models of the self and others. Attachmentfigure availability also reinforces the perceived effectiveness of the seeking of proximity and fosters the development of what we call security-based strategies. The goals of these strategies are to form and maintain close bonds with others, alleviate distress, and bolster personal adjustment through constructive, flexible, and reality-attuned mechanisms. Moreover, they create what we, following Fredrickson (2001), call a ‘‘broaden and build’’ cycle of attachment security, which builds a person’s resources for maintaining emotional equanimity in times of stress and broadens his or her perspectives and capacities. In the long run, repeated episodes of attachment-figure availability have a powerful and enduring impact on intrapsychic organization and interpersonal behavior. At the intrapsychic level, these episodes lead to the consolidation of security-based strategies as the main method of affect regulation and of secure working models of self and others as the most chronically accessible cognitive representations. At the interpersonal level, these episodes foster the enactment of a secure attachment style in social situations and close relationships. Security-based strategies and the resulting ‘‘broaden and build’’ cycle of attachment security can be viewed as the core characteristics of securely attached persons. As in the system-activation module, the answer to the question about attachment-figure availability in the second module depends on subjective appraisal of this availability and can be biased by attachment strategies. The neural networks or circuits of insecure persons make the positive appraisal of attachment-figure availability unlikely. Anxious persons’ hyperactivating strategies intensify the vigilant monitoring of attachmentfigure behaviors and slant perceptions in the direction of noticing or imagining insufficient interest, availability, and responsiveness. As a result, the likelihood of detecting signs of distance, rejection, and unavailability is increased, because the attachment figure cannot always be available and totally at the disposition of the attached person’s needs. Avoidant individuals’ deactivating strategies interfere with the monitoring of cues concerning either the availability or unavailability of the attachment figure, increasing the likelihood that genuine and clear-cut signals of attachment-figure availability will be missed. In contrast, security-based strategies facilitate the appraisal of attachment-figure availability—they reflect a positive view of relationship partners as available and supportive,

78

MIKULINCER AND SHAVER

and favor the ignoring or downplaying of episodes of momentary unavailability. These cognitive biases are amplified when attachment nodes in the memory network are preconsciously activated. At the preconscious level, appraisal of attachment-figure availability depends entirely on the type of internalized figure, available or unavailable, that is activated. Insecurely attached persons tend to give a negative answer to the question of attachment-figure availability, because they have chronic access to representations of unavailable figures and these representations become increasingly accessible when attachment nodes are activated. More securely attached people, however, tend to give a positive answer to this question, because they have chronic access to representations of available figures and their attachment nodes are linked with a host of associations and memories of security-enhancing attachment figures. Despite these cognitive biases, however, reality is still important in the appraisal of attachment-figure availability. In our view, the actual presence of an available attachment figure or contextual cues that activate representations of available attachment figures (thinking about one of these figures) can lead people to give an affirmative answer to the question of attachment-figure availability. These contexts, mainly when they are clearcut, personally meaningful, and stable over time and situations, can counteract even insecure people’s tendencies to negate the availability of attachment figures, and can set in motion the ‘‘broaden and build’’ cycle of attachment security. That is, these contextual cues can influence the appraisal of attachment-figure availability beyond the cognitive biases resulting from attachment strategies. Accordingly, these cues can activate security-based strategies even among chronically insecure persons. Security-based strategies include three core beliefs: optimistic beliefs about distress management, a sense of trust in others’ availability and good will in times of need, and a sense of self-efficacy in dealing with threats. These beliefs are a direct result of positive interactions with available attachment figures. During these interactions, individuals learn that distress is manageable and external obstacles can be overcome. Moreover, they learn about others’ positive responses to attachment behaviors and about the control one can exert over the course and outcome of threatening events. In our view, these cognitive acquisitions are the building blocks of a person’s capacity for dealing with stress. The procedural knowledge involved in security-based strategies consists of a set of rules embodied in what Waters, Rodrigues, and Ridgeway (1998) called the ‘‘secure base script.’’ This hypothetical script is organized around three regulatory tendencies—acknowledgment and display of distress, the seeking of intimacy, closeness, and support, and engagement in instrumental

THE ATTACHMENT BEHAVIORAL SYSTEM

79

problem solving. The ‘‘emotion-focused coping’’ components of this script (Lazarus & Folkman, 1984)—acknowledging and expressing feelings and seeking emotional support—work in the service of alleviating distress so that ‘‘problem-focused coping’’ components—seeking instrumental support and solving problems—can proceed successfully. Again, these tendencies seem to stem from prior appraisals of attachment-figure availability and consequent confirmations that proximity seeking results in protection, support, and relief of distress. Relatively secure individuals have learned that acknowledgment and display of distress elicit positive and supportive responses from others. They have also learned that their own actions are often able to reduce distress and remove obstacles, and that turning to others when threatened is an effective route to enhanced coping. These regulatory strategies are directly manifested in interpersonal behaviors and close relationships. One basic declarative component of security-based strategies is the knowledge, or belief, that proximity maintenance is rewarding and that relationships are an adequate framework for satisfying one’s needs. As a result, these strategies predispose people to feel comfortable with intimacy and interdependence, emphasize the benefits of being together, and organize their interaction goals around the search for intimacy. Another declarative component of security-based strategies is the belief in others’ good will, which fosters a sense of trust, gratitude, and affection toward a relationship partner as well as tolerance for ambiguous or negative partner behaviors. These strategies also include a positive attitude toward affect display, which heightens the readiness to self-disclose and share personal feelings with a relationship partner. Security-based strategies are also manifested in the process of coping with stress and the management of negative emotions. Secure individuals’ optimistic and constructive attitude toward life’s difficulties leads them to adopt what Epstein and Meier (1989) called constructive ways of coping— active attempts to remove the source of distress, manage the problematic situation, and restore emotional equanimity without creating negative socioemotional side effects. These constructive ways of coping consist of active problem solving as well as transformational strategies that involve symbolically transforming unsolvable problems into meaningful, growthpromoting challenges. They also include reliance on close relationships as an anxiety buffer and the seeking of proximity and support as a means of coping with stress. The constructive nature of security-based strategies is enhanced by their flexibility and reality-attunement. Optimistic beliefs about self-efficacy and distress management allow people to open their cognitive structures to new, even threatening, information, and then to flexibly adjust their strategies for dealing realistically with environmental demands. Experiencing, or having

80

MIKULINCER AND SHAVER

experienced, attachment figures as approving allows people to revise erroneous beliefs without excessive fear of criticism or rejection, thus avoiding the cognitive and motivational entrapment that results from being unable to revise beliefs. Furthermore, the building of a person’s resources for dealing with stress makes it less necessary to rely on maladaptive means of coping, including primitive defense mechanisms that distort perceptions and generate interpersonal conflicts. In our view, the adoption of security-based strategies epitomizes optimal functioning of the attachment system. Such strategies are expected to contribute directly to the development of generalized positive representations of others, the consolidation of a stable sense of self-efficacy and selfesteem, the maintenance of mental health and adjustment in times of stress, and the formation of long-lasting, satisfying close relationships. Moreover, these strategies facilitate the deployment of resources to other behavioral systems, such as exploration, sex, caregiving, and affiliation, and thereby contribute to the broadening of a person’s perspectives and capacities. The sense of attachment security allows people to engage in exploratory activities, enjoy sex and social interactions, and be attentive and responsive to others’ needs. Moreover, with confidence that support is available when needed, people can take risks and engage in autonomy-promoting activities. In other words, security-based strategies facilitate the development of autonomy, individuality, and self-actualization.

C. PROXIMITY-SEEKING VIABILITY AND SECONDARY ATTACHMENT STRATEGIES Attachment-figure unavailability results in attachment insecurity and compounds the distress anyone might experience when encountering a threat. We claim that this painful state forces a decision about the viability of proximity seeking as a means of self-regulation and the subsequent choice of a secondary attachment strategy. As in the two modules of Fig. 2 already described, a decision about proximity-seeking viability depends on subjective appraisal processes, which are affected by organismic and contextual cues that will be described later. 1. Hyperactivating Strategies When proximity seeking is appraised as being a viable option if only greater effort is expended, people tend to make very energetic, insistent attempts to attain security—in short, to use hyperactivating strategies (Cassidy & Kobak, 1988). The main goal of these strategies is to get an

THE ATTACHMENT BEHAVIORAL SYSTEM

81

attachment figure who is viewed as unreliable or insufficiently available and responsive to pay attention and provide protection or support. The basic means for attaining this goal is to maintain the attachment system chronically activated until an attachment figure is perceived to be available and to provide safety and security. This strategy involves exaggerating the presence and seriousness of threats and being highly vigilant regarding attachment-figure availability or unavailability, because these cues are highly relevant for security attainment. This constant worry and vigilance increases the likelihood that a person will notice real or imagined signs of a relationship partner’s disapproval, waning interest, or impending departure. In the long run, repeated episodes of attachment-figure unavailability or nonresponsiveness combined with the appraisal of proximity-seeking viability lead to the consolidation of hyperactivating strategies as the main regulatory device. At the interpersonal level, this internal process is manifested in heightened desires for closeness, merger, and security in close relationships together with an increased preoccupation with the partner’s unavailability—the two definitional components of attachment anxiety (Brennan et al., 1998). We view hyperactivating strategies as the key underlying characteristics of the anxious attachment style. Hyperactivating strategies are sustained by negative declarative beliefs about distress management, others’ good will, and one’s self-efficacy in dealing with threats. Although these beliefs are developed during negative interactions with insufficiently available, responsive, or reliable attachment figures, they also reflect cognitive biases that overgeneralize past attachment injuries and apply memories about them inappropriately to new situations and partners. They also result from the amplification of threat appraisals that maintain the attachment system chronically activated. At the interpersonal level, hyperactivating strategies manifest themselves as exaggerations of the primary attachment strategy: constant monitoring of the relationship partner and strong efforts to maintain proximity. They consist of overdependence on a relationship partner as a source of comfort; intense demands for attention and care; strong desire for enmeshment or merger; attempts to minimize cognitive, emotional, and physical distance from a partner; and clinging and controlling behaviors designed to elicit the partner’s affection and support. These strategies also reflect a malfunction of the primary attachment strategy produced by exaggerated threat appraisals (i.e., a heightened sense of vulnerability) and exaggerated appraisals of attachment figures’ unavailability. This malfunction is evident in anxious persons’ hypersensitivity to rejection cues and worries about separation and abandonment. In the process of coping with threats, hyperactivating strategies are manifested in what Lazarus and Folkman (1984) called ‘‘emotion-focused

82

MIKULINCER AND SHAVER

coping’’—hypervigilant attention to internal indications of distress. This method of coping consists of intensification of negative emotions, mental rumination on related negative cognitions, self-preoccupation, self-criticism, and overt displays of distress. As a result, anxiously attached individuals have free access to threat-related thoughts and emotions and are unable or unwilling to suppress them. Intrapsychically, amplification of threat appraisals heightens the chronic accessibility of negative thoughts and makes it likely that new sources of distress will mingle and become confounded with old accessible ones. That is, activation of one cognitive node with a negative emotional tone will automatically spread to other negatively tinged cognitive nodes, because all of these cognitions are maintained simultaneously in working memory. Moreover, this pattern of cognitive activation gives predominance to the affective implications of the information and favors the organization of cognitions in terms of simple, undifferentiated features, such as the extent to which the information is threatening or implies rejection. In this way, hyperactivating strategies create a chaotic mental architecture that is constantly pervaded by negative affect. The heightened accessibility of threat-related thoughts also creates a mental context that biases the encoding of new information. According to cognitive theories (e.g., Anderson, 1994), this context facilitates the encoding of congruent information (e.g., threat-related cues), while diverting attention away from the encoding of context-incongruent cues. This context also fosters the creation of excitatory links between threatrelated information and the large number of congruent cognitions that are highly accessible in the memory network. As a result, this information is encoded at a deep level of processing, which seems to be further exacerbated by mental rumination on threat-related material. In contrast, positive information may be encoded in a shallow way, and then easily be forgotten or recalled in an inaccurate manner. Hyperactivating strategies have negative implications for anxious persons’ self-image. Overdependence on a partner as the main source of comfort interferes with the formation of a sense of confidence in one’s own abilities and skills. Moreover, the intensification of threat appraisals causes attention to be directed to self-relevant sources of distress (e.g., thoughts about personal weaknesses, memories of personal failures), thereby fostering chronic doubts about self-worth. Frequent activation of negative cognitions and deep-level encoding of congruent information also heighten the chronic accessibility of negative self-relevant thoughts, further encouraging the formation of an undifferentiated negative self-image. Beyond these automatic processes, more controlled attempts to achieve some sort of closeness and love can add to a negative self-image—the

THE ATTACHMENT BEHAVIORAL SYSTEM

83

anxious person may emphasize his or her helplessness and present him- or herself in a negative or needy way to elicit others’ compassion and support. Through three convergent processes, hyperactivating strategies encourage negative appraisals of others: (1) The exaggerated appraisal of attachment figures’ unavailability contributes to the formation of a negative global view of others. (2) The frequent activation of negative cognitions primes negative thoughts about others’ intentions and responses and makes these thoughts accessible for social judgments. (3) The search for proximity and fusion leads anxious individuals to project their negative self-views onto relationship partners in order to create some sort of consensus and illusory closeness. This projective tendency, which object relations theorists (e.g., Klein, 1940) call ‘‘projective identification,’’ results in a negative view of others and an overestimation of self–other similarity. In this way, a negative view of others may be guided, in part, by the search for connectedness and a person’s own negative self-views. Chronic reliance on hyperactivating strategies also places anxious individuals at risk for a variety of emotional and adjustment problems. These strategies impair the ability to regulate negative emotions and lead to intense distress, which continues even after actual threats subside. As a result, the person experiences an endless and uncontrollable flow of negative thoughts and moods, which in turn may lead to cognitive disorganization and, in certain cases, culminate in psychopathology. Specifically, hyperactivating strategies can promote chronic attachment-related and attachment-unrelated anxieties, strong depressive reactions to actual or potential interpersonal losses, and intrusive symptoms following traumatic events (e.g., unwanted trauma-related thoughts). Moreover, problems in emotion control may be manifested in anger outbursts, impulsive behaviors, and the development of more severe personality disorders. Hyperactivating strategies also have several negative implications for relationship quality. First, these strategies lead a person to adopt a dependent and needy role that impairs the formation of a mature reciprocal relationship. Second, they lead the person to feel chronically frustrated due to the unfulfilled need for demonstrations of love and commitment. Third, hyperactivating strategies contribute to a catastrophic appraisal of interpersonal conflicts, the perpetuation of the resulting negative affect, and conflict escalation. Fourth, these strategies may lead partners to feel abused by a person’s endless demands for security and frequent suspicion and distrust; engulfed by the person’s desire for merger; and controlled by his or her clinging behavior and hypervigilance. All of these negative feelings may lead partners to take distance from the demanding person and reject his or her bids for proximity, which in turn may intensify the person’s insecurities and worries. In this way, a self-amplifying dyadic

84

MIKULINCER AND SHAVER

cycle of dissatisfaction is created which impairs relationship quality and stability. Of course, this cycle also depends on the functioning of the partner’s attachment system. For example, a securely attached partner can be more tolerant of a person’s demands for security, and this tolerance may reduce relational tensions. We will return to this point in a later section in which we extend our intrapersonal model of attachment-system functioning to a systemic model that takes into account the attachment systems of two interacting partners. Hyperactivating strategies also inhibit the activation of other behavioral systems. People who rely on these strategies are so focused on the seeking of protection and security that they have insufficient resources for engaging in nonattachment activities. Such people egocentrically perceive others as a source of comfort and are unable to perceive them as partners for exploration, sex, and affiliation, or as human beings who may also need help and care. They also cannot attain the relatively calm and secure state of mind that is necessary for engaging fruitfully in many nonattachment activities (Weiss, 1998). 2. Deactivating Strategies When proximity seeking is not perceived as viable, people tend to squelch their quest for proximity and inhibit the activation of the attachment system through the use of deactivating strategies (Cassidy & Kobak, 1988). These strategies include the denial of attachment needs and the pursuit of excessive self-reliance. They also include downplaying threats and blocking the monitoring of attachment-figure availability, because every thought about threats or attachment figures can reactivate the attachment system. In the long run, repeated episodes of attachment-figure unavailability combined with a negative appraisal of proximity-seeking viability lead to the consolidation of deactivating strategies as the main regulatory device. At the interpersonal level, this internal process is manifested in an avoidant style. We view deactivating strategies as the underlying core characteristics of avoidant persons. Deactivating strategies consist of negative core beliefs about the relationship partner as a source of protection, combined with positive beliefs about the self as capable of dealing with threats. During their interactions with unavailable attachment figures, individuals relying on deactivating strategies have probably experienced serious doubts about the management of distress, the partner’s good will, and the self’s capacity for exerting control over the course and outcome of threatening events. However, they must suppress any recognition of personal lack of control,

THE ATTACHMENT BEHAVIORAL SYSTEM

85

because only a powerful and competent self can handle distress alone. Moreover, such recognition entails the appraisal of the self as a source of threat and the consequent unwanted reactivation of the attachment system. As a result, deactivating strategies include heightened perceptions of selfefficacy. At the interpersonal level, deactivating strategies represent an inhibition of the primary attachment strategy. People who chronically rely on deactivating strategies have two main goals in social interactions and close relationships: (1) pursuit of distance, control, and self-reliance, and (2) avoidance of negative emotional states that demand attachment-system activation. The first goal is manifested in active attempts to maximize cognitive, emotional, and physical distance from a partner; avoidance of interactions that demand emotional involvement, intimacy, interdependence, and self-disclosure; and suppression of attachment-related thoughts and emotions that imply a sense of relational closeness, cohesion, and consensus. The second goal is manifested in reluctance to directly or symbolically confront relational tensions and conflicts; unwillingness to deal with a partner’s distress and needs for proximity and security; and suppression of thoughts and emotions related to rejection, separation, abandonment, and loss. In coping with threats, deactivating strategies lead to the adoption of what Lazarus and Folkman (1984) called ‘‘distancing coping’’—cognitive and behavioral maneuvers aimed at preventing active confrontation with threats and intrusion of threat-related thoughts into consciousness. These strategies consist of dismissal of threats; suppression of threat-related thoughts and feelings; repression of painful memories, diversion of attention away from the threat-related thoughts; and withdrawal of problem-solving efforts. They also consist of detachment from challenging and emotional interactions that potentially constitute sources of threat. These maneuvers inhibit the experience of distress, exclude threat-related cognitions from awareness, and thereby reduce the need for attachment-system activation. Deactivating strategies are also manifested in the organization of the cognitive system. The inhibitory circuits established as part of these strategies lower the accessibility of threat- and attachment-related cognitions and inhibit the spread of activation from these cognitions to other, different cognitions. These circuits also create difficulties in encoding material that is congruent with the excluded cognitions. Attention is diverted from threat-and attachment-related information, and when encoded this information is processed at a shallow level because it has no strong excitatory association with the accessible elements of the memory network. As a result, this information is not fully integrated with other parts of the network and is encapsulated in segregated structures. The resulting

86

MIKULINCER AND SHAVER

fractionated cognitive architecture resembles what Bowlby (1980) and George and West (2001) called ‘‘segregated systems.’’ These procedural rules have important implications for avoidant people’s self-image. The inhibitory circuits of deactivating strategies divert attention away from self-relevant sources of distress, and can therefore inhibit the appraisal of negative aspects of the self, suppress cognitions about personal weaknesses and imperfections, and repress memories of personal failures. In this way, deactivating strategies lead to the exclusion of negative self-aspects from consciousness and a conscious sense of high self-esteem. This defensive inflation of self-esteem is further reinforced by the adoption of a self-reliant attitude, which requires enhancement of selfworth, and by strategic attempts to convince others that one does not need their support. This defensive maintenance of a positive self-image seems to be achieved at the expense of negative beliefs about others. These negative perceptions result from the inhibitory circuits running from deactivating strategies to the monitoring and appraisal of attachment-figure availability. These circuits lead people to divert attention away from any attachment-related information, including information about others’ positive traits, intentions, and behaviors. As a result, this information is processed in a shallow manner, is easily forgotten or biased, and is not accessible when making social judgments, which results in the person maintaining a negative and inflexible image of others. In addition, the defensive exclusion of negative information about the self and attempts to maintain interpersonal distance facilitate the projection of the excluded information onto a relationship partner, thereby reinforcing the basic negative view of others. This projective tendency, which Freud (1957/1915) called ‘‘defensive projection,’’ results in a negative view of others and an overestimation of self-other dissimilarity. Thus, negative perceptions of others related to deactivating strategies may be guided, in part, by a preference for distance and a desire to view the self more positively than others. Deactivating strategies can be a source of emotional and adjustment problems. Although these strategies lead to the defensive maintenance of a faca,de of security and calmness, their inhibitory circuits leave the suppressed distress unresolved and impair the ability to directly confront life’s adversities. This impairment is particularly likely to be manifested during prolonged, highly demanding stressful experiences that require active confrontation of the problem and mobilization of external sources of support. In these cases, deactivating strategies can produce a sense of inadequacy in coping with stress, a marked decline in functioning, constricted affect, and what Horowitz (1982) called ‘‘avoidance-related’’ posttraumatic symptoms (e.g., psychic numbing, behavioral inhibition,

THE ATTACHMENT BEHAVIORAL SYSTEM

87

counterphobic activities). All of these symptoms reflect the inadequacy of deactivating strategies under certain conditions and their failure to restore emotional equanimity. In addition, although deactivating strategies involve suppressing the conscious experience and display of distress, the distress can still be indirectly manifested in somatic symptoms, sleep problems, and other health disorders. Moreover, the negative attitude toward close relationships can channel the unresolved distress into feelings of hostility, loneliness, detachment, and estrangement from others. Deactivating strategies also have negative implications for relationship quality and stability. First, they lead a person to be emotionally detached from a partner and to form superficial, cool relationships that lack the vitality and bonding power of affection and intimacy. Second, the tendency to avoid confronting relational problems may leave conflicts unresolved and increase a partner’s irritation and resentment. Third, deactivating strategies may lead partners to feel frustrated and dissatisfied due to the constant rejection of their bids for intimacy and affection as well as the person’s dismissal of, and nonresponsiveness to, their signs of distress. As a result, relationship satisfaction can decrease and the likelihood of relationship dissolution can increase. Deactivating attachment strategies maintain inhibitory links with other behavioral systems. They block cognitive exploration and the incorporation of new evidence and unusual thoughts, because such thoughts may challenge prior knowledge and lead to uncertainty, tension, and confusion. In fact, relaxed exploration and ‘‘loosening’’ of cognitive operations can open the door to threats and dangers, which activate the attachment system. Thus, deactivating strategies favor cognitive closure and rigidity, and block the full activation of the sexual and affiliation systems, because mutual exploration of sexual pleasures with a lover and having flexible, unscripted fun with friends can lead to the sort of intimacy and closeness that avoidant persons are not comfortable dealing with. These strategies also block the activation of the caregiving system, because empathic responsiveness to others’ needs entails emotional involvement, acknowledgment of others’ distress, and dealing with the closeness that this empathic reaction implies. The demands of caregiving work against the goals of deactivating strategies—to distance a person from all sources of suffering and all kinds of closeness to others.

D. SUMMARY Our model outlines the cognitive operations, response strategies, and dynamics of the attachment system in adulthood. It also describes the goals of each attachment strategy and their psychological manifestations and

88

MIKULINCER AND SHAVER

consequences. Whereas the goals of security-based strategies are to form intimate relationships, to build a person’s psychological resources, and to broaden his or her perspectives and capacities, the goal of secondary attachment strategies is to manage attachment-system activation and reduce or eliminate the pain caused by frustrated proximity-seeking attempts. Hyperactivating strategies keep the person focused on the search for love and security, and constantly on the alert for threats, separations, and betrayals. Deactivating strategies keep the attachment system in check, with serious consequences for cognitive and emotional openness. This framework serves as our ‘‘working model’’ for understanding the activation and functioning of the attachment system in adulthood. It also provides a framework for reviewing our research findings, which is the mission of the next section.

III. Empirical Assessments of the Theory In this section, we review our research program on adult attachment style. Our studies have examined the three different modules of our model and have empirically assessed our ideas concerning (1) attachment-system activation, (2) attachment-figure availability, (3) the interpersonal and intrapersonal manifestations of attachment strategies, and (4) implications of these strategies for representations of the self and others, mental health, relationship quality, and other behavioral systems. In this review, we will switch back and forth between typological and dimensional terms for attachment styles, because some of our studies were conducted when we still used typological self-report measures and others were conducted with dimensional measures of attachment anxiety and avoidance, such as the Experience in Close Relationships scale (Brennan et al., 1998). As explained earlier, most results can be summarized adequately in terms of secure, anxious, and avoidant attachment patterns, with the distinction between dismissing and fearful avoidance mattering only occasionally, when abuse or psychopathology are at issue.

A. STUDIES ON ATTACHMENT-SYSTEM-ACTIVATION Our ideas about the automatic, preconscious activation of the attachment system in adulthood have been examined in a recent series of studies (Mikulincer et al., 2000; Mikulincer, Gillath, & Shaver, 2002). These studies have exposed participants to symbolic threat contexts and assessed the

THE ATTACHMENT BEHAVIORAL SYSTEM

89

effects of threat on the accessibility of thoughts about attachment themes or mental representations of attachment figures. Specifically, they assessed the readiness of these cognitive elements to influence performance on cognitive tasks without asking participants to report whether these cognitions were active in their stream of consciousness. In the studies of Mikulincer et al. (2000), the accessibility of proximity themes was assessed in a lexical decision task following priming with a threatrelated or neutral word. The lexical decision task required participants to indicate by pressing a computer key whether a string of letters was or was not a word. Priming was accomplished by exposing threat-related or neutral words subliminally for very brief periods of time. Findings supported our hypothesis about the effects of threat on attachment-system activation: Priming with a threat word (e.g., failure, illness, death) led to faster identification of proximity-related words (e.g., love, closeness). This effect was specific to proximity-related words and did not generalize to positive affect words that have no attachment connotation. Moreover, this heightened accessibility of proximity-related thoughts occurred regardless of individual variations in attachment style, suggesting, as expected, that everyone underwent preconscious activation of the attachment system. Following these studies, Mikulincer, Gillath, and Shaver (2002) conducted three experiments focused on the accessibility of the names of people whom participants listed as security-enhancing attachment figures. Participants filled out the WHOTO scale (Fraley & Davis, 1997; Hazan & Zeifman, 1994), which identifies the names of people who serve as securityenhancing attachment figures. They also provided names of close others who were not mentioned in the WHOTO scale, as well as names of people they knew but with whom they were not close (known persons) and names of people they did not know at all. They then performed either a lexical decision task or a Stroop color-naming task, providing multiple indicators of the accessibility (as reflected by reaction times, or RTs) of representations of attachment figures, close persons, known persons, and unknown persons after subliminal exposure to either a threat prime (the word failure or separation) or a neutral word (hat, umbrella). We found that threat contexts activated mental representations of attachment figures but no one else. Across the three studies, participants reacted to threat contexts with heightened accessibility of the names of the people they listed on the WHOTO scale as attachment figures. As compared to subliminal neutral primes, subliminal priming with threat words led to (1) faster identification of names of attachment figures in the lexical decision task and (2) slower RTs in naming the color of the printed names of attachment figures in the Stroop task. In both cases, fast lexical decision RTs and slow color-naming RTs were interpreted as manifestations of

90

MIKULINCER AND SHAVER

heightened activation of representations of attachment figures in threatening contexts. Because priming with threat words increased the accessibility only of representations of attachment figures, we believe we have shown that the mind turns automatically to attachment figures under threatening conditions, as can be documented behaviorally in infants but had previously been a matter of speculation in the case of adults. The activation of representations of attachment figures occurred both when an attachment-unrelated threat word (failure) and an attachment-related threat word (separation) were presented subliminally. Beyond these normative processes, the priming studies also documented the expected individual differences in preconscious activation of the attachment system and revealed how secondary attachment strategies shape attachment-system activation. In these studies, securely attached people exhibited greater access to thoughts about proximity and love and to the names of attachment figures only in a threatening context, not in a neutral context (Mikulincer et al., 2000; Mikulincer, Gillath, & Shaver, 2002). These findings imply that secure individuals’ cognitive systems are not chronically occupied with attachment themes or attachment figures. Rather, attachment-related cognitions seem to be adaptively activated when a transaction signals that some coping action should be taken. Also important is the fact that secure individuals’ reactions to threat primes were limited to attachment themes with positive affective connotations. Mikulincer et al. (2000) found that secure individuals had relatively slow access to words connoting separation and rejection under both neutral and threatening conditions. Their positive working models apparently create a separation in their memory network between attachment-system activation and worries about rejection or separation. For them, thinking about attachment themes seems to include anticipation or attainment of relief. Anxious individuals’ pattern of cognitive activity was virtually opposite to that of secure individuals (Mikulincer et al., 2000; Mikulincer, Gillath, & Shaver, 2002). First, anxious people exhibited heightened accessibility of attachment themes and attachment figures’ names in both threatening and nonthreatening contexts. Second, these people evinced heightened access to attachment-related threats (e.g., separation). These findings fit with our ideas about the anxious person’s hyperactivating strategies, which seem to be employed even in nonthreatening contexts and to increase attention to threat-related cues. Avoidant individuals yielded a more complex set of findings. In general, their pattern of access to attachment themes resembled that observed among secure persons (Mikulincer et al., 2000; Mikulincer, Gillath, & Shaver, 2002). However, there were some important differences between secure and avoidant individuals. For avoidant persons, attachment-related worries

THE ATTACHMENT BEHAVIORAL SYSTEM

91

were relatively inaccessible even following subliminal priming with the word ‘‘death,’’ which is usually a potent activator of attachment-related fears (Kalish, 1985). Moreover, attachment-related worries did become accessible to avoidant individuals in response to threat primes when a ‘‘cognitive load’’ was added to the lexical decision task—that is, when avoidant participants had to engage in an additional cognitively demanding task, perhaps robbing mental resources from the defensive exclusion of attachment-related concerns. Mikulincer, Gillath, and Shaver (2002) reported an additional difference between secure and avoidant individuals. When the subliminal threat prime was the word separation (an attachment-related threat), secure individuals exhibited enhanced access to the names of attachment figures, whereas avoidant individuals exhibited decreased access. There was no such difference when the subliminal prime word was failure, a word not closely related to attachment. It seems, therefore, that the attachment system is preconsciously activated under attachment-unrelated threatening conditions for both avoidant and nonavoidant people, but it is preconsciously inhibited or deactivated under attachment-related threatening conditions (or at least following the threat of separation) if a person is avoidant. Under the latter conditions, avoidant individuals’ responses in the lexical decision and Stroop tasks resemble their self-reported feelings and their actual behavior in interpersonal situations. It seems that avoidant adults have learned not to appeal to attachment figures when those figures are threatening to leave—in fact, have learned to inhibit the tendency to seek proximity, which we know they possess, at least in nonattachment-related threat conditions. Our understanding of avoidant adults is compatible with the analysis by Ainsworth et al. (1978) of the behavior of avoidant infants, whose mothers seemed angrier than the mothers of infants classified as secure or anxious, less comfortable with physical contact, and less tolerant of their infants’ expressions of vulnerability and neediness. Ainsworth et al. (1978, p. 320) said, for example, ‘‘avoidance short circuits direct expression of anger to the attachment figure, which might be dangerous, and it also protects the baby from re-experiencing the rebuff that he has come to expect when he seeks close contact with his mother.’’ Perhaps when this ‘‘short circuiting’’ is practiced repeatedly, it becomes part of a person’s automatic deactivating strategies and is aroused when he or she subliminally registers the word separation. This circuit may allow avoidant people to maintain some degree of proximity to attachment figures without risking the caregivers’ disapproval and distancing (Main & Weston, 1982). Overall, the findings clearly show that threatening stimuli preconsciously activate attachment-related thoughts and mental representations of attachment figures. They also show that anxious persons’ hyperactivating strategies lead to attachment-system activation even in neutral contexts and

92

MIKULINCER AND SHAVER

color this activation with worries about separation and rejection. In contrast, avoidant persons’ deactivating strategies foster the suppression of attachment-related worries and inhibit attachment-system activation during encounters with attachment-related threats of separation and rejection.

B. STUDIES ON ATTACHMENT-FIGURE AVAILABILITY Two lines of research provide empirical support for our ideas about the appraisal of attachment-figure availability and the activation or launching of security-based strategies. One line of research has examined the contribution of attachment strategies to the subjective appraisal of the availability and supportiveness of attachment figures. The other line of research has examined effects of the actual presence of an available attachment figure or contextual cues that activate representations of available attachment figures on the enactment of security-based strategies. Overall, these studies clearly indicate that the activation of these strategies depends on the interaction between perceptions of reality and a person’s habitual attachment strategies. The first line of research has been focused mainly on attachment-style differences in the mental representations of the primary caregivers (parents). These studies have consistently shown that both attachment anxiety and attachment avoidance are involved in the appraisal of these figures as unavailable and nonsupportive. Using a 37-item adjective checklist to retrospectively assess recollections of childhood relationships with parents, Hazan and Shaver (1987) found that whereas participants who endorsed a secure style described parents as respectful, responsive, caring, accepting, and undemanding, anxious and avoidant participants described almost the opposite of this profile. These results have been conceptually replicated by Mikulincer et al. (1990) and Mikulincer and Nachshon (1991), who used a 12-item adjective checklist in samples of Israeli adults, as well as Rothbard and Shaver (1994), who assessed 15 aspects of the relationship with parents during childhood. Rothbard and Shaver (1994) also reported conceptually similar attachment-style variations in the appraisal of current relationships with parents. Secure individuals described their parents at the time the study was conducted as more available, supportive, warm, and respectful than anxious and avoidant participants. Similar attachment-style differences have been found in subsequent studies based on dimensional measures of attachment style and other techniques for assessing mental representations of attachment figures. In a nationally representative sample of American adults, Mickelson, Kessler, and Shaver (1997) reported that attachment anxiety and avoidance were

THE ATTACHMENT BEHAVIORAL SYSTEM

93

associated with memories of less parental warmth during childhood, as assessed by the Parental Bonding Scale (Parker, Tupling, & Brown, 1979). Moreover, Brennan and Shaver (1998) reported that attachment insecurity was associated with memories of less parental acceptance and less parental fostering of independence, as assessed by the Mother-Father-Peer Scale (Ricks, 1985). In an analysis of the content and structure of participants’ openended descriptions of their parents, Levy, Blatt, and Shaver (1998) found that parental descriptions of participants who reported having a secure style were characterized by greater benevolence, nonpunitiveness, differentiation, and elaboration than the descriptions given by insecure participants. The contribution of attachment strategies to the appraisal of attachmentfigure availability has also been documented in descriptions of other attachment figures. Florian, Mikulincer, and Bucholtz (1995) reported that people who endorsed a secure style perceived their close friends and romantic partner as providing more emotional and instrumental support than anxious and avoidant persons. In a recent diary study in which married participants rated the quality of their spouse’s behavior each day for a period of 3 weeks, Mikulincer, Florian, and Hirschberger (2002) reported that both attachment anxiety and attachment avoidance were associated with a daily appraisal of partners’ behaviors as reflecting greater unavailability and less supportiveness. The second line of research has provided important information about the contribution of contextual cues to the activation of security-based strategies. In a recent series of studies (Mikulincer, Hirschberger, Nachmias, & Gillath, 2001; Mikulincer, Gillath, et al., 2001; Mikulincer, Gillath, et al., in press; Mikulincer & Shaver, 2001), we found that different experimental techniques that activate mental representations of available attachment figures led to the activation of security-based strategies even among participants who scored high on attachment anxiety or attachment avoidance. These techniques consisted of subliminal presentation of security-related words such as love and proximity; subliminal presentation of security-related pictures (e.g., a Picasso drawing of a mother cradling an infant in her arms; a couple holding hands and gazing into each other’s eyes); subliminal presentation of the names of persons who were nominated by participants as security-enhancing attachment figures; guided imagery concerning the components of the secure base script; and visualization of the faces of security-enhancing attachment figures. The activation of securitybased strategies was manifested in affective ratings of neutral stimuli, responses to out-group members, attitudes toward others’ needs, and value priorities. (The specific findings are reviewed in the next section.) Two recent studies have also examined the effects of the actual presence of an available attachment figure on the activation of security-based strategies.

94

MIKULINCER AND SHAVER

In the diary study of marital interactions, Mikulincer, Florian, and Hirschberger (2002) found that a spouse’s contextual availability and supportiveness contributed to the intensity of a person’s positive feelings toward the spouse. The more often a spouse’s behaviors indicated availability and supportiveness on a given day, the stronger a person’s positive feelings toward this person on the same day. This effect was found in both husbands and wives. Importantly, contextual spouse-availability also weakened the effects of anxious individuals’ hyperactivating strategies. Although attachment anxiety was associated with less positive feelings toward a spouse during the 3-week period of the study, this association was notably weakened on days when the spouse’s behavior reflected availability and supportiveness. However, contextual spouse-availability failed to moderate the effects of attachment avoidance. That is, attachment avoidance was associated with less positive feelings toward a spouse even on days when this spouse behaved in a supportive manner. Rom and Mikulincer (in press) conducted a naturalistic study in which new recruits to the Israel Defense Forces were randomly divided into small groups and performed three group tasks. After these tasks, external observers rated participants’ instrumental functioning (the extent to which they contributed to the completion of the group task) and participants completed Smith, Murphy, and Coats’s (1999) group attachment scale, tapping attachment anxiety and avoidance toward their specific group. In addition, they rated the level of cohesiveness of their group, a group-level construct that we conceptualize as reflecting the extent to which a group serves safe haven and secure base functions. That is, high cohesive groups (assessed in terms of the average of all group members’ ratings) can contextually activate representations of attachment security and initiate security-based strategies. Scores on general attachment anxiety and attachment avoidance were associated with poorer instrumental functioning across the groups. In addition, group cohesion contributed to instrumental functioning (the higher the cohesion, the better the instrumental functioning) and to the within-group association between attachment anxiety and instrumental functioning. The higher the cohesion of a group, the weaker the association between a participant’s attachment anxiety and his or her instrumental functioning. Interestingly, group cohesion did not moderate the negative association between attachment avoidance and instrumental functioning. There was an interesting and coherent pattern of associations between general attachment scores and group attachment scores. With regard to group attachment anxiety, people who scored higher on general attachment anxiety tended also to score higher on group-specific attachment anxiety. In addition, group cohesion contributed to group attachment anxiety—the

THE ATTACHMENT BEHAVIORAL SYSTEM

95

higher the cohesion of a group, the lower the group attachment anxiety— and to the within-group association between general attachment anxiety and group attachment anxiety. The higher the cohesion of a group, the weaker the association between general attachment anxiety and group attachment anxiety. With regard to group attachment avoidance, people who scored higher on general attachment avoidance tended also to score higher on group-specific attachment avoidance. In addition, group cohesion contributed to group attachment avoidance: The higher the cohesion of a group, the lower the group attachment avoidance. However, group cohesion did not contribute to the within-group association between general avoidance and group avoidance. Overall, these studies have important implications for understanding the interface between the biasing action of chronic attachment strategies and the effects of contextual cues of attachment-figure availability on the activation of security-based strategies. Both top-down and bottom-up processes contribute to the triggering of these strategies. That is, chronic attachment-style differences in the appraisal of attachment-figure availability and contextual cues indicating the actual presence of an available attachment figure, or activation of mental representations of these figures, made unique contributions to a people’s cognitions, emotions, and functioning in a particular situation. Moreover, the actual presence of available attachment figures seemed to weaken the effects of anxious individuals’ hyperactivating strategies and to inhibit the activation of these strategies in a particular context (e.g., causing an anxious person to feel less anxious than usual toward a particular group). Interestingly, the contextual activation of attachment-figure availability failed to moderate the effects of avoidant people’s deactivating strategies. That is, deactivating strategies may continue to be used even when actual or symbolic attachment figures are available and supportive. In fact, one of the inhibitory circuits of deactivating strategies is directed against the monitoring of attachment-figure availability, leading avoidant persons to divert attention away from attachment-related cues and dismiss any positive behavior of attachment figures. As a result, these people remain suspicious about others’ intentions even when particular others are behaving in a supportive manner.

C. STUDIES ON THE INTERPERSONAL AND INTRAPERSONAL MANIFESTATIONS OF ATTACHMENT STRATEGIES In this section, we review our research findings on the interpersonal and intrapersonal manifestations of attachment strategies. We organize this

96

MIKULINCER AND SHAVER

extensive review in terms of the psychological processes in which the declarative knowledge and procedural rules associated with particular attachment strategies are manifested: the management of interpersonal behavior, the process of coping with attachment-unrelated and attachmentrelated threats, and the processing of threat-related and attachment-related information.

1. Management of Interpersonal Behavior and Close Relationships Our studies have documented the attachment strategies that are used and applied in the management of interpersonal behavior. Specifically, these studies have focused on attachment-style differences in (1) the way people construe their love experiences and beliefs, (2) the way they manage interpersonal conflicts in close relationships, (3) their proneness to disclose and share personal information and feelings (self-disclosure), and (4) their degree of trust in love relationships. In the original studies of adult attachment style, Hazan and Shaver (1987) provided initial evidence for the involvement of attachment strategies in the way people construe their experiences of romantic love. In both a newspaper survey of community members and a more typical study of university students, Hazan and Shaver (1987) found that people who classified themselves as securely attached conceptualized their love experiences in terms of the optimistic expectations and main goal of security-based strategies—the formation of intimate and supportive relationships. Specifically, they reported that their love relationships were friendly, warm, trusting, and supportive; they emphasized intimacy as the core feature of these relationships; and they believed in the existence of romantic love and the possibility of maintaining intense love over a long time period. People with an avoidant style construed their love experiences in line with their characteristic deactivating strategies. They described their romantic relationships as low in warmth, lacking friendly interactions, and low in emotional involvement; and they believed that love fades with time. In contrast, people who reported having an anxious style construed their love experiences in line with three hyperactivating strategies: (1) the search for extremely close relationships, (2) exaggerated monitoring of attachmentfigure unavailability, and (3) intensification of negative emotions. These people described their romantic relationships in terms of obsession and passion, strong physical attraction, desire for union with the partner, and proneness to fall in love quickly and perhaps indiscriminately. At the same time, they characterized their lovers as untrustworthy and nonsupportive, and reported intense bouts of jealousy and anger toward romantic partners

THE ATTACHMENT BEHAVIORAL SYSTEM

97

as well as worries about rejection and abandonment. These findings have been replicated in subsequent studies (see Feeney, 1999, for a review). In one recent study, Mikulincer and Sharir (2002) examined attachmentstyle differences in conflict resolution strategies (assessed by the Rahim Organizational Conflict Inventory) within romantic relationships. The constructive nature of security-based strategies was manifested in the tendency of secure people to report heightened reliance on effective conflict resolution strategies—compromising and integrating one’s own and one’s partner’s positions. In contrast, persons who reported having an anxious or avoidant style relied more extensively on less effective conflict resolution strategies (dominating and avoiding strategies), which leave a conflict unsolved and may even lead to conflict escalation. Whereas anxious people’s hyperactivating strategies led them to intensify the conflict and fight aggressively with a partner (dominating strategy), avoidant people’s deactivating strategies led them to distance themselves from the conflictual situation and avoid confronting the partner (avoiding strategy). These differences fit well with findings from other studies (e.g., Gaines et al., 1997; Pistole, 1989). In a series of three studies, Mikulncer and Nachshon (1991) delineated the way attachment strategies shape a person’s self-disclosure and his or her reactions to a partner’s disclosure. The main goal of security-based strategies—formation of intimate relationships—seemed to lead to what Mikulincer and Nachshon (1991) called ‘‘responsive self-disclosure.’’ Secure people were more likely to self-disclose and to be highly responsive to a partner’s disclosure. They disclosed more personal information and felt better interacting with a high than a low disclosing partner; they were attentive to the issues raised in the partner’s disclosure and expanded upon them in their own discourse (topical reciprocity). According to Berg (1987), both self-disclosure and responsiveness to partner’s disclosure are necessary for enhancing intimacy. People who wish to move toward comfortable intimacy should be responsive to a partner’s communication, reinforce the partner’s confidence in the self’s good intentions, and promote more intimate disclosure. The ‘‘responsive self-disclosure’’ exhibited by secure individuals is the best strategy for forming the kind of intimate relationships these people wish to create. The pattern of self-disclosure shown by avoidant people was a reflection of their deactivating strategies. Because their interaction goal is to maintain distance from others, persons who endorsed an avoidant style reported low proneness to self-disclose, were uncomfortable with a high disclosing partner, and did not disclose personal information even to such a partner. This interaction goal leads to what Mikulincer and Nachshon (1991) called ‘‘compulsive closure,’’ which reflects their reluctance to create open and friendly relationships with others.

98

MIKULINCER AND SHAVER

For anxious, hyperactivating individuals, what is most important about self-disclosure is that it involves a momentary dissolving of boundaries with a partner and meets a need for increased closeness (Berg, 1987). Anxious people can use self-disclosure as a means of merging with others and reducing their fear of rejection rather than as a means for enhancing reciprocal intimacy. As a result, although anxious people were found to report heightened proneness toward self-disclosure, they tended to disclose indiscriminately to people who were not prepared for intensely intimate interactions (strangers, low disclosing partners) and tended to be unresponsive to their partner’s disclosure (a pattern reminiscent of the mothers of anxious infants; Cassidy & Berlin, 1994; Main et al., 1985). In fact, anxious people did not usually deal with a partner’s disclosed information in their own disclosures, thereby violating Grice’s (1989) maxims of good communication, one of the foundations of the AAI scoring system (Hesse, 1999). These attachment-style differences have been conceptually replicated in other studies (e.g., Keelan, Dion, & Dion, 1998; Pistole, 1993). In a further attempt to delineate the interpersonal components of attachment strategies, Mikulincer (1998a) conducted five studies focused on goals, memories, and experiences related to the sense of trust in love relationships. With regard to trust-related goals, participants who endorsed a secure style reacted in line with the main goal of security-based strategies—proximity seeking. They reported intimacy as the most important trust-related goal, emphasized the impact of trust-validation and trust-violation episodes on their sense of closeness to a partner, and exhibited heightened access to thoughts about intimacy in a trust-related context (faster lexical decision times for intimacy-related words). Anxious people reacted in line with the main goal of hyperactivating strategies—the attainment of a sense of security. They reported security seeking as the most important trust-related goal and showed heightened access to thoughts about security in a trust-related context. For them, episodes in which partners behaved in a responsive way were appraised as contributing to security feelings, whereas betrayal of trust was appraised as hurting these feelings. Avoidant people reacted in line with the main goal of deactivating strategies—attainment of a sense of control. They described control seeking as the most important trust-related goal and exhibited heightened access to thoughts about control in a trust-related context. In addition, they emphasized the impact of trust-validation and trust-violation episodes on the control they exert over partners’ behaviors. Trust-related memories and experiences were also found to be direct manifestations of attachment strategies (Mikulincer, 1998a). The constructive nature of security-based strategies was evident in the following findings: Secure persons had faster access to trust-validation episodes (i.e., faster

THE ATTACHMENT BEHAVIORAL SYSTEM

99

recall latency), reported more trust-validation episodes in their current relationship, attached higher importance to these episodes, and appraised them as reflecting their partner’s beneficent disposition. The most direct expressions of this constructive attitude were found, however, in secure people’s tendency to talk openly with a partner in response to trust betrayal, their ready access to thoughts about this relationship-enhancing strategy in a trust-violation context (faster lexical decision times for the word ‘‘talk’’), and the way these persons cognitively process trust-violation episodes (Mikulincer, 1998a). Specifically, secure persons did not have ready access to trust-violation memories and tended not to appraise these episodes as important or fundamentally related to their partner’s personality. These findings reveal some of the cognitive and affective maneuvers used by secure individuals to minimize the distressing impact of a partner’s negative behaviors. Such maneuvers include the suppression of memories that raise doubts about a partner’s good will and the dismissal of trust-violation events. These reactions fit with Collins’s (1996) findings on the relationshipenhancing explanations that secure people make for a partner’s negative behaviors and suggest that attachment security is related to forgiveness, a topic receiving attention in current research (e.g., McCullough, 2001). The secondary attachment strategies of insecure individuals were evident in their reactions to trust-violation episodes (Mikulincer, 1998a). Avoidant people increased their distance from partners following a betrayal of trust and dismissed the importance of this threatening occurrence. Anxious people, in contrast, worried and ruminated during a trust-betrayal episode, reacted to it with strong negative emotion, and attached high importance to it. Interestingly, lexical decision times indicated that avoidant participants experienced activation of the word ‘‘worry’’ when primed with a trustviolation context. This discrepancy between, on the one hand, self-reports of distancing from the partner and not being bothered by betrayal and, on the other hand, experiencing an activation of ‘‘worry’’ may indicate the fragile nature of avoidant people’s deactivating strategies. In fact, when responses to a trust-violation context were assessed without the mediation of deliberate, controlled processing, avoidant people exhibited a more anxious pattern of reaction, a result found in some of our other studies as well (e.g., Berant, Mikulincer, & Florian, 2001a,b). 2. Process of Coping with Stressful Events The role of attachment strategies in coping with threats has been documented in a number of studies of a wide variety of stressful events— e.g., Iraqi Scud missile attacks on Israeli cities during the Gulf War

100

MIKULINCER AND SHAVER

(Mikulincer, Florian, & Weller, 1993), a demanding combat training exercise (Mikulincer & Florian, 1995), the process of divorce (Birnbaum, Orr, Mikulincer, & Florian, 1997), and the experience of chronic back pain (Mikulincer & Florian, 1998). These studies have also focused on the way women react to motherhood tasks, including first-time pregnancy (Mikulincer & Florian, 1999a), the birth of a first child (Mikulincer & Florian, 1998), and the birth of an infant who suffers from a congenital heart defect (CHD; Berant et al., 2001a,b). Findings concerning the appraisal of stressful events have revealed the core negative beliefs that are part of hyperactivating strategies as well as the optimistic beliefs associated with security-based strategies. Across the various stressors, anxiously attached people were found to appraise events in highly threatening terms and to hold the most serious doubts about their abilities and skills for dealing with them. In contrast, securely attached people reported the lowest levels of threat appraisal and the highest levels of perceived self-efficacy. With regard to avoidant persons, the findings are less consistent and seem to depend on contextual characteristics of the threat. For example, Berant et al. (2001a) found that avoidant mothers of either healthy infants or infants with a mild CHD held optimistic appraisals of motherhood tasks. However, avoidant mothers of infants with a severe, lifethreatening CHD appraised their plight as more threatening and viewed themselves as less able to deal with it than secure mothers. With regard to participants’ reports of coping strategies, the findings have consistently documented the expected procedural rules of attachment strategies. First, people with a secure style rely on two procedures referred to in the secure base script—support seeking and constructive attempts to solve the problem (problem-focused coping). Second, people with an anxious style score high on measures of emotion-focused coping, which is known to intensify rather than diminish the experience of distress. Third, people with an avoidant style score high on measures of distancing coping, which involves cognitive and behavioral withdrawal from the source of distress. These attachment-style differences are compatible with findings from other studies (e.g., Lussier, Sabourin, & Turgeon, 1997; Torquati & Vazsonyi, 1999). Two additional studies provided further evidence concerning secure people’s reliance on others’ support for coping with threats. First, Florian, Mikulincer, and Bucholtz (1995) found that secure individuals (compared with anxious and avoidant individuals) have a stronger tendency to seek instrumental and emotional support in times of need from parents, close friends, and a romantic partner. This finding has been replicated in other laboratories (e.g., Larose, Bernier, Soucy, & Duchesne, 1999; Ognibene & Collins, 1998). Second, Mikulincer and Florian (1997) found that secure

THE ATTACHMENT BEHAVIORAL SYSTEM

101

people were more likely than their insecure counterparts to benefit from supportive interactions while coping with stressors. Following an encounter with a distress-eliciting event (the anticipation of handling a snake), both an emotional conversation (talking about the emotions elicited by the snake) and an instrumental conversation (taking about how to deal with the snake task) were effective in restoring the emotional equanimity of secure persons. However, these techniques failed to improve the affective state of insecure persons. In fact, whereas an emotional conversation with a partner worsened avoidant individuals’ affective state, an instrumental conversation worsened the state of anxious individuals. It is important to note that characteristics of the stressful events affect attachment-style differences in coping strategies. For example, Berant et al. (2001a) found that the habitual reliance of avoidant mothers on distancing coping was observed only when their infants were healthy or had a mild CHD. However, the reported frequency of this strategy was reduced when avoidant mothers had to face an infant’s severe health condition. In such cases, these mothers, like anxious mothers, reported heightened reliance on emotion-focused coping. This finding implies that avoidant individuals’ selfreliant facade seems to be sufficient when dealing with daily hassles or minor stressors. But when they encounter uncontrollable and persisting stressors, the facade may crumble, revealing a basic insecurity that favors ineffective coping techniques. Berant et al. (2001a) also found that whereas secure mothers of both healthy infants and infants with mild CHD relied on support-seeking strategies, secure mothers of infants having a severe CHD relied on both support seeking and distancing strategies. This finding implies that secure mothers can flexibly employ distancing coping whenever thoughts about the stressful condition can impair functioning. It is possible that the suppression of painful thoughts about the infant’s severe CHD might have allowed them to maintain a positive appraisal of motherhood. As a result, the overwhelming demands of the infant’s illness may not have discouraged secure mothers and they may have been able to mobilize internal and external resources for taking caring of a vulnerable baby. Attachment-related strategies were also evident in the retrospective accounts of the experience of captivity of Israeli ex-prisoners (POWs) of the Yom Kippur War, assessed 18 years after the war (Solomon, Ginzburg, Mikulincer, Neria, & Ohry, 1998). A content analysis of these accounts revealed that securely attached ex-POWs reported having dealt with captivity by recruiting positive memories or creating positive imaginary encounters with significant others. That is, they coped with the threat by seeking symbolic proximity to, and comfort from, internalized attachment figures. In contrast, anxious persons mainly remembered suffering and pain,

102

MIKULINCER AND SHAVER

and their accounts were full of feelings of helplessness, abandonment, and loss of control, which corresponded to their hyperactivation of threatrelated feelings. The narratives of avoidant ex-POWs were shaped by deactivating strategies—they were emotionally shallow and provided little information about emotional experiences during captivity. 3. Rumination Two of our studies have documented the ruminative aspect of anxious individuals’ hyperactivating strategies during manipulated failure experiences (Mikulincer & Florian, 1998; Kogot & Mikulincer, 2002). In these studies, participants performed cognitive tasks (e.g., concept learning, probability learning) and received either failure feedback or no feedback. They then reported on the frequency of task-related, ruminative worries experienced during the experiment. Across the two studies, the experience of failure led to more frequent bursts of task-related worries (compared with the no feedback condition) mainly for people who scored high on attachment anxiety. Kogot and Mikulincer (2002) also found that anxious individuals experienced more ruminative worries even after receiving no feedback. The mere demand to solve a cognitive problem triggered their hyperactivating strategies and caused them to ruminate and worry. 4. Experience and Management of Death Anxiety While examining the manifestations of attachment strategies in the process of coping with threats, we have devoted attention to the experience and management of the threat of death, something that must be confronted by all human beings with at least normal intelligence. One line of research examined attachment-style differences in the strength of death concerns, while assessing reports of overt fear of death (Florian & Mikulincer, 1998; Mikulincer, Florian, & Tolmacz, 1990), unconscious or preconscious expressions of this fear (responses to projective death-related TAT cards; Mikulincer et al., 1990), or the accessibility of death-related thoughts (the number of death-related words a person completed in the word completion task of Greenberg et al., 1994; Mikulincer & Florian, 2000, Study 2; Mikulincer, Florian Birnbaum, & Malishkowitz, 2002). Overall, the findings reveal some of the procedural rules of hyperactivating and deactivating strategies. Anxious individuals were found to hyperactivate death concerns and keep death-related thoughts active in working memory. Specifically, attachment anxiety was associated with heightened fear of death at both conscious and unconscious levels as well as heightened accessibility of death-related thoughts even when no death

THE ATTACHMENT BEHAVIORAL SYSTEM

103

reminder was present. Avoidant individuals’ defensive attempts to suppress death-related thoughts were manifested in dissociation between their conscious claims and unconscious dynamics. Whereas attachment avoidance was related to low levels self-reported fear of death, it was also related to heightened death-related anxiety on a projective TAT measure. Attachment-style differences have also been found in the meanings people attribute to fear of death (Florian & Mikulincer, 1998; Mikulincer et al., 1990). Whereas attachment-anxious people tend to attribute this fear to the potential loss of social identity after death (e.g., ‘‘People will forget me’’), avoidant people tend to attribute it to the unknown nature of the hereafter (e.g., ‘‘Uncertainty about what to expect’’). In our view, these different meanings reflect the underlying action of attachment strategies. Anxious people hyperactivate worries about rejection and attachment-figure unavailability, viewing death as yet another relational setting in which they may be abandoned and forgotten. Avoidant deactivation of the attachment system and the consequent search for self-reliance leads to fear of the uncertain and unknown aspects of death that threaten a sense of control and mastery. That is, insecure people fear death for the same reasons they are distressed in attachment contexts (e.g., rejection, loss of control). The second line of research examined attachment-style differences in the way people manage the ‘‘terror’’ (according to terror management theory; Greenberg et al., 1997) produced by experimentally presented reminders of death. People with a secure attachment style have been found to rely on two security-based strategies—constructive transformation of threats and the seeking of proximity. Mikulincer and Florian (2000, Study 4) reported that secure people reacted to mortality salience with heightened seeking of symbolic immortality—a transformational, constructive strategy that, although not solving the unsolvable problem of death, leads a person to invest in his or her children’s care and to engage in creative, growthoriented activities whose products will live on after the person dies. That is, more secure individuals react to mortality salience by, for example, viewing procreation and parenting as important goals and by viewing their work and creative products as an important contribution to the human enterprise. Secure people have also been found to react to mortality salience with a heightened desire for intimacy in close relationships (Mikulincer & Florian, 2000, Study 5) and heightened willingness to engage in social interactions (Taubman Ben-Ari, Findler, & Mikulincer, 2002). In contrast, people with an anxious or avoidant style did not exhibit an increase in the desire for proximity or symbolic immortality following death reminders (Mikulincer & Florian, 2000; Taubman Ben-Ari et al., 2002). Rather, Mikulincer and Florian (2000, Study 1) found that insecure individuals reacted to a mortality salience induction with more severe

104

MIKULINCER AND SHAVER

judgments and punishments of moral transgressors. It seems that insecurely attached persons relied on what Greenberg et al. (1997) labeled ‘‘culturally derived defenses’’—adherence to a cultural worldview and defensive enhancement of self-esteem. For anxious persons, adhering to a shared cultural worldview may be a way to gain greater love and acceptance from members of their group. For avoidant persons, the major issues may be selfreliance and control, which benefit from the defensive enhancement of selfesteem. Integrating the entire set of findings, we see the following patterns. Secure individuals tend to rely on security-based strategies when dealing with death awareness, and these strategies seem to be highly effective in buffering death anxiety at both conscious and unconscious levels. Avoidant individuals tend to enhance their self-esteem in response to death reminders, which seems to be effective in buffering fear of death at a conscious level. However, this deactivating strategy seems to be ineffective in decreasing the unconscious fear of death and leaves a person concerned about death’s unknown and uncontrollable nature. Anxious individuals tend to bolster and express their cultural worldview as a symbolic defense against death anxiety, in the hope of gaining greater love and acceptance. But their proclivity for attachmentsystem hyperactivation sets in motion processes such as rumination on death and separation that make adherence to cultural worldviews ineffective in buffering death anxiety. 5. Management and Cognitive Processing of Attachment-Related Threats We have also examined manifestations of attachment strategies in the management and cognitive processing of attachment-related threats (e.g., separation). In a field study, Fraley and Shaver (1998) obtained evidence for the underlying action of these strategies in the behavioral reactions of couple members separating from each other in an airport. On the one hand, attachment avoidance was found to be associated with less frequent contact seeking and contact maintenance, and more frequent avoidance behavior (turning away, looking elsewhere). On the other hand, attachment anxiety was found to be associated with more intense nonverbal display of sadness and distress during the separation episode. In two laboratory studies, Fraley and Shaver (1997) used a thought suppression paradigm to examine attachment-style differences in the ability to suppress separation-related thoughts. Participants wrote continuously about whatever thoughts and feelings they were experiencing while being asked to suppress thoughts about a romantic partner leaving them for someone else. In the first study, the ability to suppress these thoughts was assessed by the number of times separation-related thoughts appeared in

THE ATTACHMENT BEHAVIORAL SYSTEM

105

participants’ stream-of-consciousness writing following the suppression period. In the second study, this ability was assessed by the level of participants’ physiological arousal (skin conductance level) during the suppression task—the lower the arousal, the higher the ability to suppress distress-eliciting thoughts about separation. The findings reflected the operation of secondary attachment strategies in the processing of separation-related thoughts. Specifically, attachment avoidance was found to be associated with greater ability to suppress separation-related thoughts—indicated by less frequent thoughts following the suppression task and lower skin conductance during the suppression task. In contrast, attachment anxiety was associated with poorer ability to suppress separation-related thoughts—more frequent thoughts following the suppression task and higher skin conductance during the suppression task. This finding is a direct indication that anxious individuals cannot ignore or minimize separation-related concerns once these are made salient. Two studies conducted by Fraley, Gardner, and Shaver (2000) further enrich our knowledge of the cognitive processes underlying avoidant individuals’ deactivation of attachment-related threats. These investigators asked whether deactivating strategies act in a preemptive manner—e.g., by deployment of attention away from, and shallow encoding of, attachmentrelated threats—or postemptive manner: repression and forgetting of material that has already been encoded. Participants listened to an interview about attachment-related threats and were later asked to recall details from the interview either immediately (Study 1) or at various delays ranging from half an hour to 21 days (Study 2). An analysis of forgetting curves over time revealed that (1) persons scoring high in attachment avoidance initially encoded less information about the interview than persons scoring low on this dimension, and (2) the two groups forgot the information they encoded at the same rate. These findings imply that avoidant persons’ deactivating strategies at least sometimes act in a preemptive manner—holding distressing attachment-related material out of awareness and memory right from the start. A recent series of three studies conducted by Mikulincer et al. (2002) documented an interesting procedural rule of anxious individuals’ hyperactivating strategies in the processing of attachment-related threats. Across studies, participants were asked to imagine being separated from a loved partner (separation reminder) and then to perform a word completion task that tapped the accessibility of death-related thoughts. Findings indicated that people high on attachment anxiety reacted to separation reminders with heightened accessibility of death-related thoughts. When given partial words and asked to complete them, anxious individuals in a separation condition produced more death-related words. This tendency was particularly strong

106

MIKULINCER AND SHAVER

when the imagined separation was long-lasting or final (Study 3) or when it involved a romantic partner (Study 2). In other words, for anxious individuals, separation reminded them of death, which is very much in line with Bowlby’s (1982/1969) original idea that the primary function of the attachment system is to protect a vulnerable person from death. 6. Experience and Management of Anger In a series of three studies, Mikulincer (1998b) examined attachment-style differences in the way people experience and manage anger-eliciting relationship episodes. Specifically, these studies examined personal recollections of feelings, goals, and responses during the arousal of anger; reactions to hypothetical anger-eliciting scenarios differing in the intentions of a romantic partner (hostile, ambiguous, nonhostile); and expectations of the responses of a partner to one’s expression of anger (e.g., he/she will accept me, he/she will leave me). Across the three studies, secure persons’ experience of anger corresponded with the declarative and procedural rules of security-based strategies. First, secure individuals held optimistic expectations of partner’s responses during anger episodes and made well-differentiated, reality-attuned appraisals of their partner’s intentions during these episodes. They tended to attribute hostility to another person and to react to him or her with anger feelings only when there were clear contextual cues about a hostile intent. Second, secure persons’ recollections of anger-eliciting episodes reflected functional, constructive attempts to rectify an undesirable relationship problem [what Bowlby (1973) called the ‘‘anger of hope’’]. Specifically, secure persons tended to adopt constructive goals aimed at repairing the relationship with the instigator of anger, to engage in adaptive problem solving, to express anger outward in a controlled and nonhostile way, and to experience more positive than negative affect following anger episodes. [Incidentally, these findings are compatible with Gottman’s (1994) findings from marital communication research, which showed that anger per se is not a predictor of relationship malfunction or subsequent dissolution.] Anxious people’s anger experiences were quite different, and fit with the declarative and procedural knowledge associated with hyperactivating strategies. First, anxious individuals held negative expectations of partner’s responses during anger episodes and tended to make undifferentiated, negatively biased appraisals of partner’s intentions. They attributed hostility to their partner and reacted to him or her with angry feelings even when there were ambiguous cues about hostile intent. Second, anxious persons’ recollections of anger episodes reflected hyperactivation of distress, which overwhelmed their cognitive system and drew resources away from adaptive

THE ATTACHMENT BEHAVIORAL SYSTEM

107

coping. Specifically, anxious individuals reported higher proneness to anger, experienced uncontrollable access to anger feelings, ruminated excessively on these feelings, and experienced more negative than positive affect following anger episodes. Avoidant people’s deactivating strategies seemed to produce what Mikulincer (1998b) called ‘‘dissociated anger.’’ Although these individuals did not report intense anger, they reported heightened hostility and exhibited intense physiological signs of arousal during anger episodes, and displayed an undifferentiated tendency to attribute hostility to a partner even when there were clear contextual cues about a partner’s nonhostile intent. This disassociated stance was also manifested in avoidant individuals’ high scores on a measure of escapist methods for dealing with anger. This pattern of findings fits Rholes, Simpson, and Orina’s (1999) observations that people scoring high on attachment avoidance displayed especially hostile behavioral responses toward a distressed romantic partner in a laboratory threat context.

7. Cognitive Consequences of Negative Affect In two experiments, Pereg and Mikulincer (2002) documented the involvement of attachment-related strategies in shaping the cognitive consequences of negative affect. In both studies, participants were randomly assigned to a negative affect condition (reading an article about a car accident) or a neutral affect condition (reading about how to construct something using a hobby kit). Following this affect induction, incidental recall or causal attributions were assessed. In the first study, participants read a booklet with positive and negative headlines, and then, without prior warning, were asked to recall as many of the headlines as possible. In the second study, participants were asked to list the causes of a hypothetical negative relationship event (‘‘your partner disclosed something you asked him to keep secret’’). The ‘‘building’’ feature of security-based strategies was manifested in a mood-incongruent pattern of cognition. Induced negative affect, as compared with a neutral condition, led participants who scored low on both attachment anxiety and avoidance to recall more positive headlines and fewer negative headlines and to attribute a negative event to less global and stable causes. Such security-based strategies are likely to inhibit the spread of negative affect throughout the working memory and to activate competing positive cognitions (positive headlines, attributions that maintain a positive view of one’s partner). These cognitions work against the pervasive effects of negative affect and facilitate mood repair.

108

MIKULINCER AND SHAVER

Anxious, hyperactivating strategies were manifested in mood-congruent cognitions. Induced negative affect, as compared with a neutral condition, led participants who scored high on attachment anxiety to recall fewer positive headlines and more negative headlines and to attribute a negative event to more global and stables causes. Hyperactivating strategies favor the spread of negative affect throughout the working memory and facilitate the processing of congruent negative cognitions (negative headlines, attributions that elicit doubts about one’s partner). These negative cognitions can exacerbate negative mood and contribute to continued activation of the attachment system. Avoidant persons’ deactivating strategies weakened the links between negative affect and cognitions. People who scored high on avoidance showed no notable effect of induced negative affect on recall or causal attributions. Deactivating strategies seem to block the acknowledgment of negative affect and prevent the use of inner-state information in cognitive processing, thereby weakening the power of negative affect to influence cognitions. In this way, deactivating strategies distance people from their own emotional world and facilitate the goal of attachment-system deactivation.

8. Cognitive Activation and Architecture of Emotional Memories Mikulincer and Orbach (1995) examined the procedural rules of attachment strategies that organize threat-related material within the associative memory network. Participants were asked to recall early experiences in which they felt anger, sadness, anxiety, or happiness, and the time for retrieving a memory was taken as a measure of cognitive accessibility. Participants also rated the intensity of dominant and nondominant emotions in each recalled event. In the memory task, people with an avoidant style exhibited the lowest accessibility (highest recall time) of memories of sadness and anxiety, people with an anxious style showed the greatest access to these painful memories, and people with a secure style fell in between. Furthermore, whereas secure individuals took more time to retrieve negative emotional memories than a happiness memory, anxious people took more time to retrieve a happiness memory than negative emotional memories. In the emotion-rating task, avoidant individuals rated dominant emotions (e.g., sadness when retrieving a sad memory) and nondominant emotions (e.g., anger when retrieving a sad memory) as less intense than secure individuals, whereas anxious individuals reported experiencing very intense dominant and nondominant emotions in the anxiety, sadness, and anger memories. Interestingly, secure

THE ATTACHMENT BEHAVIORAL SYSTEM

109

persons rated dominant emotions as far more intense than nondominant emotions. The findings provided strong support for the functional nature of security-based strategies. Secure individuals tended to acknowledge distress, have access to negative memories, and exhibit well-elaborated processing of these experiences. However, they still had better access to positive memories and tended to control the spread of negative emotions to other nondominant negative emotions. In this way, they maintained a positive cognitive context and possessed a well-differentiated emotional architecture, which probably allows them to process negative memories without being overwhelmed by distress and confounded by other negative emotions. The findings also documented the deactivating and hyperactivating strategies of insecure people. Avoidant individuals showed reduced access to negative emotional memories, and those that were recalled were psychologically shallow. In contrast, anxious individuals showed fast access to negative emotional memories and impaired control of the spread of activation from one memory with a particular negative tone to other, different negative emotions. These findings fit with our theoretical portrayal of anxious people as having an undifferentiated, chaotic emotional architecture, which makes emergence from negative emotional spirals particularly difficult.

D. STUDIES ON THE IMPLICATIONS OF ATTACHMENT STRATEGIES FOR SELF-APPRAISALS A number of our studies have documented the negative implications of anxious persons’ hyperactivating strategies for self-appraisals. Attachment anxiety has been associated with lower scores on the Rosenberg self-esteem scale (Cooper, Shaver, & Collins, 1998; Mickelson, Kessler, & Shaver, 1997; Shaver, Papalia, Clark, & Koski, 1996; Taubman Ben-Ari et al., 2002) and the endorsement and incidental recall of more negative self-relevant traits (Mikulincer, 1995). These findings are compatible with those reported in other studies (e.g., Bartholomew & Horowitz, 1991; Brennan & Morris, 1997; Griffin & Bartholomew, 1994). In a further effort to examine the implications of attachment strategies for self-appraisals, Mikulincer (1995) measured the accessibility of positive and negative self-relevant traits in a Stroop Task, the pervasiveness of affect in a task that involved sorting self-relevant traits, the level of differentiation between self-aspects in the same task, the level of integration between differentiated self-aspects, and the strength of discrepancies between selfdomains and self-standpoints as assessed by Higgins, Bond, Klein, and

110

MIKULINCER AND SHAVER

Strauman’s (1986) procedure. The findings revealed that secure individuals have ready access to both positive and negative self-attributes, possess a highly differentiated and integrated self-organization, and have relatively small discrepancies between domains and standpoints of the self. Their constructive security-based strategies seem to create a balanced, coherent, and well-organized self-view. This strategy highlights positive self-representations, consolidates boundaries between self-aspects that resist the automatic spread of negative affect, and encourage people to tolerate weak points of the self and to integrate them within the self-structure. The findings were also in line with our ideas about the cognitive operations of secondary attachment strategies. On the one hand, anxious persons were found to have access only to negative self-attributes, scored low in the differentiation and integration of self-attributes, and revealed a pervasiveness of negative affect in the sorting of these attributes. On the other hand, avoidant persons had poor access to negative selfattributes and exhibited low integration between these attributes and other self-aspects. Interestingly, both anxious and avoidant individuals had relatively large discrepancies between their self-image and self-standards (ideal-self, ought self). However, whereas anxious persons’ discrepancies resulted from their negative self-image, avoidant persons’ discrepancies resulted from high, unrealistic self-standards. It seems that avoidant people’s pursuit of a perfect and powerful self creates demanding self-standards that are discrepant even from their defensively positive self-view. This interpretation fits with results showing positive associations between attachment avoidance and traits of perfectionism and self-criticism (e.g., Rice & Mirzadeh, 2000; Zuroff & Fitzpatrick, 1995). In a series of four experimental studies, Mikulincer (1998c) provided systematic evidence that secondary attachment strategies bias self-appraisals (self-inflation, self-devaluation) as a means of dealing with threats. Participants were exposed to various experimentally induced threatening or neutral situations, and appraisals of self were measured with self-report scales and other subtler cognitive techniques, such as reaction times for trait recognition. Avoidant individuals’ deactivating strategies led them to react to threats with suppression of negative self-attributes and self-inflation. They made more explicit and implicit positive self-appraisals following threatening than neutral stimuli. In contrast, anxious individuals’ hyperactivating strategies led them to react to threats with a more intense focus on negative self-attributes and self-devaluation. They made more explicit and implicit negative self-appraisals following threatening than neutral conditions. Secure people showed no notable bias in their self-appraisals. It seems that their sense of attachment security maintains a positive, stable self-image

THE ATTACHMENT BEHAVIORAL SYSTEM

111

during the encounter with threats and renders unnecessary any defensive bias in self-appraisals. Mikulincer (1998c, Studies 3–4) also provided empirical support for our ideas that (1) avoidant people’s self-inflation bias is intended to convince other people of the strength and efficaciousness of the avoidant self, and (2) anxious people’s self-devaluation bias is intended to elicit others’ compassion and love. Specifically, findings revealed that avoidant individuals’ tendency to endorse a more positive self-view in threatening than in neutral contexts was inhibited by a message that broke the link between a positive self-view and self-reliance. The findings also revealed that anxious individuals’ tendency to endorse a less positive self-view in threatening than in neutral contexts was inhibited by a message that broke the link between self-devaluation and others’ positive responses. Insecure people’s defensively biased self-appraisals were also documented in a recent study by Kogot and Mikulincer (2002), which examined causal attribution of failure in achievement settings. In this study, anxious persons’ hyperactivating strategies led them to make what Abramson, Metalsky, and Alloy (1989) called a ‘‘hopelessness-depressive’’ pattern of attributions. Failure was attributed to more internal, stable, global, and uncontrollable causes, which, in turn, reinforced a negative view of the self. In contrast, avoidant persons’ deactivating strategies led them to make defensive, selfenhancing causal attributions. They attributed failure to less internal causes, dismissed the diagnosticity of the failure, and blamed others for it. These results fit findings reported in other studies (e.g., Kennedy, 1999; Man & Hamid, 1998).

E. STUDIES ON THE IMPLICATIONS OF ATTACHMENT STRATEGIES FOR PERSON PERCEPTION Some of our studies have documented the negative implications of secondary attachment strategies for person perception. Specifically, insecurely attached people have been found to describe relationship partners, such as parents, romantic partners, friends, and members of a group, in more negative terms than people who report having a secure style (Hazan & Shaver, 1987; Levy, Blatt, & Shaver, 1998; Mikulincer & Erev, 1991; Mikulincer & Arad, 1999; Rom & Mikulincer, in press). These findings were consistent with those reported by other researchers (e.g., Bartholomew & Horowitz, 1991; Collins, 1996; Collins & Read, 1990). In a series of six studies, Mikulincer, Orbach, and Iavnieli (1998) also provided evidence on the biases produced by secondary attachment strategies in perception of others—namely, hyperactivating strategies’

112

MIKULINCER AND SHAVER

overemphasis on self–other similarities and consensus, and deactivating strategies’ overemphasis on self–other dissimilarities and self-uniqueness. Specifically, Mikulincer et al. (1998) found that anxious people were more likely than secure people to perceive others as similar to them and to show a false consensus bias in both trait and opinion descriptions. In contrast, avoidant individuals were more likely than secure people to perceive others as dissimilar to them and to exhibit a false distinctiveness bias. Mikulincer et al. (1998) also found that anxious individuals reacted to threats by generating a self-description that was more similar to a partner’s description and by recalling more partner traits that were shared by themselves and the partner. In contrast, avoidant persons reacted to the same threats by generating a self-description that was more dissimilar to a partner’s description and by forgetting more traits that were shared by themselves and the partner. Interestingly, secure individuals’ self-descriptions and recall of partners’ traits were not affected by the encounter with threats. Following up these experiments, Mikulincer and Horesh (1999) examined the projective mechanisms that might underlie insecure people’s perceptions of others. Specifically, they examined (1) avoidant individuals’ tendency to defensively project unwanted traits of the self onto others, which increase self–other differentiation and, by comparison, enhance their sense of selfworth, and (2) anxious individuals’ tendency to project their actual-self traits onto others, which increases self–other similarity and the sense of closeness. These tendencies were examined in three two-session studies. In the first session, participants generated actual-self traits and unwanted-self traits. The second session was devoted to assessing impressions of hypothetical people, ease of retrieving memories of actual familiar people, and inferences about the learned features of hypothetical people. In each study, Mikulincer and Horesh (1999) examined the extent to which these cognitive processes were biased by participants’ self-views. The findings indeed revealed that avoidant individuals projected unwanted self-traits onto others. They were in fact likely to perceive in others the traits included in their own unwanted self, easily retrieve an example of a known person whose traits resembled those of their unwanted self, and make faulty inferences that traits taken from their unwanted self were among the features they learned about a target person whose description resembled their unwanted-self traits. It seems that avoidant individuals’ attempts to defensively exclude negative information about the self and to maintain interpersonal distance dominate their perceptions of others. The findings also provided clear-cut evidence that anxious individuals tend to project traits of their actual self onto others. They were likely to perceive in an unknown person traits that defined their actual self, easily retrieve an

THE ATTACHMENT BEHAVIORAL SYSTEM

113

example of a known person whose traits resembled their actual-self traits, and make faulty inferences that traits taken from their actual self were found among features they learned about a target person whose description resembled their actual-self traits. It seems that their intense search for connectedness dominates anxious individuals’ perception of others. Importantly, Mikulincer and Horesh (1999) found that secure persons’ representations of others seemed to be unbiased by projective mechanisms. That is, persons who reported having a secure attachment style were less prone than insecure persons to project onto others features that defined their self or that they denied having. The constructive nature of securitybased strategies may allow secure people to tolerate ambiguities and contradictions in their self-view as well as in their relationships with others, thereby reducing the threats elicited by unwanted self-aspects or by selfother discrepancies and the resulting need for psychological projection. In a recent series of five studies, Mikulincer and Shaver (2001) provided further evidence for the association between security-based strategies and a reduction of biases in representations of others. Specifically, these studies showed that security-based strategies reduce negative biases in responding to people who are different from oneself or who do not belong to one’s own social group (outgroup members). First, the findings indicated that the higher a person’s sense of chronic attachment security, the weaker his or her hostile responses to a variety of outgroups (as defined by secular Israeli Jewish students): Israeli Arabs, ultraorthodox Jews, Russian immigrants, and homosexuals. Second, different experimental priming techniques— subliminal presentation of security-related words such as love and proximity; guided imagery concerning the components of the secure base script; and visualization of the faces of security-enhancing attachment figures—that momentarily heightened the sense of attachment security were found to eliminate hostile responses to these outgroups. These effects were mediated by threat appraisal and were found even when participants were led to believe they had failed on a cognitive task or their national group had been insulted by an outgroup member. That is, a sense of attachment security reduced the threats elicited by the encounter with outgroup members and rendered unnecessary any derogative or hostile response toward them.

F. STUDIES ON THE AFFECTIVE AND ADJUSTMENT IMPLICATIONS OF ATTACHMENT STRATEGIES In this section, we review our research findings on the implications of attachment strategies for a person’s affective state. We divide this section

114

MIKULINCER AND SHAVER

into two subsections. First, we review studies that have focused mainly on a person’s transient mood and affective responses to specific events. Second, we present findings regarding attachment-style differences on more global and chronic measures of mental health, and emotional and adjustment problems. 1. Mood and Affective Responses Our studies provide empirical support for the positive contribution of security-based strategies to a person’s transient affective state. Some of our studies document associations between the sense of attachment security and mood reports. Different priming techniques (e.g., subliminal presentation of security-related words, guided imagery concerning the components of the secure base script, visualization of the faces of security-enhancing attachment figures) that momentarily heighten the sense of attachment security were found to improve participants’ mood reports during the experimental session (Mikulincer, Gillath et al., 2001; Mikulincer, Gillath et al., in press; Mikulincer & Shaver, 2001). In addition, some of these priming techniques tend to infuse even formerly neutral stimuli with positive affect without the intervention of participants’ awareness. In a series of six experiments, Mikulincer, Hirschberger, et al. (2001) reported that the subliminal presentation of security-related pictures or the names of persons who were nominated by participants as security-enhancing attachment figures in the WHOTO scale led to higher liking ratings of unknown Chinese ideographs than the subliminal presentation of neutral stimuli. Mikulincer, Hirschberger, et al. (2001) also reported that subliminal activation of the sense of attachment security led to more positive evaluations of neutral stimuli even in threat contexts and eliminated the detrimental effects that these contexts generally had on the liking of neutral stimuli. That is, whereas threat contexts lowered the liking for the ideographs, attachment-security representations counteracted this negative effect and maintained a heightened positive evaluation in these circumstances. This finding indicates the protective effect of internalized representations of attachment security. Our studies have also documented the detrimental effects of anxious persons’ hyperactivating strategies on their mood and affective responses to social interactions. Several studies have reported that the higher a person’s attachment anxiety, the more negative his or her mood during experimental sessions (e.g., Mikulincer & Florian, 1997; Mikulincer, Gillath et al., 2001; Mikulincer, Gillath, et al., in press; Mikulincer & Shaver, 2001). In a diary study in which participants completed the Rochester Interaction Record every time they had a social interaction that lasted 10 minutes or longer for

THE ATTACHMENT BEHAVIORAL SYSTEM

115

1 week, Tidwell, Reis, and Shaver (1996) found that attachment anxiety was associated with more negative affect during the reported interactions. This effect was replicated in opposite-sex partner interactions, same-sex partner interactions, and group interactions. Similarly, Rom and Mikulincer (in press) found attachment anxiety to be associated with the report of more intense negative emotions toward task-oriented groups. Some recent studies also provide initial evidence concerning the emotional detachment produced by avoidant persons’ deactivating strategies. For example, Rom and Mikulincer (2002) found that attachment avoidance was associated with less intense positive feelings toward task-oriented groups. In a diary study of marital interactions, Mikulincer, Florian, and Hirschberger (2002) found that attachment avoidance was associated with daily reports of less intense emotional responses, either negative or positive, toward a spouse. Similar findings have been reported in other studies (e.g., Pietromonaco & Feldman Barrett, 1997). 2. Mental Health and Adjustment Our research program provides important information about the implications of attachment strategies for mental health and adjustment. Data have been collected with global measures of psychological well-being and distress as well as more specific measures of emotional problems (depression, anxiety, hostility, loneliness), adjustment problems (eating disorders, substance abuse, conduct disorders), and personality disorders. In a series of studies (Berant et al., 2001a,b; Birnbaum et al., 1997; Mikulincer & Florian, 1998, 1999a; Mikulincer, Horesh, Levy-Shiff, Manovich, & Shalev, 1998), we have examined the association between attachment style and global scores of psychological well-being and distress on the Mental Health Inventory (Veit & Ware, 1983). Without exception, all of these studies have shown that secure attachment is positively related to well-being and inversely related to distress. In contrast, anxious attachment is inversely associated with well-being and positively associated with distress. These associations have been found both in community samples and in samples of people who were experiencing stressful events. They have also been found is both cross-sectional and prospective longitudinal designs. In all of these studies, avoidant attachment seems to have differential associations with mental health, depending on the presence of stressful circumstances. In community samples, weak associations have been found between attachment avoidance and mental health. However, in stressful circumstances, attachment avoidance has been strongly associated with poor mental health. For example, Berant et al. (2001b) reported that attachment avoidance of mothers of infants with CHD (measured

116

MIKULINCER AND SHAVER

immediately following the CHD diagnosis) was more strongly related to distress (measured 1 year later) than was attachment anxiety. These studies provide strong evidence for the distress-buffering effects of security-based strategies. Specifically, the encounter with stressful events has been found to heighten distress mainly among insecurely attached persons, but not among people who report a secure style. For these people, no notable difference in mental health has been found between neutral and stressful conditions. In addition, two of these studies provide support for the hypothesis that attachment-style differences in mental health result from the underlying action of attachment strategies (Birnbaum et al., 1997; Berant et al., 2001a). Structural analyses have shown that the association between secure attachment and mental health was explained by secure persons’ high appraisal of their ability to cope with stress and their reliance on support seeking. The association between attachment anxiety and distress was explained by anxious persons’ exaggerated threat appraisal and heightened reliance on emotionfocused coping. Finally, the association between avoidance and distress was explained by avoidant persons’ heightened reliance on distancing coping. Strong associations have also been found between reports of attachment style and reports of specific emotional problems. In both American and Israeli samples, symptoms of depression, anxiety, and hostility were inversely associated with the secure attachment style and positively associated with anxious and avoidant styles (Cooper et al., 1998; Mikulincer et al., 1993; Mikulincer, Horesh, Eilati, & Kotler, 1999; Mickelson et al., 1997). In addition, Hazan and Shaver (1987) and Kirkpatrick and Shaver (1990, 1992) reported that people who endorsed an avoidant or anxious style in romantic relationships or in their relationship with God reported higher levels of loneliness than people who endorsed a secure style. These attachment-style differences in emotional problems have been replicated in several studies (e.g., Carnelley, Pietromonaco, & Jaffe, 1994; Roberts, Gotlib, & Kassel, 1996). Although both anxious and avoidant people suffer from emotional problems, their reliance on hyperactivating or deactivating strategies results in different patterns of problems. For example, we have found that attachment anxiety but not attachment avoidance is related to measures of trait anxiety and neuroticism (Mikulincer Gillath, & Shaver, 2002; Mikulincer & Sheffi, 2000; Mickelson et al., 1997; Shaver & Brennan, 1992; Shaver et al., 1996). These two measures reflect heightened worries and the experience of emotional extremes and instability—core accompaniments of anxious persons’ hyperactivating strategies. In addition, Mikulincer et al. (1993) reported that anxious persons’ hyperactivating strategies led them to react to stressful events with heightened depression, anxiety, hostility, somatization, and intrusive posttraumatic symptoms. In contrast,

THE ATTACHMENT BEHAVIORAL SYSTEM

117

avoidant persons’ deactivating strategies led them to react to stressful events with heightened hostility, somatic complaints, and avoidant posttraumatic symptoms. With regard to adjustment problems, Brennan and Shaver (1995) found that both attachment anxiety and attachment avoidance were positively associated with measures of eating disorders, drinking behavior, and the use of drinking as a maladaptive means for handling distress. Furthermore, Mickelson et al. (1997) reported that both attachment anxiety and attachment avoidance were positively associated with measures of psychoactive substance use and conduct disorders. These findings were replicated by Cooper et al. (1998), who presented interesting evidence concerning the involvement of secondary attachment strategies in mediating the association between insecure attachment and maladaptive behaviors. Anxious persons’ maladaptive behaviors were mainly explained by hyperactivation of negative feelings of depression and hostility. Avoidant persons’ maladaptive behaviors were better explained by deactivating strategies that inhibit their involvement in social interactions and close relationships and impoverish their social competences and skills. These secondary attachment strategies have also been associated systematically with specific personality disorders. Brennan and Shaver (1998) reported that attachment avoidance was positively associated with the schizoid personality disorder. This disorder reflects a person’s tendency to retreat from others and avoid close relationships—a pathological result of avoidant persons’ deactivation of the attachment system. Brennan and Shaver (1998) also found that attachment anxiety was positively associated with dependent and histrionic personality disorders. The dependent disorder reflects a strong need to rely on others and the histrionic disorder reflects a tendency to exaggerate actions and emotions to attain attention from others. Both pathological tendencies seem to derive from anxious persons’ search for comfort, attention, and security.

G. STUDIES ON THE IMPLICATIONS OF ATTACHMENT STRATEGIES FOR RELATIONSHIP QUALITY The hypothesized positive association between the sense of attachment security and relationship satisfaction has been consistently documented in several of our studies (Brennan & Shaver, 1995; Hazan & Shaver, 1987, 1990; Mikulincer & Erev, 1991; Mikulincer & Florian, 1999b; Mikulincer, Horesh et al., 1998; Shaver & Brennan, 1992). These studies have focused on both dating and marital relationships and have used different self-report measures of relationship satisfaction (e.g., Relationship Rating Form,

118

MIKULINCER AND SHAVER

Dyadic Adjustment Scale). All of these studies have found that people who endorse a secure style report the highest level of relationship satisfaction and anxiously attached persons report the lowest level. This association was found in both men and women and has been replicated in a prospective longitudinal design (Shaver & Brennan, 1992). Importantly, Shaver and Brennan (1992) reported that the association between secure attachment and relationship satisfaction cannot be explained by other personality factors, such as the ‘‘big five’’ factors (McCrae & Costa, 1990). These attachmentstyle differences in relationship satisfaction have been replicated in dozens of other studies around the world (see Feeney, 1999; Koski & Shaver, 1997; and Mikulincer, Florian, Cowan, & Cowan, 2002, for extensive reviews of these studies). The sense of attachment security has also been found to contribute to other aspects of relationship quality. For example, positive associations have been found between reports of secure attachment and measures of involvement and interdependence in dating relationships (Hazan & Shaver, 1987; Mikulincer & Erev, 1991). Ratings of attachment security were significantly associated with greater commitment to a dating relationship (Mikulincer & Erev, 1991; Morgan & Shaver, 1999; Shaver & Brennan, 1992). Again, these findings have been replicated in several other studies (e.g., Bartholomew & Horowitz, 1991; Collins & Read, 1990; Levy & Davis, 1988). In a study of family dynamics, Mikulincer and Florian (1999b) found significant associations between spouses’ attachment style and their reports of marital cohesion and flexibility (FACES III). Whereas spouses with a secure style reported relatively high family cohesion and flexibility, avoidant spouses scored relatively low on these two dimensions, and anxious spouses reported high family cohesion but low family flexibility. These associations generally reflect the underlying action of attachment strategies. Secure persons’ construction of intimate relationships and the flexibility of their affective-cognitive organization were directly manifested in high levels of marital cohesion and adaptability. Avoidant persons’ construction of noncohesive and rigid relationships seems to result from their deactivating tendency to avoid both intimate interactions and flexible and open confrontation with environmental demands. The findings for anxious persons (high cohesion, low flexibility) seem to result from their compulsive search for closeness, intimacy, and connectedness. Attachment strategies have also been found to shape the quality of daily social interactions. In their diary study, Tidwell et al. (1996) asked participants to rate the intimacy, pleasure, and satisfaction they felt during social interactions. Results showed that secure participants reported more intimacy, pleasure, and satisfaction in daily interactions with opposite-sex

THE ATTACHMENT BEHAVIORAL SYSTEM

119

partners than avoidant participants. Moreover, secure participants interacted more often and for longer periods of time per partner with members of the opposite sex. Conceptually similar findings were reported by Rom and Mikulincer (in press). Participants were asked to recall task-oriented group interactions and to evaluate the quality of these interactions. Findings revealed that both attachment anxiety and attachment avoidance were significantly associated with the recall of more negative memories of group interactions.

H. STUDIES ON THE INTERPLAY BETWEEN THE ATTACHMENT SYSTEM AND OTHER BEHAVIORAL SYSTEMS 1. Exploration System Our research program offers clear-cut evidence for the hypothesized effects of the sense of attachment security on the activation of the exploration system. These effects have been documented on people’s attitudes toward work, proneness to curiosity and information search, and openness of the cognitive system to new information and changes in existing knowledge structures (cognitive openness). These effects were also illustrated in the loosening of cognitive operations and the exploration of new, unusual associations in conditions that favored a relaxed activation of the exploration system (i.e., induced positive affect). Hazan and Shaver (1990) proposed that work serves as one form of exploration in adulthood and found that secure individuals reported more positive attitudes toward work and were more satisfied with work activities than avoidant or anxious persons. Interestingly, these attitudes were found to be organized around the main goals of attachment strategies. The constructive goal of security-based strategies was directly manifested in secure persons’ perception of work as an opportunity for learning and advancement. In contrast, anxious individuals perceived work as an additional opportunity for social acceptance and approval—the main goal of their hyperactivating strategies. Avoidant individuals perceived work as an opportunity for evading close relationships—the main goal of their deactivating strategies. In a series of five studies, Mikulincer (1997) presented convincing evidence that attachment security facilitates cognitive exploration and openness to new information. First, secure persons reported more curiosityproneness than insecure persons. Similar findings were reported by Green and Campbell (2000) and Green-Hennessy and Reis (1998). Second, secure people scored lower on measures of cognitive closure, intolerance of

120

MIKULINCER AND SHAVER

ambiguity, and dogmatic thinking than insecure people. Third, secure people were less likely than their insecure counterparts to make judgments about a target person on the basis of early information and to ignore later data (primacy effect). Fourth, secure people were less likely than insecure people to make stereotype-based judgments. Specifically, insecure individuals evaluated the quality of an essay based on the supposed ethnicity of the writer rather than on exploration of the quality of the written essay: The more positive the stereotype of the writer’s ethnic group, the higher the grade assigned to the essay. In contrast, secure people were less affected by ethnic stereotypes. These attachment-style differences in cognitive openness were also examined by Mikulincer and Arad (1999), who focused on the revision of knowledge about a relationship partner following behavior on the part of the partner that seemed inconsistent with this knowledge. Compared to secure persons, both anxious and avoidant individuals displayed fewer changes in their baseline perception of the partner after being exposed to expectation-incongruent information about the partner’s behavior. They were also less able to recall the information. This finding was replicated when relationship-specific attachment orientations were assessed: The higher the attachment anxiety or avoidance toward a specific partner, the fewer the revisions people made in their perception of this partner upon receiving expectation-incongruent information (Mikulincer & Arad, 1999). Moreover, contextual heightening of the sense of attachment security (visualizing a supportive other) increased cognitive openness and led people to revise their conception of a partner based on new evidence (Mikulincer & Arad, 1999). In a series of three experiments, Mikulincer and Sheffi (2000) reported that attachment-style differences in cognitive openness can be observed even in contexts that facilitate relaxed exploration, such as subsequent to the arousal of positive affect. Participants were exposed to positive affect inductions (asking them to retrieve a happy memory or exposing them to a brief comedy film) or neutral affect conditions and their creative problemsolving performance and the breadth of their mental categorization were assessed. The typical beneficial effects of positive affect induction on creative problem solving and category breadth (reported by Isen & Daubman, 1984, and Isen, Daubman, & Nowicki, 1987) were observed only among people with a secure attachment style. For avoidant persons, no significant difference was found between positive-affect induction and neutral conditions. For anxious persons, a reverse effect was found that resembled the typical effects of negative affect induction. Specifically, these people reacted to a positive affect induction with impaired creativity and a narrowing of mental categories.

THE ATTACHMENT BEHAVIORAL SYSTEM

121

The observed reactions of secure individuals follow from their attention to affective cues. According to Schwartz and Bohner (1996), the induction of positive affect signals that ‘‘all is going well.’’ Secure people’s openness to affective cues leads naturally to considering this signal to be a relevant input for cognitive processing, and then to ‘‘loosening’’ their cognitive strategies and exploring unusual associations. These reactions are reflections of secure individuals’ confidence that letting down their guard is not dangerous and that they can deal effectively with uncertainty, novelty, and any confusion that the broadening of knowledge might create. The findings for avoidant individuals are compatible with their cognitive reactions to negative affect (Pereg & Mikulincer, 2002) and may result from their defensive exclusion of affective material. Whereas dismissal of negative affect is necessary to prevent attachment-system activation, dismissal of positive affect may be necessary to prevent a loosening of cognitive strategies that can result in uncertainty and confusion, which may bring about an unwanted reactivation of the attachment system. The findings for anxious people reveal the extent of their undifferentiated hyperactivation of negative cognitions. For them, apparently, the spread of activation across negative cognitions can begin even with positive affect. Perhaps they at first experience a positive state, but then become reminded of the down side of previous experiences that began positively and ended painfully. Once attuned to these negative memories, the anxious mind may suffer from a spread of negative associations that interferes with exploration and open information processing. Even in an experimental condition intended to induce positive affect, the hyperactivating strategies of anxious individuals prevent them from feeling safe and engaging in relaxed exploration. 2. Caregiving System Our research also provides empirical support for the facilitating effects of the sense of attachment security on the activation of the caregiving system— the system that drives us to provide protection and support to others who are either chronically dependent or temporarily in need (Bowlby, 1982/ 1969). These effects have been documented on a person’s attitudes toward caregiving in close relationships, the arousal of altruistic empathy during the encounter with others’ needs, and the endorsement of prosocial, otherfocused values. Kunce and Shaver (1994) constructed a self-report questionnaire to assess caregiving behaviors in close relationships and found that secure people were more sensitive to partners’ needs, reported more cooperative caregiving, and described themselves as more likely to provide emotional

122

MIKULINCER AND SHAVER

support than insecure persons. Moreover, whereas avoidant people’s deactivating strategies led them to maintain distance from a needy partner (less accessibility, less physical contact), anxious people’s hyperactivating strategies led them to report high levels of overinvolvement with partner’s problems and a pattern of compulsive caregiving (e.g., expressing fears that their partner would leave if they didn’t take care of him or her). These findings have been replicated using other self-report scales and behavioral observations (e.g., Carnelley, Pietromonaco, & Jaffe, 1996; Collins & Feeney, 2000). In a series of five experiments, Mikulincer, Gillath, et al. (2001) documented the facilitatory effects of attachment security on a person’s empathic, compassionate responses to others’ needs. First, both attachment anxiety and avoidance were associated with low levels of altruistic empathy. Second, the contextual heightening of the sense of attachment security (asking participants to recollect personal memories, read a story, or watch a picture of supportive others or subliminally exposing them to proximity-related words) increased reports of altruistic empathy. In contrast, the contextual activation of attachment anxiety or avoidance (asking participants to recall personal memories of relational episodes in which they felt attachment anxiety or avoidance) reduced this prosocial attitude. Importantly, these effects were not explained by mood variations and did not depend on relationship closeness. The findings also revealed that attachment anxiety but not attachment avoidance was associated with more intense personal distress responses during the encounter with others’ needs. This finding implies that avoidant people’s deactivating strategies led them to be emotionally detached while witnessing other’s plight—they react to this situation with neither empathy nor personal distress. In contrast, anxious persons become emotionally overwhelmed while witnessing another’s plight. This self-focused reaction is in line with anxious persons’ hyperactivating strategies and may inhibit empathy and caregiving reactions. While witnessing the plight of another person, anxiously attached individuals are so overwhelmed with self- related negative cognitions and emotions that they are not able to offer assistance and comfort to the needy person. There is also evidence that attachment security facilitates the development of a prosocial orientation. Across three independent studies, Mikulincer, Gillath, et al. (in press) reported that attachment avoidance was associated with low endorsement of values of universalism (concern for the welfare of all people) and benevolence (concern for the welfare of close persons). Moreover, contextual heightening of the sense of attachment security (asking participants to recall personal memories or watch a pictorial representation of supportive others) increased the endorsement of

THE ATTACHMENT BEHAVIORAL SYSTEM

123

these values. These effects were not explained by induced or reported mood. 3. Sexual System The sexual behavior system, along with attachment and caregiving, is a central component of romantic love (Shaver et al., 1988). From an evolutionary perspective, the major function of romantic attraction and sexual mating is to reproduce the mates’ genes. From a more proximal perspective, the quality of a couple’s sex life is an important contributor to relationship satisfaction—one that research shows is affected by attachment security. Our research program offers empirical evidence supporting the beneficial effects of attachment security in the functioning of the sexual system. In a recent study of adolescent sexuality, Tracy, Shaver, Albino, and Cooper (in press) found that adolescents who endorsed a secure style were open to mutually satisfying sexual exploration in the context of a stable relationship and said they engaged in sex primarily to show love for their partner. Tracy et al. (in press) also reported that if both partners were secure, initiation of sexual activity was mutual and physical closeness was enjoyed. In a study of adults’ experiences of heterosexual intercourse, Birnbaum, Gillath, and Mikulincer (2002) found that attachment insecurity contributed to the construction of sexual experience in negative and aversive terms. Specifically, both attachment anxiety and attachment avoidance were associated with higher scores on subscales tapping the experience of negative feelings and worries during heterosexual intercourse (e.g., sense of estrangement, feeling sinful, sense of sexual inadequacy). We also have initial evidence for the involvement of secondary attachment strategies in shaping the sexual attitudes, experiences, and behaviors of insecurely attached people. With regard to avoidant persons, Birnbaum et al. (2002) found that avoidant individuals tended to remain emotionally detached even during heterosexual intercourse; they scored relatively low on subscales tapping the experience of pleasure-related feelings, orgasmic experiences, and feelings of love toward their partner during sexual activity. In the analysis of reasons adolescents give for having sex, Tracy et al. (in press) found that avoidant adolescents gave more selfdefining or self-enhancing reasons, such as losing their virginity, than relationship-focused reasons. Avoidant individuals’ attempts to deactivate their attachment system are also reflected in the fact that they are more likely than anxious or secure people to approve of casual sex and engage in ‘‘one-night stands’’ (Brennan & Shaver, 1995; Fraley, Davis, & Shaver, 1998). This approach to sex may

124

MIKULINCER AND SHAVER

indicate a reluctance to get emotionally involved with or committed to any particular sexual partner (e.g., Tidwell et al., 1996; Brennan & Shaver, 1995). Consistent with this notion, Schachner and Shaver (2002) found that ‘‘mate poaching,’’ or attempting to attract someone who is already in a relationship, and being open to being ‘‘poached’’ by others—in the context of short-term but not long-term relationships—is associated with avoidance. In fact, ‘‘relationship exclusivity,’’ measured by a scale designed by Schmitt and Buss (2000), is negatively correlated with attachment avoidance, suggesting that avoidant people tend to be promiscuous and nonexclusive in their relationships (Schachner & Shaver, 2002). With regard to anxious individuals, Birnbaum et al. (2002) found clear evidence for the underlying action of their hyperactivating strategies in the construction of their sexual experiences. Specifically, they scored relatively high on subscales tapping desire for partner’s emotional involvement during heterosexual intercourse and attempts to please the partner and satisfy his or her demands during this sexual activity. Similarly, Tracy et al. (in press) found that anxious adolescents engage in sex primarily to please their partners, feel accepted, and avoid abandonment. Anxious individuals also tend to worry about losing their partners, and the ones in our recent studies (Schachner & Shaver, 2002) actually had lost their partners more often than less anxious individuals. Even if they try to ‘‘poach’’ other people’s partners, as many avoidant individuals do, they do not succeed as often as less anxious individuals. Perhaps these experiences and concerns explain why anxious adults, especially women, are more likely than their less anxious counterparts to have cosmetic surgery in an attempt to make themselves more acceptable to potential relationship partners (Davis & Lesbo, in press). 4. Affiliation System We have initial evidence suggesting interplay between attachment security and activation of the affiliation behavioral system, a system that meets an evolved need for sociability with others and motivates people to spend time in the company of others (Weiss, 1998). In a correlational study, Shaver et al. (1996) found that reports of secure attachment were positively associated with self-reports of sociability. This correlation fits with results reported by other researchers (e.g., Bartholomew & Horowitz, 1991; Griffin & Bartholomew, 1994). Mikulincer and Selinger (2001) provided further evidence for this association in their assessment of adolescents’ same-sex friendships. Findings indicated that secure adolescents placed high value on both attachment goals (support, security) and affiliation goals (accomplishment of joint projects, having fun together) in their same-sex friendships.

THE ATTACHMENT BEHAVIORAL SYSTEM

125

Moreover, they were responsive to the activation of both the attachment and affiliation systems and changed their goals in accordance with contextual cues that activate these systems. They pursued more attachment goals in attachment-activating contexts (disclosing a secret, being in a sad mood) and more affiliation goals in affiliation-activating contexts (going together to a party, being in a happy mood). In contrast, insecure adolescents showed less flexibility in the activation of the attachment and affiliation systems. Anxious adolescents were exclusively focused on attachment goals and tended to pursue more attachment than affiliation goals in both attachment- and affiliation-activating contexts. Avoidant adolescents tended to dismiss both attachment and affiliation goals in the two types of contexts. The findings for insecure individuals seemed to result from the underlying action of secondary attachment strategies. Avoidant people’s deactivating strategies were manifested not only in inhibition of the attachment system (reluctance to pursue attachment goals even when contextual cues encouraged engagement in attachment behaviors), but also in inhibition of the affiliation system. Perhaps this occurs because activation of this system counteracts avoidant peoples’ emotional detachment from a partner. Anxious people’s hyperactivating strategies were manifested in a chronic, undifferentiated search for support and security and the concomitant dismissal of affiliation goals even when contextual cues encouraged the pursuit of these goals.

I. SUMMARY Our research program has provided extensive evidence regarding the three different modules of our model. Specifically, our studies have clearly supported our hypotheses concerning (1) effects of threat contexts on attachment-system activation; (2) effects of the excitatory and inhibitory circuits associated with secondary attachment strategies on the monitoring and appraisal of threatening events, preconscious activation of attachmentrelated mental representations, and monitoring and appraisal of attachment-figure availability; (3) effects of the actual presence of an available attachment figure or of contextual cues that activate representations of available attachment figures on the enactment of security-based strategies; (4) declarative and procedural components of attachment strategies; (5) manifestations of these strategies in the management of interpersonal behavior, construction of close relationships, processing of attachment-related and -unrelated sources of threats, and regulation of distress; and (6) implications of attachment strategies for representations

126

MIKULINCER AND SHAVER

of the self and others, relationship quality, mental health and adjustment, and the functioning of other behavioral systems (exploration, caregiving, sex, affiliation). Several of our studies have also shown that the observed patterns of attachment-system activation and the documented psychological manifestations and implications of attachment strategies are unique to individual variations in attachment-related anxiety and avoidance. Specifically, the observed effects of the attachment dimensions cannot be accounted for statistically by a host of other individual-difference variables that are consistently and systematically associated with attachment-related anxiety and avoidance, such as neuroticism (Shaver & Brennan, 1992), trait anxiety (Mikulincer, Gillath, & Shaver, 2002; Mikulincer & Sheffi, 2000), global distress (Mikulincer et al., 2002), extraversion (Shaver & Brennan, 1992), self-esteem (Taubman Ben-Ari et al., 2002), reported mood (Mikulincer, Gillath, et al., 2001, in press; Mikulincer, Hirschberger, et al., 2001), social desirability (Mikulincer & Orbach, 1995; Taubman Ben-Ari et al., 2002), and sociability (Shaver & Brennan, 1992). These findings rule out several alternative explanations and argue for the distinctive dynamics and functioning of the attachment system in adulthood.

IV. Unresolved Conceptual and Empirical Issues Despite the strengths of attachment theory and research, many conceptual and empirical issues have yet to be fully tackled. We have begun to address some of the unresolved conceptual issues in recent papers (Mikulincer, Shaver, & Pereg, in press; Shaver & Mikulincer, in press 2002a, 2002b, in press) and to deal with related empirical issues in our ongoing studies. Here, we present a conceptual and empirical agenda for adult attachment theory and research in the near future.

A. COGNITIVE AND NEURAL SUBSTRATES OF ATTACHMENT-SYSTEM ACTIVATION We need to devote more attention to the first module of our model (Fig. 2), attachment-system activation. To date, most of our research efforts have been directed at understanding the psychological manifestations and consequences of people’s attachment styles, and only a few recent studies have documented the psychological reality of attachment-system activation (Mikulincer et al., 2000; Mikulincer, Gillath, & Shaver, 2002). Many

THE ATTACHMENT BEHAVIORAL SYSTEM

127

important questions about the activation process remain unanswered. For example, are proximity-seeking goals automatically activated by encounters with threats? (We have already shown that proximity-related memory nodes are activated.) Are episodic memories of attachment interactions activated by these encounters? What are the functions of such memory activation and how do they relate to representations of attachment figures and proximityseeking goals? How does attachment-system activation interact with automatic activation of other, attachment-unrelated defensive goals (e.g., self-enhancement)? What are the cognitive and affective processes that determine the passage from preconscious attachment-system activation to conscious thoughts, behavioral intentions, and actual behavior? Without knowing the answers to these questions, we cannot fully specify the cognitive substrate of attachment-system activation. Research should also examine whether the attachment system is activated under neutral, nonthreatening conditions. Proximity-seeking goals, like other well-learned goals, can presumably be activated automatically by environmental stimuli that frequently covary with them. Bargh (1990) suggested that if a goal is frequently and consistently activated in the presence of specific contextual cues (e.g., a relationship partner), the goal will come to be activated by these cues without the person consciously intending to pursue it. It therefore seems likely that encounters with an attachment figure can activate proximity-seeking goals even under nonthreatening conditions, because this person has been present during the activation of these goals in many previous threatening situations. At the cognitive level, this idea implies that the contextual activation of representations of attachment figures automatically spreads to proximityseeking goals. At the relationship level, this associative connection can explain the search for closeness and intimacy with a relationship partner under nonthreatening conditions and the establishment of threat-free attachment bonds with this partner. Research is also needed on the physiological and neural substrates of attachment-system activation. Presumably these involve autonomic nervous system activation and neuroendocrine responses. Variations in these responses are frequently observed during the arousal and alleviation of distress (Diamond, 2001), so they should be related to the regulatory functions of the attachment system. But there may also be unique attachment-related brain processes that can be identified. Research should examine the cortical and subcortical regions involved in attachment-system activation and the initiation of attachment strategies; it should map patterns of electrophysiological activity during the processing of threat-related stimuli (e.g., when pursuing hyperactivating or deactivating strategies). Researchers could use evoked reaction potential (ERP) and functional

128

MIKULINCER AND SHAVER

magnetic resonance imaging (fMRI) techniques for delineating the cognitive components (orienting response, expectation, memory) and brain regions (e.g., amygdala, anterior cingulate) involved in attachment-system activation and functioning. Some of these processes may also be identifiable, at least in a preliminary way, by using cognitive research paradigms. Cohen and Shaver (2002) recently investigated possible cerebral hemisphere differences in attachmentrelated information processing associated with avoidant individuals’ imperviousness to positive affect manipulations (see for example, Mikulincer & Sheffi, 2000, discussed earlier in this chapter). Cohen and Shaver (2002) presented positive and negative attachment-related words in the left and right visual fields of more and less avoidant individuals, asking them to make a rapid decision in each case about positivity vs. negativity. Avoidant individuals proved to be less accurate than nonavoidant individuals in detecting positive words when these were presented to the right hemisphere, the frontal (decision-making) lobe of which has been associated with negative affect and withdrawal tendencies. These results, like many of the others summarized in this chapter, strongly suggest that attachment-style differences in behavior are associated with systematic differences in brain functioning. Pursuit of more precise and powerful methods of studying these differences would be highly desirable.

B. THE CONSTRUCT OF ATTACHMENT-FIGURE AVAILABILITY Because of the importance of the attachment-figure availability module in our model (the second module in Fig. 2) in triggering different attachment strategies, we need more research that refines and elaborates this part of the activation process. According to Bowlby (1973), attachment-figure availability is a matter of the figure’s sensitivity and responsiveness to bids for proximity in times of need. However, generating a positive, optimistic attitude toward proximity seeking is not sufficient for creating a sense of attachment security in an attached individual. The partner, or attachment figure, must also collaborate effectively with the individual in coping with threats and provide effective coaching and support for removing the threat and alleviating distress. More couples research is needed to explore the contribution of one partner’s caregiving responses to the other partner’s attachment system. Recent studies conducted by Collins and her colleagues provide important initial information on this issue (Collins & Feeney, 2000; Feeney & Collins, 2001). A related issue concerns Emde’s (1980) definition of availability. For him, availability includes openness to emotional dialogue, attunement to a

THE ATTACHMENT BEHAVIORAL SYSTEM

129

relationship partner’s diverse needs and goals, and acceptance of the full range of the partner’s emotions, not just being responsive to threat-related concerns and expressions of distress. That is, availability also includes responsiveness to another’s positive affect as well as acceptance and encouragement of his or her separateness and autonomy needs. Biringen, Robinson, and Emde (1998) also added to this definition the ability to be available without intruding on the other’s autonomy and to maintain patterns of communication that are not hostile, impatient, or antagonistic. Cassidy (1994) suggested that the sense of attachment security in infancy results from interactions with a caregiver who is emotionally accessible, responsive, and expressive. In these interactions, the infant learns that emotional states can be tolerated and transformed; he or she feels comfortable exploring and learning about emotions, and expects that the caregiver will help regulate distress. Research should examine whether these ideas can be applied to adult close relationships. Specifically, studies should map out the various forms partner’s availability can take in adult relationships, explore partners’ attunement to dependence and autonomy needs as well as positive and negative emotions, and examine the effects of these interaction patterns on the formation of a within-relationship sense of attachment security. In a recent study, we (Banai, Mikulincer, & Shaver, 2002) also extended Bowlby’s availability construct to include what Kohut (1971) called the experience of mirroring (being noticed and admired for one’s qualities and accomplishments) and twinship (fitting in comfortably with others). Although receiving emotional support can strengthen one’s sense of being loved by a partner, the partner’s lack of admiration and praise can damage one’s sense of self-worth and result in the construction of negative working models of the self. Couple members’ failure to fit comfortably together (by virtue of not being similar, or simpatico, in important respects) can lead to misunderstandings and relational conflicts that foster negative working models of others. Future studies should examine whether and how the provision of mirroring and twinship experiences during the initial phases of adult relationships contributes to the formation of a within-relationship sense of attachment security.

C. SECURITY-BASED STRATEGIES, SELF-REGULATION, AND THE AUTONOMOUS SELF The conceptualization of security-based strategies raises an important issue concerning the dynamic interplay between these strategies and selfregulation. At first thought, one might expect security-based strategies,

130

MIKULINCER AND SHAVER

which lead people to rely on the help of attachment figures when coping with threats, to interfere with autonomy and independence (see Kirkpatrick, 1998, for an exposition of this reasoning). That is, attachment-figure availability might favor reliance (even over-reliance) on others—for coregulation of distress (see Tronick, 1989, for an example of this concept)—at the expense of autonomous self-regulation. As a result, securely attached individuals would be chronically driven to seek support from relationship partners instead of coping by using their own skills and relying on themselves to regulate distress. But this reasoning is at odds with our ideas and empirical findings concerning the characteristics of security-based strategies in adulthood. In our model, support seeking and effective coregulation of distress are part of only one component of security-based strategies. In fact, beyond support seeking, these strategies include a strong sense of mastery, agency, and selfdirectedness as well as reliance on instrumental problem solving and active transformational methods of coping that do not require others’ assistance. Moreover, one of the main effects of these strategies is to build a person’s inner resources for maintaining mental health even in situations where attachment figures are absent or the provision of support is blocked. The findings reviewed here also highlight the fact that secure individuals develop a caring attitude toward relationship partners and become active agents responsible for their partners’ welfare and relationship quality rather than passive recipients of caring and comfort. These qualities seem to be part of an autonomous rather than a dependent personality (Alonso-Arbiol, Shaver, & Ya´ rnoz, 2002). Overall, our model implies that attachmentfigure availability facilitates the autonomous management of distress. Recently, we (Mikulincer, Shaver, & Pereg, in press) suggested that attachment-figure availability during infancy and early childhood not only reinforces reliance on the primary attachment strategy, but also provides a basis for the development of self-regulatory skills. With the maturation of these skills during childhood and adolescence, recurrent episodes of attachment-figure availability favor the passage from coregulation to selfregulation and reliance on the self as the main executive agency of securitybased strategies. We also proposed three mechanisms that mediate the association between attachment-figure availability and self-regulation—the broadening of a person’s perspectives and capacities, expansion of the self, and internalization of functions that were originally accomplished by attachment figures. The first mechanism involves activation of other behavioral systems (e.g., exploration, caregiving, and affiliation) following attachment-figure availability and the resulting broadening of perspectives and capacities. Activation of the exploration system leads people to distance themselves

THE ATTACHMENT BEHAVIORAL SYSTEM

131

from attachment figures, explore the environment on their own, and learn new things about the world and the self that enrich their self-regulatory skills. Activation of the caregiving system allows people to learn how to help regulate others’ distress. This learning strengthens their sense of mastery and can be applied to the regulation of their own distress. Activation of the affiliation system provides an increased range of social options for exploring personal interests, developing skills, and broadening one’s self-conception. A related process is what Aron, Aron, and Norman (2001) called ‘‘expansion of the self ’’—the inclusion of a partner’s resources and strengths in one’s self-concept. Attachment-figure availability can set this process of self-expansion in motion. During effective coregulation episodes, the partner’s responses are synchronized with a person’s needs and the partner can be experienced as part of the self. As a result, the person can incorporate the partner’s resources into the self, which in turn facilitates the development of a sense of mastery and a belief that the self has capacities for handling distress alone. The process of self-expansion by including another in the self seems highly metaphorical on first encounter, but it has been well validated in both cognitive and psychophysiological research (see Aron et al., 2001, for a review). A third process is what Kohut (1971) called ‘‘transmuting internalization,’’ which involves the internalization of regulatory functions that were originally performed by the attachment figure, with the individual gradually acquiring the capacity to perform these functions autonomously. In our view, attachment-figure availability promotes a sense of self-worth and efficacy, and then makes coregulation less necessary, because people become more confident of their ability to handle distress alone. As a result, people become less dependent on external agents for dealing with threats, and the self becomes the major agent of affect regulation. This analysis implies that the development of self-regulation depends on attachment-figure availability. Without effective coregulation of distress, the activation of other behavioral systems, the expansion of the self, and the process of transmuting internalization are blocked and the development of self-regulation is disrupted. Our analysis also implies that coming to rely on oneself as the main regulatory agent need not inhibit coregulation. Rather, the self can incorporate support seeking and reliance on others within its regulatory devices and still engage in coregulation when needed. That is, secure individuals can autonomously choose to be dependent on others without feeling that support seeking implies personal weakness or helplessness. For example, a person can mobilize external sources of support during life transitions or traumatic experiences that disrupt the sense of self-worth and deplete inner resources. Furthermore, he or she can activate internal representations of attachment figures during encounters with threats in

132

MIKULINCER AND SHAVER

order to experience feelings of security and comfort that are necessary for engaging in effective self-regulatory efforts. In other words, security-based strategies in adulthood reflect both the passage from coregulation to self-regulation and the incorporation of coregulation within a person’s repertoire of self-regulatory skills. In fact, whereas anxious people’s hyperactivating strategies seem to block the transition from coregulation to self-regulation, avoidant people’s deactivating strategies interfere with reliance on others as a potential regulatory device. Future research should examine the psychological mechanisms that mediate the transition from coregulation to self-regulation and the dynamics of the integration of coregulation and self-regulation. Research should also explore the contextual and personal factors that activate coregulation in adulthood and the ways in which mental representations of attachment figures facilitate self-regulatory attempts.

D. THE INITIATION OF SECONDARY ATTACHMENT STRATEGIES In our model, the triggering of a specific secondary attachment strategy (deactivating or hyperactivating) following attachment-figure unavailability depends on the extent to which proximity seeking is appraised as a viable regulatory option. (See the third module in Fig. 2.) However, the model does not specify the situational and personal antecedents of this appraisal process in adulthood. Recently, we (Mikulincer, Shaver, & Pereg, in press) tentatively outlined the proximal and distal factors that may affect this process. Proximal factors include the states of mind produced by attachment-figure unavailability that contribute to activation of a specific secondary attachment strategy. Distal factors include the external and internal antecedents of these states of mind. According to our analysis, attachment-figure unavailability can result in two kinds of painful states of mind. One is organized around the failure of attachment behaviors to achieve a positive result (closeness, love) and the receipt of punishment (inattention, rejection. anger) following these behaviors. In this state of mind, proximity to the attachment figure is experienced as a condition of nonreward or punishment, and the individual becomes afraid of failure and punishment in future proximity-seeking attempts. The main threat here is proximity to the attachment figure; the predominant fears concern the aversive outcomes that proximity can elicit; and the person is forced to adopt a strategy that minimizes the experience of nonreward/punishment—i.e., a deactivating strategy. In other words, this

THE ATTACHMENT BEHAVIORAL SYSTEM

133

state of mind favors deactivation of the attachment system and consolidation of avoidant working models. A second state of mind is organized around a failure to coregulate distress and the need to deal with threats alone. This state of mind is based on beliefs that attachment-figure unavailability, inconsistency, or insufficiency leaves one helpless and vulnerable in a threatening world, and one must therefore try harder to attain a protective relationship. In such cases, distance from the attachment figure is experienced as dangerous, the person becomes afraid of the aversive outcomes that may result from attempts to regulate distress without the help of attachment figures, and he or she is forced to adopt a strategy that addresses the sense of helplessness and fear of being alone—i.e., a hyperactivating strategy. The person is then biased toward perceiving proximity seeking as viable: The psychological cost of recognizing the nonviability of proximity seeking is so great that the person searches for even minimal signals of attachment-figure availability and expresses anger when these signals are not forthcoming. We have attempted to specify the likely external and internal antecedents of these two states of mind (Mikulincer, Shaver, & Pereg, in press). On the one hand, any pattern of interaction that strengthens the link between proximity seeking and negative affectivity should lead to the construal of attachment-figure unavailability as a nonreward/punishment condition and to the consequent adoption of deactivating strategies. These patterns of interaction include, for example, consistent inattention, rejection, or angry responses of the attachment figure to proximity seeking; threats of punishment for the display of attachment behaviors; and traumatic/abusive experiences during proximity-seeking attempts. This state of mind is also affected by internal factors that intensify emotional reactions to attachmentfigure unavailability, such as arousability/reactivity and intolerance of frustration. On the other hand, any pattern of interaction with a frustrating attachment figure that prevents the development of self-regulation skills should strengthen a person’s sense of helplessness and contribute to the adoption of hyperactivating strategies. These interactions include, for example, a compulsive pattern of caregiving that is unrelated to the individual’s requests or need for help; attachment figures’ intrusiveness that punishes the person for engaging in autonomy-oriented activities; and attachment figures’ explicit or implicit messages that emphasize a person’s helplessness. This state of mind can also be exacerbated by temperamental deficits in self-regulation and the control of attention, memory, and behavior. This analysis is consistent with the observation of Ainsworth et al. (1978) that caregivers of avoidant infants consistently rebuffed or deflected their

134

MIKULINCER AND SHAVER

infant’s attachment behaviors and were intolerant of their infants’ expressions of vulnerability and neediness. It is also consistent with findings showing that caregivers of anxiously attached infants tend to exhibit inconsistent responsiveness to their infant’s needs, being sometimes unavailable and at other times intrusive, overprotective, and interfering with their children’s engagement in exploration (see Cassidy & Berlin, 1994, for a review). Unfortunately, adult attachment research has not provided systematic information about the specific patterns of interaction with a relationship partner in the early stages of a close relationship that contribute to the formation of within-relationship patterns of attachment anxiety or avoidance. Moreover, the evidence for temperamental contributions to attachment style is still inconsistent and incoherent (Vaughn & Bost, 1999), even though it seems likely to us that temperamental variables are involved. The attachment field still lacks qualitative studies exploring insecure persons’ states of mind concerning attachment-figure unavailability, within-relationship studies examining specific patterns of unavailability of relationship partners, and large-scale twin studies of individual differences in attachment.

E. THE NATURE OF FEARFUL AVOIDANCE Another unresolved issue related to the triggering of secondary attachment strategies concerns the dynamics of people who have high scores on both the anxiety and avoidance dimensions—i.e., the more extremely troubled individuals in the ‘‘category’’ that Bartholomew and Horowitz (1991) called fearful avoidance. Whereas our model characterizes the dynamics of people who score high on attachment anxiety (anxious individuals) and those who score high on attachment avoidance (avoidant individuals), it makes no explicit statement about how people who score especially high on both anxiety and avoidance answer the question in Fig. 2 concerning the viability of proximity seeking. This is an extremely important issue because many studies in the attachment literature have found ‘‘fearful avoidants’’ to be the least secure, least trusting, and most troubled of adolescents and adults (Shaver & Clark, 1994). In fact, although our research program has not been explicitly focused on highly fearfully avoidant individuals, who are more common in abused or clinical samples, some of our studies reveal either two main effects for the anxiety and avoidance dimensions or an interaction between the dimensions indicating that fearful avoidants have especially negative representations of their romantic partners (Mikulincer, 1998a; Mikulincer &

THE ATTACHMENT BEHAVIORAL SYSTEM

135

Arad, 1999), are unusually closed and rigid cognitively (Mikulincer, 1997), exhibit the least empathy for others who are distressed (Mikulincer, Gillath, et al., 2001), and report severe personality disorders (Brennan & Shaver, 1998), and the poorest mental health under stressful conditions (Berant et al., 2001a,b). In a recent commentary on our model, Simpson and Rholes (2002) suggested that fearful avoidant persons are unable to coherently answer the question, ‘‘Is proximity-seeking a viable option?’’ and so cannot choose between deactivating and hyperactivating strategies. As a result, Simpson and Rholes (2002) claim that these persons ‘‘may enact both strategies in a haphazard, confused, and chaotic manner . . . their behavior under stress may be an incoherent blend of contradictory, abortive approach/avoidance behaviors or perhaps paralyzed inaction or withdrawal’’ (p. 224). This description of fearful avoidants’ psychological dynamics resembles the ‘‘disorganized’’ attachment pattern observed in infants in the strange situation (Main & Hesse, 1990). These infants are characterized by simultaneous or rapidly vacillating displays of approach and avoidance behavior toward the attachment figure as well as by aimless, disoriented, and confused actions in response to attachment-figure unavailability. In our view, Simpson and Rholes’s reasoning provides an important insight into the attachment dynamics of fearful avoidant individuals. These dynamics can be incorporated into our model as an additional path that represents a failure to achieve any of the goals of the major attachment strategies: safety and security following proximity seeking (the primary, secure strategy), defensive deactivation of the attachment system (the avoidant strategy), or chronic activation of the attachment system until security-enhancing proximity is at least temporarily attained (the anxious strategy). Take, for example, the case in which deactivating strategies fail to attain their regulatory goals due to encounters with severe and prolonged traumatic events that a person cannot dismiss or deny. In this case, the attachment system is reactivated despite the deactivating strategies, attachment-related worries become accessible, and the individual becomes entrapped in a cycle of useless attempts at avoidance and the unwanted intrusion of attachment anxiety. This cycle resembles the dynamics of posttraumatic stress disorder (Horowitz, 1982) and may lead to the consolidation of an incoherent, disorganized pattern of ineffective deactivating strategies together with an unwanted activation of the attachment system. This conceptualization of fearful avoidance fits with the findings mentioned above that depict fearful avoidants as the most troubled of the four major attachment-style groups. It also fits with findings showing that the lives of fearful avoidant individuals have been scarred by physical or sexual abuse or other attachment-related traumas that a person cannot

136

MIKULINCER AND SHAVER

dismiss or deny (e.g., Brennan & Shaver, 1998; Brennan, Shaver, & Tobey, 1991; Shaver & Clark, 1994). It is important to note that when we refer to fearfully avoidant individuals, we are thinking of very high scores on both the anxiety and avoidance dimensions, not all people who might select the fearfully avoidant self-description on the Bartholomew and Horowitz (1991) categorical measure. In our opinion, that measure provides too low a hurdle for classification into the least secure category, and this creates too large a discrepancy between the clinical literature on disorganized attachment (reviewed by Lyons-Ruth & Jacobvitz, 1999) and the social-psychological literature based on Bartholomew and Horowitz’s (1991) measure and normal samples of college students. In the future, researchers should devote more attention to the measurement of fearful avoidance in community and clinical samples as well as the functioning of the attachment system of persons who score specially high on both anxiety and avoidance.

F. CONSIDERING BOTH PARTNERS IN A RELATIONSHIP: A SYSTEMIC MODEL OF ATTACHMENT DYNAMICS Although our model is focused on the intrapsychic dynamics of the attachment system, it acknowledges that these dynamics are sensitive and responsive to contextual cues in general and to a relationship partner’s behavior in particular. In fact, attachment-system activation and functioning occur in an interactional context and partly depend on the partner’s responses. In the three main components of the model, the partner’s behaviors can affect the focal individual’s attachment system. For example, the partner can be a source of threat (e.g., by threatening abandonment or violence) and can affect the appraisal of attachment-figure availability as well as the viability of proximity seeking as a means of achieving security while interacting with this partner. Therefore, the relationship partner should be introduced more explicitly into the model, and researchers should take both partners’ characteristics and behaviors into account when studying relational cognitions, emotions, and behaviors. Some of our studies have already demonstrated the importance of a partner’s attachment dynamics in explaining an individual’s construction of, and satisfaction with, close relationships. Brennan and Shaver (1995) reported that both partners’ attachment security added to the prediction of each partner’s reports of satisfaction in a romantic relationship, after controlling for variance in relationship satisfaction due to a person’s own attachment security. Similarly, Mikulincer and Erev (1991) found that a person’s reported romantic love was directly affected by his or her partner’s

THE ATTACHMENT BEHAVIORAL SYSTEM

137

attachment style. Mikulincer, Florian, and Hirschberger (2002) reported that a person’s scores on attachment anxiety and avoidance had differential effects on daily relational emotions, cognitions, and behaviors depending on his or her partner’s attachment scores. The importance of a partner’s attachment dynamics has also been documented in other studies using observational techniques, diary keeping, and narrative accounts (e.g., Collins & Feeney, 2000; Feeney, 2002; Feeney & Hohaus, 2001; Simpson, Rholes, & Nelligan, 1992). When the partner is fully introduced into the model, and is considered as an additional focal individual, a foundation is provided for a systemic model of attachment dynamics at both the personal and interpersonal levels (see also Feeney, in press, and Mikulincer, Florian, Cowan, and Cowan, 2002, for discussions of systemic attachment models). As a result, Fig. 2 should be duplicated in order to represent the attachment dynamics of the two partners, and bidirectional arrows should be drawn connecting the components and strategies of the two partners’ attachment systems. In this representation, a person’s cognitions, emotions, and behaviors are organized around intrapsychic and interpersonal regularities related to both partners’ attachment dynamics. Specifically, the partners reciprocally influence each other, and changes in the attachment dynamics of one partner are likely to have effects throughout the attachment system of the other partner. In the future, researchers should devote more attention to the complex connections between partners’ attachment dynamics and the ways in which both partners’ attachment systems alter the quality of their relationship.

G. ADULT ATTACHMENT BONDS Although most of the research on the interpersonal manifestations of attachment dynamics in adulthood has been conducted within romantic and marital relationships, one cannot completely equate attachment bonds to romantic love. First, attachment strategies can be directed toward nonromantic partners such as friends, counselors, teachers, therapists, and even groups and organizations. Second, although all children are attached to their primary caregivers (Cassidy, 1999), romantic partners do not necessarily function as attachment figures for each other. The formation of an attachment bond within a romantic relationship depends on both partners’ attachment styles and the extent to which one person serves attachment functions of safe haven and secure base for the other (see Fraley & Shaver, 1999, 2000, for more detailed discussions). Third, even when partners function as attachment figures, some dyadic interactions reflect the underlying action of other behavioral systems rather than the attachment system.

138

MIKULINCER AND SHAVER

The diversity of relationships in which the attachment system is activated and the diversity of behavioral systems that are involved in romantic relationships raise several issues that are still unresolved. One issue concerns the formation, dissolution, and reorganization of attachment bonds. Because adults can have multiple attachment figures and can transfer attachment functions from one partner to another, more conceptual and empirical effort should be directed toward understanding why and how a specific partner is targeted as a potential attachment figure. Specifically, what partner characteristics and behaviors make him or her more likely to became an attachment figure? How does a person become attached to such a partner? How are the diverse attachment figures organized into what Bowlby (1979) called a ‘‘hierarchy’’ of attachment figures? In addition to grappling with these matters, researchers should examine the dissolution of attachment bonds: Why and how does a relationship partner stop functioning as an attachment figure? Special attention should be given to reorganization of attachment bonds following the dissolution of a romantic relationship (e.g., divorce, separation, or death) and to the phenomenon of continuous attachment to a lost partner (see Fraley & Shaver, 1999; Mikulincer & Florian, 1996, for discussions). Of course, research should also examine the involvement of attachment strategies in these processes and the ways in which people with different attachment styles negotiate the formation, dissolution, and reorganization of attachment bonds. Another underdeveloped research area concerns the interplay of the attachment system and other behavioral systems within romantic relationships. Shaver et al. (1988) argued that romantic love involves the integration of the attachment, caregiving, and sexual systems. However, most subsequent studies have focused mainly on the pivotal role of the attachment system and the contribution of attachment security to caregiving and sexuality (see Feeney & Collins, 2001, for an exception). We lack information about the contribution of individual differences in the caregiving and sexual systems to partners’ attachment dynamics as well as the way the three behavioral systems are coordinated and integrated. Researchers should also look more deeply into the interplay between the attachment and exploration systems, and between dependency and autonomy needs within romantic relationships.

H. DEVELOPMENT, STABILITY, AND CHANGE Attachment theory is a theory of personality functioning and intrapsychic structure, a theory about the construction of interpersonal behavior, emotional bonds, and close relationships, and a theory of personality

THE ATTACHMENT BEHAVIORAL SYSTEM

139

development. In this chapter and in most of the research on which it is based, we have attempted to understand attachment-system functioning and attachment-related variations in intrapsychic organization and close relationships. We have devoted less attention to the developmental origins of these variations and to Bowlby’s (1973) contention that early childhood experiences shape attachment dynamics. Although our model does not deal directly with personality development, we have explicitly posited that the chronically accessible working models of attachment that result from the internalization of past experiences affect the three main components of the model. As such, a person’s attachment dynamics in adulthood should reflect his or her past experiences, perhaps beginning in infancy. Of course, this simple contention needs elaboration and an explicit developmental model of attachment dynamics (Fraley, 2002). Although not developmentalists by training, we have conducted a number of studies in which adults were asked about childhood relationships with parents and about experiences that might be expected to have a long-term impact on attachment style. Brennan et al. (1991) found, for example, that attachment insecurity in adulthood was related to childhood experiences with one or more parents who had a serious drinking problem. Shaver and Clark (1994) found that fearful avoidant attachment was related to childhood experiences of physical and sexual abuse. Brennan and Shaver (1993, 1998) revealed that experiencing the death of one’s father or the divorce of one’s parents early in childhood was associated with self-reports of insecure attachment in adulthood. Of course, these studies are not sufficient to make a strong case for the developmental origins of adult attachment style. More longitudinal studies should be conducted, linking self-report measures of adult attachment to measures of attachment-related experiences in infancy and childhood. Even after several good longitudinal studies have been conducted, however, we do not expect the developmental trajectory of attachment orientations to be linear or in any other way simple. We believe that attachment orientations are not based only on childhood experiences. Current attachment dynamics are also likely to be affected by a broad array of contextual factors that moderate or even override the effects of internalized representations of past experiences. In fact, Bowlby (1988) claimed that these representations can be updated throughout life, and he selected the word ‘‘working’’ in the phrase ‘‘internal working models’’ partly to represent the changing nature of these cognitive-affective structures. The unstable nature of working models is made more likely when one considers recent cognitive theories (e.g., Anderson, 1994) that portray mental representations as associative neural networks. These networks can change, subtly or dramatically, depending on context and recent

140

MIKULINCER AND SHAVER

experiences. In such theories, psychological reality is constituted by a series of neural networks, each containing millions of cells, which together can represent either specific experiences or averages of representations of specific experiences, depending on what is being thought about or has been mentioned or experienced recently. Just as repeated experiences in other domains lead to excitatory and inhibitory connections between many nodes in a network, repeated experiences with attachment figures, especially the most important and lasting ones, forge central tendencies in neural networks that make some working models chronically accessible. These tendencies, however, can be modified by many contextual factors, such as current interactions with a partner, the attachment dynamics of the partner, a person’s current life situation, and contextual activation of less accessible attachment-related mental representations. In line with this reasoning, every model of attachment dynamics should include ideas concerning contextual and more long-lasting changes in the functioning of the attachment system. This issue is particularly relevant for creating adequate attachment-oriented psychotherapies and for understanding factors in the client–psychotherapist relationship that foster symptom alleviation and personality change (Shaver & Norman, 1995). In this context, our studies on the positive effects of the contextual activation of security-enhancing mental representations (e.g., Mikulincer, Gillath, et al., 2001; Mikulincer & Shaver, 2001) can be viewed as an initial step in developing a research program on attachment and psychotherapy. This program should examine the construct of ‘‘working alliance,’’ the functioning of the psychotherapist as a security-enhancing attachment figure, changes in the client’s attachment dynamics caused by attachment to and internalization of this figure, and the ways in which these processes advance psychotherapeutic aims. A neural network conceptualization of working models has important implications for measurement. The outcome of measurement depends on which networks are accessed by or constructed during the measurement procedure. If we ask research participants to describe a particular ongoing relationship, we may get different results than if we ask about close relationships in general. If we ask about an adult’s relationship with an attachment figure now, we may get different results than if we ask about the earliest memories the person has of relating to this figure during childhood. Therefore, although our studies have obtained systematic, theory-consistent findings using generic self-report attachment measures, it does not mean that such measurement is equivalent to measures of childhood or current relationships with specific relationship partners. In general, much important work remains to be done in the area of measuring attachment phenomena. The self-report measures on which our

THE ATTACHMENT BEHAVIORAL SYSTEM

141

research is based were created through a combination of research on infants, intuition, psychometrics, and convenience. Our present measures (Brennan et al., 1998) were created by factor analysis of many previous measures, and the items consequently range from ones concerned with relationships in general to ones concerned with a particular partner. Some deal with ‘‘comfort’’ and other feelings; some deal with desires and motives. They were all based on item writers’ exposure to the attachment literature, but in most cases were not designed component-by-component with a coherent theoretical model in mind. Given their relative crudeness, it is remarkable how systematic and cumulative our research findings have been.

V. Concluding Remarks We began this chapter with personal observations and will conclude in the same vein. Neither of us realized or dreamed when we began studying individual differences in adult attachment 15 years ago that by now we would have accumulated so many diverse, intriguing, and coherent findings. Bowlby’s theory has clearly been generative and Ainsworth’s methodological creativity has been inspiring. Today, a literature search on ‘‘attachment’’ turns up thousands of studies motivated by attachment theory, including a few hundred related to our own measures and models, and there is still no end in sight. Research in this area confirms what we said at the outset, that the human mind/brain is a social machine. The brain evolved in highly social contexts and was ‘‘designed’’ by evolution to solve important problems of survival and reproduction within those contexts. The size of the brain and its incomplete programming at birth, combined with the highly immature and incompetent body in which it is born, make each person dependent on ‘‘older and wiser’’ caregivers from the beginning. This need for others slowly gives way, in a ‘‘good enough’’ relational environment, to increased selfregulation and autonomy as well as mutually supportive couple relationships. The need for others, both real and symbolic, is perpetual, and the possibilities for attachment injuries all through life are numerous, so there are many opportunities to get shunted off onto an insecure path. We hope our research will stimulate others, as Bowlby and Ainsworth’s work inspired us, and that together we will create a more complete and powerful theory of the social mind, one that provides an improved foundation for individual and couples therapy—and perhaps contributes eventually to a more tolerant and supportive world.

142

MIKULINCER AND SHAVER

Acknowledgments The authors would like to thank all of the people, too numerous to mention, who helped design and conduct the research summarized here. Mikulincer thanks, in approximate historical order, Victor Florian, Rami Tolmacz, Israel Orbach, Gurit Birnbaum, Ety Berant, Neta Horesh, Orit Nachmias, Gilad Hirschberger, Dana Pereg, and Omri Gillath. Shaver thanks, in approximate historical order, Cindy Hazan, Lee Kirkpatrick, Kelly Brennan, Linda Kunce, Marie Tidwell, Harry Reis, Julie Rothbard, Daria Papalia, Ken Levy, Catherine Clark, Lilah Koski, Hillary Morgan, Chris Fraley, Mike Cohen, and Dory Schachner. Preparation of this chapter was facilitated by a grant from the Fetzer Institute.

References Abramson, L. Y., Metalsky, G. I., & Alloy, L. B. (1989). Hopelessness depression: A theorybased subtype of depression. Psychological Review, 96, 358–372. Ainsworth, M. D. S. (1967). Infancy in Uganda: Infant care and the growth of love. Baltimore: Johns Hopkins University Press. Ainsworth, M. D. S. (1991). Attachment and other affectional bonds across the life cycle. In C. M. Parkes, J. Stevenson-Hinde, & P. Marris (Eds.), Attachment across the life cycle (pp. 33–51). New York: Routledge. Ainsworth, M. D. S., Blehar, M. C., Waters, E., & Walls, S. (1978). Patterns of attachment: Assessed in the strange situation and at home. Hillsdale, NJ: Erlbaum. Alonso-Arbiol, I., Shaver, P.R., & Ya´ rnoz, S (2002). Insecure attachment, gender roles, and interpersonal dependency in the Basque Country. Personal Relationships, 9, 479–490. Anderson, J. (1994). Cognitive psychology. New York: Academic Press. Aron, A., Aron, E. N., & Norman, C. (2001). Self-expansion model of motivation and cognition in close relationships and beyond. In G. J. O. Fletcher & M. S. Clark (Eds.), Blackwell handbook of social psychology: Interpersonal processes (pp. 478–501). Malden, MA: Blackwell Publishers. Banai, E., Mikulincer, M., & Shaver, P. R. (2002). ‘‘Selfobject’’ needs in Kohut’s self psychology: Links with attachment, self-cohesion, affect regulation, and adjustment. Submitted for publication. Bargh, J. A. (1990). Auto-motives: Preconscious determinants of social interaction. In E. T. Higgins & R. M. Sorrentino (Eds.), Handbook of motivation and cognition: Foundations of social behavior (Vol. 2, pp. 93–130). New York: Guilford Press. Bartholomew, K. (1990). Avoidance of intimacy: An attachment perspective. Journal of Social and Personal Relationships, 7, 147–178. Bartholomew, K., & Horowitz, L. M. (1991). Attachment styles among young adults: A test of a four-category model. Journal of Personality and Social Psychology, 61, 226–244. Bartholomew, K., & Shaver, P. R. (1998). Methods of assessing adult attachment: Do they converge? In J. A. Simpson & W. S. Rholes (Eds.), Attachment theory and close relationships (pp. 25–45). New York: Guilford Press. Berant, E., Mikulincer, M., & Florian, V. (2001a). The association of mothers’ attachment style and their psychological reactions to the diagnosis of infant’s congenital heart disease. Journal of Social and Clinical Psychology, 20, 208–232.

THE ATTACHMENT BEHAVIORAL SYSTEM

143

Berant, E., Mikulincer, M., & Florian, V. (2001b). Attachment style and mental health: A oneyear follow-up study of mothers of infants with congenital heart disease. Personality and Social Psychology Bulletin, 8, 956–968. Berg, J. H. (1987). Responsiveness and self-disclosure. In V. J. Derlega & J. H. Berg (Eds.), Selfdisclosure (pp. 101–130). New York: Plenum Press. Biringen, Z., Robinson, J., & Emde, R. N. (1998). Emotional availability scales (3rd ed.). Unpublished manual, Department of Human Development and Family Studies, Colorado State University, Fort Collins, CO. Birnbaum, G. E., Gillath, O., & Mikulincer, M. (2002). Adult attachment style and the subjective experience of heterosexual intercourse. Manuscript in preparation. Birnbaum, G. E., Orr, I., Mikulincer, M., & Florian, V. (1997). When marriage breaks up: Does attachment style contribute to coping and mental health? Journal of Social and Personal Relationships, 14, 643–654. Bowlby, J. (1956). The growth of independence in the young child. Royal Society of Health Journal, 76, 587–591. Bowlby, J. (1973). Attachment and loss: Vol. 2. Separation: Anxiety and anger. New York: Basic Books. Bowlby, J. (1979). The making and breaking of affectional bonds. London: Tavistock. Bowlby, J. (1980). Attachment and loss: Vol. 3. Sadness and depression. New York: Basic Books. Bowlby, J. (1982/1969). Attachment and loss: Vol. 1. Attachment (2nd ed.). New York: Basic Books. Bowlby, J. (1988). A secure base: Clinical applications of attachment theory. London: Routledge. Brennan, K. A., Clark, C. L., & Shaver, P. R. (1998). Self-report measurement of adult attachment: An integrative overview. In J. A. Simpson & W. S. Rholes (Eds.), Attachment theory and close relationships (pp. 46–76). New York: Guilford Press. Brennan, K. A., & Morris, K. A. (1997). Attachment styles, self-esteem, and patterns of seeking feedback from romantic partners. Personality and Social Psychology Bulletin, 23, 23–31. Brennan, K. A., & Shaver, P. R. (1993). Attachment styles and parental divorce. Journal of Divorce and Remarriage, 21, 161–175. Brennan, K. A., & Shaver, P. R. (1995). Dimensions of adult attachment, affect regulation, and romantic relationship functioning. Personality and Social Psychology Bulletin, 21, 267–283. Brennan, K. A., & Shaver, P. R. (1998). Attachment styles and personality disorders: Their connections to each other and to parental divorce, parental death, and perceptions of parental caregiving. Journal of Personality, 66, 835–878. Brennan, K. A., Shaver, P. R., & Tobey, A. E. (1991). Attachment styles, gender and parental problem drinking. Journal of Social and Personal Relationships, 8, 451–466. Carnelley, K. B., Pietromonaco, P. R., & Jaffe, K. (1994). Depression, working models of others, and relationship functioning. Journal of Personality and Social Psychology, 66, 127–140. Carnelley, K. B., Pietromonaco, P. R., & Jaffe, K. (1996). Attachment, caregiving, and relationship functioning in couples. Personal Relationships, 3, 257–277. Cassidy, J. (1994). Emotion regulation: Influence of attachment relationships. In N. A. Fox (Ed.). The development of emotion regulation: Biological and behavioral considerations. Monographs of the Society for Research in Child Development, 59, (2–3, Seriel No. 240), 228–249. Cassidy, J. (1999). The nature of the child’s ties. In J. Cassidy & P. R. Shaver (Eds.), Handbook of attachment: Theory, research, and clinical applications (pp. 3–20). New York: Guilford Press. Cassidy, J., & Berlin, L. J. (1994). The insecure/ambivalent pattern of attachment: Theory and research. Child Development, 65, 971–981.

144

MIKULINCER AND SHAVER

Cassidy, J., & Kobak, R. R. (1988). Avoidance and its relationship with other defensive processes. In J. Belsky & T. Nezworski (Eds.), Clinical implications of attachment (pp. 300–323). Hillsdale, NJ: Erlbaum. Cohen, M., & Shaver, P. R. (2002). Avoidant attachment and hemispheric lateralization of the processing of attachment- and emotion-related words. Submitted for publication. Collins, N. L. (1996). Working models of attachment: Implications for explanation, emotion, and behavior. Journal of Personality and Social Psychology, 71, 810–832. Collins, N. L., & Feeney, B. C. (2000). A safe haven: An attachment theory perspective on support seeking and caregiving in intimate relationships. Journal of Personality and Social Psychology, 78, 1053–1073. Collins, N. L., & Read, S. J. (1990). Adult attachment, working models, and relationship quality in dating couples. Journal of Personality and Social Psychology, 58, 644–663. Collins, N. L., & Read, S. J. (1994). Cognitive representations of attachment: The structure and function of working models. In K. Bartholomew & D. Perlman (Eds.), Attachment processes in adulthood (pp. 53–92). London: Jessica Kingsley. Cooper, M. L., Shaver, P. R., & Collins, N. L. (1998). Attachment styles, emotion regulation, and adjustment in adolescence. Journal of Personality and Social Psychology, 74, 1380–1397. Crowell, J. A., Fraley, R. C., & Shaver, P. R. (1999). Measurement of adult attachment. In J. Cassidy & P. R. Shaver (Eds.), Handbook of attachment: Theory, research, and clinical applications (pp. 434–465). New York: Guilford Press. Davis, D., & Lesbo, M. (in press). Sculpting the body beautiful: Attachment style, neuroticism, and use of cosmetic surgeries. Sex Roles. Diamond, L. M. (2001). Contributions of psychophysiology to research on adult attachment: Review and recommendations. Personality and Social Psychology Review, 5, 276–295. Emde, R. N. (1980). Emotional availability: A reciprocal reward system for infants and parents with implications for prevention of psychosocial disorders. In P. M. Taylor (Ed.), Parentinfant relationships (pp. 87–115). Orlando, FL: Grune & Stratton. Epstein, S., & Meier, P. (1989). Constructive thinking: A broad coping variable with specific components. Journal of Personality and Social Psychology, 57, 332–350. Feeney, B. C., & Collins, N. L. (2001). Predictors of caregiving in adult intimate relationships: An attachment theoretical perspective. Journal of Personality and Social Psychology, 80, 972–994. Feeney, J. A. (1999). Adult romantic attachment and couple relationships. In J. Cassidy & P. R. Shaver (Eds.), Handbook of attachment: Theory, research, and clinical applications (pp. 355–377). New York: Guilford Press. Feeney, J. A. (2002). Attachment, marital interaction, and relationship satisfaction: A diary study. Personal Relationships, 9, 39–55. Feeney, J.A. (in press). The systemic nature of couple relationships: An attachment perspective. In P. Erdman & T. Caffery (Eds.), Attachment and family systems: Conceptual, empirical and therapeutic relatedness. New York: Brunner/Mazel. Feeney, J. A., & Hohaus, L. (2001). Attachment and spousal caregiving. Personal Relationships, 8, 21–39. Florian, V., & Mikulincer, M. (1998). Symbolic immortality and the management of the terror of death: The moderating role of attachment style. Journal of Personality and Social Psychology, 74, 725–734. Florian, V., Mikulincer, M., & Bucholtz, I. (1995). Effects of adult attachment style on the perception and search for social support. Journal of Psychology, 129, 665–676. Fraley, R. C. (2002). Attachment stability from infancy to adulthood: Meta-analysis and dynamic modeling of developmental mechanisms. Personality and Social Psychology Review, 6, 123–151.

THE ATTACHMENT BEHAVIORAL SYSTEM

145

Fraley, R. C., & Davis, K. E. (1997). Attachment formation and transfer in young adults’ close friendships and romantic relationships. Personal Relationships, 4, 131–144. Fraley, R. C., Davis, K. E., & Shaver, P. R. (1998). Dismissing-avoidance and the defensive organization of emotion, cognition, and behavior. In J. A. Simpson & W. S. Rholes (Eds.), Attachment theory and close relationships (pp. 249–279). New York: Guilford Press. Fraley, R. C., Garner, J. P., & Shaver, P. R. (2000). Adult attachment and the defensive regulation of attention and memory: Examining the role of preemptive and postemptive defensive processes. Journal of Personality and Social Psychology, 79, 816–826. Fraley, R. C., & Shaver, P. R. (1997). Adult attachment and the suppression of unwanted thoughts. Journal of Personality and Social Psychology, 73, 1080–1091. Fraley, R. C., & Shaver, P. R. (1998). Airport separations: A naturalistic study of adult attachment dynamics in separating couples. Journal of Personality and Social Psychology, 75, 1198–1212. Fraley, R. C., & Shaver, P. R. (1999). Loss and bereavement: Attachment theory and recent controversies concerning grief work and the nature of detachment. In J. Cassidy & P. R. Shaver (Eds.), Handbook of attachment: Theory, research, and clinical applications (pp. 735–759). New York: Guilford Press. Fraley, R. C., & Shaver, P. R. (2000). Adult romantic attachment: Theoretical developments, emerging controversies, and unanswered questions. Review of General Psychology, 4, 132–154. Fraley, R. C., Spieker, S. J. (2003). Are infant attachment patterns continuously or categorically distributed? A taxometric analysis of strange situation behavior. Developmental Psychology. In press. Fraley, R. C., & Waller, N. G. (1998). Adult attachment patterns: A test of the typological model. In J. A. Simpson & W. S. Rholes (Eds.), Attachment theory and close relationships (pp. 77–114). New York: Guilford Press. Fredrickson, B. L. (2001). The role of positive emotions in positive psychology: The broadenand-build theory of positive emotions. American Psychologist, 56, 218–226. Freud, S. (1957/1915). Instincts and their vicissitudes. In J. Strachey (Ed. and Trans.), The standard edition of the complete psychological works of Sigmund Freud (Vol. 3, pp. 157–186). London: Hogarth Press. Freud, S. (1961/1930). Address delivered [in 1930] at the Goethe House at Frankfurt. In J. Strachey (Ed. and Trans.), The standard edition of the complete psychological works of Sigmund Freud (Vol. 22, pp. 208–212). London: Hogarth Press. Gaines, S. O. J., Reis, H. T., Summers, S., Rusbult, C. E., Cox, C. L., & Wexler, M. O. (1997). Impact of attachment style on reactions to accommodative dilemmas in close relationships. Personal Relationships, 4, 93–113. George, C., Kaplan, N., & Main, M. (1985). The Adult Attachment Interview. Unpublished manuscript, University of California, Berkeley, Department of Psychology. George, C., & West, M. (2001). The development and preliminary validation of a new measure of adult attachment: The Adult Attachment Projective. Attachment and Human Development, 3, 30–61. Gottman, J. M. (1994). What predicts divorce? The relationship between marital processes and marital outcomes. Hillsdale, NJ: Erlbaum. Green, J. D., & Campbell, W. K. (2000). Attachment and exploration in adults: Chronic and contextual accessibility. Personality and Social Psychology Bulletin, 26, 452–461. Green-Hennessy, S., & Reis, H. T. (1998). Openness in processing social information among attachment types. Personal Relationships, 5, 449–466. Greenberg, J., Pyszczynski, T., & Solomon, S. (1997). Terror management theory of self-esteem and cultural worldviews: Empirical assessments and conceptual refinements. In M. P.

146

MIKULINCER AND SHAVER

Zanna (Ed.), Advances in Experimental Social Psychology (Vol. 29, pp. 61–141). San Diego, CA: Academic Press. Greenberg, J., Pyszczynski, T., Solomon, S., Simon, L., & Breus, M. (1994). The role of consciousness and accessibility of death-related thoughts in mortality salience effects. Journal of Personality and Social Psychology, 67, 627–637. Grice, P. (1989). Studies in the way of words. Cambridge, MA: Harvard University Press. Griffin, D. W., & Bartholomew, K. (1994). Models of the self and other: Fundamental dimensions underlying measures of adult attachment. Journal of Personality and Social Psychology, 67, 430–445. Harlow, H. F. (1959). Love in infant monkeys. Scientific American, 200, 68–86. Hazan, C., & Shaver, P. R. (1987). Romantic love conceptualized as an attachment process. Journal of Personality and Social Psychology, 52, 511–524. Hazan, C., & Shaver, P. R. (1990). Love and work: An attachment-theoretical perspective. Journal of Personality and Social Psychology, 59, 270–280. Hazan, C., & Shaver, P. R. (1994). Attachment theory as an organizational framework for research on close relationships. Psychological Inquiry, 5, 1–22. Hazan, C., & Zeifman, D. (1994). Sex and the psychological tether. In K. Bartholomew & D. Perlman (Eds.), Advances in personal relationships: Vol. 5. Attachment processes in adulthood (pp. 151–177). London: Jessica Kingsley. Hesse, E. (1999). The Adult Attachment Interview: Historical and current perspectives. In J. Cassidy & P. R. Shaver (Eds.), Handbook of attachment: Theory, research, and clinical applications (pp. 395–433). New York: Guilford Press. Higgins, E. T., Bond, R. N., Klein, R., & Strauman, T. J. (1986). Self discrepancies and emotional vulnerability: How magnitude, accessibility, and type of discrepancy influence affect. Journal of Personality and Social Psychology, 51, 5–15. Horowitz, M. J. (1982). Psychological processes induced by illness, injury, and loss. In T. Millon, C. Green, & R. Meagher (Eds.), Handbook of clinical health psychology (pp. 53–68). New York: Plenum Press. Isen, A. M., & Daubman, K. A. (1984). The influence of affect on categorization. Journal of Personality and Social Psychology, 47, 1206–1217. Isen, A. M., Daubman, K. A., & Nowicki, G. P. (1987). Positive affect facilitates creative problem solving. Journal of Personality and Social Psychology, 52, 1122–1131. Kalish, R. A. (1985). Death, grief, and caring relationships. New York: Cole. Keelan, J. P. R., Dion, K. K., & Dion, K. L. (1998). Attachment style and relationship satisfaction. Canadian Journal of Behavioral Science, 30, 24–35. Kennedy, J. H. (1999). Romantic attachment style and ego identity, attributional style, and family of origin. College Student Journal, 33, 171–180. Kirkpatrick, L. A. (1998). Evolution, pair-bonding, and reproductive strategies: A reconceptualization of adult attachment. In J. A. Simpson & W. S. Rholes (Eds.), Attachment theory and close relationships (pp. 353–393). New York: Guilford Press. Kirkpatrick, L. A., & Shaver, P. R. (1990). Attachment theory and religion: Childhood attachments, religious beliefs, and conversion. Journal for the Scientific Study of Religion, 29, 315–334. Kirkpatrick, L. A., & Shaver, P. R. (1992). An attachment-theoretical approach to romantic love and religious belief. Personality and Social Psychology Bulletin, 18, 266–275. Klein, M. (1940). Mourning and its relationship with manic-depressive states. International Journal of Psychoanalysis, 12, 47–82. Kogut, E., & Mikulincer, M. (2002). Adult attachment style and cognitions, affect, and behavior in achievement settings. Manuscript in preparation. Kohut, H. (1971). The analysis of the self. New York: International Universities Press.

THE ATTACHMENT BEHAVIORAL SYSTEM

147

Koski, L. R., & Shaver, P. R. (1997). Attachment and relationship satisfaction across the lifespan. In R. J. Sternberg & M. Hojjat (Eds.), Satisfaction in close relationships (pp. 26–55). New York: Guilford Press. Kunce, L. J., & Shaver, P. R. (1994). An attachment-theoretical approach to caregiving in romantic relationships. In K. Bartholomew & D. Perlman (Eds.), Advances in personal relationships (Vol. 5, pp. 205–237). London, UK: Kingsley. Larose, S., Bernier, A., Soucy, N., & Duchesne, S. (1999). Attachment style dimensions, network orientation, and the process of seeking help from college teachers. Journal of Social and Personal Relationships, 16, 225–247. Lazarus, R. S., & Folkman, S. (1984). Stress, appraisal, and coping. New York: Springer. Levy, K. N., Blatt, S. J., & Shaver, P. R. (1998). Attachment styles and parental representations. Journal of Personality and Social Psychology, 74, 407–419. Levy, M. B., & Davis, K. E. (1988). Love styles and attachment styles compared: Their relations to each other and to various relationship characteristics. Journal of Social and Personal Relationships, 5, 439–471. Lussier, Y., Sabourin, S., & Turgeon, C. (1997). Coping strategies as moderators of the relationship between attachment and marital adjustment. Journal of Social and Personal Relationships, 14, 777–791. Lyons-Ruth, K., & Jacobvitz, D. (1999). Attachment disorganization: Unresolved loss, relational violence, and lapses in behavioral and attentional strategies. In J. Cassidy & P. R. Shaver (Eds.), Handbook of attachment: Theory, research, and clinical applications (pp. 520–554). New York: Guilford Press. Main, M. (1990). Cross-cultural studies of attachment organization: Recent studies, changing methodologies, and the concept of conditional strategies. Human Development, 33, 48–61. Main, M. (1995). Attachment: Overview with implications for clinical work. In S. Goldberg, R. Muir, & J. Kerr (Eds.), Attachment theory: Social, developmental, and clinical perspectives (pp. 407–474). Hillsdale, NJ: Erlbaum. Main, M., & Hesse, E. (1990). Parents’ unresolved traumatic experiences are related to infant disorganized attachment status: Is frightened and/or frightening parental behavior the linking mechanism?. In M. T. Greenberg, D. Cicchetti, & E. M. Cummings (Eds.), Attachment in the preschool years: Theory, research, and intervention (pp. 161–182). Chicago: University of Chicago Press. Main, M., Kaplan, N., & Cassidy, J. (1985). Security in infancy, childhood, and adulthood: A move to the level of representation. Monographs of the Society for Research in Child Development, 50, 66–104. Main, M., & Weston, D. R. (1982). Avoidance of the attachment figure in infancy: Descriptions and interpretations. In C. Parkes & J. Stevenson-Hinde (Eds.), The place of attachment in human behavior (pp. 31–59). New York: Basic Books. Man, K., & Hamid, P. N. (1998). The relationship between attachment prototypes, self-esteem, loneliness, and causal attributions in Chinese trainee teachers. Personality and Individual Differences, 24, 357–371. McCrae, R. R., & Costa, P. T., Jr. (1990). Personality in adulthood. New York: Guilford Press. McCullough, M. E. (2001). Forgiveness: Who does it and how do they do it? Current Directions in Psychological Science, 10, 194–197. Mickelson, K. D., Kessler, R. C., & Shaver, P. R. (1997). Adult attachment in a nationally representative sample. Journal of Personality and Social Psychology, 73, 1092–1106. Mikulincer, M. (1995). Attachment style and the mental representation of the self. Journal of Personality and Social Psychology, 69, 1203–1215.

148

MIKULINCER AND SHAVER

Mikulincer, M. (1997). Adult attachment style and information processing: Individual differences in curiosity and cognitive closure. Journal of Personality and Social Psychology, 72, 1217–1230. Mikulincer, M. (1998a). Attachment working models and the sense of trust: An exploration of interaction goals and affect regulation. Journal of Personality and Social Psychology, 74, 1209–1224. Mikulincer, M. (1998b). Adult attachment style and individual differences in functional versus dysfunctional experiences of anger. Journal of Personality and Social Psychology, 74, 513–524. Mikulincer, M. (1998c). Adult attachment style and affect regulation: Strategic variations in self-appraisals. Journal of Personality and Social Psychology, 75, 420–435. Mikulincer, M., & Arad, D. (1999). Attachment working models and cognitive openness in close relationships: A test of chronic and temporary accessibility effects. Journal of Personality and Social Psychology, 77, 710–725. Mikulincer, M., Birnbaum, G., Woddis, D., & Nachmias, O. (2000). Stress and accessibility of proximity-related thoughts: Exploring the normative and intraindividual components of attachment theory. Journal of Personality and Social Psychology, 78, 509–523. Mikulincer, M., & Erev, I. (1991). Attachment style and the structure of romantic love. British Journal of Social Psychology, 30, 273–291. Mikulincer, M., & Florian, V. (1995). Appraisal of and coping with a real-life stressful situation: The contribution of attachment styles. Personality and Social Psychology Bulletin, 21, 406–414. Mikulincer, M., & Florian, V. (1996). Emotional reactions to interpersonal losses over the life span: An attachment theoretical perspective. In C. Magai & S. H. McFadden (Eds.), Handbook of emotion, adult development, and aging (pp. 269–285). San Diego, CA: Academic Press. Mikulincer, M., & Florian, V. (1997). Are emotional and instrumental supportive interactions beneficial in times of stress? The impact of attachment style. Anxiety, Stress and Coping: An International Journal, 10, 109–127. Mikulincer, M., & Florian, V. (1998). The relationship between adult attachment styles and emotional and cognitive reactions to stressful events. In J. A. Simpson & W. S. Rholes (Eds.), Attachment theory and close relationships (pp. 143–165). New York: Guilford Press. Mikulincer, M., & Florian, V. (1999a). Maternal-fetal bonding, coping strategies, and mental health during pregnancy: The contribution of attachment style. Journal of Social and Clinical Psychology, 18, 255–276. Mikulincer, M., & Florian, V. (1999b). The association between spouses’ self-reports of attachment styles and representations of family dynamics. Family Process, 38, 69–83. Mikulincer, M., & Florian, V. (2000). Exploring individual differences in reactions to mortality salience: Does attachment style regulate terror management mechanisms? Journal of Personality and Social Psychology, 79, 260–273. Mikulincer, M., & Florian, V. (2001). Attachment style and affect regulation: Implications for coping with stress and mental health. In G. J. O. Fletcher & M. S. Clark (Eds.), Blackwell handbook of social psychology: Interpersonal processes (pp. 537–557). Oxford, UK: Blackwell Publishers. Mikulincer, M., Florian, V., Birnbaum, G., & Malishkowitz, S. (2002). The death-anxiety buffering function of close relationships: Exploring the effects of separation reminders on death-thought accessibility. Personality and Social Psychology Bulletin, 28, 287–299. Mikulincer, M., Florian, V., Cowan, P.A., & Cowan, C.P. (2002). Attachment security in couple relationships: A systemic model and its implications for family dynamics. Family Process, 341, 405–434.

THE ATTACHMENT BEHAVIORAL SYSTEM

149

Mikulincer, M., Florian, V., & Hirschberger, G. (2002). The dynamic interplay of global, relationship-specific, and contextual representations of attachment security. Paper presented at the annual meetings of the Society for Personality and Social Psychology conference, Savannah, GA. Mikulincer, M., Florian, V., & Tolmacz, R. (1990). Attachment styles and fear of personal death: A case study of affect regulation. Journal of Personality and Social Psychology, 58, 273–280. Mikulincer, M., Florian, V., & Weller, A. (1993). Attachment styles, coping strategies, and posttraumatic psychological distress: The impact of the Gulf War in Israel. Journal of Personality and Social Psychology, 64, 817–826. Mikulincer, M., Gillath, O., Halevy, V., Avihou, N., Avidan, S., & Eshkoli, N. (2001). Attachment theory and reactions to others’ needs: Evidence that activation of the sense of attachment security promotes empathic responses. Journal of Personality and Social Psychology, 81, 1205–1224. Mikulincer, M., Gillath, O., Sapir-Lavid, Y., Yaakobi, E., Arias, K. Tal-Aloni, L., & Bor, G. (in press). Attachment theory and concern for others’ welfare: Evidence that activation of the sense of secure base promotes endorsement of self-transcendence values. Basic and Applied Social Psychology. Mikulincer, M., Gillath, O., & Shaver, P.R. (2002). Activation of the attachment system in adulthood: Threat-related primes increase the accessibility of mental representations of attachment figures. Journal of Personality and Social Psychology, 83, 881–895. Mikulincer, M., Hirschberger, G., Nachmias, O., & Gillath, O. (2001). The affective component of the secure base schema: Affective priming with representations of attachment security. Journal of Personality and Social Psychology, 81, 305–321. Mikulincer, M., & Horesh, N. (1999). Adult attachment style and the perception of others: The role of projective mechanisms. Journal of Personality and Social Psychology, 76, 1022–1034. Mikulincer, M., Horesh, N., Eilati, I., & Kotler, M. (1999). The association between adult attachment style and mental health in extreme life-endangering conditions. Personality and Individual Differences, 27, 831–842. Mikulincer, M., Horesh, N., Levy-Shiff, R., Manovich, R., & Shalev, J. (1998). The contribution of adult attachment style to the adjustment to infertility. British Journal of Medical Psychology, 71, 265–280. Mikulincer, M., & Nachshon, O. (1991). Attachment styles and patterns of self-disclosure. Journal of Personality and Social Psychology, 61, 321–331. Mikulincer, M., & Orbach, I. (1995). Attachment styles and repressive defensiveness: The accessibility and architecture of affective memories. Journal of Personality and Social Psychology, 68, 917–925. Mikulincer, M., Orbach, I., & Iavnieli, D. (1998). Adult attachment style and affect regulation: Strategic variations in subjective self-other similarity. Journal of Personality and Social Psychology, 75, 436–448. Mikulincer, M., & Selinger, M. (2001). The interplay between attachment and affiliation systems in adolescents’ same-sex friendships: The role of attachment style. Journal of Social and Personal Relationships, 18, 81–106. Mikulincer, M., & Sharir, S. (2002). Adult attachment style and strategies of conflict resolution in marital relationships. Manuscript in preparation. Mikulincer, M., & Shaver, P. R. (2001). Attachment theory and intergroup bias: Evidence that priming the secure base schema attenuates negative reactions to out-groups. Journal of Personality and Social Psychology, 81, 97–115.

150

MIKULINCER AND SHAVER

Mikulincer, M. Shaver, P. R., & Pereg, D. (in press). Attachment theory and affect regulation: The dynamics, development, and cognitive consequences of attachment-related strategies. Motivation and Emotion. Mikulincer, M., & Sheffi, E. (2000). Adult attachment style and cognitive reactions to positive affect: A test of mental categorization and creative problem solving. Motivation and Emotion, 24, 149–174. Morgan, H. J., & Shaver, P. R. (1999). Attachment processes and commitment to romantic relationships. In J. M. Adams & W. H. Jones (Eds.), Handbook of interpersonal commitment and relationship stability (pp. 109–124). New York: Plenum. Ognibene, T. C., & Collins, N. L. (1998). Adult attachment styles, perceived social support, and coping strategies. Journal of Social and Personal Relationships, 15, 323–345. Parker, G., Tupling, H., & Brown, L. B. (1979). A parental bonding instrument. British Journal of Medical Psychology, 52, 1–10. Pereg, D., & Mikulincer, M. (2002). Mood and cognition: The moderating role of attachment style. Submitted for publication. Pietromonaco, P. R., & Feldman Barrett, L. (1997). Working models of attachment and daily social interactions. Journal of Personality and Social Psychology, 73, 1409–1423. Pistole, M. C. (1989). Attachment in romantic relationships: Style of conflict resolution and relationship satisfaction. Journal of Social and Personal Relationships, 6, 505–512. Pistole, M. C. (1993). Attachment relationships: Self-disclosure and trust. Journal of Mental Health Counseling, 15, 94–106. Rholes, W. S., Simpson, J. A., & Orina, M. M. (1999). Attachment and anger in an anxietyprovoking situation. Journal of Personality and Social Psychology, 76, 940–957. Rice, K. G., & Mirzadeh, S. A. (2000). Perfectionism, attachment, and adjustment. Journal of Counseling Psychology, 47, 238–250. Ricks, M. H. (1985). The social transmission of parental behavior: Attachment across generations. In I. Bretherton & E. Waters (Eds.). Growing points in attachment theory and research. Monographs of the Society for Research in Child Development, 50, (1-2, Serial No. 209), 211–227. Roberts, J. E., Gotlib, I. H., & Kassel, J. D. (1996). Adult attachment security and symptoms of depression: The mediating roles of dysfunctional attitudes and low self-esteem. Journal of Personality and Social Psychology, 70, 310–320. Rom, E., & Mikulincer, M. (2002). Attachment theory and group processes: The association between attachment style and group-related representations, goals, memories, and functioning. Submitted for publication. Rothbard, J. C., & Shaver, P. R. (1994). Continuity of attachment across the life span. In M. B. Sperling & W. H. Berman (Eds.), Attachment in adults: Clinical and developmental perspectives (pp. 31–71). New York: Guilford Press. Rubenstein, C., & Shaver, P. R. (1982). The experience of loneliness. In L. A. Peplau & D. Perlman (Eds.), Loneliness: A sourcebook of current theory, research, and therapy (pp. 206–223). New York: Wiley-Interscience. Schachner, D. A., & Shaver, P. R. (2002). Attachment style and human mate poaching. Submitted for publication. Schmitt, D. P., & Buss, D. M. (2000). Sexual dimensions of person description: Beyond or subsumed by the Big Five? Journal of Research in Personality, 34, 141–177. Schwarz, N., & Bohner, G. (1996). Feelings and their motivational implications: Moods and the action sequence. In P. M. Gollwitzer & J. A. Bargh (Eds.), The psychology of action: Linking cognitions and motivation to behavior (pp. 119–145). New York: Guilford Press.

THE ATTACHMENT BEHAVIORAL SYSTEM

151

Shaver, P. R., Belsky, J., & Brennan, K. A. (2000). The adult attachment interview and selfreports of romantic attachment: Associations across domains and methods. Personal Relationships, 7, 25–43. Shaver, P. R., & Brennan, K. A. (1992). Attachment styles and the ‘‘big five’’ personality traits: Their connections with each other and with romantic relationship outcomes. Personality and Social Psychology Bulletin, 18, 536–545. Shaver, P. R., & Clark, C. L. (1994). The psychodynamics of adult romantic attachment. In J. M. Masling & R. F. Bornstein (Eds.). Empirical perspectives on object relations theory (Empirical studies of psychoanalytic theories, Vol. 5, pp. 105–156). Washington, DC: American Psychological Association. Shaver, P. R., Collins, N., & Clark, C. L. (1996). Attachment styles and internal working models of self and relationship partners. In G. J. O. Fletcher & J. Fitness (Eds.), Knowledge structures in close relationships: A social psychological approach (pp. 25–61). Hillsdale, NJ: Erlbaum. Shaver, P. R., & Hazan, C. (1984). Incompatibility, loneliness, and limerence. In W. Ickes (Ed.), Compatible and incompatible relationships (pp. 163–184). New York: Springer. Shaver, P. R., & Hazan, C. (1993). Adult romantic attachment: Theory and evidence. In D. Perlman & W. Jones (Eds.), Advances in personal relationships (Vol. 4, pp. 29–70). Jessica Kingsley. Shaver, P. R., Hazan, C., & Bradshaw, D. (1988). Love as attachment: The integration of three behavioral systems. In R. J. Sternberg & M. Barnes (Eds.), The psychology of love (pp. 68–99). New Haven, CT: Yale University Press. Shaver, P.R., & Mikulincer, M. (2002a). Attachment-related psychodynamics. Attachment and Human Development, 4, 133–161. Shaver, P.R., & Mikulincer, M. (2002b). Dialogue on adult attachment: Diversity and integration. Attachment and Human Development, 4, 243–257. Shaver, P.R., & Mikulincer, M. (in press). The psychodynamics of social judgments: An attachment theory perspective. In J. P. Forgas, K. D. Williams, & W. von Hippel (Eds.), Responding to the social world: Implicit and explicit processes in social judgments and decisions. London: Cambridge University Press. Shaver, P. R., & Norman, A. J. (1995). Attachment theory and counseling psychology: A commentary. Counseling Psychologist, 23, 491–500. Shaver, P. R., Papalia, D., Clark, C. L., & Koski, L. R. (1996). Androgyny and attachment security: Two related models of optimal personality. Personality and Social Psychology Bulletin, 22, 582–597. Simpson, J.A., & Rholes, W.S. (2002). Fearful-avoidance, disorganization, and multiple working models: Some directions for future theory and research. Attachment and Human Development, 4, 223–229. Simpson, J. A., Rholes, W. S., & Nelligan, J. S. (1992). Support seeking and support giving within couples in an anxiety-provoking situation: The role of attachment styles. Journal of Personality and Social Psychology, 62, 434–446. Smith, E. R., Murphy, J., & Coats, S. (1999). Attachment to groups: Theory and measurement. Journal of Personality and Social Psychology, 77, 94–110. Solomon, Z., Ginzburg, K., Mikulincer, M., Neria, Y., & Ohry, A. (1998). Coping with war captivity: The role of attachment style. European Journal of Personality, 12, 271–285. Sroufe, L. A., & Waters, E. (1997). Attachment as an organizational construct. Child Development, 48, 1184–1199. Taubman Ben-Ari, O., Findler, L., & Mikulincer, M. (2002). The effects of mortality salience on relationship strivings and beliefs: The moderating role of attachment style. British Journal of Social Psychology, 41, 419–441.

152

MIKULINCER AND SHAVER

Tidwell, M. C. O., Reis, H. T., & Shaver, P. R. (1996). Attachment, attractiveness, and social interaction: A diary study. Journal of Personality and Social Psychology, 71, 729–745. Torquati, J. C., & Vazsonyi, A. T. (1999). Attachment as an organizational construct for affect, appraisals, and coping of late adolescent females. Journal of Youth and Adolescence, 28, 545–562. Tracy, J. L. Shaver, P. R., Albino, A. W. & Cooper, M. L. (in press). Attachment styles and adolescent sexuality. In P. Florsheim (Ed.), Adolescent romance and sexual behavior: Theory, research, and practical implications. Mahwah, NJ: Erlbaum. Tronick, E. Z. (1989). Emotions and emotional communication in infants. American Psychologist, 44, 112–119. Vaughn, B. E., & Bost, K. K. (1999). Attachment and temperament: Redundant, independent, or interacting influences on interpersonal adaptation and personality development? In J. Cassidy & P. R. Shaver (Eds.), Handbook of attachment: Theory, research, and clinical applications (pp. 198–225). New York: Guilford Press. Veit, C. T., & Ware, J. E. (1983). The structure of psychological stress and well being in general populations. Journal of Counseling and Clinical Psychology, 51, 730–742. Waters, H. S., Rodrigues, L. M., & Ridgeway, D. (1998). Cognitive underpinnings of narrative attachment assessment. Journal of Experimental Child Psychology, 71, 211–234. Wegner, D. M., & Smart, L. (1997). Deep cognitive activation: A new approach to the unconscious. Journal of Consulting and Clinical Psychology, 65, 984–995. Weiss, R. S. (1973). Loneliness: The experience of emotional and social isolation. Cambridge, MA: MIT Press. Weiss, R. S. (1998). A taxonomy of relationships. Journal of Social and Personal Relationships, 15, 671–683. Zuroff, D. C., & Fitzpatrick, D. K. (1995). Depressive personality styles: Implications for adult attachment. Personality and Individual Differences, 18, 253–365.

STEREOTYPES AND BEHAVIORAL CONFIRMATION: FROM INTERPERSONAL TO INTERGROUP PERSPECTIVES

Olivier Klein Mark Snyder

I. Introduction Consider the case of a Belgian employer interviewing an applicant of Moroccan ancestry to determine whether this person is suitable for a highly competitive job. This employer may adhere to the stereotype, widely shared in Belgian society, that Moroccans are unreliable and not committed to their jobs. Guided by this stereotype, the interviewer may employ an interviewing strategy for evaluating this candidate’s suitability for the job in question by asking him to talk about negative experiences in the job market or about interests outside of the workplace on the assumption that unreliable employees will have numerous negative work-related experiences to report and those not involved in their jobs will be most eager to talk about their nonwork interests and activities. This information-gathering strategy is confirmatory in that the evidence that it attempts to gather would tend to be supportive of the interviewer’s belief. Moreover, the candidate, in an effort to be responsive to the interviewer’s questions, may answer these questions as fully and precisely as possible, reporting negative workplace experiences and describing extracurricular interests and activities. In doing so, the candidate may provide evidence in support of the interviewer’s expectations and display behavior that would tend to confirm the stereotype held about people of Moroccan origin in Belgium. In this situation, the very existence of an expectation regarding the candidate set in motion a chain of events that eventually induced behavior consistent with this expectation. This example, loosely based on existing research, illustrates the process known as behavioral confirmation. 153 ADVANCES IN EXPERIMENTAL SOCIAL PSYCHOLOGY, VOL. 35

Copyright 2003 Elsevier Science (USA). All rights reserved. 0065-2601/03 $00.00

154

KLEIN AND SNYDER

In the past several decades, researchers in social psychology have demonstrated, in a variety of laboratory and field contexts, that expectations about other persons can actually induce these persons to adopt behaviors consistent with these expectations (Jussim, Palumbo, Chatman, Madon, & Smith, 2000; Miller & Turnbull, 1986; Neuberg, 1994; Snyder, 1992; Snyder & Stukas, 1999). In one commonly used procedural paradigm (e.g., Snyder, Tanke, & Berscheid, 1977), two participants placed in separate experimental rooms communicate via a telephone system. Prior to the interaction, the experimenter provides one of the participants (the perceiver) with information regarding the personal attributes of the other participant (the target), either directly (such as by providing access to personality test information indicating that the target is, for example, an extravert or an introvert) or indirectly by revealing the category membership of the target (such as by providing a photograph that reveals, for example, the appearance, ethnicity, gender, or weight of the target) and relying on stereotypes about that category to generate expectations about the target. Actually, in investigations of behavioral confirmation, the expectation is defined randomly and is independent of the actual characteristics of the target; thus, perceivers are randomly assigned to conditions in which they are led to expect that their interaction partners are, for example, extraverts or introverts, attractive or unattractive, obese or normal weight, females or males. The ensuing interaction between perceiver and target is tape-recorded for later rating by independent judges, blind to conditions, of the contributions of perceiver and target to the interaction. Perceptual confirmation is said to occur when, after the interaction, the perceiver views the target in a direction consistent with initial expectations. Behavioral confirmation is evidenced if the target’s personality, as rated by judges who listen to tape recordings of the target’s contributions to the interaction, differs in the two experimental conditions in the same direction. Both perceptual and behavioral confirmation effects have been documented, although there are limiting conditions to their occurrence (Miller & Turnbull, 1986; Snyder, 1984, 1992). Two types of behavioral confirmation processes can and have been investigated in such procedural paradigms, and these two types of behavioral confirmation can be distinguished as a function of the source and the nature of the expectation. In the example involving the Belgian employer and the Moroccan candidate, the expectations that we have focused on derive directly from a social categorization of the target by the employer and of the activation of social stereotypes associated with the target’s category membership. Such an interaction, although it involves only two people, can be regarded as essentially an intergroup situation because

STEREOTYPES AND BEHAVIORAL CONFIRMATION

155

the perceiver views the target in terms of his or her social identity (that is, as a prototypical member of a social category) rather than as an idiosyncratic individual (Brown, 1988; Moya, 1998; Tajfel & Turner, 1986). We shall label such a case of behavioral confirmation as an instance of social stereotype confirmation. Of course, stereotypes based on a target’s category membership are but one source of expectations (Olson, Roese, & Zanna, 1996). Quite possibly, our hypothetical Belgian interviewer, for idiosyncratic reasons unrelated to the job candidate’s national origin, may expect the Moroccan candidate to be a shy person, rather than uncommitted to his job. In this instance, the perceiver’s expectations are unrelated to the target’s group membership because the trait ‘‘shy’’ is not stereotypical of Moroccans. It can therefore be viewed as a purely interpersonal expectation. Yet, by adopting a confirmatory information-gathering strategy in the job interview, the interviewer may lead the applicant to behave in accordance with this expectation as well. We shall label this case of behavioral confirmation an instance of personal expectation confirmation.1 Much of the interest in behavioral confirmation processes derives from the role these phenomena may play in the maintenance of stereotypes and in the perpetuation or reproduction of the social structure (Claire & Fiske, 1998; Deaux & Major, 1987; Jussim & Fleming, 1996; Merton, 1948). Indeed, if members of advantaged groups can influence members of disadvantaged groups into performing the behaviors that confirm their negative expectations, they may thereby reinforce their privileged status in society. For example, if, time and again, Belgian interviewers could systematically influence North African interviewees into providing evidence of unreliability or incompetence, they could then use these behaviors as evidence that North Africans deserve their disadvantaged position in society and that equal opportunity policies and practices should not be implemented. When considering these possible societal consequences of behavioral confirmation for understanding intergroup relations and the relative positions of groups within the structural organization of society, the distinction between personal expectation confirmation and social stereotype confirmation takes on particular importance. The personal expectations that are brought to bear on social interactions may be as diverse as the individuals who hold those expectations and the individual targets of 1

Some instances of behavioral confirmation lie between these two extremes. This is the case when the source of the expectation is influenced by the category membership of the target whereas the perceiver is not. For example, our Belgian interviewer may have heard from a prejudiced colleague that the target possessed stereotypical traits. A perceiver possessing such stereotypical expectations may still view the target in terms of his or her personal identity.

156

KLEIN AND SNYDER

those expectations. Hence, the consequences of the confirmation of personal expectations may be confined to individual pairings of holders and targets of expectations and, as such, will not be generalized to large segments of low status groups. On the other hand, social stereotypes are widely shared, collective representations (Schaller & Conway, 2001; Stangor & Schaller, 1996; Tajfel, 1981b). If they are repeatedly confirmed in social interaction, they can contribute to the persistence of intergroup stereotypes and behaviors based on those stereotypes, such as discrimination, that serve to maintain existing patterns of intergroup relations. For this reason, in our theoretical analysis, we shall mainly be concerned with social stereotype confirmation as a form of behavioral confirmation. In spite of the assumed ‘‘intergroup’’ implications of behavioral confirmation processes, the methodological options generally pursued in research on behavioral confirmation, especially in laboratory experiments, do pose some difficulties for, and hence place some limits on, our ability to make the transition from the interpersonal level at which research is typically conducted (e.g., studies of interaction between individual perceivers and individual targets) to the intergroup level of analysis (e.g., the implications of behavioral confirmation for understanding intergroup relations, including the perpetuation of widely shared social stereotypes and the maintenance of the relative positions of power and influence of groups within society). First, studies relying on variations in the (expected) social category membership of the target (that is, studies of social stereotype confirmation) have been comparatively rare in comparison with those manipulating expectations regarding personality traits (that is, studies of personal expectation confirmation). For example, membership in an ethnic minority has only rarely been manipulated (for exceptions, see Chen & Bargh, 1997; Chidester, 1986; Word, Zanna, & Cooper, 1974) and to our knowledge, on experiment on behavioral confirmation has manipulated expectations about the sexual orientation of the target. Second, even when categories have been used to define expectations in studies of behavioral confirmation, such studies have generally relied on situations in which the target did not truly belong to the stigmatized group. For example, targets described as ‘‘obese’’ (on the basis of a snapshot) in the Snyder and Haugen studies (1994, 1995) were not more likely to be overweight than other students and the supposedly black targets in the second study of Word et al. (1974) actually were white. Yet, when targets actually are nonstigmatized group members, their reactions to interactions with individuals who hold stereotyped expectations about them may differ from those of a truly stigmatized target. They are likely to be shaped in a large part by their personal history as a group member and

STEREOTYPES AND BEHAVIORAL CONFIRMATION

157

by the position of their group in the social structure (Claire & Fiske, 1998; Miller & Myers, 1998; Pinel, 1999). For example, our Moroccan job candidate may have been repeatedly confronted with prejudiced Belgians and may therefore have developed interaction strategies and tactics that allow him or her to project a favorable image in such contexts. Or, conversely, membership in a disadvantaged group may limit one’s opportunities to enact these strategies and tactics, perhaps because the interviewer may be motivated to end the interaction more quickly as a result of prejudiced attitudes toward members of that group (Devine, Evett, & Vasquez-Suson, 1996; Hebl & Kleck, 2000). Membership in ‘‘real’’ groups may therefore expand or constrain the behavioral opportunities available to them when interacting with others who hold stereotyped expectations of them. In this chapter, we seek to address these gaps in existing theory and research on behavioral confirmation, in order to further understand behavioral confirmation as an intergroup phenomenon, with its attendant implications for the perpetuation of social stereotypes, especially those about disadvantaged groups within society, and the maintenance of relations between groups within society. To do so, we will build a bridge between the processes of stigmatization and those of behavioral confirmation, drawing on the large body of literature examining dyadic interactions between members of stigmatized and nonstigmatized groups. More precisely, we shall try to answer two key questions. First, how does stigmatization of one of the parties to a social interaction affect stereotype confirmation? Hence, functionally, we shall focus on situations in which the nonstigmatized party is the perceiver, holding stereotypes, and the stigmatized party is a target, or ‘‘victim’’ of stereotypes. In line with Crocker, Major, and Steele’s definition, we shall define a stigmatized individual as ‘‘possessing (or believed to possess) some attribute, or characteristic, that conveys a social identity that is devalued in a particular social context’’ (Crocker, Major, & Steele, 1998, p. 505). In this respect, our analysis will be broad, and will not make explicit distinctions between different types of stigmatizing attributes that can be brought to bear on individuals as a result of their membership in groups about whom negative social stereotypes are held. Second, we shall ask: How does stereotype confirmation contribute to the persistence of stereotypes and oppressive intergroup relations guided by these stereotypes? As a corollary to these considerations of how the largescale consequences of behavioral confirmation can potentially constrain the opportunities of disadvantaged groups in society, we will also consider the implications of behavioral disconfirmation for the modification of social stereotypes and for changes in patterns of intergroup relations.

158

KLEIN AND SNYDER

Our plan for addressing these two questions is to begin by proposing and articulating some key conceptual and theoretical distinctions. These distinctions concern the processes involved in behavioral confirmation, and the behavioral styles displayed in interactions between the nonstigmatized and the stigmatized. These distinctions will then be used to ascertain the impact of the target’s membership in a stigmatized group on the occurrence of behavioral confirmation. We shall try to assess this impact in terms of three types of variables—variables related to the nonstigmatized perceiver, variables related to the stigmatized target, and sociostructural variables related to the perceiver–target dyad considered as a unit. Based on these considerations, we will then try to draw implications for understanding when and why the confirmation of expectations can contribute to the persistence of social stereotypes, as well as when and why their disconfirmation will lead to the modification of social stereotypes.

II. Processes Involved in Behavioral Confirmation In accord with the plan that we have just laid out, let us begin by delineating two processes that, based on relevant theorizing and research, may underlie behavioral confirmation when it occurs in interactions between perceivers and the targets of their stereotype-based expectations.

A. RECIPROCATION STRATEGY According to interaction adaptation theories (e.g., Argyle & Dean, 1965; Burgoon, 1978; Giles, Coupland, & Coupland, 1991; Knowles, 1980), interactants can adopt two main strategies to respond to their partner’s behavior. Reciprocation involves matching the partner’s behavior by displaying a similar level of friendliness and warmth. Compensation involves, on the contrary, moving away from one’s partner’s interpersonal style (e.g., by behaving more warmly as a response to a partner’s cold responses). Generally, reciprocation is thought to be the default strategy, but compensation can and does occur when the partner greatly violates expectations (Burgoon, Le Poire, & Rosenthal, 1995). For example, if a partner who is expected to be warm and friendly suddenly behaves coldly, displays of increased warmth can be used to restore the quality of the interaction. According to an interpretation of behavioral confirmation based on these concepts, behavioral confirmation in social interaction can precisely be

STEREOTYPES AND BEHAVIORAL CONFIRMATION

159

described as involving reciprocal patterns of behaviors on the part of the perceiver and target (see, e.g., Burgoon et al., 1995; Jones & Panitch, 1971; Snyder, 1984; Word et al., 1974). For example, in the study by Snyder et al. (1977), expectations were manipulated by presenting to perceivers a picture of an attractive or unattractive target (this picture was of course independent of the real physical characteristics of the targets). Perceivers in the ‘‘attractive condition’’ may have relied on their stereotypes to anticipate friendly and warm behavior on the part of the target. In anticipation of such behavior, they may have made warm and friendly overtures to the target, thereby reciprocating their expectations with these behaviors. In turn, the target reciprocated the friendly behavior of the perceiver, which yielded an impression of the target as actually friendly and warm. A parallel process can explain the target’s behavior in the ‘‘unattractive’’ condition, in which the perceiver’s cool and distant overtures are reciprocated by cool and distant reactions from the target. Evidence that the perceiver’s behavior, as evidenced by both verbal and nonverbal indices, is matched by the target’s corresponding behavior can be taken as supportive of this interpretation of the dynamics of behavioral confirmation in social interaction.

B. CONFIRMATORY STRATEGY A second intepretation of behavioral confirmation is based on the finding that subject to certain limiting conditions, perceivers generally attempt to confirm their initial expectations of their targets (Erber & Fiske, 1984; Fiske & Neuberg, 1990; Snyder & Campbell, 1980; Snyder, Campbell, & Preston, 1982; Snyder & Cantor, 1979). This confirmatory orientation may lead perceivers to ‘‘bias’’ their interaction strategy such that targets have relatively great opportunities to behave in accord with the perceivers’ expectations. Thus, in interview formats for studying behavioral confirmation, perceivers are likely to ask leading questions that provide targets with opportunities to talk about themselves in ways that would tend to confirm the expectation at hand but that are difficult to answer in a ‘‘disconfirming’’ manner (Neuberg, 1994; Snyder & Campbell, 1980; Snyder et al., 1982; Snyder & Cantor, 1979). The success of the perceiver’s confirmatory strategy depends, to some extent, on rules of social etiquette and norms of conversational practice that favor a smoothly flowing and responsive pattern of conversation, one in which the target, in response to the topics of conversation laid down by the perceiver, answers these questions with docility, never trying to assert her own selfviews. Such an interactional orientation on the part of the target is called ‘‘deferential’’ (Smith, Neuberg, Judice, & Biesanz, 1997) and is thought to be an expression of a ‘‘getting along agenda’’ motivated by the desire to have a

160

KLEIN AND SNYDER

smoothly flowing and pleasing interaction with the perceiver (Snyder, 1992; Snyder & Haugen, 1995). The success of the confirmatory strategy in eliciting behavioral confirmation is premised on the existence of a power differential between perceiver and target, as the perceiver needs to be able to impose his or her ‘‘script’’ on the target for behavioral confirmation to occur (Copeland, 1994; Neuberg, 1994, 1996; Snyder & Kiviniemi, 2001). This power differential can be the case either because the roles occupied by perceiver and target are associated with specific prescriptions or because the perceiver controls outcomes valued by the target. This power differential, however, need not be a formal one: A perceiver can, for example, feel subjectively that he or she deserves more power in the interaction because he or she feels more intelligent or competent than the target. Manifestations of this claim to greater power can be accepted and go unchallenged by the target. Indeed, empirical evidence shows that when the perceiver has greater power (Copeland, 1992, 1994; Harris, Lightner, & Manolis, 1998) or status (Virdin & Neuberg, 1990), behavioral confirmation is more likely to occur. When the target is deprived of power, she or he is even more likely to pursue a ‘‘getting along’’ agenda (Copeland, 1994), thereby facilitating behavioral confirmation.

III. Interactions between Stigmatized and Nonstigmatized Individuals Next, following along with the plan we have laid out, let us examine two styles of interactions that nonstigmatized individuals can and do adopt in their dealings with the stigmatized, and consider how these two styles may lead the targets of stigmatizing expectations (especially those based on stereotypes about the groups to which the targets belong) to provide behavioral confirmation for these expectations. These two styles of interaction we label, respectively, avoidance and dominance.

A. AVOIDANCE The ‘‘avoidance style’’ is a pattern of behaviors, both verbal and nonverbal, that tends to increase the perceiver’s psychological and interpersonal distance from the target. Verbal behaviors, such as low self-disclosure or early interruption of the interaction, can be characterized as avoidant to the extent that they serve to increase the symbolic distance between two interactants. As well, nonverbal behaviors such as reduced eye contact, large interaction distance, backward lean, and silence are indicative of avoidance. The

STEREOTYPES AND BEHAVIORAL CONFIRMATION

161

avoidant style, when practiced by the nonstigmatized in their dealings with the stigmatized, can be contrasted with the moderately friendly behavior generally displayed toward other nonstigmatized group members. For example, when interacting with disabled individuals, able-bodied individuals tend to terminate interviews sooner (Kleck, 1969) and to distance themselves more from their partner (Kleck, 1968) than when interacting with other ablebodied individuals. They are also more likely to avoid the interaction altogether if it is possible to do so (Snyder, Kleck, Strenta, & Mentzer, 1979). To the extent that the nonstigmatized manage to escape contact with the stigmatized, their stereotypes about them will, of course, go unchallenged and will, by default, persist. Hence, the absence of disconfirmation is often similar, in its effects, to confirmation. Moreover, even when the nonstigmatized have contact with the stigmatized, they may succeed in psychologically distancing themselves from the stigmatized through the use of avoidant styles. In doing so, they may also be functionally engaging in a confirmatory interactional strategy that will set the stage for behavioral confirmation of negative stereotypes about the stigmatized. For, if in accord with the principle of reciprocation that we have already articulated, the stigmatized target reciprocates this interactional style and matches the level of avoidance, detachment, and distancing displayed by the nonstigmatized, the perceiver’s behavior will indirectly lead to the behavioral confirmation of negative expectations regarding the sociability of the target. For example, if a European-American perceiver expects an AfricanAmerican target to be hostile and hence behaves in an avoidant way, the African-American target may then respond by avoidant behavior as well (for an empirical illustration of this chain of events, see Word et al., 1974). This avoidant behavior can be interpreted as diagnostic of ‘‘hostility’’ or ‘‘coldness’’ and hence confirming of the perceiver’s expectations. More generally, in such situations, negative expectations about a target individual, stigmatized because of his or her membership in a category about whom negative stereotypes are held, are likely to be confirmed. And, by extension, so too are the more general social stereotypes held by the nonstigmatized about the entire group to which the stigmatized target belongs confirmed by the events of an encounter built around an avoidant style of interaction

B. DOMINANCE Another style of interaction, likely to be adopted by members of nonstigmatized groups in dealing with members of stigmatized groups, can be characterized as ‘‘dominant’’—that is, one involving attempts to control or manipulate the behavior of the other person (Maccoby & Jacklin, 1974).

162

KLEIN AND SNYDER

Behaviors characteristic of the dominant interactional style are visual dominance (i.e., looking at the other person more when speaking than when listening), rapid speech, a relaxed posture, a firm and loud tone of voice, orders, and interruptions of the partner. This style also involves attempts to influence the other person and to be less likely to listen to him or her. The counterpart of this style is a submissive style marked by little talking, tentative speech, little eye contact when speaking, and deference to the partner’s injunctions.2 Empirical evidence suggests that members of dyads and small groups or dyads tend to behave submissively in the presence of an interactional partner who displays a dominant style (e.g., Berger, Cohen & Zelditch, 1972; Ridgeway, 1987; Ridgeway & Berger, 1988). If stereotypes of a stigmatized social group depict members of this group as low in competence, for example (as do stereotypes of many disadvantaged groups, such as African-Americans in the United States and North Africans in Belgium), adoption of a submissive style by members of these groups in response to a dominant interactional style on the part of the nonstigmatized may lead to stereotype confirmation. Indeed, a submissive style is typically viewed as indicative of low intelligence and low task competence; for example, Ridgeway (1987) observed that individuals adopting this style were judged as having a lower GPA than those adopting a dominant style. Moreover, adoption of this submissive style in response to dominant overtures can also be considered as indicative of an inherent lack of assertiveness or leadership, a trait that is associated with many stigmatized groups (Fiske, Cuddy, Glick, & Xu, 2002; Fiske, Xu, & Cuddy, 1999). Most importantly, acceptance of the perceiver’s attempts at asserting his or her power may contribute to the success of the perceiver’s confirmatory strategy of acting on his or her expectations about the members of stigmatized groups. For, when the target defers to the perceiver, and explicitly or implicitly accepts his or her power by behaving submissively, this strategy is most likely to be successful at eliciting behavioral confirmation. Considered together, the avoidant and dominant interactional styles of members of nonstigmatized groups in their dealings with members of stigmatized groups may be seen as attempts to exercise power—the power to control whether or not to have any dealings with the targets of stigmatized groups and the power to dominate and control one’s dealings with them. 2 Unlike Ridgeway (1987), and to simplify our terminology, when we refer to dominant behaviors, we include not only behaviors that directly attempt to control the partner’s behavior but also those that do so indirectly, and sometimes unintentionally, by manifesting the actor’s high level of competence or status. Conversely, submissive behavior includes not only behavior manifesting an acceptance of this control but also behavior manifesting low task competence or low status.

STEREOTYPES AND BEHAVIORAL CONFIRMATION

163

IV. The Perceiver’s Perspective Now that we have presented the building blocks of our analysis— reciprocation and confirmatory strategies as mechanisms of behavioral confirmation, and avoidance and dominance as interactional styles that members of nonstigmatized groups bring to bear on their interactions with members of stigmatized groups—let us proceed to examine how stigmatization of one of the interaction partners (the ‘‘target’’) can affect the occurrence of behavioral confirmation in dyadic interactions. To do so, we shall extrapolate from lessons learned from the literature on interactions between the stigmatized and the nonstigmatized and, by building on the distinctions that we have already articulated, examine factors facilitating the confirmation of the stereotypes held by the nonstigmatized about the stigmatized. To accomplish these goals, we first concentrate on factors related to the nonstigmatized person’s (that is, the perceiver’s) perspective and examine which factors can lead him or her to adopt behavioral styles conducive to behavioral confirmation, especially avoidance and dominance. Specifically, we will give special attention to the role of social categorization, a process that we view as necessary for the activation of group-based expectations. But, as well, we shall examine the role of factors such as the purpose of the interaction, the levels of prejudice and anxiety of the perceiver, and the content of the stereotypes being brought to bear on the interaction between perceiver and target—all of which may also contribute to the confirmation of social stereotypes.

A. SELF- AND OTHER-CATEGORIZATION AND BEHAVIORAL CONFIRMATION In intergroup contexts, the activation of stigmatizing expectations about the target based on stereotypes about the group to which the target belongs requires, of course, that this target first be categorized as a member of the stigmatized group. If, as we have proposed, the use of avoidant and dominant behavioral styles by the perceiver derives from the activation of stereotypes based on this categorization, it is important to consider the factors that may lead to construe the stigmatized as a member of the social category to which these stereotypes are associated. Accordingly, let us examine relevant theoretical perspectives on the process of categorization as it occurs in interactions between the nonstigmatized and the stigmatized. As well, let us examine the empirical evidence relevant to the proposition that

164

KLEIN AND SNYDER

perceivers do indeed rely on avoidant and dominant behavioral repertories when interacting with a target who has been categorized as a member of a stigmatized group. To understand the determinants and consequences of the categorization process, we shall draw on self-categorization theory (SCT; Turner, Hogg, Oakes, Reicher, & Wetherell, 1987). The main assumption of this theory is that social categorization is a flexible process and depends on the social context. Individuals can define themselves and others as members of different groups. They can also view themselves and others as individuals and define them in terms of their ‘‘personal’’ identity. For example, our Belgian interviewer may either perceive the Moroccan candidate as a typical Moroccan, interchangeable with other Moroccans, or as a unique individual, possessing an idiosyncratic personality that differentiates him from other candidates. What are the consequences of categorizing the target as a member of an outgroup? According to SCT, such a categorization will be associated with expectations that will tend to involve traits differentiating members of the outgroup from those of the ingroup. In this case, members of the ingroup will be perceived as similar in terms of a common ‘‘self-stereotype’’ whereas outgroup members will be viewed in terms of an outgroup stereotype. Thus, a correlate of categorizing the target as a member of an outgroup involves the tendency to self-categorize as a member of an ingroup (Hogg & Abrams, 1988; Hogg & Turner, 1987; Simon & Hamilton, 1994) and to behave in terms of the norms of this ingroup. By contrast, when the perceiver views himself or herself at a personal level, as an idiosyncratic individual, he is expected to view the target at the same level of abstraction. A theoretical implication of this analysis is that behavioral confirmation processes will take the form of personal expectation confirmation when the perceiver’s personal identity is salient, whereas behavioral confirmation processes will concern stereotype confirmation when the perceiver defines himself or herself at a group level. According to SCT, individuals defining themselves in terms of a social category tend to adopt behaviors, and to expect other ingroup members to adopt behaviors, construed as typical of the ingroup at the same time as they expect outgroup members to adopt behaviors typical of their group, that is to enact the outgroup stereotype (Turner et al., 1987; Turner, 1991). Different aspects of these stereotypes may be activated as a function of the social context as it affects the salience of the perceiver’s self-categorization. For example, in the context of sports, a white perceiver may expect an African-American target to be particularly athletic, and self-define as not particularly gifted in sports. This categorization and the attendant activation of stereotyped-based expectations may lead to a form of

STEREOTYPES AND BEHAVIORAL CONFIRMATION

165

submissiveness on the part of the white perceiver, for example, if both individuals are part of a sports team. By contrast, in the context of a mathematics contest, traits related to African-Americans’ athletic skills may not be activated whereas traits related to academic achievement (in particular, those relative to poor performance in mathematics on the part of the out-group) will be activated. In such a context, the white perceiver may self-define as able in mathematics and adopt a dominant behavioral style of dealing with an African-American partner. SCT also suggests that attraction is influenced by the use of different selfand other-categorizations. According to this theory, attraction to other group members is a function of their similarity to the ingroup prototype. This prototype is made of the traits that best differentiate the ingroup from the relevant outgroup. An implication of this assumption is that individuals tend to experience positive affects toward prototypical ingroup members and aversion toward individuals they view as stereotypical members of an outgroup. Consistent with this assumption, attraction to other ingroup members is a function of the extent to which they match the groups’ prototypical norms and values (for reviews, see Hogg, 1987, 1992; Hogg & Hardie, 1991). Hence, if the target is categorized as a member of an outgroup, perceivers may both activate negative expectations and be motivated to distance themselves from the target whereas the reverse should occur if the target is construed as a member of a psychological ingroup. According to this logic, an avoidance-oriented behavioral style should particularly be present in perceivers dealing with targets perceived as prototypical outgroup members. To more precisely specify the applicability of the assumptions and propositions of SCT to our present concerns with the confirmation of social stereotypes in intergroup interactions, we find it useful to make a distinction between two types of interactions between stigmatized and nonstigmatized individuals, both of which are represented in the empirical literature: In the first type of interaction, individuals engage in ‘‘getting-acquainted’’ interactions, in which the purpose of the interaction is specifically to meet and to get to know each other. Sometimes, although not always, this purpose is explicitly conveyed to the perceiver. For example, a participant designated as the ‘‘interviewer’’ may be asked to interview another participant, the ‘‘candidate,’’ in order to form an impression of his personality. In such an instance, the perceiver is said to be in an ‘‘assessment set’’ (Hilton & Darley, 1991). In other instances, participants find themselves in a laboratory with the opportunity to converse with another person; although not necessarily explicitly described as such, it is not unreasonable to assume that one goal of interactions that occur in such a situation is to get acquainted with one’s conversational partner (see, e.g.,

166

KLEIN AND SNYDER

Ickes, 1984). In such getting-acquainted situations, perceivers are generally primarily interested in information regarding the sociability and morality of the target; they search for this type of information first and devote more attention to it (De Bruin & Van Lange, 2000). In the second type of interaction, the purpose of the interaction is to cooperatively perform a task. In such interactions, in which task performance leads the agenda, impression formation is only a secondary goal of the interaction. Perceivers in such interactions may be said to be in an ‘‘action set’’ (Hilton & Darley, 1991). Here, with the success of the task being the chief purpose of the interaction, individuals may be particularly interested in traits relevant to the success of the task. If the task requires intellectual skills (as is usually the case in such studies), information regarding the intellectual capacities of the target may be particularly likely to be sought. These two kinds of interactions provide a context for articulating the role of social categorization processes in the confirmation of social stereotypes. According to SCT (but see also, other theoretical frameworks: Fiske, 1998; Hilton, 1998; Snyder, 1998), perceivers should define the target in terms of trait dimensions that are relevant to the interaction’s goal, and are thereby made more accessible. If a social categorization can account meaningfully for differences and similarities between the perceiver and the target on these dimensions, this category should then become salient (Oakes, 1987; Simon, Hastedt, & Aufderheide, 1997; van Knippenberg & Dijksterhuis, 2001) and, for our purposes, more likely to engage interactional styles that culminate in behavioral confirmation. The question becomes: What trait dimensions are particularly likely to become salient in interactions that revolve around considerations of getting acquainted and those that focus on considerations of task performance? In getting-acquainted interactions, social categorization may be organized around the dimensions of sociability. For example, if the target is black and the perceiver is white, and if blacks are expected to be hostile or otherwise undesirable interaction partners, a categorization in terms of race may make stereotypical attributes associated with racial categorization salient. In this case, the black target will be perceived as different from the white perceiver in terms of sociability (e.g., as hostile or cold). Avoidance is typically an anticipated reciprocation of these traits (Burgoon et al., 1995; Ickes, Patterson, Rajecki, & Tanford, 1982). Besides, when the purpose of the interaction is simply to get acquainted, perhaps as a prelude to developing some form of social relationship, attraction toward one’s ingroup and aversion for the outgroup may be likely to guide the nonstigmatized person’s behavior toward outgroup members. Accordingly, an avoidanceoriented interactional style can be an expression of the aversion toward prototypical outgroup members. We expect this scenario involving an

STEREOTYPES AND BEHAVIORAL CONFIRMATION

167

avoidance-oriented interactional style on the part of the perceiver to be particularly prevalent in ‘‘getting-acquainted’’ interactions, as sociability is the focus of these interactions. Consider now an interaction focusing on performance of a task requiring intellectual skills. If the target is black, and if blacks are expected to be less intelligent than whites, categorization in terms of race may be meaningful and make racial categorization salient. The perceiver will then define himself as an ‘‘intelligent’’ white interacting with an ‘‘incompetent’’ black. According to SCT, salience of this categorization should then shape the perceiver’s behavior and lead to the adoption of behavior that is stereotypical of the ingroup. The activation of stereotypes of the ingroup as competent would then tend to produce a dominant behavioral style (Ridgeway, 1991). More generally, we would expect this pattern to be particularly likely in ‘‘task-oriented’’ interactions, as these interactions generally demand intellectual skills. Altogether, this analysis suggests that as compared with situations in which they interact with other ingroup members, nonstigmatized perceivers should be more likely to display an avoidant style in getting-acquainted interactions and a dominant style in task-oriented interactions with outgroup members. As we have seen, these styles favor the emergence of stereotype confirmation if the target adopts the complementary avoidant or submissive style. So far, we recognize that this analysis has been somewhat speculative. Therefore, we shall now review evidence relevant to the predictions derived from it by examining studies of factors affecting category salience (from the perceiver’s perspective). It follows from our analysis that these factors should enhance the use of avoidant behaviors in getting-acquainted interactions and the use of dominant behavior in taskoriented interactions. How do perceivers behave in the presence of an outgroup, as opposed to an ingroup member? In mixed settings, involving interactions of ingroup members with outgroup members, perceivers are more likely to define themselves in terms of their social identity than when interacting with other members of their ingroup (Haslam & Turner, 1992). We shall therefore examine whether perceivers behave differently when interacting with stigmatized partners than other members of their own (nonstigmatized) ingroup. Then, we shall consider the influence of factors that should enhance category salience over and above the influence of the target’s membership in a stigmatized outgroup. Consider first studies that involve getting-acquainted interactions. Vorauer and Kumyhr (2001) noted that white Canadians were more likely to experience negative feelings oriented toward others (such as hostility or anger at others) when interacting with an aboriginal Canadian than with another white Canadian. Similarly, women can elicit distancing behaviors

168

KLEIN AND SNYDER

from men in unstructured getting-acquainted interactions. For example, Saris, Johnson, and Lott (1995) found that men tended to distance themselves more from a woman when approached by a (male) investigator wearing a feminist T-shirt than when this investigator wore a blank T-shirt. It is likely that wearing this T-shirt made the gender categorization salient and elicited distancing from women. The literature on interactions between able-bodied and disabled individuals provides convergent findings. Individuals are more likely to display avoidant behavior in the presence of a disabled than an able-bodied individual (for a review, see Hebl & Kleck, 2000). In addition, able-bodied individuals show lesser variability in their interactional styles when they interact with disabled persons than they do with other able-bodied interaction partners (Kleck, Ono, & Hastorf, 1966), suggesting that they may be particularly likely to categorize disabled persons as interchangeable members of an outgroup to be treated one and all in the same ways. Several studies have documented the presence of avoidant nonverbal behavior on the part of whites in studies of unstructured getting-acquainted interactions between American whites and blacks (Crosby, Bromley, & Saxe, 1980; Hendricks & Bootzin, 1976; Ickes, 1984; Weitz, 1972; Word et al., 1974). For example, Ickes found that both members of a dyad including a prejudiced white behaved less warmly when the experimenter was black than white (Ickes, 1984). Presumably, being in a minority increased the salience of the ingroup vs. outgroup categorization, from the white’s perspective (Oakes, 1987; Simon & Brown, 1987; Taylor, 1979). This factor may in turn have elicited negative behaviors in the black target. What about task-oriented interactions? Consistent with our predictions, interactions between blacks and whites in task groups also reveal differences in dominance. In task-oriented situations, blacks tend to behave less assertively in the presence of whites (Adams, 1980; Cohen & Roper, 1972; Cohen, 1982; Katz & Benjamin, 1960) and tend to rely on more submissive behavior than their white partners. Conversely, white members of a mixed task group tend to become the leader of their group even if they are a minority of one (Kelsey, 1998). Studies conducted with other ethnic minorities reveal a similar pattern (for a review, see Ridgeway, 1991). Similarly, the literature on gender in task situations suggests that men are more likely to adopt dominant behavioral styles when interacting with women than when interacting with other men. For example, they make more suggestions and engage in more active task behavior (Wood & Karten, 1986), they are more likely to be selected leader than women (Eagly & Wood, 1991), they show more visual dominance (Ellyson, Dovidio, & Brown, 1992), and they interrupt women more (Smith-Lovin & Brody, 1989; Zimmerman & West, 1975). These differences are typically absent in same-sex discussions

STEREOTYPES AND BEHAVIORAL CONFIRMATION

169

(Carli, 1990, 1991; Ridgeway & Smith-Lovin, 1999). Hence, these studies suggest that in task groups, men tend to use more dominant behaviors in the presence of women than in the presence of other men. This dominant behavioral style can also be coupled with manifestations of avoidance: For example, in a study using dyads cooperating on a task, Lott (1987) found that men simultaneously distanced themselves from their partner more if this partner was a woman than a man (a manifestation of avoidance) but followed their advice less and made more negative comments (manifestations of dominance). Now that we have reviewed studies comparing nonstigmatized group members’ behaviors in the presence of ingroup versus outgroup members, we shall consider several factors that are likely to enhance the salience of the perceiver’s self-categorization when he or she interacts with an outgroup member. One factor likely to affect the self-categorization process on the part of the perceiver is the typicality of the target in regard to the perceiver’s self-categorization. A target displaying traits that are perceived to be typical of his or her group is more likely to be categorized as a member of this group. In a study of interactions between Hindus and Muslims in India, Islam and Hewstone (1993) found that interactions with typical outgroup members are associated with more intergroup anxiety, typically a source of avoidant behavior (Daly, 1978; Pancer, McMullen, Kabatoff, Johnson, & Pond, 1979; Stephan & Stephan, 1985). Similarly, Wilder (1984, experiment 3) observed that a target was rated as a less desirable interaction partner when that target was a typical rather than an atypical member of a rival college. Of course, these results need to be considered in the context of the competitive relations between the groups involved in these studies, the next factor that we shall consider here. Intergroup competition tends to make group membership salient: Sherif’s classic boy camps studies (Sherif, 1966) show that group members develop more avoidant behaviors when interacting with outgroup members in the context of competitive rather than cooperative relations. As a function of competition, group members not only displayed hostile behaviors but also developed attitudes of social distance toward outgroup members. These results have been replicated in organizational contexts (Blake & Mouton, 1964) and more recent studies also have provided consistent findings (Gaertner et al., 1990, 1999). A third factor affecting the salience of an ingroup versus an outgroup categorization on the part of the perceiver is the typicality of the interactional task between perceiver and target in regard to this categorization. If the interaction involves a task construed as typical of one of the two groups, this categorization may be particularly salient and therefore generate behaviors typical of the ingroup. Thus, although stereotypes of

170

KLEIN AND SNYDER

housewives generally portray this group as less intellectually able than men (Fiske et al., 1999), they are thought to be more competent than men in their own area of expertise (e.g., child care, home duties, or handling of emotions). If a woman and a man interact on a task demanding skills related to these areas, and assuming that their gender identities are salient, the man may activate self-stereotypes of men as unable to handle such tasks. The activation of such self-stereotypes may lead a man in such a situation to behave submissively and yield to the woman’s greater (expected) competence. Consistent with this proposition, Dovidio (1988) has found, for example, that men’s visual dominance increased when the two partners engaged in a masculine task as compared to a gender-neutral task, but that it decreased when the task was typically feminine (for other relevant evidence, see Eagly & Wood, 1991). Hence, in these cases, dyad members’ behavior is shaped by their stereotypes about the relative competence of ingroup and outgroup members in their own areas of expertise. By adopting a dominant style when they expect to be competent, men encourage women to behave submissively and contribute to the confirmation of their stereotypes, whereas the reverse occurs when women are thought to be superior to men. Taken together, these studies suggest that in interactions between members of nonstigmatized ingroups and stigmatized outgroups, selfcategorization by the perceiver as a member of an ingroup coupled with categorization of the target as a member of an outgroup can encourage, depending on the purpose of the interaction, either dominant or avoidant behavior on the part of the nonstigmatized perceiver. Whereas a dominant interactional style is typically present in task-oriented interactions, avoidance is often present in getting-acquainted interactions. Moreover, factors increasing the salience of the perceiver’s self-categorization, such as the typicality of the task with respect to the perceiver’s self-categorization, the typicality of the target in relation to her group membership and perceived intergroup competition may moderate these effects. By encouraging behaviors likely to elicit stereotype confirmation, salience of the perceiver’s group membership, coupled with categorization of the target into an outgroup, may therefore facilitate the occurrence of this phenomenon.

B. PREJUDICE OF THE PERCEIVER AND BEHAVIORAL CONFIRMATION As much as the self-categorization processes that we have discussed may constrain the interactions between perceivers who are members of nonstigmatized ingroups and targets who are members of stigmatized outgroups, so too may the stable and enduring levels of prejudice that

STEREOTYPES AND BEHAVIORAL CONFIRMATION

171

perceivers brings to bear on their dealings with targets. As perceivers differ in their level of prejudice, they may also interact differently with members of the group against which their prejudice is directed. According to a classic definition, prejudice is ‘‘an antipathy based upon a faulty and inflexible generalization’’ (Allport, 1954, p. 9). Accordingly, we shall now consider some of the consequences of prejudice for the perceiver’s interaction strategies and the occurrence of behavioral confirmation of social stereotypes associated with the prejudice of the perceiver. First, prejudiced individuals should be especially likely to experience negative emotions when interacting with members of the groups toward which they are prejudiced and about which they apply stereotypes associated with their prejudices. These negative emotions may be expressed through avoidant behavioral styles. Following the now familiar scenario, targets may then reciprocate these behaviors, which could then be interpreted as indicative of coldness or hostility. If the negative stereotypes about the targets of prejudice concern these dimensions, which is typically the case for many targets of prejudice, such as blacks and Jews (see, e.g., Fiske et al., 1999, 2002), this scenario would be indicative of behavioral confirmation. Is this intuitively plausible scenario supported by existing research? Based on Allport’s definition of prejudice, the answer is generally ‘‘yes.’’ Prejudiced individuals tend to experience negative emotions toward members of stigmatized groups (Devine & Monteith, 1993; Fiske, 1998; Gaertner & Dovidio, 1986; Monteith, Devine, & Zuwerink, 1993; Stephan & Stephan, 2000; Vanman, Paul, Ito, & Miller, 1997). For example, according to Gaertner and Dovidio (1986), the most prevalent form of racism toward blacks involves a combination of egalitarian values and negative feelings toward blacks, such as discomfort and uneasiness. These feelings are hard to acknowledge and are therefore most likely to express themselves through avoidant behaviors if perceivers can attribute their behavior to factors other than prejudice. Consistent with this view, prejudiced individuals wish to limit their interactions with the targets of their prejudices (Pettigrew, 1998) and if they are forced to interact with them, they generally want to shorten the interaction (Devine et al., 1996). These avoidant tendencies are generally manifested by less friendliness in interactions with outgroup than ingroup members (Dovidio, Kawakami, & Gaertner, 2002; Ickes, 1984). Research investigating the impact of prejudice on unstructured interracial interactions (e.g., Ickes, 1984) has found that when in the presence of African-Americans, whites who display avoidant tendencies toward blacks tended to elicit avoidant behavior as well. Similarly, heterosexuals high in prejudice toward homosexuals are more likely to be motivated to shorten an interaction with a homosexual than individuals low in prejudice toward this group (Devine et al., 1996). In the same vein, Dovidio, Kawakami, and

172

KLEIN AND SNYDER

Gaertner (2002) found that (explicit) prejudice level was negatively correlated with nonverbal friendliness. Generally, then, this analysis suggests a role for the prejudice of the perceiver in the confirmation of social stereotypes. For the evidence suggests that prejudice indeed produces an avoidant style of interaction and that, if reciprocated by the target, this style can be instrumental in confirming the perceiver’s stereotype-based negative expectations about targets and the stigmatized groups to which they belong.

C. INTERGROUP ANXIETY AND BEHAVIORAL CONFIRMATION Although prejudice may tend to be associated with increases in avoidant behavior, avoidant behaviors may be present even among people low in prejudice, perhaps due to anxiety stemming from contact with outgroup members. In this section, we shall examine the role of the perceiver’s level of intergroup anxiety (Stephan & Stephan, 1985) on the occurrence of stereotype confirmation. Regardless of their level of prejudice, interactions with the stigmatized seem to be threatening for nonstigmatized group members (Blascovich, Mendes, Hunter, & Lickel, 2000; Blascovich, Mendes, Hunter, Lickel, & Kowai-Bell, 2001; Devine et al., 1996; Vanman et al., 1997). Indeed, the evidence suggests that members of nonstigmatized groups experience emotions such as feelings of threat (Blascovich et al., 2000) and anxiety (Devine et al., 1996; Greenland & Brown, 1999; Islam & Hewstone, 1993; Stephan & Stephan, 1985, 2000) when in the presence of members of stigmatized groups. For example, Ickes (1984) has found that regardless of their level of prejudice, whites perceived interactions with blacks as less comfortable and as more strained and awkward than did their interactional partners. Moreover, in a study simultaneously measuring prejudice level and manipulating the composition of the interacting dyad (i.e., as involving two nonstigmatized versus one nonstigmatized and one stigmatized group member), European Canadian participants experienced more negative emotions (such as remorse, guilt, anger, and hostility) when interacting with aboriginal Canadians than with other European Canadians (Vorauer & Kumhyr, 2001). These studies, it would seem, suggest that both low- and high-prejudiced individuals can experience intergroup anxiety. However, this intergroup anxiety may stem from different sources, prejudice being one of them. First, prejudiced individuals may simply experience antipathy, or even disgust, toward members of the stigmatized groups and may therefore view contact as an uncomfortable experience. Second, anxiety may be due to the

STEREOTYPES AND BEHAVIORAL CONFIRMATION

173

perceptions of psychological danger elicited in nonstigmatized individual by stigma (Blascovich et al., 2000). Third, group members may expect to be viewed as prejudiced by their audiences (Vorauer, Main, & O’Connell, 1998) and may therefore be anxious about appearing nonprejudiced to their audiences (Devine et al., 1996). Fourth, they may simply be stressed because they do not know which interaction pattern to adopt with members of the stigmatized group (Hebl & Kleck, 2000; Langer, Fiske, Taylor, & Chanowitz, 1976; Moghaddam, Taylor, & Wright, 1993; Stephan & Stephan, 1985). In which ways could these various forms of intergroup anxiety, when experienced by members of nonstigmatized groups in their dealings with members of stigmatized groups, affect the occurrence of behavioral confirmation? We suggest three routes by which intergroup anxiety can and does lead to behavioral confirmation of social stereotypes. First, anxiety has cognitive consequences that may affect the perceiver’s interaction strategy in ways that facilitate stereotype confirmation. That is, anxiety on the part of the perceiver increases the likelihood of categorization and stereotyping of the target (Greenland & Brown, 1999, 2000; Islam & Hewstone, 1993), perhaps because the arousal elicited by anxiety prevents perceivers from concentrating on individuating information (Baron, Inman, Kao, & Logan, 1992; Wilder & Shapiro, 1989). As the likelihood of categorization increases, so does the likelihood that expectations regarding the target’s group will be activated and used by the perceiver in his or her interaction strategies. In line with this view, lack of cognitive resources does tend to increase the likelihood of behavioral confirmation (Biesanz, Neuberg, Smith, Asher, & Judice, 2001). This is so because trying to elicit information inconsistent with expectations and conducting the interaction in a way that allows the target’s personality to express itself as richly as possible are relatively effortful. Thus, when cognitive resources are limited, it is tempting to rely on categorical information (Fiske & Neuberg, 1990; Macrae, Milne, & Bodenhausen, 1994). It follows from this line of reasoning, and the associated evidence, that it may be much less taxing for perceivers to adopt a stereotype-confirming than a stereotypedisconfirming interaction strategy with the targets of their stereotype-based expectations. Second, out of anxiety, even low prejudiced perceivers may become highly self-conscious and behave awkwardly (Devine et al., 1996; Stephan & Stephan, 1985) in the presence of targets who belong to groups about which stigmatizing stereotypes exist. For example, as they are busy controlling their self-presentation, they may produce speech errors, talk less, avoid direct eye contact (Daly, 1978), or maintain a greater interpersonal distance (Pancer et al., 1979). Kleck et al. (1996) observed that the likelihood of

174

KLEIN AND SNYDER

displaying avoidant behavior when interacting with disabled individuals was greater among able-bodied individuals who perceived the interaction as uncomfortable. In the same vein, Weitz’ repressed affect model of interracial interactions (Weitz, 1972) suggests that even if they try to behave in a friendly way with blacks, whites’ negative emotions are communicated covertly and detected. Similarly, Devine et al. (1996) have argued that behavioral correlates of anxiety could be interpreted as interpersonal distance. We would suggest that these correlates are functionally equivalent to (what we have labeled) an avoidant interactional style, one that may culminate in behavioral confirmation in social interaction. Indeed, members of stigmatized groups tend to be alert, and to try to actively detect any sign of hostility from the nonstigmatized (Crocker, Major, & Steele, 1998; Crocker, Voelkl, Testa, & Major, 1991). A compelling example of this tendency has been provided by Kleck and Strenta (1980), who arranged for participants having cosmetic scars applied to their faces to interact with another participant. Actually, the scar had been removed unbeknownst to the participant. In spite of this removal, participants reported more negativity and behavioral discrimination than control individuals (who did not have any cosmetic scars applied to their faces). This analysis suggests that avoidant behavior driven by anxiety can be interpreted by the stigmatized target as diagnostic of prejudice or hostility. According to the reciprocation hypothesis, this target may also respond by avoidant behavior, which could then be construed as confirmation of negative stereotypes about the group to which the target belongs. Third, according to Stephan and Stephan (1985), intergroup anxiety may induce a rigid adherence to the normative standards required by the situation in which individuals interact. In task groups, such a tendency may produce excessively dominant behavior if the perceiver has higher formal power than the target. If the perceiver and target interact in the context of specific role relationships, this may also induce them to stick to their role rather than express their ‘‘personal’’ identity. To the extent that the roles occupied by members of stigmatized groups are often associated with their stereotypes (Eagly, 1987; Eagly & Steffen, 1984; Hoffman & Hurst, 1990), this may facilitate the behavioral confirmation of these stereotypes. Although this analysis applies to all perceivers, those low in prejudice may be more able to overcome the debilitating aspects of their anxiety on the interaction. Consider, for example, Vorauer and Khumyr’s study (2001): In spite of the negative emotions that were pervasive across all levels of prejudice, interactions seemed to be much smoother when they involved low prejudice than high prejudice whites. Low prejudice participants reported more positive emotions after interacting with a stigmatized than with a nonstigmatized person. Moreover, low prejudice participants thought that they were viewed

STEREOTYPES AND BEHAVIORAL CONFIRMATION

175

less in accordance with the white stereotype when interacting with an aboriginal than with a white. Conversely, aboriginal participants interacting with a high prejudice partner experienced more discomfort and self-directed negative emotions than those interacting with a low prejudice partner. Hence, it seems that in spite of their ambivalence (demonstrated by a combination of negative and positive emotions), low prejudice participants managed to make their partner feel comfortable in the interaction. As this study did not incorporate measures of expectations and ratings of the stigmatized person’s behavior, it is difficult to draw definite conclusions regarding the occurrence of behavioral confirmation. However, it seems reasonable to infer that the aboriginal Canadians interacting with a low prejudice partner exhibited more warmth and friendliness, as a result of their greater comfort, than those who interacted with high prejudice participants, who experienced negative selfdirected emotions and discomfort during the interaction. As we have seen, this discomfort can lead to the adoption of behaviors confirming negative stereotypes about the target’s level of sociability. In sum, based on the existing evidence, it seems that overall, across all levels of prejudice, intergroup anxiety may contribute to behavioral confirmation by making categorical perceptions and behavior that could be interpreted as a sign of avoidance more likely to occur, with the attendant consequences of these avoidant styles of interaction on the part of the perceiver for behavioral confirmation on the part of the target. However, it also seems to be the case that as they are not exclusively motivated to distance themselves from the target and want to present an image of themselves as tolerant, low prejudice individuals are often able to overcome the negative impact of their anxiety.

D. STEREOTYPE CONTENT AND BEHAVIORAL CONFIRMATION Let us now turn to the last of the factors that we propose to affect the perceiver’s use of the avoidant and dominant styles of interaction that may generate behavioral confirmation—the very content of the perceiver’s stereotype-based expectation about the target. According to the reciprocation interpretation, the avoidance style can be triggered by an anticipated reciprocation of behaviors diagnostic of the traits attributed to the target (Burgoon et al., 1995; Ickes et al., 1982). A more cognitive variation of this interpretation suggests that the activation of a stereotype automatically elicits behavior consistent with this stereotype (Bargh, Chen, & Burrows, 1996). As they categorize the target as ‘‘black,’’ for example, perceivers would activate the traits of coldness and hostility, and would perform behavior consistent with these traits of behavior thereby triggering the

176

KLEIN AND SNYDER

reciprocation route to behavioral confirmation (Chen & Bargh, 1997). Accordingly, stereotype content would directly determine the perceiver’s behavior (and may do so especially for prejudiced perceivers). In addition, the avoidant and dominant styles of interaction can also be triggered by the content of the perceiver’s stereotyped expectations of the target. For example, avoidance can also be triggered by the hostile attitudes and anxiety that may be elicited by many stigmatized groups. In these cases, stereotype content may play a role in these reactions as well, as the attribution of specific traits to a group can be associated with more negative attitudes and emotions toward these groups (Eagly & Mladinic, 1989; Esses, Haddock, & Zanna, 1993; Stephan & Stephan, 1993). For example, Mackie, Devos, and Smith (2000) have found that stereotype content (e.g., strength) predicted specific emotions (e.g., anger) and behavioral tendencies (e.g., aggression) toward members of outgroups. Similarly, in task-oriented interactions, the dominant style can be triggered by the expectation that a member of the nonstigmatized perceiver’s ingroup is more intelligent or competent than the stigmatized outgroup member, whereas a target adhering to the stereotype that the nonstigmatized group is more competent can adopt a submissive behavior (Berger, Cohen, & Zelditch, 1972; Berger, Rosenholtz, & Zelditch, 1980; Ridgeway, 1987; Ridgeway & Berger, 1988). Findings consistent with this hypothesis have been observed in interview settings; for example, Rudman and Borgida (1995) found that men primed with ads depicting women in a stereotypic way (as ‘‘sexual objects’’) displayed more dominant behaviors in a subsequent interview with women than unprimed men. In sum, it seems that stereotypes of the outgroup as unsociable seem to elicit reciprocation (i.e., a behavior in line with the trait attributed to the target) whereas stereotypes of the outgroup as incompetent or submissive seem to elicit a behavioral style complementing the target’s expected behavior (i.e., a dominant style). That is, it would appear that stereotypes may elicit from perceivers the very behavioral tendencies that make confirmation of those stereotypes likely.

E. CONCLUSION Overall, this analysis suggests that when the target is categorized as a member of a stigmatized group, the nonstigmatized perceiver is likely to display avoidance or dominance. If the interaction is a ‘‘getting-acquainted’’ interaction, expectations regarding the sociability of the target are more likely to be activated and the main determinant of the perceiver’s behavior will be attraction toward this target. In the presence of a stigmatized group

STEREOTYPES AND BEHAVIORAL CONFIRMATION

177

member, less attraction should be evident than in the presence of another nonstigmatized and the outcome may be avoidant behavior. This will be particularly true if the perceiver has a high level of prejudice and experiences intergroup anxiety. We have also seen that stereotype content could directly affect the perceiver’s behavioral style, in such a way that negative expectations regarding the target’s level of sociability tend to elicit avoidance. If the interaction involves the joint performance of a task, on the other hand, expectations regarding competence will be activated and the nonstigmatized group member is more likely to behave in a dominant way. These behavioral styles on the part of perceivers are, of course, precisely the ones that set the stage for behavioral confirmation on the part of the targets of their stereotype-based stigmatizing expectations.

V. The Target’s Perspective Now that we have examined how the perceiver’s self-categorization can and does affect the occurrence of stereotype confirmation, let us turn to the target’s perspective and try to understand how stigmatization can affect his or her interaction strategies in ways that may influence the occurrence of stereotype confirmation. To date, most theorizing about behavioral confirmation has, either by design or by default, viewed the target as if he or she docilely responds to the perceiver’s behavior. For example, it has been demonstrated that the typical interactional strategy used by targets involves trying to get along well with the perceiver by tuning their behavior to the perceiver’s behavior and rendering the flow of conversation as smooth as possible (Snyder, 1992). As we have seen, this strategy generally results in behavioral confirmation. Of course, this strategy is not the only one that targets could use. They could, instead of reciprocating the perceiver’s overtures, compensate for them, responding, for example, to a perceiver’s cold behavior with an increased level of friendliness (Burgoon et al., 1995; Ickes et al., 1982). Or, they could refuse to abide by the perceiver’s script and impose their own self-presentational agenda (Neuberg, 1996; Smith et al., 1997). Accordingly, we shall now examine factors that affect the target’s strategies for interacting with perceivers who hold stereotype-based expectations about them, focusing especially on those defined at an intergroup level. We propose that the primary determinant of the target’s strategies is whether the target is aware that her or his social identity as a member of a stigmatized group is known by the perceiver. When this is the case, targets may be said to be in a state of ‘‘stigma consciousness’’ (Pinel, 1999), a term

178

KLEIN AND SNYDER

that denotes the belief that one is viewed as a stereotypical member of the stigmatized group. Although this concept has typically been defined as an individual differences variable (Pinel, 1999, 2002), it can also be viewed as being determined by features of the situation in which perceiver and target interact. Depending on the context in which an individual is placed, his or her degree of stigma consciousness may vary (Shih, Pittinsky, & Ambady, 1999). Several factors, each of which has its particular relevance to behavioral confirmation scenarios, may influence stigma consciousness. The first such factor is visibility to the perceiver: If the target does not believe that his or her stigmatized attributes are readily visible to the perceiver, he or she will not expect to be treated in terms of his or her group membership. In fact, stigmatized individuals may develop strategies aimed at dissimulating their stigmata. When they do not desire to be viewed in terms of a devalued social identity, one of the most common strategies involves ‘‘passing’’ for a member of the nonstigmatized, advantaged group (Goffman, 1963). Passing can involve trying to eliminate the features that mark the individual as a member of a stigmatized group (thus, an immigrant can change nationality, an obese person can try to lose weight), to conceal it (e.g., a closeted gay man may tell fictional accounts about his success with women; a facially disfigured person can use special headwear dissimulating the disfigurement), or to deny it (e.g., a deaf person can act as if his or her hearing was good, a former delinquent may refuse to acknowledge his or her past). Second, stigma consciousness may depend on the salience of the target’s group identity. As a function of the intergroup context, targets may vary in the extent to which their identity is salient (Oakes, 1987; Turner et al., 1987; van Knippenberg & Dijksterhuis, 2001). For example, in the context of a discussion on abortion rights, religious identity may be a particularly salient category and induce a target, who happens to be highly religious, to be treated in terms of this identity. By contrast, this identity may be less salient in the context of a discussion on affirmative action. Third, repeated and chronic exposure to situations evoking stigma consciousness makes it especially likely that targets will expect to be stereotyped in their interactions with members of a nonstigmatized group. In this regard, perceived personal and group discrimination seem to be strong predictors of stigma consciousness (Pinel, 1999), and a history of experiencing discrimination increases the likelihood that people will expect to be the targets of stereotypes and prejudice (Crocker et al., 1998). Now, based on these considerations of stigma consciousness, we can articulate the features of two scenarios in which stigma consciousness on the part of targets of stereotype-based expectations will or won’t be present, with attendant consequences for whether or not these scenarios will culminate in stereotype confirmation.

STEREOTYPES AND BEHAVIORAL CONFIRMATION

179

A. THE TARGET DOES NOT EXPECT TO BE VIEWED AS A MEMBER OF A STIGMATIZED GROUP Consider the case of a second-generation Mexican newly hired to work in an American firm that proclaims its commitment to diversity and tolerance. This person, however, views himself or herself as a dedicated member of his or her profession before being someone of Mexican ancestry. When interacting with Anglo employees, he or she may not expect to be categorized as ‘‘Hispanic’’ and perceived through the lens of stereotyped beliefs and expectations about members of this group. Being unsuspicious, our new member of the firm may try to get along and to have smooth interpersonal dealings with others in the firm. But, in spite of the tolerant values proclaimed by the firm, many of its employees may still expect their new Mexican colleague to be a typical Hispanic and not try to perceive him instead in terms of his or her unique personal and professional identity. In such a situation, the new employee, as a target of stereotypes about his or her nationality, may not actively try to self-present in ways that would disconfirm such stereotypes and may instead tend to fall prey to the confirmatory strategy of those co-workers who, as perceivers, treat him or her in accord with their stereotype-based expectations. Hence, at least in this example, the absence of stigma consciousness may place the target in an ideal position to display stereotype confirmation. In fact, to the extent that targets in such situations repeatedly and chronically enact such getting along strategies, and do so regularly and consistently with many different perceivers who share the same social stereotype about the targets’ group membership, not only will such group stereotypes appear to be confirmed but they will be reinforced and maintained. One key feature of this example is the asymmetry between the perceiver’s and the target’s levels of categorization—whereas the target self-categorizes at the individual level, the perceiver categorizes at the group level. This type of asymmetry between the perceiver’s and the target’s levels of categorizations, we would suggest, is often inscribed in the typical behavioral confirmation scenario. In its basic form, this paradigm involves two different expectations. Usually one of these expectations is positive (e.g., intelligent, extraverted, normal-weight, attractive) whereas the other expectation is negative and associated with a stigmatizing, or a negatively valued, characteristic (e.g., black, obese, introverted, unintelligent). In the ‘‘negative’’ expectation condition of the typical behavioral confirmation experiment, the only information that the perceiver possesses about the target is that he or she belongs to a stigmatized outgroup. However, targets in the typical behavioral confirmation experiment do not truly belong to the categories associated with the expectations. That is, the

180

KLEIN AND SNYDER

target is not, in fact, black, obese, or introverted, but rather has been labeled as such by the random assignment of expectations to perceivers. Accordingly, targets in the negative expectation condition of the typical behavioral confirmation experiment, although stigmatized by the expectations assigned to the perceiver, are not in a position to actually be ‘‘stigma conscious.’’ Therefore, it is also impossible for them to define themselves in terms of a collective identity and to try to enact positive self-stereotypes associated with this group membership. They may, however, try to simply reciprocate the perceiver’s behavior and abide by his or her script. On the other hand, in the ‘‘positive expectation’’ condition of the typical behavioral confirmation experiment, the target can usually be construed as a member of the perceiver’s own ingroup. Targets in behavioral confirmation studies of expectations based on racial categories have, in fact, been, like their perceivers, white, of average weight, and more likely to regard themselves as extraverts than introverts (Klein & Snyder, 2000). Thus, in the ‘‘positive expectation’’ condition of the typical behavioral confirmation experiment, such an intragroup interaction should therefore follow standards of interpersonal conversational friendliness, with the perceiver treating the target as an individual member of his or her own ingroup, and the target in fact self-perceiving and self-categorizing in like terms. Taken together, these considerations of the levels of self-categorization and other-categorization on the part of perceivers and targets suggest that the perceiver is more likely to perceive the interaction as intergroup when the expectation is negative or concerns a stigmatized group membership than when the expectation is a positive one. The target, on the other hand, should always perceive the interaction as an interpersonal one because he or she has no reason to suspect that the perceiver holds an expectation, whether positive or negative, and because the target in the negative expectation condition does not in fact belong to the group associated with the perceiver’s expectation, is in no position to be or to become stigma conscious. Hence, the target should follow the interpersonal strategy of interpersonal adjustment (‘‘getting along’’) that typically leads to behavioral confirmation (Snyder & Haugen, 1995). Behavioral confirmation, it follows from this line of argument, is facilitated when the perceiver treats the target in terms of his or her group membership and when the target is not ‘‘stigma conscious.’’ Hence, this discrepancy between the perceiver and the target’s levels of categorization facilitates behavioral confirmation. But what about studies in which the target actually belongs to the stigmatized category associated with the perceiver’s expectations? Although such studies are rare, they do tend to offer findings consistent with our analysis. When the target is a member of a stigmatized category, and is not stigma conscious, stereotype confirmation is often exacerbated (Miller &

STEREOTYPES AND BEHAVIORAL CONFIRMATION

181

Myers, 1998; Miller, Rothblum, Barbour, Brand, & Felicio, 1990; Miller, Rothblum, Felicio, & Brand, 1995; see also Jussim, Eccles, & Madon, 1996). Perceivers tend to evaluate stigmatized persons, such as the obese (Miller et al., 1990, 1995) and unattractive women (Goldman & Lewis, 1977), more negatively than nonstigmatized persons even when these targets know that the perceiver is unaware of their stigmatized status. Miller et al suggest that, as compared to situations in which their social identity is visible, the stigmatized underestimate the social skills needed for making the interaction smooth and pleasant, and therefore do not mobilize sufficient energy for implementing a favorable impression. Thus, being unaware of how difficult it is to overcome the impact of stigma on their self-presentation, they may infer that achieving their impression management goals is a relatively easy task when their stigma is not visible. Hence, they may remain somewhat passive and aloof, not suspecting that achieving a positive self-presentation is an effortful task even in the absence of visible stigma. Overall, the results of Miller et al. suggest that the stigmatized may be especially vulnerable to the confirmation of negative stereotypes when they are not stigma conscious. Note, however, that such situations are exceptional. As stigmatized individuals tend to be constantly on the lookout for evidence of prejudiced behavior (Crocker & Major, 1989; Crocker et al., 1991; Goffman, 1963; Kleck & Strenta, 1980), they may only rarely be stigma ‘‘unconscious.’’

B. THE TARGET DOES EXPECT TO BE PERCEIVED AS A MEMBER OF A STIGMATIZED GROUP In the typical behavioral confirmation study, targets are unaware of the perceiver’s expectations and are therefore not prepared to counteract a confirmatory strategy by displaying disconfirming behavior. However, when targets in behavioral confirmation studies are made aware of negative expectations, they may try to dispel them especially if they are negative: For example, Hilton and Darley (1985) found that targets who were aware that they were perceived as cold tried to disconfirm this expectation by behaving warmly. Nevertheless, other data suggest that targets do not necessarily compensate for negative expectations. Thus, Curtis and Miller (1986) found that targets who erroneously thought that a perceiver disliked them disclosed less to their partner and behaved less warmly, which had the effect of making the perceiver’s expectation become true (for convergent findings, see also Farina, Gliha, Boudreau, Allen, & Sherman, 1971; Farina, Sherman, & Allen, 1968).

182

KLEIN AND SNYDER

These studies, in which targets are aware of the perceivers’ expectations, it should be noted, all concern expectations about the traits of the targets as individuals; that is, in terms of the distinction that we offered early on in our analysis, they concern personal expectation confirmation rather than social stereotype confirmation. The question thus arises of whether the processes at work in these studies can be generalized from these relatively interpersonal situations to the intergroup level of analysis, in which the expectations at issue derive from the membership in groups about which potentially stigmatizing stereotypes exist. That is, what will be the effects of the target becoming aware that the perceiver has categorized the target as a member of a stigmatized group? In answer to this question, the relevant evidence indicates that if the target expects to be viewed as a typical member of the stigmatized category, he or she does become aware of the metastereotype (Vorauer et al., 1998), that is, the stereotype held by the perceiver’s group about the stigmatized group. This meta-stereotype determines how the target expects to be treated by the nonstigmatized perceiver. Hence, the meta-stereotype plays the same role as the information about individual traits that Hilton and Darley (1985) communicated to their targets regarding the expectations induced in the perceiver. To further explicate the role of stigma consciousness on the part of targets who do expect to be perceived as members of stigmatized groups, and the implications of these perceptions for behavioral confirmation, let us focus on two types of situations. The first type of situation is that in which the target performs behaviors that are consistent with the perceiver’s stereotypes regarding the stigmatized group (which we will refer to as ‘‘stereotype enactment’’). The second type of situation is that in which the target purposefully adopts behaviors that contradict the perceiver’s stereotypes (which we will refer to as ‘‘stereotype compensation’’)3.

3 A third option can be considered. Stigmatized individuals who are high in stigma consciousness (because, for example, their stigma is visible) may simply wish to avoid contact with the nonstigmatized because of the threat and anxiety it creates. Goffman (1963) calls this strategy defensive cowering. If a total absence of contact is impossible, a similar strategy involves avoiding contact situations in which the negative stereotypes may be applied to the self. This type of strategy seems particularly common among people high in stigma consciousness: According to Pinel (1999, 2002), these people are particularly reluctant to being stereotyped because the stereotype provides self-discrepant feedback. Hence, they may forego opportunities to disconfirm stereotypes. She found, for example, that women who expected to compete on a jeopardy-like task were less likely to choose stereotypically male topics if they expected to compete with a man than a woman. There was no difference among women low in stigmaconsciousness. If generalized, both defensive cowering and this avoidance of opportunities to disconfirm stereotypes when interacting with the nonstigmatized will contribute to the persistence of stereotypes regarding the stigmatized group.

STEREOTYPES AND BEHAVIORAL CONFIRMATION

183

1. Enactment of the Perceivers’ Stereotype When targets enact behaviors in accordance with stereotypical expectations, they do, by definition, provide behavioral confirmation for the perceivers’ expectations based on those stereotypes. For the most part, as we have argued, targets will be particularly likely to confirm perceivers’ stereotypes when they are not stigma conscious. Why, it should be asked, would targets ever enact the perceivers’ stereotypes when they are, in fact, stigma conscious? One determinant of stereotype confirmation in the presence of stigma consciousness is lack of capacity to enact positive self-views, either because of a constraining situation, or because of lack of skills. Even if the stigmatized target does not adhere to the meta-stereotype, the stigma in itself can be threatening and may be a source of anxiety and discomfort (Steele, 1998; Steele & Aronson, 1995). These states can have detrimental effects on both performance (a phenomenon known as ‘‘stereotype threat’’) and the warmth of interpersonal behavior. For example, in a study by Farina, Gliha, Boudreau, Allen, and Sherman (1971), psychiatric patients who thought that perceivers were aware of their stigma performed less well on a task and were perceived as more anxious and less adjusted than those who interacted with a perceiver who was unaware of their stigma. Similarly, Comer and Piliavin (1972) reported that when interacting with able-bodied persons, physically disabled participants displayed more signs of avoidance (as revealed by quicker termination of the interaction, greater interpersonal distance, and reduced eye contact) than when interacting with another disabled person. According to Comer and Piliavin, this avoidant behavior seems to be due to the anxiety and discomfort elicited by interactions with nonstigmatized individuals. Such situations may place a heavy burden on the stigmatized, who often have to simultaneously pursue multiple impression management goals such as, using Goffman’s terminology (Goffman, 1963), ‘‘carrying their lot lightly,’’ or ‘‘being well adjusted’’ but ‘‘not behaving inappropriately for a person with a disability.’’ In a very different context, Pinel (2002) obtained quite similar findings. In her study, women, whose level of stigma consciousness had been previously measured, interacted with a man whom they expected to be sexist or not. Following the interaction, women high in stigma consciousness, but not those low in stigma consciousness, were rated as less friendly and warm than their male partner if they expected him to be sexist. Hence, they enacted the negative views hostile sexists may hold about women. In this case, stigma conscious women experienced discomfort engendered by the idea of having to interact with a sexist man. They therefore behaved in a more avoidant way.

184

KLEIN AND SNYDER

Note that, consistent with self-categorization theory, stigma consciousness is context dependent and very subtle contextual changes may affect the occurrence of behavioral confirmation. In a study of stereotype threat, Shih et al. (1999) have observed that Asian-American women performed better on a mathematical test when their Asian identity (stereotypically associated with success on mathematical tasks) was salient than in a comparison condition in which no identity was made salient. Conversely, they performed less well when their female identity (stereotypically associated with poor performance in mathematics) was salient than in the comparison condition. Another reason why targets may enact unfavorable stereotypes, even when stigma conscious, is that they expect rewards to accrue to them from confirming expectations that others hold for them (Miller & Turnbull, 1986). Thus, if the target wishes to have a pleasing interaction with the perceiver (perhaps, in hopes of winning acceptance, gaining affection, or obtaining a job), he or she may not be motivated to dispel negative stereotypes regarding his or her group. For example, women have been found to strategically conform to the sexist stereotypes presumed to be held by a man if they found this man attractive (Zanna & Pack, 1975) or if he was a job interviewer (von Baeyer, Sherk, & Zanna, 1981). A further reason why targets, even when stigma conscious, may nonetheless confirm stereotypes held about their groups, is that, for members of stigmatized groups, stereotypes are not necessarily perceived uniformly negatively. In fact, stereotypes of many stigmatized groups are actually ambivalent (Fiske et al., 1999; Glick & Fiske, 2001): At the same time as they characterize stigmatized groups as incompetent, they often portray them as sociable and warm. Thus, the enactment of the ‘‘positive’’ aspects of these stereotypes may be rewarding and help the target develop a smooth interaction with members of the nonstigmatized group. For example, Goffman (1963) cites the example of a female dwarf who, in spite of her introverted character, was always joyful and merry in the presence of people of normal size in order to have positive interactions with them. More generally, stigmatized individuals typically have few relationships with the nonstigmatized, often because they are not valued by the latter (McPherson, Smith-Lovin, & Cook, 2001) or because the nonstigmatized feel threatened and anxious in their presence (Blascovich et al., 2000; Goffman, 1963; Stephan & Stephan, 1985). Hence, they may find themselves motivated to present the most acceptable, and unthreatening, social face rather than disconfirming negative stereotypes about their incompetence, all in hopes of increasing their chances of initiating and maintaining social contact with the nonstigmatized. Finally, stereotypes may be enacted by targets, even the stigma conscious, simply because they are thought to be true. Stigmatized groups have been

STEREOTYPES AND BEHAVIORAL CONFIRMATION

185

known to apply the negative views of society concerning their groups to themselves (Crandall, 2000; Wright, 1983), although this may actually be a rare occurrence (Crocker & Major, 1989; Crocker et al., 1998). According to Swann (1983; Swann & Read, 1981), individuals are motivated to verify their self-conceptions. Interactions with members of a nonstigmatized group may actually constitute an opportunity to verify their self-views, which may at times actually lead to behavioral confirmation of stereotypes. Consider, for example, a woman who, in accordance with stereotypes about her gender (Eagly & Mladinic, 1989; Kray, Thompson, & Galinsky, 2001), views herself as unassertive. In the presence of a male perceiver who holds stereotypical expectations about women as a group, she may try to affirm her self-views. For example, she may try to elicit questions that will help her confirm her self-views, or she may lead the perceiver to provide feedback consistent with her self-views (Swann & Read, 1981). Although she may be aware of the perceiver’s expectations, such a behavior can be interpreted as consistent with the stereotype of women, and hence be viewed as an example of stereotype confirmation. 2. Stereotype Compensation Although, as we have seen, there are a variety of reasons why targets, even when their stigma consciousness is high, will confirm stereotypes about their groups, it is possible to specify circumstances in which targets will try to actively show that they do not possess the negative traits stereotypically attributed to their group. This strategy, which we refer to as stereotype compensation, may serve the same purpose as ‘‘passing’’ as a member of the nonstigmatized group. However, it is used in situations in which the target’s stigma is visible; in such circumstances, although the target cannot conceal his or her membership in the stigmatized group, he or she can create the appearance of not possessing the attributes stereotypically associated with membership in that group. As stereotypes and their associated traits are often negative, they may often hinder the accomplishment of their targets’ interaction goals. Consider the case of an obese woman who expects to be viewed as introverted; for her, the goal of having a pleasing interaction with a normal-weight person may require her to show that she is socially skilled and enthusiastic. Indeed, obese targets have been shown to use compensatory strategies when aware that a normal-weight perceiver could see them, behaving in a more cheerful way than when they thought that the perceiver could not see them (Miller et al., 1995). That is, rather than simply reciprocating their partner’s level of friendliness, they appear to have purposefully compensated for the coldness initially displayed by their perceivers by engaging in a friendlier interactional style.

186

KLEIN AND SNYDER

This strategy of self-enhancement is likely to differ from the ‘‘getting along’’ strategy. It will often require that the target actively disrupt the flow of conversation if the conversational script imposed by the perceiver does not enable the target to express her self-views (Neuberg, 1994). By contrast, a target who hopes to ingratiate herself by ‘‘getting along’’ with the perceiver will try to keep the flow of conversation as smooth as possible. Hence, the use of such nondeferential behavior may serve to obstruct both the confirmatory strategy and the reciprocation routes to behavioral confirmation. When the perceiver’s behavior toward the target is cold or unfriendly, the choice of a strategy of compensation, rather than of reciprocation, may reflect the target’s desire to form or maintain a relationship with a perceiver who is appealing as a relationship partner. Thus, Burgoon, LaPoire, and Rosenthal (1995) propose that compensation is likely to be chosen when targets wish to promote a relationship with a perceiver who is valued but who behaves (or who is expected to behave) coldly. On the other hand, according to Burgoon et al. (1995), when the perceiver is negatively valenced, unfriendly or cold behavior on her part should be reciprocated. For example, from this perspective, it is conceivable that targets in the study by Hilton and Darley (1985) expected the perceivers to be likable and wished to maintain a positive relationship with them; after all, the perceiver was another student (like the target) and had received a random profile that depicted the target as cold and introverted. Hence targets, wanting to forge a positive relationship with an attractive perceiver, may have been motivated to refute the negative attributes imputed to them in the perceiver’s expectation, about which they had become aware during the experiment. Similarly, obese targets in the study by Miller et al. (1995) had no reason to believe that the perceivers they were interacting with were dislikable and, certainly having no reason to think otherwise, most likely assumed by default that they were likable. They may have feared that the information regarding their obesity could lead to inaccurate impressions on their part. They may therefore have wanted to correct the biasing influence of negative stereotypes associated with their social category. However, it is important to recognize that being motivated to compensate for, and therefore disconfirm negative stereotypes is not the same thing as actually succeeding in dispelling those negative stereotypes. For example, consider the nonstigmatized participants in the study by Farina, Allen, and Saul (1968) who thought they were viewed by a perceiver as either homosexual or mentally ill. Compared with a control condition, in which no such meta-perception was induced, participants were actually viewed more negatively although they tried to dispel the negative image associated with the stigma. As we have noted, to be able to overcome the negative

STEREOTYPES AND BEHAVIORAL CONFIRMATION

187

consequences of stigma consciousness, stigmatized group members need to have developed skills that enable them to compensate for the situational demands posed by interaction with potentially prejudiced members of nonstigmatized groups (Miller & Myers, 1998; Miller et al., 1995). For example, as overweight people may be ignored or treated negatively by others, they have to respond to others’ behavior with more outgoingness and warmth than those of normal weight. This requires the development of special skills of sociability. Participants in the study by Farina et al. (1971) were psychiatric patients whereas those in the study by Farina et al. (1968) were nonstigmatized Ivy League students. Hence, it is unlikely that either group had developed sufficient skills to allow them to compensate for the demands placed on them by the situations created in these studies. For the target to effectively compensate for negative expectations, it is important that the meta-stereotype be clearly activated, that is, that the target be fully aware of the stereotype held by the perceiver’s group about the social group to which the target belongs. In a recent experiment by Kray, Thompson, and Galinsky (2001, study 2), women and men were informed that traits such as assertiveness and rationality (associated stereotypically with men) contributed positively to performance in negotiations, whereas traits such as emotionality and accommodatingness (stereotypically associated with women) were associated with poor negotiation performances. In such circumstances, women performed less well than when the stereotype was not primed. In a following study (Kray et al., 2001, study 3), when participants were explicitly informed that women performed less well than men because of gender differences on these same dimensions, women performed better than in a control condition in which this stereotype was not activated. According to Kray et al., explicit priming of the stereotype enables targets to react to its adverse effects. When the stereotype is subtly primed, individuals are not able to effectively correct its biasing influence and tend to conform to the stereotype (see: Steele, 1998; Steele & Aronson, 1995). Similarly, in Curtis and Miller’s study (1986), targets who thought that they were disliked may not have been able to respond effectively to the perceiver’s behavior as there was no clear expectation to disconfirm. Altogether, our analysis of the role of stigma consciousness in targets’ interactions with perceivers who hold negative stereotypes about them shows that even in the face of stigma consciousness, targets can fall prey to negative stereotypes. For targets to be able to disconfirm these negative stereotypes, several conditions must be met. First, they must view the stereotypes as not applying to themselves. Second, they must be motivated to present themselves in stereotype-inconsistent ways and they must have the power and the opportunity to do so. And, third, they must have the

188

KLEIN AND SNYDER

necessary skills to implement these stereotype-disconfirming behaviors and the opportunity to exercise those skills. In the next stage of our analysis, we will examine, in keeping with the group level perspective that we seek to apply, how these conditions can be fulfilled through the pursuit of collective strategies. 3. Stereotype Change as a Collective Strategy When asked what a feminist is, journalist Anne Marlowe (2002) responded ‘‘someone who believes that she should pay for her own dinner.’’ As this example illustrates, what is traditionally seen as an interpersonal encounter, a dinner date in this example, can be an opportunity to engage in intergroup behavior, in this case refusing to be treated to a dinner in order to manifest the independence of women. In doing so, the ‘‘someone’’ of Marlowe’s example may hope to disconfirm a sexist man’s view of women as a group. So far, we have treated compensation as primarily an interpersonal strategy used by members of disadvantaged groups to show that they were not typical members of their groups. By using self-presentations that differed markedly from the stereotype of their group, they seek to be viewed as atypical members of their group, or even as members of a more prestigious group. For example, our job applicant of Moroccan ancestry may display clear signs of a work ethic in an effort to show that he is a ‘‘true’’ Belgian. Similarly, an obese person may behave in a most cheerful manner as evidence that she should not be categorized as an obese but as a ‘‘normal’’ ingroup member. In this regard, passing and compensation can be viewed as individual strategies of upward mobility, following social identity theory’s classifications of identity management strategies (Tajfel, 1975; Tajfel & Turner, 1986). According to this theory, individuals are motivated to have a positive social identity, which depends on the existence of favorable comparisons to outgroups. By definition, members of stigmatized groups possess a negative social identity. So far, the examples of stereotype compensation that we have considered involved situations in which members of the stigmatized group responded to their predicament by dissociating themselves from the devalued ingroup and trying to acquire membership in (or at least association with) a more prestigious outgroup. However, as Anne Marlowe’s example suggests, members of stigmatized groups may also try to disconfirm the negative stereotypes held about their group by embracing and enacting more positive self-stereotypes rather than psychologically escaping from their group. When members of a stigmatized group have developed a collective identity, their self-stereotype generally differs from the stereotype held by the dominant group about their ingroup

STEREOTYPES AND BEHAVIORAL CONFIRMATION

189

(Judd, Park, Ryan, Brauer, & Kraus, 1995; Klein & Azzi, 2001; Krueger, 1996). In this case, the target’s behavior will not be directed at defending the individual self through disconfirming expectations about himself or herself as an individual, but rather at upgrading the position of the group as a whole by changing collective perceptions of the group itself. In this regard, stereotype change can be viewed as part of a ‘‘collective’’ strategy, in Tajfel’s classification (Tajfel, 1981a; Tajfel & Turner, 1986). Such a strategy will manifest itself by the enactment of the group’s self-stereotypes (Turner et al., 1987). For example, women who have a high feminist consciousness may try to display particularly assertive behavior when interacting with men in an effort to change their view of women as dependent and submissive. This strategy may express itself in several ways. One such manifestation consists of openly acknowledging one’s stigma in the presence of nonstigmatized individuals. This ‘‘breaking through’’ can dispel the discomfort created by the stigma and implicitly allow both the nonstigmatized and the stigmatized to talk about the stigmatizing condition (Hebl & Kleck, 2000). For example, consider a disabled individual who talks openly about the problems posed by his or her disability. Implicitly, this means that the stigma can be talked about and does not need to be eluded. Talking openly about the stigma should reduce anxiety on the part of both participants and, to the extent that the manifestations of anxiety are often taken as confirmation of negative stereotypes, make the confirmation of negative stereotypes less likely. Consistent with this analysis, perceivers do view a person who acknowledges his or her stigma in a more positive light than one who does not (Hastorf, Wildfogel, & Cassman, 1979; Hebl & Kleck, 2000). A similar strategy can also be used when the stigma is invisible. For example, gay people may ‘‘come out’’ and openly acknowledge their sexual orientation. In the same vein, Goffman (1963) cites the example of secondgeneration immigrants who interlace their speech with Jewish idiom and accent. When the stigma is not directly visible, however, acknowledgment of it is not intended to reduce the perceiver’s discomfort. It may, instead, actually serve the function of stereotype disconfirmation. For, acknowledgment forces the perceiver to categorize the target as a member of the stigmatized group. Hence, acknowledgment may be part of a strategy of influence aimed at inducing the perceiver to view behavior consistent with the target’s positive self-stereotype as characteristic of the stigmatized group. In other words, the stigmatized target may try to change the perceiver’s stereotype, which would be impossible had the perceiver not initially categorized the target as a member of the stigmatized group. For example, the Jews described by Goffman may want to affirm not only that they are Jews, which is not directly visible, but that Jews are proud of their

190

KLEIN AND SNYDER

group membership and the many positive attributes that they associate with their group membership. More directly, members of stigmatized groups can purposefully display counternormative behavior as a way of affirming their rejection of behavioral standards associated with their groups. One such example is Anne Marlowe’s feminist who refuses to allow anyone to pick up her check. In such instances, members of stigmatized groups indirectly challenge stereotypes and the behavioral norms that are often inscribed in the group’s stereotype (e.g., the stereotype of women as financially dependent and therefore in need of others to pay their way). Now that we have viewed some of the forms such a collective strategy may take in interpersonal interactions, let us examine how it may affect stereotype confirmation. We have seen that one of the main factors determining stereotype confirmation is the use by the target of a deferential behavioral style, which allows the perceiver to implement a confirmatory strategy. The examples of collective strategies we have encountered, such as refusing a free dinner or openly acknowledging one’s stigma, reveal that the target’s strategy can express itself through assertive, nondeferential, behavior. We shall now discuss several of the processes through which a collective identity may influence the use of nondeferential behavior, and thereby encourage behavioral disconfirmation. First, highly identified members of stigmatized groups are more likely to view their position as illegitimate or as the result of discrimination (Postmes, Branscombe, Spears, & Young, 1999). This means that they are less likely to attribute their disadvantage to the inner dispositions and capacities of their group in comparison with outgroup members, viewing it instead as the outcome of illegitimate behaviors and acts. These individuals are generally more willing to take risks on behalf of their group (Doosje, Ellemers, & Spears, 1999; Ellemers, Spears, & Doosje, 1997; Ellemers, Van Rijswijk, Bruins, & De Gilder, 1998) in order to respond to this perceived injustice, perhaps because they ‘‘depersonalize’’ their interests and identify with the well-being of the group as a whole (Simon, 1998). As a consequence, even if they are powerless and even if it could be costly, they may be more likely to engage in stereotype-disconfirming behavior. In addition, groups provide a sense of social support that enables their members, not only to perceive their situation in group terms, but also to claim the group norms and values, even if doing so entails risks (Doosje et al., 1999; Ellemers, Barreto, & Spears, 1999). For example, in studies by Reicher, Levine, and Gordijn (1998), making students visible to each other increased their propensity to publicly endorse student norms (cheating) that were punishable by the staff. The knowledge that others support the target’s action may be central in this regard. In the typical behavioral confirmation

STEREOTYPES AND BEHAVIORAL CONFIRMATION

191

scenario, the target is alone. A sense of collective identity is associated with the awareness that one is not an isolated group member but that others can support your action. Hence, through its empowering influence, collective identity may lead to less deference toward perceivers belonging to advantaged groups. For example, in a large-scale study, Gruber and Smith (1995) have shown that women were more likely to respond assertively to sexual harassment attempts if they had a high feminist consciousness. Not only may group support empower individuals, but it may also contribute to skill development. Group members can help the stigmatized by instigating techniques and behaviors enabling them to behave appropriately in their interactions with the nonstigmatized and to do so in ways that express the group’s view of itself. For example, in Britain, several feminist organizations have centered their campaigns on the idea that the oppression of women occurred in personal relationships, and that therefore ‘‘the personal was political’’ (Charles, 1993). Hence women were invited to live their politics in their homes. They were given guidelines on how to proceed for doing so, such as by negotiating (or even doing battle) with their partners to share child care and housework. If the relations between ingroup and outgroup are perceived as conflictual, another consequence of the target’s self-definition in terms of a social category can be greater interpersonal distance, and a more conflictual stance toward a perceiver categorized as an outgroup member. Depending on the expectations held by the perceiver, this type of behavior can be construed as consistent with these expectations, especially if these expectations concern sociability (such as coldness, aggression, hostility). For some suggestive evidence in support of this assumption, consider the research on linguistic accommodation (for a review, see: Giles, Coupland, & Coupland, 1991), which shows that individuals with an insecure social identity are likely to adopt a ‘‘competitive’’ linguistic style (such as the ingroup’s idiom or accent) when interacting with members of relevant outgroups. A study conducted in Belgium by Bourhis, Giles, Leyens, and Tajfel (1979) illustrates this process. In this study, Flemish (Dutch-speaking) participants were addressed in French by a Walloon (French-speaking) confederate after having spoken in English for a few minutes. When this happened, participants tended to respond in their own idiom, Flemish, as a way of affirming their threatened identity. Following the terminology of Smith et al. (1997), linguistic divergence is an instance of ‘‘nondeferential’’ behavior in which participants refuse to abide by the conversational script imposed by their partner. Nondeferential behaviors, like divergence, may have clear implications for the occurrence of behavioral confirmation and discomfirmation of social stereotypes. As we have already noted, nondeferential behavior on the part

192

KLEIN AND SNYDER

of the target prevents the perceiver from imposing his ‘‘script’’ on the interaction, and it is therefore likely to disrupt the strategy of confirmatory hypothesis testing (Neuberg, 1996; Smith et al., 1997). The consequences of such disruptions may depend on the content of the perceivers’ expectation about the target’s group. When the target is seen as belonging to a low status group that has a cooperative relation with the higher status perceiver’s group, stereotypes may represent the stigmatized as inherently friendly and docile (Fiske et al., 2002). In this case, an outcome of the adoption of a nondeferential, challenging, behavior may be behavioral disconfirmation. When the target is seen as belonging to a group perceived as having a competitive relation with the high status perceiver’s group, stereotypes may represent the stigmatized as hostile and aggressive (such as black professionals, Jews, or rich people). In this case, the outcome of the adoption of a nondeferential behavior may be behavioral confirmation.

C. THE CHOICE OF AN INDIVIDUAL VERSUS COLLECTIVE STRATEGY Having identified both individual and collective strategies that may be employed in attempts to disconfirm stereotypes, the following question naturally arises: What determines the type of strategy that targets will adopt? A factor that may increase the likelihood of adopting a collective strategy when one suffers from a negative or stigmatized identity is the perception that boundaries between the stigmatized group and more prestigious other groups are impermeable (Tajfel, 1975; Tajfel & Turner, 1986; Taylor & McKirnan, 1984; Wright, Taylor, & Moghaddam, 1990). Thus, if members of a social group believe that it is possible to move individually from one group to another, they may favor a strategy of individual mobility in order to obtain a satisfactory social identity. Beyond the objective stratification of society, legitimizing ideologies (such as the Protestant work ethic) and the presence of successful tokens (Wright & Taylor, 1998, 1999) influence this perception. Because of these factors, boundaries that are almost impermeable can be perceived as permeable. On the other hand, when members of a disadvantaged group perceive that barriers between their own ingroup and other, more prestigious outgroups are impermeable, they are likely to believe that a satisfying social identity cannot be achieved unless a collective effort is undertaken (Tajfel, 1975; Tajfel & Turner, 1986; Taylor & McKirnan, 1984). When group boundaries are perceived to be impermeable, other factors may come into play, including the perceived legitimacy and stability of existing status differentials. If group members feel that their disadvantaged

STEREOTYPES AND BEHAVIORAL CONFIRMATION

193

status is legitimate, they may view collective action as unwarranted and, if they view the dominant outgroup as sufficiently powerful to maintain the status quo, they may refuse to engage in such action fearing that it will be unsuccessful. In such a situation, they may choose other comparison groups, or rely on intragroup comparisons to achieve a positive social identity. They are then likely to accept the stereotypes legitimizing their disadvantage (Jost & Banaji, 1994; Tajfel & Turner, 1986), and to enact them in interpersonal interactions with members of the advantaged group. Perceived instability and illegitimacy may be viewed as the ‘‘cognitive alternatives’’ to the existing situation, alternatives that enable targets to believe in the feasibility and legitimacy of this strategy (Ellemers, 1993; Wright & Taylor, 1998; Wright et al., 1990). When they are present, strategies such as social competition (competing with the outgroup on existing dimensions of comparisons) or social creativity (defining new dimensions of comparisons favoring the ingroup) are likely to be implemented. In such a case, targets will be motivated to change the stereotype of their group in order to create more favorable comparisons to relevant outgroups (Tajfel, 1981b).

D. CONCLUSION Altogether, our analysis of the situation of the target of stigmatizing social stereotypes suggests that when intergroup boundaries are perceived to be permeable, targets may engage in a strategy of individual mobility if they are stigma conscious. If successful, the outcome of this strategy should be stereotype disconfirmation in interactions with the nonstigmatized. However, we have seen that there may exist powerful barriers to the implementation of this strategy, especially lack of skills and lack of power. When targets are not stigma conscious, they may develop a strategy of interpersonal adjustment, which leads to stereotype confirmation. By contrast, when group boundaries are perceived as impermeable, targets may engage in a collective strategy of stereotype change, if they perceive alternatives to the existing status system. In the absence of alternatives, real or perceived, targets are likely to endorse the stereotype and to enact it. This analysis suggests that targets who identify with the stigmatized group may be able to resist the perceiver’s expectations because their identification makes them more likely to be stigma conscious. Moreover, group consciousness may provide a sense that their disadvantage is illegitimate and does not reflect the inner dispositions and traits of members of the target group. The group may also empower them by providing a perception of social support and skills that can enable them to interact with outgroup

194

KLEIN AND SNYDER

members. All of these factors may make them more willing to endure personal risks on behalf of the group. Hence, a sense of collective identity on the part of members of groups that are the targets of stigmatizing social stereotypes may undermine and overcome the influence of those factors that tend to facilitate, at an individual level, stereotype enactment even in the presence of stigma consciousness.

VI. Power and Status Differences in the Dyad So far, in our analysis, we have considered the perceiver and the target separately. Now, let us build on these considerations to examine the dyad composed of the perceiver and the target as a structural unit. Specifically, we shall examine how stigmatization can affect power and status relations within this dyad in ways that affect the occurrence of behavioral confirmation of social stereotypes. We have already seen that as the perceiver’s power in relation to the target increases, so does the likelihood of behavioral confirmation (Copeland, 1994; Harris et al., 1998). Several authors have argued that power differences combined with an implicit or explicit acceptance of it by the target are necessary for behavioral confirmation to occur (Neuberg, 1996; Snyder & Kiviniemi, 2001). Our analysis of the intergroup dimension of behavioral confirmation therefore must attend to the ways that the relations between the groups to which the target and the perceiver belong affect the balance of power in their interaction. First, regarding stigmatized groups are more likely to be behaviorally confirmed if members of those groups tend to occupy positions of low power in relation to members of nonstigmatized groups who adhere to these stereotypes. This position of lesser power seems to be the rule for members of stigmatized groups. For example, in a study by Ramirez and Soriano (1993), European Americans reported being less likely to encounter members of ethnic minorities in equal and higher power positions than in lower power positions. The reverse held for participants belonging to such minorities (blacks, Hispanics, Asian-Americans). A similar pattern obtains for women in relation to men (Smith-Lovin & McPherson, 1991) as well as for other minorities in relation to the majority group (McPherson et al., 2001). One of the implications of these findings is that stereotype confirmation may sometimes be directly due to differences in status and power, and not only indirectly through the acceptance of the perceiver’s script by the target. For example, Henley and colleagues have suggested that gender differences in nonverbal behavior were due to women’s lesser status

STEREOTYPES AND BEHAVIORAL CONFIRMATION

195

and power, but that these differences disappeared when power and status were controlled (Henley, 1977, 1995; LaFrance & Henley, 1994). If this is the case, stereotypes could be confirmed only because men tend to have a higher status than women in interpersonal interactions. Although Henley’s hypothesis has been contested (Hall & Friedman, 1999), status differences seem to indeed contribute to behavioral differences that may, falsely, be attributed to women’s inner traits and dispositions. Moreover, power differentials are not distributed evenly across groups. Some groups tend to accumulate power and status on many important dimensions whereas other groups accumulate ‘‘powerlessness’’ on these same dimensions (Fiske, 2001; Pratto & Walker, 2001; Sidanius & Pratto, 1999). A consequence of these differences in power is that members of a subordinate group may be deprived of power in interpersonal interactions even if the power differential along which this difference is defined is not directly relevant to the interaction. Pratto and Walker (2001) note, for example, that constraints on women’s power in their families limit their power with their employers. If a professional woman is constrained to perform most of the housekeeping and child-rearing activities, it will be generally more difficult for her to commit herself as much to an organization as her male domestic partner who does not experience such constraints. In turn, this lesser commitment may deteriorate her potential value for the employer, and hence her power within the organization. Third, belonging to a low status group can, in and of itself, be a source of low power in interactions with nonstigmatized people. The studies that we have considered so far rely on a formal definition of power, generally defined as ‘‘legitimate’’ power in the terminology of French and Raven (1959). Nevertheless, informal differences in power can also be present in dyads. Consider, for example, the case of two members of a dyad cooperating on a task in order to obtain financial rewards. One of the dyad members may, for whatever reason, consider that his or her partner is more task-competent and therefore choose to follow the partner’s guidelines for performing the task. In such a situation, this dyad member voluntarily yields control of the valued reward to his or her partner who, therefore, has more power. How does such an imbalance in power emerge? According to expectation states theory (EST: Balkwell & Berger, 1996; Berger et al., 1972; Ridgeway & Berger, 1988; Wagner & Berger, 1993), members of such task groups form performance expectations about the other group members. An informal ‘‘status and prestige order’’ typically emerges as a function of these expectations. Individuals who are expected to perform best have greater power and influence over others. These expectations are informed by ‘‘status characteristics’’ that may include actual evidence of skills at performing the

196

KLEIN AND SNYDER

task (e.g., diplomas, prior experiences), but they can also depend on stereotypes associated with the group members’ social categories. Thus, women are often expected to be less competent than men in mechanical tasks and may therefore be given lesser status than men in interactions revolving around task performance. In this case, stereotypes serve to legitimize interpersonal power differences. However, sometimes, there is no a priori link between a social category membership and performance expectations. For example, if the task involves solving an anagram, women may not be expected to be particularly less efficient than men (and vice versa). In such instances, however, EST argues that people will act under the assumption that this social category is relevant and can be used for making performance expectations unless the evidence shows that it is irrelevant. In this case, they will use the position of the group in the social structure as a guide for forming performance expectations. When this group membership is salient in the interaction, and when these beliefs are relevant to the purpose of the interaction, it leads to consensual expectations that the participant belonging to the high-status group will be more competent, and perform better, than the other participant. These expectations may then shape behavior in a self-fulfilling way, such that the high-status group member will tend to behave more assertively and influence the low-status group member whereas the lowstatus group member will behave with deference and accept influence attempts. In other words, status determines power in these interactions. High status will be associated with patterns of dominant behavior, indicative of high task competence (such as a fast speech rate, a firm tone of voice, few hesitations) whereas low status will be associated with behaviors indicative of low competence (Ridgeway, 1987, 1991). The highstatus member’s power can materialize itself by greater control over the task and more influence over the choices made by the dyad. Hence, if both participants share the same cultural beliefs regarding the performance expectations associated with their respective group memberships, these expectations can become self-fulfilling. EST, which has received ample empirical support (for reviews, see Berger, 1992; Berger et al., 1972, 1980; Ridgeway, 1991, 2001), therefore suggests that members of stigmatized groups may similarly tend to adopt submissive behavior even when they are formally as powerful as their nonstigmatized partner. Consider, for example, a study conducted in an Australian context by Riches and Foddy (1989). These authors manipulated the accent used by a confederate to address participants in the context of a joint decisionmaking task. Participants behaved with greater deference and were more likely to accept influence attempts from the confederate if he spoke with the dominant Anglo-Australian accent than if he spoke with a Greek accent

STEREOTYPES AND BEHAVIORAL CONFIRMATION

197

(Greeks constitute a low-status ethnic minority in Australia) although the task was unrelated to the stereotype of Greeks in Australia. When the stigmatized group members try to adopt a ‘‘more’’ dominant style in such groups, they are often met with resistance on the part of the nonstigmatized. For example, in cooperative tasks involving cognitive problem solving, attempts at achieving equal status interactions between whites and blacks by training the latter to use a more assertive behavior failed because whites resisted these attempts (Katz, 1970; Katz & Cohen, 1962). Similar findings have been reported in the context of task groups involving men and women (Ridgeway, 1982; Ridgeway & Berger, 1988) as well as other ethnic minorities (Cohen, 1982; Ridgeway, 1991). In line with EST, these findings suggest that intergroup differences in power can result in interpersonal differences in power. Overall, in our considerations of the structural qualities of perceiver– target dyads, we have seen that members of stigmatized groups (as targets) tend to have lower power and status in their interactions with members of nonstigmatized groups (as perceivers), which increases the likelihood of behavioral confirmation of stereotypes about the stigmatized group. In line with this analysis, Madon, Jussim, and Eccles (1997) have shown that disadvantaged groups such as African-Americans and people low in socioeconomic status (SES) were most vulnerable to stereotype confirmation. Similarly, women are more vulnerable to stereotype confirmation than men (Christensen & Rosenthal, 1982; Johanson, 1999; Snyder & Oyamot, 2001) and less likely to induce it (Dvir, Eden, & Banjo, 1995). This evidence suggests indeed that disadvantaged groups are more vulnerable to behavioral confirmation when they are in contact with members of an advantaged group.

VII. How Does Behavioral Confirmation Affect Stereotypes and Intergroup Relations? Now, building on our considerations of the interactional strategies of perceivers and targets, we shall turn to our ultimate question, which concerns the impact of behavioral confirmation and disconfirmation on social stereotypes and intergroup relations. When are the stereotypes held by a nonstigmatized group about a stigmatized group likely to be maintained or to be changed as a result of behavioral confirmation or disconfirmation of expectations based on these stereotypes in dyadic encounters between members of these two groups? And what is the impact of behavioral confirmation, as it occurs in interpersonal interactions

198

KLEIN AND SNYDER

between individual members of these groups, on the structure and dynamics of intergroup relations between the groups themselves? Thus, although much of what we have to say in answer to these questions is admittedly speculative, we are attempting to address the question of how the ‘‘micro’’ or interindividual level of analysis can affect the ‘‘macro’’ or sociostructural level of analysis. Consider the case of a white settler interacting with his local servant in colonial Africa. He may display the typical paternalistic attitude, combining a form of benevolence with power and authority. Conversely, the African servant may adopt a complementary behavioral style, combining signs of gratitude and submission to his master’s authority. In such a situation, the adoption of this style by the servant may be construed as confirming the white man’s stereotype of Africans as inherently stupid and submissive. As these types of relations were common in colonial Africa, such a process of stereotype confirmation may have served to reinforce the regime by legitimizing the domination of the Europeans over the Africans. But, how readily can such bridges from the interpersonal to the intergroup level of analysis be built, especially when they are to be built on the foundations provided by social psychological research on the confirmation of stereotype-based expectations in dyadic social interaction? Although this question has rarely been addressed, it has not been completely neglected either. One answer has been offered by Jussim (Jussim & Fleming, 1996; Jussim et al., 2000), who has drawn a distinction between dyadic selffulfilling prophecies (the typical interactions studied in research on behavioral confirmation) and sociological self-fulfilling prophecies, which require the action of many people (often in the form of social/cultural institutions). In terms of this distinction between dyadic and sociological self-fulfilling prophecies, it is the sociological ones, rather than their dyadic counterparts, that are thought to contribute to the maintenance of social stereotypes: Although self-fulfilling prophecies clearly occur in dyadic interactions, they may have only limited involvement in many of the deepest and most intractable social problems associated with stereotypes, prejudice, and discrimination. For example, the ghettoization of Jews in Europe, the Hindu caste system, American slavery and South African apartheid could not have been maintained by the actions of a handful of individuals. (Jussim & Fleming, 1996, p. 179)

In this view, dyadic interactions cannot contribute to intergroup relations because they do not involve a large number of people. For example, in our example, the interaction between the white settler and his African servant has, according to this point of view, no bearing on the social stereotypes

STEREOTYPES AND BEHAVIORAL CONFIRMATION

199

held by each group regarding his or her counterparts; what matters, instead, is colonialism as an institutional and political endeavor. To assess the validity of this argument, we shall delineate three conditions that are necessary for behavioral confirmation processes to affect the actual content of social stereotypes. First, the nonstigmatized perceiver must view the stigmatized target’s behavior as either confirming or disconfirming his or her expectation and attribute this behavior to the target’s inherent dispositions. Second, the nonstigmatized perceiver must maintain or change his or her stereotype of the target’s group in a direction consistent with his or her perception of the target’s behavior. Thus, not only must the stereotype inform his or her expectation, but behavioral confirmation of this expectation derived from the stereotype must ‘‘feed back’’ to the stereotype itself and either strengthen it or modify it. In other words, perceptual confirmation or disconfirmation needs to be generalized to the stigmatized group as a whole and become actual stereotype confirmation or disconfirmation. And, third, this process must occur at a ‘‘macro’’ level, such that the stereotype, viewed as a collective representation shared by the members of the nonstigmatized group, needs to be affected in the same way. That is, the perceptual and behavioral confirmation or disconfirmation of expectations and their consequent feedback to the general stereotypes from which they are generated need to be consistent across a broad array of individual nonstigmatized perceivers. Let us consider each of these three conditions in turn, and examine whether it can plausibly be fulfilled. Once we have examined each of these conditions, we shall be in a position to address the relevance of dyadic self-fulfilling prophecies to the ‘‘macro’’ context of intergroup relations.

A. PERCEIVING AND EXPLAINING THE TARGET’S BEHAVIOR Even in the presence of disconfirming behaviors, perceivers may nonetheless manifest perceptual confirmation and fail to modify their expectations regarding the target’s traits and dispositions (Bond, 1972; Hilton & Darley, 1985; Ickes et al., 1982; Jones & Panitch, 1971; Miller & Turnbull, 1986; Neuberg, 1994; Snyder & Haugen, 1995; Swann & Snyder, 1980). Thus, perceivers tend to maintain their expectations. As Miller and Turnbull (1986) have noted, this effect may be due to lower level encoding processes in which perceivers selectively focus on expectation-consistent behavior and use this behavior when forming an impression of the target. Consistent with this assumption, expectation-congruent behaviors tend to be better remembered and integrated in judgments than expectationincongruent behaviors (Rothbart, Evans, & Fulero, 1979).

200

KLEIN AND SNYDER

Second, expectation-congruent behaviors are more likely than expectationincongruent behaviors to be attributed to internal and stable dispositions of the actor; expectation-incongruent behaviors, by contrast, are preferentially attributed to external and unstable factors (Bond, Omar, Pitre, & Lashley, 1992; Miller & Ross, 1975; Olson et al., 1996; Weiner, 1986). Similarly, the prevalent process of social attribution involves the tendency to attribute stereotype-consistent behavior more so than stereotype-inconsistent behavior to inner dispositions (Deschamps, 1973–1974; Duncan, 1976; Yzerbyt, Corneille, & Estrada, 2001). For example, if the imaginary Moroccan job candidate of our opening example conveys an interest in extracurricular activities, and expresses correspondingly little professional ambition, the Belgian interviewer may be much more likely to attribute these behaviors to inner dispositions (e.g., laziness) than if the job candidate displayed stereotype-inconsistent behavior. But stereotypes may even reinforce this tendency. According to Leyens, Yzerbyt, and Schadron (1994), stereotypes are ‘‘naive theories’’ that can be used for explaining the behavior of members of the target group. People tend to attribute stereotype-consistent behavior to the underlying essence of the group, or its deep psychological properties (Rogier, 1999; Yzerbyt et al., 2001; Yzerbyt, Rocher, & Schadron, 1997). Thus, our interviewer may not only view the Moroccan candidate as lazy, but consider this laziness as reflecting the underlying essence of Moroccans, their ‘‘Morocanness.’’ To the extent that stereotypeinconsistent behavior cannot be linked to an underlying essence, it may be more easily discarded as a consequence of the situation. For example, the interviewer may attribute any claims the job candidate makes about commitment to work as reflecting a self-presentational agenda rather than any inner motivations and capacities. This analysis suggests that when nonstigmatized group members hold stereotypic views of the stigmatized group, interactions between members of stigmatized and nonstigmatized groups are much more likely to contribute to the confirmation of these views than to their modification. It follows, too, from this line of argument that the target may actually need to explicitly dissociate himself or herself from the stigmatized group in order to escape perceptual confirmation. For example, the Moroccan candidate may try to display evidence of many attitudes and behaviors thought by the interviewer to be atypical of Moroccans as a group (e.g., not practicing Islam, being favorable to the emancipation of women, or being very individualistic). If the job candidate pursues this strategy, he is likely to be viewed as a very untypical Moroccan and the stereotype will be perceived as less relevant in judging him and his suitability for the job in question. So far, this analysis suggests that stereotypical information seems to be more easily integrated in the judgment of the target than

STEREOTYPES AND BEHAVIORAL CONFIRMATION

201

counterstereotypical information. However, when behavior is extremely incongruent, the advantage of congruent information on the encoding process tends to disappear (Hashtroudi, Mutter, Cole, & Green, 1984; Hastie & Kumar, 1979; Stern, Marrs, Millar, & Cole, 1984). When they are very surprising, these events may attract attention and be integrated in the judgment. Thus, our Moroccan candidate may adopt a second strategy, one that consists of remaining a ‘‘true Moroccan’’ but display extremely counterstereotypical behavior on the few dimensions of comparison that may be relevant to the interviewer’s judgment. For example, he may show his adherence to family values and to Islam at the same time as he displays ample evidence (far beyond that expected of a non-Moroccan candidate) of his commitment to work (e.g., through letters of recommendation, diplomas, awards). Thus, although a variety of processes work together to make behavioral disconfirmation go unnoticed much of the time, behavioral disconfirmation will attract attention to the extent that the target succeeds in dissociating his or her image from the group or displays extreme instances of disconfirming behavior.

B. MAINTENANCE AND CHANGE OF THE PERCEIVER’S STEREOTYPES According to the logic of essentialist theorizing (Yzerbyt et al., 1997, 2001; Yzerbyt & Rogier, 2001), stereotype-confirming behavior is not only attributed internally, but serves to reinforce the stereotype. The explanatory value of the stereotype is bolstered by the evidence it serves to explain. Thus, in our example of the Moroccan job candidate, evidence indicating lack of commitment or initiative may not only be viewed as reflecting the Moroccan’s inherent essence but as confirming that laziness is part of this essence. Thus, there may be an asymmetry between confirmation and disconfirmation, with stereotype-consistent behavior being more likely to lead to the maintenance of the perceiver’s stereotype than stereotype disconfirmation is to contribute to changing the perceiver’s stereotype. As we have seen, the target may need to dissociate himself or herself from the group if he or she wishes to succeed in changing the perceiver’s mind. But as a consequence, the perceiver may not view this target’s behavior as relevant to the group stereotype. After all, as our hypothetical job interviewer might reason, if this candidate is such an idiosyncratic Moroccan, his commitment should not be taken as evidence that Moroccans in general can be committed to their job.

202

KLEIN AND SNYDER

The literature on stereotype change is consistent with this line of reasoning. Although there seems to be a natural tendency for group members to generalize the behavioral information they possess regarding an individual outgroup member to his or her group as a whole (Quattrone & Jones, 1980), this target needs to be viewed as a typical exemplar of the group for this to be the case (Hewstone, 1996; Hewstone & Brown, 1986; Rothbart & John, 1985; Weber & Crocker, 1983; Wilder, 1984). Changing the stereotype of the group may therefore demand that the stigmatized be perceived as a typical member, that is, as ‘‘fitting’’ the category (Rothbart & John, 1985). Hence, the most fruitful strategy for changing negative stereotypes, and associated prejudicial attitudes, toward the target group may involve the display of extreme counterstereotypic behavior on an important dimension of comparison rather than a total dissociation from the stereotype. A study by Brown, Vivian, and Hewstone (1999: study 1) illustrates this strategy for changing negative stereotypes. In this study, British participants interacted with a German target (actually, a confederate) who was either typical or atypical of stereotypes about Germans, based on an alleged personal profile read by the British participant prior to the interaction. After the interaction, participants reported that their perception of Germans as a group had changed, on both stereotype-relevant and stereotype-irrelevant attributes, but these changes were especially pronounced when the confederate was described as a typical German. Thus, only when the target was perceived as a typical member of the outgroup were ratings of the individual German target generalized to the outgroup as a whole. In their interpretation of these findings, Brown et al. (1999) proposed that typicality enhances the salience of the target’s group membership. And, as we have already seen, SCT holds that category salience entails perceptions of the outgroup as undifferentiated and homogeneous (for a review of empirical evidence, see Haslam, Oakes, & Turner, 1996), with each group member viewed as a prototypical and interchangeable member of the group. Hence, under conditions of category salience, perceptions of individual group members are more likely to be generalized to the group as a whole. So far, we have treated category salience from the nonstigmatized perceiver’s perspective. But the salience of the stigmatized target’s selfcategorization may also contribute to the generalization of the nonstigmatized perceiver’s beliefs and expectations about individual group members to their stereotypes and attitudes toward the target’s group as a whole entity. First, when his or her social identity is salient, the stigmatized target is more likely to self-stereotype himself or herself, and thereby adopt the attitudes or behaviors stereotypically ascribed to his or her group. Such self-stereotyping may make the target less likely to explicitly dissociate himself or herself from

STEREOTYPES AND BEHAVIORAL CONFIRMATION

203

the target’s group and, thus, may be associated with a greater likelihood of generalization of the exemplar to the group. Moreover, when the stigmatized target’s identity is salient, not only does the target display behavior that he or she views as prototypical of his or her group, but he or she is motivated to be viewed as a typical member of this group (Klein, Licata, Azzi, & Durala, in press; Noel, Wann, & Branscombe, 1995). Hence, even if the target’s self-stereotype may differ in many ways from the perceiver’s view of the stigmatized group, the target’s motivation and efforts to show that he or she is a representative and ‘‘true’’ member of the stigmatized group should minimize the perceiver’s tendency to view him or her as an atypical exemplar. Finally, the salience of the targets’ self-category may also be associated with a process of ‘‘depersonalization’’ in which targets will view themselves as interchangeable members of the group and identify with the interests of the group as a whole (Simon, 1998; Simon et al., 1997). Hence, rather than modifying the view held by the perceiver about himself or herself as an idiosyncratic individual, the target may be motivated to change the view and attitudes held by the perceiver regarding the stigmatized group as a whole. In other words, under such circumstances, the target is more likely to follow a collective strategy of stereotype change. Conversely, the pursuit of an individual strategy by the target may contribute to the persistence of the perceiver’s stereotype. Group members who succeed in their attempts at social mobility not only tend to dissociate themselves from their group, but they are particularly likely to endorse negative stereotypes concerning their group. For example, Ellemers (2001) has found that female, but not male, university professors described women Ph.D. students as less committed to the university and their career than male students, thereby perpetuating the stereotype of women as less ambitious than men. Altogether, the foregoing analysis suggests that salience of the perceiver’s identity facilitates the generalization of the target’s behavior to the stigmatized group by inducing a view of the target’s group as undifferentiated. Moreover, salience of the target’s identity also facilitates this process by making it more likely that the target will behave as a typical ingroup member, and be motivated to be viewed as such by other people.

C. MAINTENANCE AND CHANGE IN GROUP PERCEPTIONS AND ATTITUDES So far, we have examined how confirmation or disconfirmation could affect the nonstigmatized group member’s stereotype of the stigmatized group. In doing so, we have considered the stereotype at the level of the

204

KLEIN AND SNYDER

individual perceiver. But what about the impact of stereotype confirmation and disconfirmation on intergroup perceptions—that is, on the collective representations held by the nonstigmatized group as a whole? We shall address this question in two steps: first, by examining stereotype maintenance and, second, by examining stereotype change. 1. Stereotype Maintenance A simple way to address the transition from the interpersonal to the intergroup perspectives on stereotypes and their confirmation involves considering that collective representations will be maintained to the extent that it is the rule for the nonstigmatized perceivers’ stereotypes regarding the stigmatized group to be subjectively confirmed, or at least not disconfirmed, when they interact with members of this group. This condition, of course, can hold only if the stereotypes held by individual perceivers are shared with great regularity and consistency across members of the nonstigmatized group when they interact with members of the stigmatized group. If not, each instance of behavioral confirmation or disconfirmation will concern different stereotypes and, considered globally, these episodes are unlikely to have an impact at a group level. Stereotypes, it has often been noted, are flexible and their content can vary as a function of the social context, even among the same perceivers (see, e.g., Devine & Elliot, 1995; Diab, 1962; Haslam & Turner, 1992, 1995; Haslam, Turner, Oakes, & McGarty, 1992; Karlins, Coffman, & Walters, 1969). Hence, it is imperative to specify those conditions in which there will exist such great consistency and regularity in nonstigmatized group members’ view of stigmatized targets. Again, category salience seems to provide an answer. Group members are most likely to agree on a consensual representation of a target group when their own identity is salient (Haslam, 1997; Haslam, Oakes, Reynolds, & Turner, 1999; Haslam, Turner, Oakes, McGarty, & Reynolds, 1998). For example, Haslam et al. (1999) found that Australians were more likely to use the same traits to describe their ingroup when their social identity had been made salient (i.e., by asking them to list activities they and Australians did well, badly, frequently, and rarely) than when their personal identity was made salient (i.e., by asking them to list activities they personally did well, badly, frequently, and rarely). If perceivers tend to share a common self-categorization when they interact with members of the stigmatized group, their stereotypes will be shared. But, even more importantly, their behavior as perceivers will be affected in similar ways by the salience of their common group identity. For individuals adopt what they perceive to be the shared and normative patterns of action when their social identity is salient (Turner et al., 1987). In

STEREOTYPES AND BEHAVIORAL CONFIRMATION

205

the present case, their actions will tend to be shared to the extent that they are informed by common, shared expectations and that they reflect consensual norms toward members of the target group. For example, the dominant style of the European colonialist reflects a form of normative ‘‘etiquette’’ indicating how relationships with Africans should be handled. Through these shared patterns of action, the nonstigmatized group may exert a truly collective influence on the stigmatized group by inciting its members to perform stereotype-consistent behaviors, even if they are physically isolated. Can this condition be fulfilled? Is it likely that perceivers would separately tend to use a common self-categorization when they interact with members of a stigmatized group? For this to be the case, factors that transcend the context of the interaction need to uniformly affect perceivers, as a group, in such a way that they tend to rely on the same stereotypes to describe members of the target group. For example, when Belgians interact with Moroccans, they should tend to define them as Moroccans (rather than as men, or as fish lovers, or any of a variety of idiosyncratic categories) and they should activate the same expectations regarding Moroccans. In fact, there are several reasons to believe that nonstigmatized perceivers are likely to view stigmatized targets in categorical terms rather than in terms of their idiosyncratic traits and dispositions. First, there seems to be a general tendency to view outgroup members in categorical terms and ingroup members in terms of personal categories (Park & Rothbart, 1982; Quattrone & Jones, 1980; Sedikides, 1997). As outgroup members are generally appraised in intergroup contexts and ingroup members in intragroup contexts, this is consistent with SCT4 (Haslam et al., 1996). This effect, known as the differential processing effect, is more likely when the perceiver belongs to a higher status group than the target (Sedikides, 1997). Thus, high-status group members tend to view members of disadvantaged groups as prototypical and interchangeable members of their group and other advantaged group members as idiosyncratic individuals. Hence, nonstigmatized perceivers should tend to uniformly view their stigmatized targets as interchangeable prototypical exemplars of their category. However, for a common category to be used across perceivers to define targets belonging to a stigmatized group, this category may need to be a particularly meaningful and global category, so that it is likely to be salient across a variety of contexts. For example, categories based on sex, race, and 4

Note that Halsam et al. (1996) have shown that the ingroup and the outgroup are perceived as equally homogeneous when they are both appraised in intergroup rather than intragroup contexts.

206

KLEIN AND SNYDER

age define such categories in American society (Fiske, 1998) just as language does in Canada and Belgium, or as caste does in India. The importance of specific categorical criteria is mainly a function of the history of intergroup relations, as these relations define which categories have an important meaning within a particular society. But even if perceivers all categorize the low-status target in the same way, are they likely to use the same stereotypes to describe this target, and hence to develop the same expectations to guide their treatment of the target? This is, of course, our second condition for making the transition from the interpersonal to the intergroup level. As we have seen, stereotypes tend to be known and shared across group members (Gordijn, Koomen, & Stapel, 2001; Haslam, 1997; Madon et al., 2001). Nevertheless different traits might be activated depending on the context in which the interaction takes place. Hence, this condition is most likely to be fulfilled if the contexts of the interactions between members of the advantaged and disadvantaged group tend to be consistent. Such a homogeneity in the contexts of interactions will tend to lead to the activation of the same traits. For example, the white settler of our example meets blacks in only limited social roles and may therefore activate a small set of traits concerning blacks, whereas another person who meets blacks in a variety of roles and statuses may find a variety of different traits or subtypes become relevant to perceiving blacks. For many minority groups in the United States, the fulfillment of this condition seems to be facilitated by the presence of group segregation in many important contexts of social life. For example, a recent study using a representative sample of African-Americans (Brown, 2001) indicates that most respondents lived most of their life in all-black, or predominantly black, contexts (e.g., school, church, neighborhood). The only mixed settings were usually the workplace and college. This finding is important in itself as the influence of stereotypes on majority group’s judgments of minority groups is proportional to the degree of segregation in this setting (Kraiger & Ford, 1985; Reskin, McBrier, & Kmec, 1999; Sackett, DuBois, & Noe, 1991). Studies in workplace settings, one of the few contexts in which mixed interactions regularly occur, indicate that such interactions are likely to be very stereotyped for at least two reasons. First, groups tend to have homophilous social networks in organizations; that is, they tend to develop relationships with members of the same ethnic group (for a review, see McPherson et al., 2001), which suggests that intergroup interactions are generally formal ones that involve strong role constraints. Second, blacks are often segregated in a limited array of low power positions and social roles within contemporary American organizations, especially small ones (Reskin et al., 1999), which means that white employees may interact with

STEREOTYPES AND BEHAVIORAL CONFIRMATION

207

blacks in the context of only a handful of subordinate social roles. This pattern of segregation should lead whites, as perceivers, to consistently rely on similar expectations in their interactions with blacks (Eagly, 1987). Even when blacks manage to access higher power positions, they may tend to become ‘‘tokens’’ and their relationships with whites often become even more stereotypic. According to Ibarra (1993), ‘‘the presence of tokens produced boundary-heightening processes by which dominants exaggerate group differences to reinforce their common bonds’’ (p. 69). The foregoing considerations may apply to other ethnic minorities (Ibarra, 1995) and to women as well (Smith-Lovin & McPherson, 1991). Where, then, do we stand with respect to the conditions for the maintenance of collective stereotypes as a result of stereotype confirmation processes that we set forth, namely that nonstigmatized perceivers adopt a common self- and other-categorization when interacting with members of the stigmatized groups, and that they rely on the same stereotypes during these interactions? We have seen that such conditions could plausibly be fulfilled, and even facilitated, in at least some types of intergroup relations, such as those between ethnic majority and some minorities in the United States. Let us now consider whether stereotype change could be plausibly affected by stereotype disconfirmation. 2. Stereotype Change Stereotype change should occur to the extent that the stereotypes held by the nonstigmatized group tend to be perceptually disconfirmed when they interact with members of the stigmatized group. Any generalized transition from the micro- to the macrolevel may be harder to accomplish as the multiple evidences of stereotype disconfirmation would need to be consistent and follow a regular pattern if they were to affect consensual stereotypes. Yet, several roadblocks may stand in the way of such an outcome. There is, of course, the well-documented reluctance of perceivers to abandon the stereotypes that they hold, even in the face of disconfirming evidence; accordingly, instances of disconfirmation may need to be extremely systematic, consistent, and frequent if they are to affect perceivers (see, for example, Garcia-Marques & Mackie, 2001; Kashima, Woolcock, & Kashima, 2000; Weber & Crocker, 1983). Moreover, to the extent that perceivers acknowledge instances of disconfirmation, but treat each instance of stereotype disconfirmation as if it were an isolated and idiosyncratic occurrence with each one believed to concern a different specific aspect of the overall general stereotype, their impact will not be systematic and may have inconsistent effects on shared views of the target group. This may likely be the case if the target chooses to ‘‘pass’’ as a member of the

208

KLEIN AND SNYDER

nonstigmatized group, or to adopt an individual self-presentational agenda of emphasizing his or her own individual achievements and qualities, which are likely to differ from those of other members of stigmatized groups pursuing this strategy. For a widespread pattern of disconfirmation of stereotypes to occur, and for it to be regularly and consistently acknowledged and accepted by perceivers, several conditions would seem to need to be met. Members of the stigmatized group would need to consistently follow a shared and common self-presentational agenda geared at advertising new self-stereotypes of the target group. This would, in keeping with our earlier considerations, require the salience of a common self-category. As we have seen, to the extent that their social identity is salient, each group member would then endorse the consensual self-stereotype, adopt shared patterns of behavior, and expect other group members to do so (see the section on collective strategy). By adopting the same behaviors, the stigmatized group can therefore also exert a collective influence on the nonstigmatized group. Is the pursuit of a collective strategy by nonstigmatized group members likely, or even possible? As we have argued, targets may pursue this strategy when they perceive intergroup boundaries as impermeable but when, simultaneously, they view cognitive alternatives to this situation. Such perceptions are common in social movements and are likely to arise among members of stigmatized groups (see, e.g., Kelly & Breinlinger, 1996; Tajfel, 1975). The main difficulty, however, lies in the enactment of the collective strategy of stereotype change. As we have seen, when an individual’s social identity is salient, such an enactment is mainly a problem of power. If members of a disadvantaged group have a collective consciousness but only encounter members of the advantaged group as their supervisors or their bosses, it may be extremely costly for them to enact positive self-stereotypes. As we have seen, this consciousness and the support provided by other group members can help these individuals overcome these risks. In fact, they may sometimes endure personal costs in the interest of the group. An extreme, but relevant, historical example is Rosa Parks, who refused to sit in the ‘‘black’’ section of a bus in Alabama in December 1955. By denouncing the illegitimacy of segregation, her behavior, later imitated by numerous blacks, contributed to the success of the Civil Rights movement in the United States. In less extreme ways, disadvantaged group members may express their identity in ways that do not entail risks for the self, such as the expression of collective beliefs and attitudes that are not punishable by powerful others (Reicher & Levine, 1994). Hence, a salient social identity may often find avenues for expression itself even in the face powerful opposition. In addition, the existence of a collective consciousness often creates opportunities for mixed interactions in new contexts. Collective movements

STEREOTYPES AND BEHAVIORAL CONFIRMATION

209

can indirectly affect stereotype change by helping to modify the contexts in which mixed interactions take place and in which stereotype disconfirmation is therefore facilitated. For example, as women massively entered British universities as a result of feminist movements (Charles, 1993), interactions between men and women took place in new and different contexts. These new contexts allowed women to enact self-stereotypes different from those of the loving, caring wife or mother. In turn, and by extension, the interactions that occurred in these new contexts then empowered them in other spheres of their lives. Existing studies suggest, for example, that as the number of women in an organization increases, the more likely it is that men will hold nonsexist attitudes and stereotypes regarding women in general (Heilman, 1980; Konrad, Winter, & Gutek, 1992; Reskin et al., 1999).

D. INTERACTIVE LEVELS OF ANALYSIS IN STEREOTYPE CONFIRMATION AND SOCIAL CHANGE Our attempt to build a bridge between interpersonal and intergroup levels of analysis suggests that dyadic behavioral confirmation will preserve general stereotypes to the extent that members of the nonstigmatized group tend to view their stigmatized interaction partners as typical members of their groups and share common expectations across perceivers regarding the traits displayed by members of this group. We further suggested that this was particularly likely to be the case if their identity as a member of the nonstigmatized group was salient during these interactions. Conversely, we proposed that behavioral disconfirmation would contribute to changes in general stereotypes to the extent that members of the stigmatized group tended to adopt a shared pattern of action and try to advertise common aspects of their self-stereotypes when interacting with members of the nonstigmatized group. This outcome would be most likely, we suggested, when their self-categorization as members of the stigmatized group was salient during interactions with members of the nonstigmatized group. Our analysis of the relations between behavioral confirmation and the associated consequences of social reproduction and social change (in general stereotypes and in the relations between groups who hold and who are the targets of social stereotypes) is compatible with a symbolic interactionist perspective (Blumer, 1969). According to this view, the large-scale institutions that form the structure of society at a macrolevel are partly maintained and reproduced, through face-to-face interactions (Becker, 1963; Blumer, 1969; Couch & Hintz, 1975; Maines, 1982; Maines &

210

KLEIN AND SNYDER

Charlton, 1985; see also Cohen, 2001 for a similar perspective on cultural variation). For example, colonialism was, in part, reproduced and maintained as an institution through the kinds of face-to-face contacts that we used to open this section. Joint activity in face-to-face contact is interwoven within other types of activities, of increasing complexity and scale but all these layers, by their very interactive nature, contribute to social reproduction and social change. This view does not consider social institutions as independent of face-to-face contacts, or as determining them. Rather, it suggests that face-to-face contacts can serve to maintain and change institutions even if they are constrained by larger scale factors. In this regard, we are not proposing that dyadic behavioral confirmation can, in and by itself, fully account for stereotype maintenance and change. Other layers of social organization constrain the contexts in which interactions tend to take place and, by the same token, determine the likelihood of confirmation or disconfirmation occurring, whether at the interpersonal or at the intergroup levels. For example, in a colonial society, contact between the colonizers and the colonized tended to take place in very specific social contexts, associated with role prescriptions that allow only for a limited range from both groups, making behavioral confirmation particularly likely. These contexts may limit the very occurrence of a contact: In a segregated society, for example, opportunities for contact between the races may be altogether absent. The very nature of these contexts is determined by the power and status relationships between groups. As we have seen, though, as relations between groups evolve, these contexts are likely to change, and so will the opportunities for stereotype confirmation and disconfirmation in social interactions. In turn, stereotype confirmation and disconfirmation may facilitate change at a higher level and make it more acceptable (witness, for example, the Rosa Parks example). Hence, our analysis, compatible both with SCT and a symbolic interactionist perspective, proposes that there is no discontinuity between the individual and the collective level of analysis—both levels interact and mutually influence each other.

VIII. Stereotype Confirmation, Stereotype Maintenance, and Stereotype Change: An Integration In our considerations of how targets would respond to perceivers who act toward them in terms of their membership in a devalued social category, we have envisaged several strategies that targets may use to respond to

STEREOTYPES AND BEHAVIORAL CONFIRMATION

211

perceivers. In the present section, we shall integrate these identity management strategies and, by building on the previous sections, examine how the use of these strategies can affect stereotype confirmation in interpersonal encounters as well as contribute to stereotype maintenance and change in intergroup relations. Critical determinants of the strategies chosen by targets, we have argued, are the permeability of the boundaries between the groups to which stigmatized targets and nonstigmatized perceivers belong, the stigma consciousness of the target, and the availability of cognitive alternatives to the existing position of the stigmatized group. Figure 1 presents the four strategies elicited by the interactive operation of these factors, as well as the outcomes of these strategies at the levels of personal expectations (what the perceiver believes about the target as an individual), individual stereotypes (what the perceiver believes about the target’s group), and collective stereotypes (what the perceiver’s group believes about the target’s group). Let us consider these four strategies in turn, beginning with those situations in which group boundaries are perceived as permeable. In such situations, stigmatized group members, we have argued, will seek to upgrade their status as individuals; however, stigma consciousness will play a key role in determining which strategy they will pursue.

A. INTERPERSONAL ADJUSTMENT: ‘‘GETTING ALONG’’ If their group membership is not salient, if they have rarely been confronted with racism or discrimination, or if they enter a context that they expect to be tolerant, members of stigmatized groups may not be stigma conscious. In such circumstances, as we have seen, they may be particularly likely to rely on the ‘‘getting along’’ strategy of ensuring a smooth and pleasant relationship with the perceiver. This strategy, we have seen, is conducive to both perceptual and behavioral confirmation of the perceiver’s expectations. Such confirmation may in turn serve to bolster the perceiver’s stereotypes concerning the group to which the target belongs. Theoretically, if stigmatized group members consistently follow this strategy in the presence of members of the nonstigmatized group, this outcome should contribute to the maintenance of social stereotypes. However, it is unlikely that members of the stigmatized group would continually ignore the facts that they may be the targets of stereotypes and prejudice. For, as we have seen, members of stigmatized groups tend to be constantly on the lookout for evidence of discriminatory behavior, not least because it protects their self-esteem (see, e.g., Crocker & Major, 1989; Crocker et al.,

Outcome at the level of Strategy

Personal expectation

Individual stereotype

Collective stereotype

Getting along

Perceptual Confirmation Behavioral Confirmation

Stereotype confirmation

Stereotype maintenance

Stereotype Compensation/ Dissociation

Behavioral disconfirmation Perceptual disconfirmation?

Irrelevant

Irrelevant

No

Stereotype enactment

Perceptual Confirmation Behavioral Confirmation

Stereotype Confirmation

Maintenance of the Stereotype

Yes

Stereotype change

Perceptual disconfirmation Behavioral disconfirmation

Stereotype disconfirmation

Stereotype Change

No

No

Is the target stigma conscious? Yes

Are group boundaries perceived as impermeable?

Yes

Are there cognitive alternatives to the existing position of the stigmatized group?

Fig. 1. Influence of the target’s strategies on stereotype confirmation at the individual and group level.

STEREOTYPES AND BEHAVIORAL CONFIRMATION

213

1991). Accordingly, the adoption of such strategies by targets may not necessarily result in the perpetuation and persistence of stereotypes.5

B. STEREOTYPE COMPENSATION When targets are stigma conscious, we have argued, they may be likely to engage in strategies in which they try to access a more prestigious group by actively dissociating themselves from their stigmatizing identity. If targets possess sufficient skills and power, they will try to show that they do not possess the attributes that are stereotypical of their group. By dissociating themselves from their group, the stigmatized may disconfirm the perceiver’s expectations regarding themselves as individuals. If this strategy is successful, the perceiver will view the target as an atypical exemplar and may not consider the stereotype as applicable to her. As a consequence, the target’s behavior may not elicit stereotype disconfirmation and will not contribute to stereotype change. This strategy of stereotype compensation is sometimes unsuccessful, especially if the target lacks the necessary skills or power for disconfirming the expectations. In this case, its net result may be stereotype confirmation, or at least the absence of disconfirmation. Failures of this strategy may also be due to the perceivers’ cognitive biases in favor of confirmatory information and their tendency to discount stereotype-inconsistent information as due to situational factors.

C. STEREOTYPE ENACTMENT When intergroup boundaries are perceived as rigid and impermeable, stigmatized targets are generally stigma conscious when interacting with members of nonstigmatized groups. When these targets perceive that there are no alternatives to their current situations, they generally will not try to challenge the power and status of the stigmatized group. Neither are targets likely to challenge the stereotype of their group held by nonstigmatized advantaged groups (Jost & Banaji, 1994; Jost, Burgess, & Mosso, 2001). On the other hand, in such a situation, targets may view these stereotypes as legitimate and enact them, especially their positive aspects. As we have seen, they may be encouraged to do so because this behavior is rewarded by powerful perceivers. This strategy facilitates behavioral and perceptual 5 For this reason, we have inserted a question mark after ‘‘Maintenance of the stereotype’’ in the relevant box of Fig. 1.

214

KLEIN AND SNYDER

confirmation. In turn, the perceiver is likely to view the target’s behavior as confirming his or her stereotypes concerning the stigmatized group. Finally, in a society characterized by such a structure, targets are likely to interact with members of the nonstigmatized group in a limited array of context and social roles, and also to consistently enact common elements of the stereotypes. Hence, multiple and consistent instances of individual stereotype confirmation contribute to the maintenance of collective views of the stigmatized group.

D. STEREOTYPE CHANGE Finally, when cognitive alternatives to the current position of the stigmatized group are present, targets may be motivated to improve their social identity by modifying the position of the group as a whole. In this case, targets will self-categorize in terms of the target group and view themselves as a typical member of this group. They are then likely to enact and advertise the self-stereotype endorsed by other ingroup members, especially if they have the power and the skills to do so. In this regard, the group may play an important role in empowering the target and providing guidelines for action. The target may try to challenge the dominant group on existing dimensions of comparisons or seek other dimensions of comparisons allowing the ingroup to compare positively to the outgroup. In the context of an interpersonal interaction, this strategy may lead to behavioral disconfirmation. It may also lead to stereotype disconfirmation, especially if the perceiver repeatedly encounters targets who display stereotype-inconsistent behavior and if the stigmatized is perceived to be a typical member of his or her group. These two conditions are facilitated by the tendency for members of a common salient ingroup to define themselves as prototypical members of their group and, therefore, to adopt common patterns of behaviors. Finally, if targets consistently advertise these alternative self-stereotypes in the presence of members of the advantaged group, they may successfully implement stereotype change.

IX. Summary and Conclusions ‘‘The organization of society is the framework inside of which social action takes place . . . . Such organization and changes in it are the product of the activity of acting units and not of ‘forces’ which leave such units out of account.’’ —Herbert Blumer (1962, p. 189)

STEREOTYPES AND BEHAVIORAL CONFIRMATION

215

The longstanding interest aroused by research on behavioral confirmation in social interaction has been motivated, in part, by the implications of this phenomenon for interactions at an intergroup level. For if members of disadvantaged or stigmatized groups can be led to conform to the derogatory stereotypes held by a dominant group, the latter can maintain their positions of privilege and power. However, most studies on behavioral confirmation have concerned interactions between individual perceivers and individual targets. Moreover, these interactions have generally not been defined as involving intergroup relations, as the expectations induced in the perceiver have generally concerned personality traits attributed to the target as an individual rather than to the target’s group membership per se. Hence, although it is quite possible that behavioral confirmation may take place during intergroup contact, the relevance of existing studies of interpersonal interactions to these settings needs to be established, both theoretically and empirically. Accordingly, our primary goal in this chapter has been to address this transition from the interpersonal to the intergroup levels of interaction by trying to answer the following question: How does stigmatization of one of the group members affect the confirmation of stereotypical expectations (a process that we labeled as ‘‘stereotype confirmation’’) during contact between members of these groups? To address this question, we began by articulating two processes believed to underlie stereotype confirmation in interpersonal settings: (1) the reciprocation of the perceiver to the target’s anticipated behavior, and the target’s own reciprocation of the perceiver’s overtures; and (2) the pursuit of a confirmatory strategy by the perceiver, coupled with deference on the target’s part. Then, we drew on the literature on interactions between nonstigmatized individuals (as perceivers) and stigmatized individuals (as targets) to suggest two behavioral styles likely to be adopted by the former— avoidance and dominance—in their dealings with the latter. Specifically, we reviewed relevant empirical data suggesting that avoidance coupled with reciprocation is particularly likely to lead to the confirmation of stereotypes regarding the sociability of the target, whereas the dominant style will facilitate the implementation of a confirmatory strategy and the consequent confirmation of expectations regarding the stigmatized target’s competence to the extent that the target adopts a complementary submissive style. Based on these distinctions, we next identified which aspects of the interactions between members of stigmatized and nonstigmatized groups may facilitate stereotype confirmation and disconfirmation, respectively. Among these aspects, we first considered those that are related to the perceiver. We hypothesized that in unstructured interactions, category salience, intergroup anxiety, and prejudice each should facilitate the use of the avoidant style by the perceiver, whereas in structured interactions

216

KLEIN AND SNYDER

(involving the performance of a joint task), category salience should facilitate the use of a dominant style by the perceiver. A review of the relevant literature on the interactions between members of various stigmatized and nonstigmatized groups proved to be generally consistent with this prediction. Second, we considered aspects related to the stigmatized target, and how the target’s response to the perceiver’s behavior could affect stereotype confirmation. We proposed that this response may depend on the perceived permeability of group boundaries. When these boundaries are perceived as permeable, we proposed that the awareness that the perceiver may stereotype oneself (‘‘stigma consciousness’’) is the crucial factor determining the occurrence of behavioral confirmation or disconfirmation. When not stigma conscious, targets are especially vulnerable to stereotype confirmation. When they are stigma conscious, we draw on social identity theory to suggest that they may try to dissociate themselves from their group in order to show that the stereotype does not apply to themselves. These actions, we have argued, may lead to perceptual disconfirmation. By contrast, when intergroup boundaries are perceived as impermeable, we suggested that targets may respond differently as a function of whether they perceive cognitive alternatives to the existing status differences. If there are none, targets may simply endorse the negative stereotypes and enact them. If there are such alternatives, targets may engage in a collective strategy of stereotype change. They are then likely to enact more positive stereotypes of their group and elicit behavioral disconfirmation. Third, we examined how the target’s membership in a stigmatized group can affect the very structure of the interaction with a nonstigmatized perceiver. Specifically, we reviewed empirical evidence showing that members of stigmatized groups tended to occupy positions of lesser interpersonal status and power when interacting with nonstigmatized group members and that this inferior position makes them particularly vulnerable to behavioral confirmation. Based on evidence drawn from expectation states theory, we showed that this power imbalance concerns both formal and informal aspects of status and power. Finally, after having considered the impact of stigmatization on stereotype confirmation during interpersonal and intergroup interactions, we addressed the question of how stereotype confirmation and disconfirmation during these contacts can actually affect stereotype persistence and change. Based on theories of intergroup contact, we proposed that such an influence of the interpersonal on the intergroup level will hold only to the extent that the perceiver construes the contact as intergroup rather than interpersonal. This, we suggested, will be facilitated to the extent that the target pursues a collective strategy.

STEREOTYPES AND BEHAVIORAL CONFIRMATION

217

This analysis has been consistent with the idea that stereotype confirmation processes are heavily influenced by factors defined at an intergroup level (such as permeability of group boundaries, group differences in power and status, as well as the perceived legitimacy and stability of these differences). Further, we have also suggested that these processes could play a role in the maintenance and change of intergroup relations. Nevertheless, in accord with the quotation that opens this section, we have argued that to understand this influence of the ‘‘micro’’ on the ‘‘macro’’ level, one must take into account the interactions between the sociostructural level of analysis as they affect the contexts in which dyadic interactions take place and the processes occurring at the level of these interactions themselves. This interactive standpoint has at least two major implications with respect to traditional social psychological research and theory on behavioral confirmation. First, it implies that the study of behavioral confirmation processes can, at least partially, inform an understanding of social processes occurring at superordinate levels of analysis and, hence, serve to address global social problems. Second, it implies that conceptualizations of behavioral confirmation processes should take into account the intergroup contexts in which the confirmation of stereotypes may take place. For such intergroup contexts can and do constrain the settings in which interactions between individuals take place as well as the expectations held by the interacting partners. Obviously, these two implications are complementary. It is only to the extent that it is informed by an understanding of the societal context in which interaction takes place that behavioral confirmation research can contribute to an understanding of these wider social problems. For example, an issue that may benefit from such an interactive analysis is soccer violence, a phenomenon that is seen as extremely prevalent among English supporters. Typically, the mass media portray supporters of the English soccer team as dangerous ‘‘hooligans’’ (see, for example, Murphy, Dunning, & Williams, 1991). In turn, these stereotypes may become the basis of the actions performed by non-English people when they interact with their targets out of their country. For example, Stott, Hutchinson, and Drury (2001) have shown that during the World Cup held in France in 1998, local youth, but also the police, tended to behave aggressively and indiscriminately toward these English fans, probably as a result of these stereotypes. This observation is consistent with a laboratory study showing that perceivers expecting a target to be hostile tended to behave more aggressively toward this target than perceivers interacting with a target who had not been described as such (Snyder & Swann, 1978). In this study, targets who had been randomly labeled as hostile did indeed behave more aggressively toward the perceivers who had so labeled them. Just like these targets, the supporters of the English team, who until their interaction with the local

218

KLEIN AND SNYDER

youth had been peaceful, felt unjustly treated and retaliated (Stott et al., 2001). Their behavior, which was publicized in the media, therefore served to reinforce the stereotype of the English supporter as a ‘‘hooligan.’’ The analysis of this social problem by Stott et al. (2001) illustrates the relevance of a dynamic analysis of the relations between the local level of analysis (here, specific supporters interacting with specific police officers and local youth) and the global level of analysis (the mass media, English supporters in general, the police, and so on). Methodologically, our analysis suggests that efforts can and should be made to increase the relevance of the typical behavioral confirmation scenario in which an individual perceiver interacts with an individual target, to larger intergroup phenomena; the same, of course, can be recommended more generally of studies of individual and dyadic level phenomena and processes. First, the analysis that we have presented here points to the necessity of using the laboratory to examine the impact of expectations on individuals who truly belong to the groups targeted by expectations, a step that has already been taken in field studies (Dougherty, Turban, & Callender, 1994; Harris Kern & Perkins, 1995; Harris, 1994; Jussim, 1989; Jussim, Eccles, & Madon, 1996). As we have pointed out, people who have a history of living with the burden of expectations may react very differently to a perceiver than individuals who are not even aware that they could be treated as a member of a group (of which they are generally not members). Second, explicit comparisons should be made between situations in which the target belongs and does not belong to the stigmatized group, a path that has almost never been taken (Miller & Myers, 1998). The target’s ‘‘real’’ traits, as revealed by personality assessment devices, could then be entered as a covariate in an analysis of the impact of different expectations on ratings of the personality of the target. Doing so would allow the researcher to disentangle the impact of present expectations and the role of more chronic dispositions in determining the target’s behavior. Third, explicit attempts to study the impact of interpersonal behavioral confirmation processes at the group level should be pursued. This could be accomplished, for example, by examining how a nonstigmatized perceiver communicates his or her perception of the target’s behavior, that is, whether perceptual confirmation can be communicated to third parties (Kashima, 2000; Ruscher, 1998, 2001; Thompson, Judd, & Park, 2000). More generally, it is apparent from our analysis that many aspects of group relations that shape interpersonal interactions are often implicit in studies using the behavioral confirmation paradigm. Although the setting may seem somewhat artificial and impoverished, perhaps even vacuous, when compared to naturally occurring social interactions, its interest may actually stem precisely from this ‘‘bareness’’—if such a minimally defined

STEREOTYPES AND BEHAVIORAL CONFIRMATION

219

situation can contribute to reproducing stereotypical expectations, and make them become true, the power of expectations must indeed be noteworthy. In this regard, it may be worth drawing an analogy with the minimal group paradigm (Tajfel, Billig, Bundy, & Flament, 1971); just as this paradigm illustrates the minimal conditions for intergroup bias to occur, so too may the behavioral confirmation paradigm highlight the minimal conditions for the realization of expectations. In closing, we note that the great sociologist Georg Simmel (1971) compared the structure of society to that of the natural world, in which ‘‘not a single grain of sand could have a shape different from what it has, or be in a position different from its actual position, without first conditioning the alteration by a change of the whole, and without entailing such a change in the whole’’ (p. 19). Individuals taking part in dyadic interactions are, in terms of Simmel’s analogy, adjacent grains of sand within the deserts and beaches constituted by groups, institutions, and societies. In this chapter we have attempted to offer a novel perspective on the interplay between interpersonal and intergroup processes involved in stereotype confirmation. We have done so by integrating behavioral confirmation processes with separate strands of relevant theory and research (expectation states theory, the literature on the influence of stigma on social interaction, the social identity perspective). In this view, stereotype confirmation in interpersonal settings is considered both as an outcome and as a determinant of specific patterns of intergroup relations. In developing this perspective, we hope that we have convincingly shown that the study of stereotype confirmation can contribute meaningfully to understanding larger societal phenomena. If there is a grain of truth in Simmel’s analogy, this may not be wishful thinking.

Acknowledgments The authors gratefully acknowledge the constructive commentary and helpful suggestions of Marc Kiviniemi, Stefan Stu¨ rmer, Hilary Ammazzalorso, and Mark Zanna on earlier versions of this manuscript. The writing of this chapter was supported in part by a grant of the Belgian American Educational Foundation to Olivier Klein. Mark Snyder’s research on behavioral confirmation in social interaction has been supported by the National Science Foundation.

References Adams, K. A. (1980). Who has the final word? Sex, race, and dominance behavior. Journal of Personality and Social Psychology, 38, 1–8. Allport, G. W. (1954). The nature of prejudice. Camridge, Mass: Addison-Wesley.

220

KLEIN AND SNYDER

Argyle, M., & Dean, J. (1965). Eye-contact, distance, and affiliation. Sociometry, 28, 289–304. Balkwell, J. W., & Berger, J. (1996). Gender, status, and behavior in task situations. Social Psychology Quarterly, 59, 273–283. Bargh, J. A., Chen, M., & Burrows, L. (1996). Automaticity of social behavior: Direct effects of trait construct and stereotype priming on action. Journal of Personality and Social Psychology, 71, 230–244. Baron, R. S., Inman, M. L., Kao, C. F., & Logan, H. (1992). Negative emotion and superficial social processing. Motivation and Emotion, 16, 323–346. Becker, H. S. (1963). Outsiders. New York: Macmillan. Berger, J. (1992). Expectations, theory, and group processes. Social Psychology Quarterly, 55, 3–11. Berger, J., Cohen, B. P., & Zelditch, M. (1972). Status characteristics and social interaction. American Sociological Review, 37, 241–255. Berger, J., Rosenholtz, S. J., & Zelditch, M. (1980). Status organizing processes. Annual Review of Sociology, 6, 479–508. Biesanz, J. C., Neuberg, S. L., Smith, D. M., Asher, T., & Judice, T. N. (2001). When accuracymotivated perceivers fail: Limited attentional resources and the reemerging self-fulfilling prophecy. Personality and Social Psychology Bulletin, 27, 621–629. Blake, R. R., & Mouton, J. S. (1964). The managerial grid: Key orientations for achieving production through people. Houston, TX: Gulf. Blascovich, J., Mendes, W. B., Hunter, S. B., & Lickel, B. (2000). Stigma, threat, and social interactions. In T. F. Heatherton & R. E. Kleck (Eds.), The social psychology of stigma (pp. 307–333). New York: Guilford Press. Blascovich, J., Mendes, W. B., Hunter, S. B., Lickel, B., & Kowai-Bell, N. (2001). Perceiver threat in social interactions with stigmatized others. Journal of Personality and Social Psychology, 80, 253–267. Blumer, H. (1962). Society as symbolic interaction. In A. Rose (Ed.), Human behavior and social processes (pp. 92–179). Boston: Houghton Mifflin. Blumer, H. (1969). Symbolic interactionism: Perspective and method. Berkeley, CA: University of California Press. Bond, C. F., Omar, A., Pitre, U., & Lashley, B. R. (1992). Fishy-looking liars: Deception judgment from expectancy violation. Journal of Personality and Social Psychology, 63, 969–977. Bond, M. H. (1972). Effect of an impression set on subsequent behavior. Journal of Personality and Social Psychology, 24, 303–312. Bourhis, R. Y., Giles, H., Leyens, J. P., & Tajfel, H. (1979). Psycholinguistic distinctiveness: Language divergence in Belgium. In H. Giles & R. N. S. Clair (Eds.), Language and social psychology (pp. 158–185). Baltimore, MD: University Park Press. Brown, R. (1988). Group processes: Dynamics within and between groups. Oxford, UK: Blackwell. Brown, R., Vivian, J., & Hewstone, M. (1999). Changing attitudes through intergroup contact: The effects of group membership salience. European Journal of Social Psychology, 29, 741–764. Brown, T. R. (2001). Exposure to all-black contexts and psychological consequences on wellbeing: The benefits of racial concentration. African-American Research Perspectives, 7, 157–172. Burgoon, J. K. (1978). A communication model of personal space violations: Explication and an initial test. Human Communication Research, 4, 129–142. Burgoon, J. K., Le Poire, B. A., & Rosenthal, R. (1995). Effects of preinteraction expectancies and target communication on perceiver reciprocity and compensation in dyadic interaction. Journal of Experimental Social Psychology, 31, 287–321.

STEREOTYPES AND BEHAVIORAL CONFIRMATION

221

Carli, L. L. (1990). Gender, language, and influence. Journal of Personality and Social Psychology, 59, 941–951. Carli, L. L. (1991). Gender, status, and influence. In E. J. Lawler, B. Markovsky, C. L. Ridgeway, & H. Walker (Eds.), Advances in Group Processes (Vol. 8, pp. 89–113). Greenwich, CT: JAI. Charles, N. (1993). Gender divisions and social change. Hemel Hempstead, UK: Harvester Wheatsheaf. Chen, M., & Bargh, J. A. (1997). Nonconscious behavioral confirmation processes: The selffulfilling consequences of automatic stereotype activation. Journal of Experimental Social Psychology, 33, 541–560. Chidester, T. R. (1986). Problems in the study of interracial interaction: Pseudo-interracial dyad paradigm. Journal of Personality and Social Psychology, 50, 74–79. Christensen, D., & Rosenthal, R. (1982). Gender and nonverbal decoding skill as determinants of interpersonal expectancy effects. Journal of Personality and Social Psychology, 42, 75–87. Claire, T., & Fiske, S. T. (1998). A systemic view of behavioral confirmation: Counterpoint to the individualist view. In C. Sedikides & J. Schopler (Eds.), Intergroup cognition and intergroup behavior (pp. 205–231). Mahawah, NJ: Erlbaum. Cohen, D. (2001). Cultural variation: Considerations and implications. Psychological Bulletin, 127, 451–471. Cohen, E., & Roper, S. (1972). Modification of interracial interaction disability. American Sociological Review, 37, 643–647. Cohen, E. G. (1982). Expectation states and interracial interaction in school settings. Annual Review of Sociology, 8, 209–235. Comer, R. J., & Piliavin, J. A. (1972). The effects of physical deviance upon face-to-face interaction: The other side. Journal of Personality and Social Psychology, 23, 33–39. Copeland, J. T. (1992). Motivational implications of social power for behavioral confirmation. Doctoral dissertation, University of Minnesota, Minneapolis, MN. Copeland, J. T. (1994). Prophecies of power: Motivational implications of social power for behavioral confirmation. Journal of Personality and Social Psychology, 67, 264–277. Couch, C. J., & Hintz, R. A. (1975). Constructing social life: Readings in behavioral sociology from the Iowa School. Greenwich, CT: Stripes. Crandall, C. S. (2000). Ideology and lay theories of stigma: The justification of stigmatization. In T. F. Heatherton & R. E. Kleck (Eds.), The social psychology of stigma (pp. 126–150). New York: Guilford Press. Crocker, J., & Major, B. (1989). Social stigma and self-esteem: The self-protective properties of stigma. Psychological Review, 96, 608–630. Crocker, J., Major, B., & Steele, C. (1998). Social stigma. In D. T. Gilbert & S. T. Fiske (Eds.), The handbook of social psychology (Vol. 2, pp. 504–553). Boston: McGraw-Hill. Crocker, J., Voelkl, K., Testa, M., & Major, B. (1991). Social stigma: The affective consequences of attributional ambiguity. Journal of Personality and Social Psychology, 60, 218–228. Crosby, F., Bromley, S., & Saxe, L. (1980). Recent unobtrusive studies of black and white discrimination and prejudice: A literature review. Psychological Bulletin, 87, 546–563. Curtis, R. C., & Miller, C. T. (1986). Believing another likes or dislikes you: Behaviors making the belief come true. Journal of Personality and Social Psychology, 51, 284–290. Daly, S. (1978). Behavioural correlates of social anxiety. British Journal of Social and Clinical Psychology, 17, 117–120. Deaux, K., & Major, B. (1987). Putting gender into context: An interactive model of genderrelated behavior. Psychological Review, 94, 369–389.

222

KLEIN AND SNYDER

De Bruin, E. N., & Van Lange, P. A. (2000). What people look for in others: Influences of the perceiver and the perceived on information selection. Personality and Social Psychology Bulletin, 26, 206–219. Deschamps, J.-C. (1973–1974). L’attribution, la cate´ gorisation sociale et les relations intergroupes [Attribution, social categorization, and intergroup relations]. Bulletin de Psychologie, 27, 710–721. Devine, P. G., & Elliot, A. J. (1995). Are racial stereotypes really fading? The Princeton trilogy revisited. Personality and Social Psychology Bulletin, 21, 1139–1150. Devine, P. G., Evett, S. R., & Vasquez-Suson, K. A. (1996). Exploring the interpersonal dynamics of intergroup contact. In R. M. Sorrentino & E. T. Higgins (Eds.), Handbook of motivation and cognition, Vol. 3: The interpersonal context. Handbook of motivation and cognition (pp. 423–464). New York: Guilford Press. Devine, P. G., & Monteith, M. J. (1993). The role of discrepancy-associated affect in prejudice reduction. In D. M. Mackie & D. L. Hamilton (Eds.), Affect, cognition, and stereotyping: Interactive processes in group perception (pp. 317–344). San Diego, CA: Academic Press. Diab, L. N. (1962). National stereotypes and the ‘‘reference group’’ concept. Journal of Social Psychology, 57, 339–351. Doosje, B., Ellemers, N., & Spears, R. (1999). Commitment and intergroup behavior. In N. Ellemers & R. Spears (Eds.), Social identity: Context, commitment, content (pp. 84–106). Oxford, UK: Blackwell. Dougherty, T. W., Turban, D. B., & Callender, J. C. (1994). Confirming first impressions in the employment interview: A field study of interviewer behavior. Journal of Applied Psychology, 79, 659–665. Dovidio, J. F. (1988). Power displays between women and men in discussions of gender-linked tasks: A multichannel study. Journal of Personality and Social Psychology, 55, 580–587. Dovidio, J. F., Kawakami, K., & Gaertner, S. L. (2002). Implicit and explicit prejudice and interracial interaction. Journal of Personality and Social Psychology, 82, 62–68. Duncan, B. L. (1976). Differential social perception and attribution of intergroup violence: Testing the lower limits of stereotyping of Blacks. Journal of Personality and Social Psychology, 34, 590–598. Dvir, T., Eden, D., & Banjo, M. L. (1995). Self-fulfilling prophecy and gender: Can women be Pygmalion and Galatea? Journal of Applied Psychology, 80, 253–270. Eagly, A. H. (1987). Sex differences in social behavior: A social-role Interpretation. Hillsdale, NJ: Erlbaum. Eagly, A. H., & Mladinic, A. (1989). Gender stereotypes and attitudes toward women and men. Personality and Social Psychology Bulletin, 15, 543–558. Eagly, A. H., & Steffen, V. J. (1984). Gender stereotypes stem from the distribution of women and men into social roles. Journal of Personality and Social Psychology, 46, 735–754. Eagly, A. H., & Wood, W. (1991). Explaining sex differences in social behavior: A metaanalysis. Psychological Bulletin, 17, 306–315. Ellemers, N. (1993). The influence of socio-structural variables on identity management strategies. In W. Stroebe & M. Hewstone (Eds.), European review of social psychology (Vol. 4, pp. 27–57). Chichester, UK: Wiley. Ellemers, N. (2001). Individual upward mobility and the perceived legitimacy of intergroup relations. In J. T. Jost & B. Major (Eds.), The psychology of legitimacy: Emerging perspectives on ideology, justice, and intergroup relations (pp. 205–222). New York: Cambridge University Press. Ellemers, N., Barreto, M., & Spears, R. (1999). Commitment and strategic responses to social context. In N. Ellemers & R. Spears (Eds.), Social identity: Context, commitment, content (pp. 127–146). Oxford, UK: Blackwell.

STEREOTYPES AND BEHAVIORAL CONFIRMATION

223

Ellemers, N., Spears, R., & Doosje, B. (1997). Sticking together or falling apart: In-group identification as a psychological determinant of group commitment versus individual mobility. Journal of Personality and Social Psychology, 72, 617–626. Ellemers, N., Van Rijswijk, W., Bruins, J., & De Gilder, D. (1998). Group commitment as a moderator of attributional and behavioural responses to power use. European Journal of Social Psychology, 28, 555–573. Ellyson, S. L., Dovidio, J. F., & Brown, C. E. (1992). The look of power: Gender differences and similarities. In C. Ridgeway (Ed.), Gender, interaction and inequality (pp. 50–80). New York: Springer. Erber, R., & Fiske, S. T. (1984). Outcome dependency and attention to inconsistent information. Journal of Personality and Social Psychology, 47, 709–726. Esses, V. M., Haddock, G., & Zanna, M. P. (1993). Values, stereotypes, and emotions as determinants of intergroup attitudes. In D. M. Mackie & D. L. Hamilton (Eds.), Affect, cognition, and stereotyping: Interactive processes in group perception (pp. 137–166). San Diego, CA: Academic Press. Farina, A., Allen, J. G., & Saul, B. B. (1968). The role of the stigmatized person in affecting social relationships. Journal of Personality, 36, 169–182. Farina, A., Gliha, D., Boudreau, L. A., Allen, J. G., & Sherman, M. (1971). Mental illness and the impact of believing others know about it. Journal of Abnormal Psychology, 77, 1–5. Farina, A., Sherman, M., & Allen, J. G. (1968). Role of physical abnormalities in interpersonal perception and behavior. Journal of Abnormal Psychology, 73, 590–593. Fiske, S. T. (1998). Stereotyping, prejudice, and discrimination. In D. T. Gilbert & S. T. Fiske (Eds.). The handbook of social psychology (4th ed., Vol. 2, pp. 357–411). Boston, MA: McGraw-Hill. Fiske, S. T. (2001). Power explains and maintains individual, group, and societal disparities. In A. Y. Lee-Chai & J. A. Bargh (Eds.), The use and abuse of power: Multiple perspectives on the causes of corruption (pp. 181–193). Philadelphia, PA: Psychology Press. Fiske, S. T., Cuddy, A. C., Glick, P., & Xu, J. (2002). A model of (often mixed) stereotype content: Competence and warmth respectively follow from perceived status and competition. Journal of Personality and Social Psychology, 82, 878–902. Fiske, S. T., & Neuberg, S. L. (1990). A continuum of impression formation from categorybased to individuating processes: Influences of information and motivation on attention and interpretation. In M. P. Zanna (Ed.), Advances in experimental social psychology (Vol. 23, pp. 1–74). New York: Academic Press. Fiske, S. T., Xu, J., & Cuddy, A. C. (1999). (Dis)respecting versus (dis)liking: Status and interdependence predict ambivalent stereotypes of competence and warmth. Journal of Social Issues, 55, 473–489. French, J. R., & Raven, B. (1959). The bases of social power. In D. Cartwright (Ed.), Studies in social power (pp. 150–167). Ann Arbor, MI: University of Michigan Press. Gaertner, S. L., & Dovidio, J. F. (1986). The aversive form of racism. In J. F. Dovidio (Ed.), Prejudice, discrimination, and racism (pp. 61–89). San Diego, CA: Academic Press. Gaertner, S. L., Dovidio, J. F., Rust, M. C., Nier, J. A., Banker, B. S., Ward, C. M., Mottola, G. R., & Houlette, M. (1999). Reducing intergroup bias: Elements of intergroup cooperation. Journal of Personality and Social Psychology, 76, 388–402. Gaertner, S. L., Mann, J. A., Dovidio, J. F., Murrell, A. J., et al. (1990). How does cooperation reduce intergroup bias? Journal of Personality and Social Psychology, 59, 692–704. Garcia-Marques, L., & Mackie, D. M. (2001). Not all stereotype-incongruent information is created equal: The impact of sample variability on stereotype change. Group Processes and Intergroup Relations, 4, 5–19.

224

KLEIN AND SNYDER

Giles, H., & Coupland, N. (1991). Language: Contexts and consequences. Pacific Grove, CA: Brooks/Cole. Giles, H., Coupland, N., & Coupland, J. (1991). Accommodation theory: Communication, context, and consequence. In H. Giles & J. Coupland (Eds.), Contexts of accommodation: Developments in applied sociolinguistics. Studies in emotion and social interaction (pp. 1–68). New York, Paris: Cambridge University Press, Editions De La Maison Des Sciences De L’Homme. Glick, P., & Fiske, S. T. (2001). Ambivalent stereotypes as legitimizing ideologies: Differentiating paternalistic and envious prejudice. In J. T. Jost & B. Major (Eds.), The psychology of legitimacy (pp. 278–306). New York: Cambridge University Press. Goffman, E. (1963). Stigma: Notes on the management of identity. Englewood Cliffs, NJ: Prentice-Hall. Goldman, W., & Lewis, P. (1977). Beautiful is good: Evidence that the physically attractive are more socially skilled. Journal of Experimental Social Psychology, 13, 125–130. Gordijn, E. H., Koomen, W., & Stapel, D. A. (2001). Level of prejudice in relation to knowledge of cultural stereotypes. Journal of Experimental Social Psychology, 37, 150–157. Greenland, K., & Brown, R. (1999). Categorization and intergroup anxiety in contact between British and Japanese nationals. European Journal of Social Psychology, 29, 503–521. Greenland, K., & Brown, R. (2000). Categorization and intergroup anxiety in intergroup contact. In D. Capozza & R. Brown (Eds.), Social identity processes: Trends in theory and research (pp. 167–183). London: Sage. Gruber, J. E., & Smith, M. (1995). Women’s responses to sexual harassment: A multivariate analysis. Basic and Applied Social Psychology, 17, 543–562. Hall, J. A., & Friedman, G. B. (1999). Status, gender, and nonverbal behavior: A study of structured interactions between employees of a company. Personality and Social Psychology Bulletin, 25, 1082–1091. Harris, M. J. (1994). Self-fulfilling prophecies in the clinical context: Review and implications for clinical practice. Applied and Preventive Psychology, 3, 145–158. Harris, M. J., Lightner, R. M., & Manolis, C. (1998). Awareness of power as a moderator of expectancy confirmation: Who’s the boss around here? Basic and Applied Social Psychology, 20, 220–229. Harris Kern, M. J., & Perkins, R. (1995). Effects of distraction on interpersonal expectancy effects: A social interaction test of the cognitive busyness hypothesis. Social Cognition, 13, 163–182. Hashtroudi, S., Mutter, S. A., Cole, E. A., & Green, S. K. (1984). Schema-consistent and schema-inconsistent information: Processing demands. Personality and Social Psychology Bulletin, 10, 269–278. Haslam, S. A. (1997). Stereotyping and social influence: Foundations of stereotype consensus. In R. Spears & P. J. Oakes (Eds.), The social psychology of stereotyping and group life (pp. 119–143). Oxford, UK: Blackwell. Haslam, S. A., Oakes, P. J., Reynolds, K. J., & Turner, J. C. (1999). Social identity salience and the emergence of stereotype consensus. Personality and Social Psychology Bulletin, 25, 809–818. Haslam, S. A., Oakes, P. J., & Turner, J. C. (1996). Social identity, self-categorization, and the perceived homogeneity of ingroups and outgroups: The interaction between social motivation and cognition. In R. M. Sorrentino & E. T. Higgins (Eds.), Handbook of motivation and cognition, Vol. 3: The interpersonal context (pp. 182–222). New York: Guilford Press. Haslam, S. A., & Turner, J. C. (1992). Context-dependent variation in social stereotyping 2: The relationship between frame of reference, self-categorization and accentuation. European Journal of Social Psychology, 22, 251–277.

STEREOTYPES AND BEHAVIORAL CONFIRMATION

225

Haslam, S. A., & Turner, J. C. (1995). Context-dependent variation in social stereotyping: III. Extremism as a self-categorical basis for polarized judgement. European Journal of Social Psychology, 25, 341–371. Haslam, S. A., Turner, J. C., Oakes, P. J., & McGarty, C. (1992). Context-dependent variation in social stereotyping: I. The effects of intergroup relations as mediated by social change and frame of reference. European Journal of Social Psychology, 22, 3–20. Haslam, S. A., Turner, J. C., Oakes, P. J., McGarty, C., & Reynolds, K. J. (1998). The group as a basis for emergent stereotype consensus. In W. Stroebe & M. Hewstone (Eds.), European Review of Social Psychology (Vol. 8, pp. 203–239). New York: John Wiley & Sons. Hastie, R., & Kumar, P. A. (1979). Person memory: Personality traits as organizing principles in memory for behaviors. Journal of Personality and Social Psychology, 37, 25–38. Hastorf, A. H., Wildfogel, J., & Cassman, T. (1979). Acknowledgment of handicap as a tactic in social interaction. Journal of Personality and Social Psychology, 37, 1790–1797. Hebl, M. R., & Kleck, R. E. (2000). The social consequences of physical disability. In T. F. Heatherton & R. E. Kleck (Eds.), The social psychology of stigma (pp. 419–439). New York: Guilford Press. Heilman, M. E. (1980). The impact of situational factors on personnel decisions concerning women: Varying the sex composition of the applicant pool. Organizational Behavior and Human Decision Processes, 26, 386–395. Hendricks, M., & Bootzin, R. (1976). Race and sex as stimuli as stimuli for negative affect and physical avoidance. Journal of Social Psychology, 98, 111–120. Henley, N. (1977). Body politics: Power, sex, and nonverbal communication. Englewood Cliffs, NJ: Prentice-Hall. Henley, N. (1995). Body politics revisited: What do we know today? In P. J. Kalbfeisch & M. J. Cody (Eds.), Gender, power and communication in human relationships. Hillsdale, NJ: Erlbaum. Hewstone, M. (1996). Contact and categorization: Social psychological interventions to change intergroup relations. In C. N. Macrae, C. Stangor, & M. Hewstone (Eds.), Foundations of stereotypes and stereotyping (pp. 323–368). New York: Guilford Press. Hewstone, M., & Brown, R. (1986). Contact and conflict in intergroup encounters. Oxford, UK: Blackwell. Hilton, J. L (1998). Interaction goals and person perception. In J. M. Darley & J. Cooper (Eds.), Attribution and social interaction: The legacy of Edward E. Jones (pp. 127–197). Washington, DC: American Psychological Association. Hilton, J. L., & Darley, J. M. (1985). Constructing other persons: A limit on the effect. Journal of Experimental Social Psychology, 21, 1–18. Hilton, J. L., & Darley, J. M. (1991). The effect of interaction goals in person perception. In M. P. Zanna (Ed.), Advances in experimental social psychology (Vol. 24, pp. 236–262). Orlando, FL: Academic Press. Hoffman, C., & Hurst, N. (1990). Gender stereotypes: Perception or rationalization? Journal of Personality and Social Psychology, 58, 197–208. Hogg, M. A. (1987). Social identity and group cohesiveness. In J. C. Turner, M. A. Hogg, P. J. Oakes, S. D. Reicher, & M. S. Wetherell (Eds.), Rediscovering the social group (pp. 89–116). Oxford, UK: Blackwell. Hogg, M. A. (1992). The social psychology of group cohesiveness: From attraction to social identity. New York: New York University Press. Hogg, M. A., & Abrams, D. (1988). Social identifications: A social psychology of intergroup relations and group processes. London: Routledge.

226

KLEIN AND SNYDER

Hogg, M. A., & Hardie, E. A. (1991). Social attraction, personal attraction and self-categorization: A field study. Personality and Social Psychology Bulletin, 17, 175–180. Hogg, M. A., & Turner, J. C. (1987). Intergroup behaviour, self-stereotyping and the salience of social categories. British Journal of Social Psychology, 26, 325–340. Ibarra, H. (1993). Personal networks of women and minorities in management: A conceptual framework. The Academy of Management Review, 18, 56–87. Ibarra, H. (1995). Race, opportunity, and diversity of social circles in managerial networks. The Academy of Management Journal, 38, 673–703. Ickes, W. (1984). Compositions in black and white: Determinants of interaction in interracial dyads. Journal of Personality and Social Psychology, 47, 330–341. Ickes, W., Patterson, M. L., Rajecki, D. W., & Tanford, S. (1982). Behavioral and cognitive consequences of reciprocal versus compensatory responses to preinteraction expectancies. Social Cognition, 1, 160–190. Islam, M. R., & Hewstone, M. (1993). Dimensions of contact as predictors of intergroup anxiety, perceived out-group variability, and out-group attitude: An integrative model. Personality and Social Psychology Bulletin, 19, 700–710. Johanson, J. C. (1999). Effect of target’s sex on manipulations of self-fulfilling prophecy. Psychological Reports, 84, 413–423. Jones, S. C., & Panitch, D. (1971). The self-fulfilling prophecy and interpersonal attraction. Journal of Personality and Social Psychology, 7, 356–366. Jost, J. T., & Banaji, M. R. (1994). The role of stereotyping in system-justification and the production of false consciousness. British Journal of Social Psychology, 33, 1–27. Jost, J. T., Burgess, D., & Mosso, C. O. (2001). Conflicts of legitimation among self, group, and system: The integrative potential of system justification theory. In J. T. Jost & B. Major (Eds.), The psychology of legitimacy: Emerging perspectives on ideology, justice, and intergroup relations (pp. 363–388). New York: Cambridge University Press. Judd, C. M., Park, B., Ryan, C. S., Brauer, M., & Kraus, S. (1995). Stereotypes and ethnocentrism: Diverging interethnic perceptions of African American and white American youth. Journal of Personality and Social Psychology, 69, 460–481. Jussim, L. (1989). Teacher expectations: Self-fulfilling prophecies, perceptual biases, and accuracy. Journal of Personality and Social Psychology, 57, 469–480. Jussim, L., Eccles, J., & Madon, S. (1996). Social perceptions, social stereotypes, and teacher expectations: Accuracy and the quest for the powerful self-fulfilling prophecy. In M. P. Zanna (Ed.), Advances in Experimental Social Psychology (Vol. 28, pp. 281–388). San Diego, CA: Academic Press. Jussim, L., & Fleming, C. (1996). Self-fulfilling prophecies and the maintenance of social stereotypes: The role of dyadic interactions and social forces. In C. N. Macrae, C. Stangor, & M. Hewstone (Eds.), Foundations of stereotypes and stereotyping (pp. 161–193). New York: Guilford Press. Jussim, L., Palumbo, P., Chatman, C., Madon, S., & Smith, A. (2000). Stigma and self-fulfilling prophecies. In T. F. Heatherton & R. E. Kleck (Eds.), The social psychology of stigma (pp. 374–418). New York: Guilford Press. Karlins, M., Coffman, T. L., & Walters, G. (1969). On the fading of social stereotypes: Studies in three generations of college students. Journal of Personality and Social Psychology, 13, 1–16. Kashima, Y. (2000). Maintaining cultural stereotypes in the serial reproduction of narratives. Personality and Social Psychology Bulletin, 26, 594–604.

STEREOTYPES AND BEHAVIORAL CONFIRMATION

227

Kashima, Y., Woolcock, J., & Kashima, E. S. (2000). Group impressions as dynamic configurations: The tensor product model of group impression formation and change. Psychological Review, 107, 914–942. Katz, I. (1970). Experimental studies in negro-white relationships. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 5, pp. 71–117). New York: Academic Press. Katz, I., & Benjamin, L. (1960). Effects of white authoritarianism in biracial work groups. Journal of Abnormal and Social Psychology, 61, 448–456. Katz, I., & Cohen, M. (1962). The effects of training negroes upon cooperative problem solving in biracial teams. Journal of Abnormal and Social Psychology, 64, 319–325. Kelly, C., & Breinlinger, S. (1996). The social psychology of collective action: Identity, injustice, and gender. London: Taylor & Francis. Kelsey, B. L. (1998). The dynamics of multicultural groups: Ethnicity as a determinant of leadership. Small Group Research, 29, 602–623. Kleck, R. (1968). Effects of stigmatizing conditions on the use of personal space. Psychological Reports, 23, 111–118. Kleck, R. (1969). Physical stigma and task oriented interactions. Human Relations, 22, 53–60. Kleck, R., Ono, H., & Hastorf, A. H. (1966). The effect of physical deviance upon face-to-face interaction. Human Relations, 19, 425–437. Kleck, R. E., & Strenta, A. (1980). Perceptions of the impact of negatively valued physical characteristics on social interaction. Journal of Personality and Social Psychology, 39, 861–873. Klein, O., & Azzi, A. (2001). How group members attempt to tailor an out-group’s representation of themselves. British Journal of Social Psychology, 40, 279–294. Klein, O., Licata, L., Azzi, A., & Durala, I. (In press.) ‘‘How European am I?’’: Prejudice expression and the strategic presentation of social identity. Self and Identity. Klein, O., & Snyder, M. (2000). Perceivers’ expectations regarding the introversion of a target and group identification. Unpublished data: University of Minnesota. Knowles, E. S. (1980). An affiliative conflict theory of personal and group spatial behavior. In P. B. Paulus (Ed.), Psychology of group influence (pp. 133–188). Hillsdale, NJ: Erlbaum. Konrad, A. M., Winter, S., & Gutek, B. A. (1992). Diversity in work group sex composition: Implications for majority and minority workers. In S. Bacharach & P. Tolbert (Eds.), Research in the Sociology of Organizations (Vol. 10, pp. 115–140). Greenwich, CT: JAI. Kraiger, K., & Ford, J. K. (1985). A meta-analysis of ratee race in performance ratings. Journal of Applied Psychology, 70, 56–65. Kray, L. J., Thompson, L., & Galinsky, A. (2001). Battle of the sexes: Gender stereotype confirmation and reactance in negotiations. Journal of Personality and Social Psychology, 80, 942–958. Krueger, J. (1996). Personal beliefs and cultural stereotypes about racial characteristics. Journal of Personality and Social Psychology, 71, 536–548. LaFrance, M., & Henley, N. (1994). On oppressing hypotheses: Or differences in nonverbal sensitivity revisited. In H. L. Radtke & H. J. Stam (Eds.), Power/gender: Social relations in theory and practice (pp. 287–311). London: Sage. Langer, E. J., Fiske, S., Taylor, S. E., & Chanowitz, B. (1976). Stigma, staring, and discomfort: A novel-stimulus hypothesis. Journal of Experimental Social Psychology, 12, 451–463. Leyens, J.-P., Yzerbyt, V., & Schadron, G. (1994). Stereotypes and social cognition. Thousand Oaks, CA: Sage. Lott, B. (1987). Sexist discrimination as distancing behavior: I. A laboratory demonstration. Psychology of Women Quarterly, 11, 47–58. Maccoby, E., & Jacklin, C. (1974). The psychology of sex differences. Stanford, CA: Stanford University Press.

228

KLEIN AND SNYDER

Mackie, D. M., Devos, T., & Smith, E. R. (2000). Intergroup emotions: Explaining offensive action tendencies in an intergroup context. Journal of Personality and Social Psychology, 79, 602–616. Macrae, C. N., Milne, A. B., & Bodenhausen, G. V. (1994). Stereotypes as energy-saving devices: A peek inside the cognitive toolbox. Journal of Personality and Social Psychology, 66, 37–47. Madon, S., Jussim, L., & Eccles, J. (1997). In search of the powerful self-fulfilling prophecy. Journal of Personality and Social Psychology, 72, 791–809. Madon, S., Guyll, M., Aboufadel, K., Montiel, E., Smith, A., Palumbo, P., & Jussim, L. (2001). Ethnic and national stereotypes: The Princeton trilogy revisited. Personality and Social Psychology Bulletin, 27, 996–1010. Maines, D. (1982). Myth, text and interactionist complicity in the neglect of Blumer’s macrosociology. Symbolic Interaction, 11, 43–57. Maines, D., & Charlton, J. (1985). The negotiated order approach to the analysis of social organization. In H. Farbermann & R. S. Perinbanayagam (Eds.), Foundations of interpretive sociology, studies in symbolic interaction (pp. 271–308). London: JAI press. Marlowe, A. (2002). Check please. Retrieved from the World Wide Web: http://www.salon. com/mwt/feature/2002/05/04/who_pays/?x. McPherson, M., Smith-Lovin, L., & Cook, J. M. (2001). Birds of a feather: Homophily in social networks. Annual Review of Sociology, 27, 415–444. Merton, R. K. (1948). The self-fulfilling prophecy. Antioch Review, 8, 193–210. Miller, C. T., & Myers, A. M. (1998). Compensating for prejudice: How heavyweight people (and others) control outcomes despite prejudice. In J. K. Swim & C. Stangor (Eds.), Prejudice: The target’s perspective (pp. 191–218). San Diego, CA: Academic Press. Miller, C. T., Rothblum, E. D., Barbour, L., Brand, P. A., & Felicio, D. M. (1990). Social interactions of obese and nonobese women. Journal of Personality, 58, 365–380. Miller, C. T., Rothblum, E. D., Felicio, D., & Brand, P. (1995). Compensating for stigma: Obese and nonobese women’s reactions to being visible. Personality and Social Psychology Bulletin, 21, 1093–1106. Miller, D. T., & Ross, M. (1975). Self-serving biases in the attribution of causality: Fact or fiction? Psychological Bulletin, 82, 213–225. Miller, D. T., & Turnbull, W. (1986). Expectancies and interpersonal processes. Annual Review of Psychology, 37, 233–256. Moghaddam, F. M., Taylor, D. M., & Wright, S. C. (1993). Social psychology in cross-cultural perspective. New York: Freeman. Monteith, M. J., Devine, P. G., & Zuwerink, J. R. (1993). Self-directed versus other-directed affect as a consequence of prejudice-related discrepancies. Journal of Personality and Social Psychology, 64, 198–210. Moya, M. (1998). Social identity and interpersonal relationships. In S. Worchel, J. F. Morales, D. Paez, & J.-C. Deschamps (Eds.), Social identity: International perspectives (pp. 154–166). London: Sage. Murphy, P., Dunning, E. G., & Williams, J. (1988). Soccer crowd disorder and the press: Processes of amplification and deamplification in historical perspective. Theory Culture and Society, 5, 645–673. Neuberg, S. L. (1994). Expectancy-confirmation processes in stereotype-tinged social encounters: The moderating role of social goals. In M. P. Zanna & J. M. Olson (Eds.), The psychology of prejudice (pp. 103–130). Hillsdale, NJ: Erlbaum. Neuberg, S. L. (1996). Social motives and expectancy-tinged social interactions. In R. M. Sorrentino & E. T. Higgins (Eds.), Handbook of motivation and cognition, Vol. 3: The interpersonal context (pp. 225–261). New York: Guilford Press.

STEREOTYPES AND BEHAVIORAL CONFIRMATION

229

Noel, J. G., Wann, D. L., & Branscombe, N. R. (1995). Peripheral ingroup membership status and public negativity toward outgroups. Journal of Personality and Social Psychology, 68, 127–137. Oakes, P. J. (1987). The salience of social categories. In J. C. Turner, M. A. Hogg, P. J. Oakes, S. Reicher, & M. S. Wetherell (Eds.), Rediscovering the social group: A self-categorization theory (pp. 117–141). Oxford, UK: Blackwell. Olson, J. M., Roese, N. J., & Zanna, M. P. (1996). Expectancies. In E. T. Higgins & A. W. Kruglanski (Eds.), Social psychology: Handbook of basic principles (pp. 211–238). New York: Guilford Press. Pancer, S. M., McMullen, L. M., Kabatoff, R. A., Johnson, K. G., & Pond, C. A. (1979). Conflict and avoidance in the helping situation. Journal of Personality and Social Psychology, 37, 1406–1411. Park, B., & Rothbart, M. (1982). Perception of out-group homogeneity and levels of social categorization: Memory for the subordinate attributes of in-group and out-group members. Journal of Personality and Social Psychology, 42, 1051–1068. Pettigrew, T. F. (1998). Intergroup contact theory. Annual Review of Psychology, 49, 65–85. Pinel, E. C. (1999). Stigma consciousness: The psychological legacy of social stereotypes. Journal of Personality and Social Psychology, 76, 114–128. Pinel, E. C. (2002). Stigma consciousness in intergroup contexts: The power of conviction. Journal of Experimental Social Psychology, 38, 178–185. Postmes, T., Branscombe, N. R., Spears, R., & Young, H. (1999). Comparative processes in personal and group judgments: Resolving the discrepancy. Journal of Personality and Social Psychology, 76, 320–338. Pratto, F., & Walker, A. (2001). Dominance in disguise: Power, beneficence, and exploitation in personal relationships. In A. Y. Lee-Chai & J. A. Bargh (Eds.), The use and abuse of power: Multiple perspectives on the causes of corruption (pp. 93–114). Philadelphia, PA: Psychology Press. Quattrone, G. A., & Jones, E. E. (1980). The perception of variability within in-groups and outgroups: Implications from the law of small numbers. Journal of Psychology and Social Psychology, 38, 141–152. Ramirez, A., & Soriano, F. I. (1993). Differential patterns of intra- and interethnic interaction in social power systems. Journal of Social Psychology, 133, 307–316. Reicher, S., & Levine, M. (1994). Deindividuation, power relations between groups and the expression of social identity: The effects of visibility to the out-group. British Journal of Social Psychology, 33, 145–163. Reicher, S., Levine, R. M., & Gordijn, E. (1998). More on deindividuation, power relations between groups and the expression of social identity: Three studies on the effects of visibility to the in-group. British Journal of Social Psychology, 37, 15–40. Reskin, B. F., McBrier, D. B., & Kmec, J. A. (1999). The determinants and consequences of workplace sex and race composition. Annual Review of Sociology, 25, 335–361. Riches, P., & Foddy, M. (1989). Ethnic accent as a status cue. Social Psychology Quarterly, 52, 197–206. Ridgeway, C. (1982). Status in groups: The importance of motivation. American Sociological Review, 47, 175–188. Ridgeway, C. L. (1987). Nonverbal behavior, dominance, and the basis of status in task groups. American Sociological Review, 52, 683–694. Ridgeway, C. (1991). The social construction of status value: Gender and other nominal characteristics. Social Forces, 70, 367–386. Ridgeway, C. L. (2001). The emergence of status beliefs: From structural inequality to legitimizing ideology. In J. T. Jost & B. Major (Eds.), The psychology of legitimacy:

230

KLEIN AND SNYDER

Emerging perspectives on ideology, justice, and intergroup relations (pp. 257–277). New York: Cambridge University Press. Ridgeway, C. L., & Berger, J. (1988). The legitimation of power and prestige orders in task groups. In M. Webster, Jr. & M. Foschi (Eds.), Status generalization: New theory and research (pp. 207–231). Stanford, CA: Stanford University Press. Ridgeway, C. L., & Smith-Lovin, L. (1999). The gender-system and interaction. Annual Review of Sociology, 25, 191–216. Rogier, A. (1999). Stereotypes as legitimizing tools: Social attribution and essentialism. Doctoral dissertation, Universite´ Catholique de Louvain, Louvain-La-Neuve, Belgium. Rothbart, M., Evans, M., & Fulero, S. (1979). Recall for confirming events: Memory processes and the maintenance of social stereotypes. Journal of Experimental Social Psychology, 15, 353–355. Rothbart, M., & John, O. P. (1985). Social categorization and behavioral episodes: A cognitive analysis of the effects of intergroup contact. Journal of Social Issues, 41, 81–104. Rudman, L. A., & Borgida, E. (1995). The afterglow of construct accessibility: The behavioral consequences of priming men to view women as sexual objects. Journal of Experimental Social Psychology, 31, 493–517. Ruscher, J. B. (1998). Prejudice and stereotyping in everyday communication. In M. P. Zanna (Ed.), Advances in experimental social psychology (Vol. 30, pp. 241–307). San Diego, CA: Academic Press. Ruscher, J. B. (2001). Prejudiced communication: A social psychological perspective. New York: Guilford Press. Sackett, P. R., DuBois, C. L. Z., & Noe, A. W. (1991). Tokenism in performance evaluation: The effects of work group representation on male-female and white-black differences in performance ratings. Journal of Applied Psychology, 76, 263–267. Saris, R. N., Johnston, I., & Lott, B. (1995). Women as cues for men’s approach or distancing behavior: A study of interpersonal sexist discrimination. Sex Roles, 33, 289–298. Schaller, M., & Conway, L. G. (2001). From cognition to culture: The origins of stereotypes that really matter. In G. Moscowitz (Ed.), Cognitive social psychology: On the tenure and future of social cognition (pp. 163–176). Mahwah, NJ: Erlbaum. Sedikides, C. (1997). Differential processing of ingroup and outgroup information: The role of relative group status in permeable boundary groups. European Journal of Social Psychology, 27, 121–144. Sherif, M. (1966). In common predicament; social psychology of intergroup conflict and cooperation. Boston: Houghton Mifflin. Shih, M., Pittinsky, T., & Ambady, N. (1999). Stereotype susceptibility: Identity salience and shifts in quantitative performance. Psychological Sciences, 10, 80–83. Sidanius, J., & Pratto, F. (1999). Social dominance: An intergroup theory of social hierarchy and oppression. New York: Cambridge University Press. Simmel, G. (1971). On individuality and social form. Chicago: University of Chicago Press. Simon, B. (1998). Individuals, groups, and social change: On the relationship between individual and collective self-interpretations and collective action. In C. Sedikides & J. Schopler (Eds.), Intergroup cognition and intergroup behavior (pp. 257–282). Mahwah, NJ: Erlbaum. Simon, B., & Brown, R. (1987). Perceived intragroup homogeneity in minority-majority contexts. Journal of Personality and Social Psychology, 53, 703–711. Simon, B., & Hamilton, D. L. (1994). Self-stereotyping and social context: The effects of relative in-group size and in-group status. Journal of Personality and Social Psychology, 66, 699–711.

STEREOTYPES AND BEHAVIORAL CONFIRMATION

231

Simon, B., Hastedt, C., & Aufderheide, B. (1997). When self-categorization makes sense: The role of meaningful social categorization in minority and majority members’ self-perception. Journal of Personality and Social Psychology, 73, 310–320. Smith, D. M., Neuberg, S. L., Judice, T. N., & Biesanz, J. C. (1997). Target complicity in the confirmation and disconfirmation of erroneous perceiver expectations: Immediate and longer term implications. Journal of Personality and Social Psychology, 73, 974–991. Smith-Lovin, L., & Brody, C. (1989). Interruptions in group discussion and the influence of women in mixed-sex groups. American Sociological Review, 54, 424–435. Smith-Lovin, L., & McPherson, J. M. (1991). You are who you know: A network perspective on gender. In P. England (Ed.), Theory on gender/feminism on theory (pp. 223–251). New York: Aldine. Snyder, M. (1984). When belief creates reality. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 18, pp. 248–305). Orlando, FL: Academic Press. Snyder, M. (1992). Motivational foundations of behavioral confirmation. In M. P. Zanna (Ed.), Advances in experimental social psychology (Vol. 25, pp. 67–114). San Diego, CA: Academic Press. Snyder, M. (1998). Interaction goals: Their structure and function. In J. M. Darley & J. Cooper (Eds.), Attribution and social interaction: The legacy of Edward E. Jones (pp. 127–197). Washington, DC: American Psychological Association. Snyder, M., & Campbell, B. (1980). Testing hypotheses about other people: The role of the hypothesis. Personality and Social Psychology Bulletin, 6, 421–426. Snyder, M., Campbell, B. H., & Preston, E. (1982). Testing hypotheses about human nature: Assessing the accuracy of social stereotypes. Social Cognition, 1, 256–272. Snyder, M., & Cantor, N. (1979). Testing hypotheses about other people: The use of historical knowledge. Journal of Experimental Social Psychology, 15, 330–342. Snyder, M., & Haugen, J. A. (1994). Why does behavioral confirmation occur? A functional perspective on the role of the perceiver. Journal of Experimental Social Psychology, 30, 218–246. Snyder, M., & Haugen, J. A. (1995). Why does behavioral confirmation occur? A functional perspective on the role of the target. Personality and Social Psychology Bulletin, 21, 963–974. Snyder, M., & Kiviniemi, M. T. (2001). Getting what they came for: How power influences the dynamics and outcomes of interpersonal interaction. In A. Y. Lee-Chai & J. A. Bargh (Eds.), The use and abuse of power: Multiple perspectives on the causes of corruption (pp. 133–155). Philadelphia, PA: Psychology Press. Snyder, M., & Oyamot, C. M. (2001). Self-fulfilling prophecies. In J. Worell (Ed.), Encyclopedia of Women and Gender (pp. 945–953). San Diego, CA: Academic Press. Snyder, M., & Stukas, A. A., Jr. (1999). Interpersonal processes: The interplay of cognitive, motivational, and behavioral activities in social interaction. Annual Review of Psychology, 50, 273–303. Snyder, M., & Swann, W. B. (1978). Behavioral confirmation in social interaction: From social perception to social reality. Journal of Experimental Social Psychology, 14, 148–162. Snyder, M., Tanke, E. D., & Berscheid, E. (1977). Social perception and interpersonal behavior: On the self-fulfilling nature of social stereotypes. Journal of Personality and Social Psychology, 35, 656–666. Snyder, M. L., Kleck, R. E., Strenta, A., & Mentzer, S. J. (1979). Avoidance of the handicapped: An attributional ambiguity analysis. Journal of Personality and Social Psychology, 37, 2297–2306. Stangor, C., & Schaller, M. (1996). Stereotypes as individual and collective representations. In C. N. Macrae, C. Stangor, & M. Hewstone (Eds.), Stereotypes and stereotyping (pp. 3–40). New York: Guilford Press.

232

KLEIN AND SNYDER

Steele, C. M. (1998). A threat in the air: How stereotypes shape intellectual identity and performance. In J. L. Eberhardt & S. T. Fiske (Eds.), Confronting racism: The problem and the response (pp. 202–233). Thousand Oaks, CA: Sage. Steele, C. M., & Aronson, J. (1995). Stereotype threat and the intellectual test performance of African Americans. Journal of Personality and Social Psychology, 69, 797–811. Stephan, W. G., & Stephan, C. W. (1985). Intergroup anxiety. Journal of Social Issues, 41, 157–175. Stephan, W. G., & Stephan, C. (1993). Cognition and affect in stereotyping: Parallel interactive networks. In D. M. Mackie & D. L. Hamilton (Eds.), Affect, cognition and stereotyping: Interactive processes (pp. 111–136). Orlando, FL: Academic Press. Stephan, W. G., & Stephan, C. W. (2000). An integrated threat theory of prejudice. In S. Oskamp (Ed.), Reducing prejudice and discrimination (pp. 23–45). Mahwah, NJ: Erlbaum. Stern, L. D., Marrs, S., Millar, M. G., & Cole, E. (1984). Processing time and the recall of inconsistent and consistent behaviors of individuals and groups. Journal of Personality and Social Psychology, 47, 253–262. Stott, C. J., Hutchinson, P., & Drury, J. (2001). ‘‘Hooligans’’ abroad? Inter-group dynamics, social identity and participation in collective ‘‘disorder’’ at the 1998 World Cup Finals. British Journal of Social Psychology, 40, 359–384. Swann, W. B. (1983). Self-verification: Bringing social reality into harmony with the self. In J. Suls (Ed.), Psychological Perspectives on the self (Vol. 2, pp. 33–36). Hillsdale, NJ: Erlbaum. Swann, W. B., & Read, S. J. (1981). Self-verification processes: How we sustain our selfconceptions. Journal of Experimental Social Psychology, 17, 351–372. Swann, W. B., & Snyder, M. (1980). On translating beliefs into action: Theories of ability and their application in an instructional setting. Journal of Personality and Social Psychology, 38, 879–888. Tajfel, H. (1975). The exit of social mobility and the voice of social change. Social Science Information, 14, 101–118. Tajfel, H. (1981a). Human groups and social categories. Cambridge, UK: Cambridge University Press. Tajfel, H. (1981b). Social stereotypes and social groups. In J. C. Turner & H. Giles (Eds.), Intergroup Behavior (pp. 144–167). Oxford, UK: Blackwell. Tajfel, H., Billig, M. G., Bundy, R. P., & Flament, C. (1971). Social categorization and intergroup behaviour. European Journal of Social Psychology, 1, 149–178. Tajfel, H., & Turner, J. C. (1986). The social identity theory of intergroup behavior. In S. Worchel & W. G. Austin (Eds.), The psychology of intergroup relations (pp. 7–24). Chicago: Nelson-Hall. Taylor, D. M., & McKirnan, D. J. (1984). Theoretical contributions: A five-stage model of intergroup relations. British Journal of Social Psychology, 23, 291–300. Taylor, S. E. (1979). The generalizability of salience effects. Journal of Personality and Social Psychology, 37, 357–368. Thompson, M. S., Judd, C. M., & Park, B. (2000). The consequences of communicating social stereotypes. Journal of Experimental Social Psychology, 36, 567–599. Turner, J. C. (1991). Social influence. Pacific Grove, CA: Brooks/Cole. Turner, J. C., Hogg, M. A., Oakes, P. J., Reicher, S. D., & Wetherell, M. S. (1987). Rediscovering the social group: A self-categorization theory. New York: Basil Blackwell. van Knippenberg, A., & Dijksterhuis, A. (2001). Social categorization and stereotyping: A functional perspective. In W. Stroebe & W. Hewstone (Eds.), European Review of Social Psychology (pp. 105–144). Chichester, UK: Wiley.

STEREOTYPES AND BEHAVIORAL CONFIRMATION

233

Vanman, E. J., Paul, B. Y., Ito, T. A., & Miller, N. (1997). The modern face of prejudice and structural features that moderate the effect of cooperation on affect. Journal of Personality and Social Psychology, 73, 941–959. Virdin, L. M., & Neuberg, S. L. (1990). Perceived status: A moderator of expectancy confirmation effects. Paper presented at the 98th annual convention of the American Psychological Association, Boston. von Baeyer, C. L., Sherk, D. L., & Zanna, M. P. (1981). Impression management in the job interview: When the female applicant meets the male (chauvinist) interviewer. Personality and Social Psychology Bulletin, 7, 45–51. Vorauer, J. D., & Kumhyr, S. M. (2001). Is this about you or me? Self-versus other-directed judgments and feelings in response to intergroup interaction. Personality and Social Psychology Bulletin, 27, 706–719. Vorauer, J. D., Main, K. J., & O’Connell, G. B. (1998). How do individuals expect to be viewed by members of lower status groups? Content and implications of meta-stereotypes. Journal of Personality and Social Psychology, 75, 917–937. Wagner, D. G., & Berger, J. (1993). Status characteristics theory: The growth of a program. In J. Berger & M. Zelditch, Jr. (Eds.), Theoretical research programs: Studies in the growth of theory (pp. 23–63). Stanford, CA: Stanford University Press. Weber, R., & Crocker, J. (1983). Cognitive processes in the revision of stereotypic beliefs. Journal of Personality and Social Psychology, 45, 961–977. Weiner, B. (1986). Attribution, emotion, and action. In R. M. Sorrentino & E. T. Higgins (Eds.), Handbook of motivation and cognition: Foundations of social behavior (pp. 281–312). New York: Guilford Press. Weitz, S. (1972). Attitude, voice, and behavior: A repressed affect model of interracial interaction. Journal of Personality and Social Psychology, 24, 14–21. Wilder, D. A. (1984). Intergroup contact: The typical member and the exception to the rule. Journal of Experimental Social Psychology, 20, 177–194. Wilder, D. A., & Shapiro, P. (1989). Effects of anxiety on impression formation in a group context: An anxiety-assimilation hypothesis. Journal of Experimental Social Psychology, 25, 481–499. Wood, W., & Karten, S. J. (1986). Sex differences in interaction style as a product of perceived sex differences in competence. Journal of Personality and Social Psychology, 50, 341–347. Word, C. O., Zanna, M. P., & Cooper, J. (1974). The nonverbal mediation of self-fulfilling prophecies in interracial interaction. Journal of Experimental Social Psychology, 10, 109–120. Wright, B. A. (1983). Physical disability: A psychological approach. New York: Harper & Row. Wright, S. C., & Taylor, D. M. (1998). Responding to tokenism: Individual action in the face of collective injustice. European Journal of Social Psychology, 28, 647–667. Wright, S. C., & Taylor, D. M. (1999). Success under tokenism: Co-option of the newcomer and the prevention of collective protest. British Journal of Social Psychology, 38, 369–396. Wright, S. C., Taylor, D. M., & Moghaddam, F. M. (1990). Responding to membership in a disadvantaged group: From acceptance to collective protest. Journal of Personality and Social Psychology, 58, 994–1003. Yzerbyt, V., Corneille, O., & Estrada, C. (2001). The interplay of subjective essentialism and entitativity in the formation of stereotypes. Personality and Social Psychology Review, 5, 141–155. Yzerbyt, V., Rocher, S., & Schadron, G. (1997). Stereotypes as explanations: A subjective essentialistic view of group perception. In R. Spears & P. J. Oakes (Eds.), The social psychology of stereotyping and group life (pp. 20–50). Oxford, UK: Blackwell. Yzerbyt, V., & Rogier, A. (2001). Blame it on the group: Entitativity, subjective essentialism, and social attribution. In J. Jost & B. Major (Eds.), The psychology of legitimacy: Emerging

234

KLEIN AND SNYDER

perspectives on ideology, justice, and intergroup relations (pp. 103–134). New York: Cambridge University Press. Zanna, M. P., & Pack, S. J. (1975). On the self-fulfilling nature of apparent sex differences in behavior. Journal of Experimental Social Psychology, 11, 583–591. Zimmerman, D. H., & West, C. (1975). Sex roles, interruptions and silences in conversation. In B. Thorne & Henley (Eds.), Language and sex: Differences and dominance (pp. 105–129). Rowley, MA: Newbury.

MOTIVATIONAL BASES OF INFORMATION PROCESSING AND STRATEGY IN CONFLICT AND NEGOTIATION

Carsten K. W. De Dreu Peter J. Carnevale

I. Introduction and Overview ‘‘Had we begun negotiations then, I think at least half of the casualties and injuries on both sides could have been avoided . . . (after meeting recently) the Hanoi officials agreed that they also missed these opportunities.’’ (58,000 US soldiers perished, as did 3.8 million Vietnamese.)’’ —Former Secretary of Defense Robert McNamara, in a discussion of his book Argument Without End: Searching for Answers to the Vietnam Tragedy, The Brown Daily Herald, Tuesday, April 27, 1999

The death and destruction of unremitting social conflict, and the waste of it, pose a challenge for analysis. For social psychologists, the challenge entails an understanding of underlying processes—cognition, affect, motivation—and, in particular, how these processes play out in negotiation. Negotiation can provide a means of escaping social conflict. It entails a discussion between two or more parties about differences of interest, and its aim is agreement that reconciles incompatible preferences among a set of options. Negotiation has tremendous consequences for social outcomes at all levels of society, including marital satisfaction and child development (Amato & Booth, 2001), work team performance and organizational functioning (Jehn, 1995), community and environment development (Susskind, McKearnan, & Thomas-Larmer, 1999), and international relations (Berton, Kimura, & Zartman, 1999). Much of the social psychological analysis has, in the past two decades, focused on cognitive processes responsible for suboptimal negotiation 235 ADVANCES IN EXPERIMENTAL SOCIAL PSYCHOLOGY, VOL. 35

Copyright 2003 Elsevier Science (USA). All rights reserved. 0065-2601/03 $35.00

236

DE DREU AND CARNEVALE

outcomes (for reviews, see Bazerman, Curhan, Moore, & Valley, 2000; Carnevale & Pruitt, 1992; Thompson, 1990). This work has tended to explain much of what happens in conflict and negotiation in terms of cognitive deficiencies and erroneous reasoning. For example, Neale and Bazerman (1991, p. 170) state that ‘‘a great deal of sub-optimality in negotiated outcomes is attributable to deviations from rationality in negotiators’ cognitive processes.’’ The focus on cognitive processes should not be taken as indicating that motivation and affect are unimportant. In 1960, Thomas Schelling, the economist, introduced the concept of mixed-motive interaction to describe situations in which two or more parties simultaneously experience the motivation to cooperate and compete with each other, and argued that these situations present a particularly difficult cognitive problem. No doubt the decision makers in Hanoi and the United States, in 1968, faced a difficult cognitive problem, a situation that comprised a mixture of cooperative and competitive motivation. Others have argued that disputants want to be consistent and to reduce cognitive dissonance (Ross & Ward, 1995), to protect their face (Brown, 1977), to develop and maintain a positive self-image (Kramer, Pommerenke, & Newton, 1993), to be treated with respect and to obtain a fair distribution of outcomes (Tyler, Huo, & Lind, 1999), and to take other’s perspective to better understand the other (Kemp & Smith, 1994; Pruitt, 1981). Recently the importance of motivation was recognized by Chaiken, Gruenfeld, and Judd (2000, p. 158) in their comment that ‘‘in negotiation . . . motives are likely to be intense.’’ We argue that many of these needs and desires may be subsumed under the header of one of two broad classes of motivation that affect conflict and negotiation, and that will be the focus of the current analysis. Motives that have to do with outcomes, and its distribution between conflict parties, fall under social motivation, or the desire to attain certain distributions of outcomes between oneself and the other party (Deutsch, 1949, 1973; Messick & McClintock, 1968; McClintock, 1977; Pruitt, 1967). Motives that have to do with understanding the task and one’s opponent, to obtain and maintain cognitive consistency, and to reduce dissonance can be subsumed under the broad header of epistemic motivation, or the need to develop a rich and accurate understanding of the world (see Kruglanski, 1989). Ross and Ward (1995) argued that dispute resolution is impeded by naive realism, or the tendency to assume the world is as it is perceived, and ego defensiveness, or the tendency to develop and uphold a positive self-image. We will review recent work to demonstrate that social motive moderates the influence of ego defensiveness and naive realism on information processing and strategic choice in negotiation. Further, we will demonstrate that

MOTIVATION IN CONFLICT AND NEGOTIATION

237

epistemic motives moderate the impact of naive realism and the information processing biases it engenders. This analysis results in a motivated information processing model of negotiation. It captures the influence of many variables—situational features such as power differences or time pressure, and individual differences in social value, cultural background, or need for cognitive closure—on cognitive processes and economic and psychological outcomes in conflict and negotiation. Moreover, it points to four archetypes of disputants in conflict and negotiation—prosocial or selfish misers and prosocial or selfish thinkers. Kelley and Thibaut (1978) argued that to understand behavior in interdependent situations, one needs to consider the structure of the situation within which interaction takes place. Indeed, research has revealed that minor changes in the structure of the task can have important consequences for the way people think and act, as well as for the outcomes from the negotiation (Beersma & De Dreu, 2002; Carnevale & Probst, 1998; De Dreu & McCusker, 1997; Polzer, Mannix, & Neale, 1998; Pruitt, 1967). Accordingly, we begin with a review of the settings and research paradigms to which our analysis applies, to illustrate that in some conflicts and in some negotiation settings some motives are more likely to emerge than in other settings. Subsequently we review past research on cognitive and motivational processes in conflict and negotiation, and we will introduce in more detail ego defensiveness and naive realism as barriers to dispute resolution. In the fourth and fifth sections we discuss the antecedents and consequences of social and epistemic motivation, respectively, and consider their influence on information processing biases and strategic choice. In the sixth section of this chapter we present the motivated information processing model in its entirety, we discuss the possibility that motives change during the negotiation, and we deal with some unresolved issues. We end with a section in which we consider other motives, and alternative models of information processing and strategic choice in conflict and negotiation.

II. Issues, Outcomes, and Tasks in the Study of Conflict and Negotiation The central element of negotiation and related situations is the reward structure, which is the set of alternatives that people must choose among, and the outcomes that are the possible results of these choices. The alternatives in negotiation are represented as the topics under discussion, and can usually be divided into one or more issues that require a decision by the parties.

238

DE DREU AND CARNEVALE

A major strength of the social psychological approach to negotiation is the development of laboratory tasks that model issues and that allow tests of causal hypotheses about underlying process. The participants in these laboratory studies are usually undergraduate volunteers. Two or more true participants may deal with each other, or a single participant may (without knowing it) deal with a confederate or a computer program. Communication is sometimes face-to-face, sometimes by means of note passing, and sometimes over a computer network. The tasks are often simplified versions of reality, yet retain key elements of the structure or processes of negotiation. To appreciate the variety of tasks and methods, a bit of history is required. Much of today’s social psychology of negotiation reflects a behavioral science tradition that seeks to develop and test predictive theory about the impact of environmental conditions on negotiator behavior and the impact of these conditions and behaviors on outcomes (see Pruitt, 1981; Rubin & Brown, 1975). There are important early influences from outside social psychology: mathematical models of rational decision making in negotiation developed by economists and game theorists (e.g., Luce & Raiffa, 1957; Raiffa, 1982; Rapoport, 1960; Roth, 1994), and theories about the origins and impact of negotiator tactics developed in particular by industrial relations specialists (Walton & McKersie, 1965) and the economist Schelling (1960). In addition, there have always been books and manuals that provide advice to negotiators (e.g., de Callieres, 1716; Fisher & Ury, 1981; Thompson, 1998), which often serve as useful foils for testing hypotheses. The important role of economic ideas in the social psychology of negotiation is clearly seen in the classic work of Siegel and Fouraker (1960; the former a psychologist and the latter an economist; see Smith, 2001), which played a heavy role in important work by Kelley and Schenitzki (1972). These authors, and others, made an important distinction between negotiation tasks that have a single dimension of value, and multiple dimensions of value (see Kelley, 1966; Beckman, & Fischer, 1967; Froman & Cohen, 1970; Pruitt & Lewis, 1975). Consider the task used by Kelley et al. (1967), the ‘‘Game of Nines.’’ Here, two people divide 9 points (points that have a real-money equivalent), by holding up cards that show numbers from 1 to 8. If the sum of the two cards held up on any trial is 9 or less, then an agreement occurs, and the agreement is that each person gets the amount showing on their card (e.g., 3 for one person, 6 for the other). If there is no agreement, each gets 0 (or some prespecified minimum). Clearly there is a competitive incentive (I would rather have 6 than 3; or 5 than 4; or 8 than 1) but there is a cooperative incentive as well: we both should prefer any agreement than

MOTIVATION IN CONFLICT AND NEGOTIATION

239

no agreement. Thus the task is mixed-motive in the sense identified by Schelling (1960). A single-issue negotiation task (e.g., the price of a car in a buyer–seller negotiation) has a single dimension of value. Table I shows a single-issue negotiation task that simulates a union–management negotiation. The participants play a role, here a management and a union negotiator dealing with a wage issue. The management negotiator is given the issue chart shown at the top of Table I, and the union negotiator is given the issue chart shown at the bottom. Each chart lists five possible settlement points on the issue, represented by the labels in the column. Next to each label is the value of settlement at that point (numbers typically indexed to money). Each side sees only their own issue chart, but typically they are told they can say anything they want during the negotiation about the charts. As can be seen, the value of each of the five possible settlement points indicates that the parties are dividing a fixed sum of value (in this case, 240), and the task is therefore ‘‘fixed-sum.’’ Note that the existence of a no-agreement alternative with, for example, a value of 0 to both parties, alters the structure of the game, making it ‘‘variable sum’’ (since the outcome values are either 240, or 0; see below). Note also that the role labels on the task have largely been considered unimportant by researchers, and interchangeable (e.g., replace task labels ‘‘union/management’’ and ‘‘wages’’ with ‘‘buyer/seller’’ and ‘‘price,’’ or with

TABLE I The Singe Issue Negotiation: Issue Charts for Management and Union Negotiatorsa Annual raise (%)

(Value)

Management Issue Chart 15 12 9 6 3

(00) (60) (120) (180) (240)

Union Issue Chart 15 12 9 6 3

(240) (180) (120) (60) (00)

a All agreements sum to a constant for the dyad, 240, making the task ‘‘fixed-sum.’’

240

DE DREU AND CARNEVALE

‘‘country A/country B’’ and ‘‘territory’’). There is, however, reason to suspect that such labels may have important cuing effects (Rettinger & Hastie, 2001) that can be quite subtle (Burnham, McCabe, & Smith, 2000). Consider a second type of task developed by Siegel and Fouraker (1960), Kelley (1966), and Pruitt and Lewis (1975). Here, there are multiple issues, each having several possible settlement points, and each side having a different preference ordering on which issue is most important. This kind of task allows the development of ‘‘integrative’’ or win–win agreements. The values of the possible outcomes are variable, which means some agreements represent better outcomes for both parties than other agreements. An example is given in Table II. The participants again play roles of management and union negotiators faced with the task of reaching agreement, here on four issues. The management negotiator is given the issue chart shown at the top of Table II, and the union negotiator is given the issue chart shown on the bottom. These charts list five possible settlement points on each issue, as in the task shown in Table I. Again, each side sees only their own chart, and typically they get the instruction that they can say anything to one another about it. As can be seen in the Table II charts, the issue represented on the left is important to one side (has greater value) and not important to the other side (has low value), and the opposite for the issue to the far right. This means that there are many possible agreements, and they vary in their value to the negotiators, both individually and collectively. (This example stems from TABLE II Issue Charts for Management and Union Negotiators a Salary

Vacation days

Annual raise

Medical coverage

Management Issue Chart F70.000 (00) 3 week (00) F65.000 (15) 2.5 week (30) F60.000 (30) 2 weeks (60) F55.000 (45) 1.5 week (90) F50.000 (60) 1 week (120)

15% (00) 12% (60) 9% (120) 6% (180) 3% (240)

100% (00) 80% (100) 60% (200) 40% (300) 20% (400)

Union Issue Chart F70.000 (400) F65.000 (300) F60.000 (200) F55.000 (100) F50.000 (00)

15% (240) 12% (180) 9% (120) 6% (60) 3% (00)

100% (60) 80% (45) 60% (30) 40% (15) 20% (00)

3 week (120) 2.5 week (90) 2 week (60) 1.5 week (30) 1 week (00)

a Agreement values for the dyad vary from 820 to 1160 making the task ‘‘variable-sum.’’ Source: De Dreu, Giebels, and Van de Vliert (1998).

MOTIVATION IN CONFLICT AND NEGOTIATION

241

Pruitt, 1981, and has been widely adopted, e.g., Bazerman, Magliozzi, & Neale, 1985; Carnevale & Lawler, 1986; Thompson & Hastie, 1990.) The two issues in the center of the task have equal value for both parties and, in the parlance of negotiation research, are ‘‘distributive’’ issues; the other two issues that can be traded for one another are called ‘‘integrative.’’ The laboratory paradigm for modeling this form of negotiation is called ‘‘integrative bargaining,’’ reflecting Walton and Mckersie’s (1965) observation that ‘‘Distributive bargaining is the process by which each party attempts to maximize his own share in the context of fixed-sum payoffs. Integrative bargaining is the process by which the parties attempt to increase the size of joint gain without respect to the division of the payoffs’’ (p. 8). Joint gain is reflected in the task in that a settlement of F70 on ‘‘salary’’ and 20% on ‘‘medical’’ provide an outcome that has greater value—to each individual and to the pair together—than an agreement that split these two issues down the middle (joint gain of 800 versus 460 on these two issues). A plethora of laboratory tasks have been devised by psychologists, economists, political scientists, and others for the study of strategic interaction, negotiation, and social conflict (see, for example, Pruitt & Kimmel, 1977; Rubinstein, 1999). Table III presents a simple taxonomy of such tasks. One aspect of the taxonomy is whether the task models a single dimension of value, or multiple dimensions. The former include one-issue negotiation tasks (as shown in Table I), whereas the latter includes integrative bargaining tasks and its variations (as shown in Table II, including contingent agreement games and the integrative ultimatum game). The other feature of the taxonomy of laboratory tasks is whether the task is a game of agreement or a game of coordination. A game of agreement allows the behavioral act of agreeing, coming to an accord, consensus, TABLE III Taxonomy of Laboratory Tasks for the Study of Mixed-Motive Interaction Coordination games

Agreement games

Single dimension of value

Prisoner’s dilemma Chicken N-Person dilemma Locomotion games Coalition games Trust game Etc.

1-Issue negotiation Ultimatum bargaining

Multiple dimensions of value

Not done

Integrative bargaining Integrative ultimatum

242

DE DREU AND CARNEVALE

shared understanding, or to terms that reflect a shared commitment, harmony of opinion, or course of action on the matters at hand. Consider the ultimatum game, where a ‘‘proposer’’ makes an offer of an exchange and the ‘‘responder’’ either accepts or rejects it, then game over (Guth, Schmittberger, & Schwarze, 1982). Agreement games allow nonagreement, which is often determined by one or more participant saying no, withdrawing, or by a deadline reached that terminates negotiation; no agreement implies that the status quo is maintained. Agreements often represent a change in the status quo, for example, terms of an exchange (a new joint venture; change in ownership) or a new common understanding, or new shared perception, belief, attitude, or feeling (Harinck, De Dreu, & Van Vianen, 2000; Rapoport, 1960; Druckman & Zechmeister, 1973; Levine & Thompson, 1996). Games of coordination model one, or a series of many, individual decisions about a distribution of outcomes. These include, as examples, the prisoner’s dilemma-type game (Rubin & Brown, 1975), locomotion and coalition games (Pruitt & Kimmel, 1977), and trust games (Berg, Dickhaut, & McCabe, 1995). Coordination games typically generate simple response variables. The notion of disagreement is usually not clearly represented, and coordination games typically do not model more than one dimension of value. Nevertheless, studies of coordination games are highly important for understanding strategic interaction; indeed, there is often a structural similarity among games and often one can be transformed into another, for example, negotiation can be seen as a prisoner’s dilemma as well as a game of chicken (Carnevale & Pruitt, 2003; see Kelley & Thibaut, 1978). An important dependent variable in many negotiation studies is the quality of the agreement. Researchers have examined agreements in terms of how much profit a focal party made (individual outcomes), in terms of how much profit both parties made together (joint outcomes), and sometimes in terms of how outcome was distributed between the parties (outcome difference). In essence, these approaches correspond to three basic motivational goals in negotiation, namely to do well personally (egoistic motivation), to do well collectively (prosocial motivation), and to achieve distributive fairness. In addition, researchers have examined agreements in terms of its deviation from some normative standard, such as Pareto Optimality—the point at which no party can do better without the other party doing worse. Interesting discussions about how to code and evaluate negotiation agreements have been published by Tripp and Sondak (1992) and Clyman (1995). Although in theory large discrepancies are possible, empirically joint outcomes appear strongly correlated with more sophisticated and mathematically more complex outcomes measures such as deviations from Pareto Optimality (De Dreu, Giebels, & Van de Vliert, 1998; Weingart, Hyder, & Prietula, 1996).

MOTIVATION IN CONFLICT AND NEGOTIATION

243

III. Cognition and Motivation in Conflict and Negotiation Conflict research and theory have been concerned with cognition and information processing long before the cognitive revolution swept its way through social psychology. The early work by Siegel and Fouraker (1960), for example, was strongly concerned with the ways negotiators exchange information, and with its consequences for perception and financial gain. In the 1960s and 1970s, many authors considered how misperceptions contribute to the escalation of conflict (e.g., Hammond, 1965; Jervis, 1976; see also Kemmelmeier & Winter, 2000). Many researchers, including Cross (1977), characterize negotiators as adaptive decision makers who base their strategic choices on expectations about and perceptions of their opponent and who adapt these perceptions when they appear erroneous. The cognitive processes perspective on conflict and negotiation gained momentum with the work by Bazerman and Neale (1983; Neale & Bazerman, 1991). Their research built on the seminal work by Simon (1957) on rationality and problem solving, and subsequent studies on biases and heuristics in judgment and decision making by Kahneman and Tversky (1973, 1979). In essence, this work considers negotiation behavior and failure to reach integrative agreements to be a function of the individual’s tendency to rely on cognitive heuristics, and to engage in erroneous reasoning. The heuristics and biases that have been shown to affect integrative negotiation may be due primarily to (1) cognitive limitations that cause erroneous reasoning, (2) ego defensiveness, or the desire to develop and maintain a positive self-image, and (3) naive realism, or the tendency to assume the world as it is perceived (Kruglanski & Ajzen, 1983; Ross, 1977; Ross & Ward, 1995).1 In this section we discuss several heuristics and biases that prohibit or undermine constructive negotiation processes, and reduce the likelihood that negotiators reach an integrative agreement. We begin with a discussion of bias due to cognitive limitations, such as anchoring-and-adjustment, and the use of stereotypes. We then move on to bias due to naive realism, and discuss fixed-pie bias and confirmatory information search. Finally, we consider ego defensiveness, and discuss self-serving evaluations of conflict behavior, and reactive devaluation. We will conclude this section with a critical look at the generality of these cognitive and motivational biases, and 1

Cognition and motivation are intimately connected and difficult to separate (Higgins & Kruglanski, 2001; Tetlock & Levi, 1983). Indeed, the biases identified in the (conflict) literature have a cognitive and a motivational component. This notwithstanding, most bias in conflict and negotiation can be distinguished in terms of the relative importance of cognitive limitations, ego defensiveness, or naive realism.

244

DE DREU AND CARNEVALE

we will argue that naive realisms and ego defensiveness may, in fact, constitute but one side of the coin in conflict and negotiation.

A. BIAS DUE TO COGNITIVE LIMITATIONS Some bias in human thinking originates in the limitations of otherwise reasonable information-processing strategies. People possess a large amount of suboptimal strategies that they bring to the task of predicting and explaining events (Kruglanski & Ajzen, 1983). Kahneman and Tversky (1973) identified three types of heuristics that people use to make sense of their complex environment and that lead to suboptimal, even downright incorrect conclusions. These are anchoring-and-(insufficient) adjustment— the tendency to overly rely on an arbitrarily chosen reference point, the availability heuristic—the tendency to overly rely on information that is salient in memory, and the representativeness heuristic—the tendency to make judgments solely based on the most obvious features of the stimulus. Anchoring-and-adjustment and the representativeness heuristic have been shown to be a barrier to constructive negotiation and dispute resolution. 1. Anchoring-and-Adjustment and Gain–Loss Framing Negotiation is all about relative values, and comparison standards have an important influence when it comes to setting one’s aspiration or determining one’s reservation price or limit—the point below which one refuses to concede (Pruitt, 1981; Raiffa, 1982). Unfortunately, humans have a tendency to base estimates on irrelevant anchor information, and adjust their estimates inappropriately (Kahneman & Tversky, 1973). For example, Northcraft and Neale (1987) showed that real estate agents insufficiently adjust for irrelevant anchor information when providing estimates of real estate value. In their study, real estate agents visited a piece of property currently for sale and then provided an estimate of the value of the property. Each participant was provided with an information package that contained information about the property’s square footage, characteristics of the property, similar property recently sold, similar property currently for sale, and the listing price. Although the actual listing price and appraised value was US$74,900, some real estate agents were provided with a listing price of US$83,900 (12% above the appraised value) and others were given a listing price of US$65,900 (12% below the appraised value). Although real estate agents claimed that they would notice and mark any deviation over 5% as unrealistic, they were substantially influenced by the anchor information provided—those in the low anchor condition appraised the property at

MOTIVATION IN CONFLICT AND NEGOTIATION

245

US$67,811 and those in the high anchor condition provided an appraisal value of US$75,190. Related to anchoring is that people use reference points and code prospective outcomes above the reference point as gains, and those below the reference point as losses (Kahneman & Tversky, 1979). In conflict and negotiation, relevant reference points may be an alternative to a negotiated agreement, one’s initial aspiration level, the settlement reached in adjacent markets, or the outcome reached in a previously conducted and highly similar negotiation. People are loss averse—losses loom larger than equivalent gains (Kahneman & Tversky, 1979), and framing outcomes as losses rather than gains has been shown to make people resistant to concession making (Bottom & Studt, 1993; De Dreu, Carnevale, Emans, & Van de Vliert, 1994; Kristensen, & Ga¨ rling, 1997; Shalev, 2002). Also, people become more risk-seeking when they frame outcomes as losses rather than gains, and this may translate into more contentious negotiation behavior and less integrative agreements (Bazerman et al., 1985; Bottom, 1998). 2. Stereotypes An instance of the representativeness heuristic that has received quite some attention in the negotiation literature is stereotyping. To predict other’s behavior efficiently, individuals may rely on stereotypes (Fiske & Neuberg, 1990). Conflict and negotiation research has revealed that individuals rely on stereotypic information when predicting their opponent’s tendency to compete, and the concomitant expectations predict one’s own behavior (Morris, Larrick, & Su, 1999; Ross & Ward, 1995). For example, De Dreu, Yzerbyt, and Leyens (1995) engaged psychology students in a prisoner’s dilemma game, and led them to believe that their opponent majored in business studies, or in theology. Pilot research had revealed that psychology students stereotyped business students as competitive and opportunistic, and theology students as cooperative and ethical. The results of two experiments showed that participants predicted more competition, and made more competitive choices when they believed the other was a business rather than theology student. In a third experiment, they found that participants construed ambiguous information about the other (he is member of a fraternity) as consistent with the stereotype (i.e., the Christian Ethics Club, in the case of a religion major). Such confirmatory construal even strengthened the tendency to compete or cooperate with a business or theology student, respectively. Consistent with this, Kray, Thompson, and Galinsky (2001) found that negotiators not only stereotype their opponent, but sometimes also engage in self-stereotyping, which might explain why gender sometimes does, and sometimes does not affect negotiation.

246

DE DREU AND CARNEVALE

B. NAIVE REALISM Some bias originates in the individual’s needs and desires that direct attention, encoding, and retrieval processes. Ross and Ward (1995) trace some of these biases to naive realism—the individual’s tendency to assume that he or she sees the world as it is, and that other rational perceivers will therefore share these perceptions. When others fail to see the world as it is, this either reflects others had different information, are lazy, or biased by ideology or self-interest (Ross & Ward, 1995; see also Keltner & Robinson, 1993; Robinson, Keltner, Ward, & Ross, 1995). Naive realism manifests itself in a variety of biases that have been shown to interfere with constructive conflict management and integrative negotiation. An example is optimistic overconfidence, which can lead disputants to forego attempts at settlement, ‘‘believing that time is on their side, and that complete, unilateral victory is just around the corner’’ (Ross & Ward, 1995, p. 266; see also Neale & Bazerman, 1985; Van Boven, Dunning, & Loewenstein, 2000). In the case of arbitration, such optimistic overconfidence may lead both parties to assume that the arbitrator will side with their case, and therefore prefer costly arbitration to constructive negotiation (Ross & Conlon, 2001). 1. Fixed-Pie Assumption A well-known reflection of naive realism that is related to the correspondence bias (Ross, 1977) is the fixed-pie assumption. When parties to a conflict lack information about their opponent’s preferences and priorities, which they often do, they may make a fixed-pie assumption. This means that individuals assume that their own and the other’s preferences are diametrically opposed. Indeed, negotiation research suggests that most people entering a negotiation make such a fixed-pie assumption (Thompson & Hastie, 1990). This shortcut is suboptimal because negotiation settings often contain integrative potential (Pruitt, 1981), and making a fixed-pie assumption leads negotiators to engage in distributive bargaining. They fail to exchange information, and they reach relatively few integrative agreements (e.g., Bottom & Paese, 1997; Harinck, De Dreu, & Van Vianen, 2000; Gelfand & Christakoupoulo, 1999; Pinkley, Griffith, & Northcraft, 1995; Thompson & Hastie, 1990; Thompson, 1991; for a review, see Thompson & Hrebec, 1996). 2. Confirmatory Information Search Naive realism is also reflected in confirmatory information search, in which individuals rely on their own initial beliefs and cognitive structures when searching new and additional information about the task, or about their opponent. Information that is or becomes available during the conflict

MOTIVATION IN CONFLICT AND NEGOTIATION

247

is often the result of the questions disputants ask one another, and the type of questions they ask to gather information may have important consequences for the amount and kind of information that becomes available for subsequent processing. Rubin, Pruitt, and Kim (1994) hypothesize that disputants seek confirmation rather than disconfirmation of their initial beliefs, plans, and strategies: ‘‘By framing questions in ways that could only confirm their hypotheses, both sides would be likely to discover only what they wanted to find’’ (p. 105). Confirmatory information search may easily lead to a self-fulfilling prophecy (Snyder, 1992; Klein & Snyder, this volume) and if the initial hypotheses were incorrect, confirmatory search exacerbates the problems associated with building a strategy on inadequate and incorrect assumptions, beliefs, and cognitive structures (Neale & Bazerman, 1991). Consider, for example, the study by De Dreu, Yzerbyt et al. (1995) on the use of stereotypes described earlier. In the prisoner’s dilemma game used in this study individuals experience a strong incentive to reciprocate other’s noncooperative behavior (see Axelrod, 1984). If one party’s noncooperative choice were based on this party’s (erroneous) belief that the other is about to make a noncooperative choice, the party is bound to make a noncooperative choice to protect against other’s (assumed) noncooperativeness. This noncooperative decision is, most likely, reciprocated by other, thereby confirming party’s (potentially erroneous) belief (see also De Dreu & Van Kleef, 2002).

C. EGO DEFENSIVENESS Not all bias can be traced to naive realism. In fact, some powerful deviations from what would be normatively correct or objectively accurate can be traced to the fact that humans are motivated to develop, maintain, and protect a positive self-concept and that, as a result, evaluations of the self are positively biased (e.g., Greenwald, 1980; Taylor & Brown, 1988). It manifests itself in a variety of biases that have been shown to interfere with constructive conflict management and integrative negotiation. Examples that have received much attention in the conflict literature are egocentric assessments of oneself and what one deserves (De Dreu, Nauta, & Van de Vliert, 1995; Paese & Yonker, 2001; Thompson & Loewenstein, 1992) and reactive devaluation (Ross & Stillinger, 1991; Ross & Ward, 1995). 1. Self-serving Evaluation In social dilemmas, conflict, and negotiation, individuals evaluate themselves as fairer, more moral, and more honest than the average other

248

DE DREU AND CARNEVALE

(Kramer et al., 1993; Messick, Bloom, Boldizar, & Samuelson, 1985; Rothbart & Hallmark, 1988). Several studies have shown that this selfserving tendency hinders negotiation and the resolution of conflict. Thompson and Loewenstein (1992; see also Paese & Yonker, 2001) showed that negotiators tend to read information about their own and their opponent’s position as suggesting that they deserve more than their opponent. Such egocentric assessment of fairness was related to the duration of impasse. De Dreu, Nauta et al. (1995) examined self-serving evaluation and its consequences in a study involving managers and professional negotiators. Study 1 and 2 revealed that negotiators recalled more constructive and less destructive tactics when asked how they behaved in a recent negotiation, and less constructive and more destructive tactics when asked how their opponent behaved in a recent negotiation. Study 3 revealed that such selfserving recall was related to less integrative negotiation behavior, greater frustration and annoyance, and greater likelihood of future conflict. 2. Reactive Devaluation Reactive devaluation refers to the tendency to reduce the value of a concession or proposal received just because the concession was made or the proposal was offered (i.e., ‘‘they wouldn’t have offered those terms if those terms didn’t advance their interests’’). Ross and Stillinger (1991) describe a study in which American college students were asked to state their opinion about a disarmament proposal, allegedly from the United States, the Soviet Union, or some neutral third party. The terms of this proposal were seen as much more unfavorable (to the United States) when the proposal was attributed to the Soviet Union than when it was attributed to the third party or to the United States. In other words, the same terms in a proposed agreement are viewed entirely differently depending on who proposes it, and are downplayed when being proposed by the adversary.

D. AFTERTHOUGHTS ABOUT NAIVE REALISM AND EGO DEFENSIVENESS Naive realism and ego defensiveness may reflect only one side of the coin. The concepts are useful for explaining the harsh, aggressive, and competitive disputant. The concept of naive realism cannot, however, account for the fact that adversaries sometimes engage in deep and systematic processing of information about the task (De Dreu, Koole, & Steinel, 2000), or other’s communication (Chaiken et al., 2001). Also, it cannot account for the fact

MOTIVATION IN CONFLICT AND NEGOTIATION

249

that negotiators take other’s perspective so as to better understand why the other acts the way he or she does (Kemp & Smith, 1994), and sometimes develop quite accurate perceptions of the structure of the negotiation (De Dreu, Koole, et al., 2000; Thompson, 1995). The concept of ego defensiveness cannot account for the fact that disputants may be inclined to view their opponent as trustworthy (Parks, Henager, & Scamahorn, 1996), see other’s concessions as genuine and deserving a response in kind (Pruitt & Drews, 1969), and seek a fair distribution of outcomes (Ohbutsu & Kameda, 1998; Tyler et al., 1999). In the next two sections, we introduce social motivation and epistemic motivation as two critical moderators of the biases and heuristics reviewed so far. We will review evidence that suggests that social motives moderate the impact of ego defensiveness and naive realism. Further, we review evidence that suggests that epistemic motives moderate the impact of cognitive limitations and naive realism on information processing and strategic choice in conflict and negotiation.

IV. Social Motivation In Leviathan, Thomas Hobbes (1651) pondered why societies and collectivities were able to function at all, if humankind is basically selfinterested. About two centuries later, Adam Smith (1872) sought to solve the Hobbesian problem by his notion of the beneficent ‘‘invisible hand,’’ assuming that private and collective interests tend to correspond rather than conflict. Indeed, Adam Smith basically claimed that collectivities and societies are well-functioning because individuals pursue their self-interest (which in turn enhances collective interest). Decades of theorizing and research within psychology and related disciplines followed, questioning Smith’s idea on two grounds. First, as we have seen in Section II, many settings are structured such that serving one’s immediate, personal interests often harms collective welfare (e.g., Deutsch, 1973; Kelley & Thibaut, 1978; Nash, 1954; Rapoport, 1960; Von Neumann & Morgenstern, 1947). Second, scholars have been questioning whether humans are fundamentally self-interested, or whether they have instead prosocial motives oriented toward the well-being of others (i.e., what Adam Smith called the ‘‘fellow feeling’’) (see, e.g., Batson, 1998; Batson & Shaw, 1991; Caporael, Dawes, Orbell, & Van der Kragt, 1989; Eisenberg & Miller, 1987). This research on social motives falls into one of three classes. First, there is research, some of which we will review below, that considers the sources of egoistic and prosocial motivation in daily life (Batson, 1998) and in organizational

250

DE DREU AND CARNEVALE

settings (Van Dyne & LePine, 1998). Second, there is research dealing with the relationship between particular sources of prosocial motivation, such as empathy, and various instances of prosocial behavior such as donating money to UNICEF (for a review, see Eisenberg & Miller, 1987). Third, and most relevant for current purposes, there is research examining the influence of egoistic and prosocial motives on cooperation in social dilemmas, conflict, and negotiation. At the outset, social motives were defined as the individual’s preference for a particular outcome-distribution between oneself and the opponent (cf. McClintock, 1977). A variety of social motives can be distinguished, including altruistic, competitive, individualistic, and cooperative (MacCrimmon & Messick, 1976). Many studies on social dilemmas, conflict, and negotiation have, however, relied on the more global distinction between selfish and prosocial motivation (e.g., Carnevale & Lawler, 1986; De Dreu & Van Lange, 1995; Van Lange & Kuhlman, 1994; Weingart, Bennet, & Brett, 1993). Selfish motivation comprises both competitive and purely individualistic goals, and prosocial motivation comprises both cooperative and purely altruistic goals. In the case of selfish motivation, individuals try to maximize their own outcomes, and they have no (or negative) regard for the outcomes obtained by their opposing negotiator. Individuals with a selfish motive tend to see the negotiation as a competitive game in which power and personal success are key. Negotiators with a prosocial motive try to establish a fair distribution that maximizes both their own and other’s outcomes; they see the negotiation as a collaborative game in which fairness, harmony, and joint welfare are key. In this section we first review personality characteristics and situations that give rise to selfish or prosocial motives. Subsequently, implications of social motives for strategic choice in negotiation are discussed in terms of three related theories—economic models of bargaining (e.g., Roth, 1994), the theory of cooperation and competition (Deutsch, 1949, 1973), and dual concern theory (Pruitt & Rubin, 1986). Finally, we review recent research on social motives and information processing to argue that social motivation reduces the detrimental influence of ego defensiveness and naive realism in conflict and negotiation.

A. ORIGINS OF SOCIAL MOTIVES Social motives may be rooted in individual differences (Deutsch, 1982; Van Lange, 1999; Van Lange, Otten, De Bruin, & Joireman, 1997). Relevant individual differences variables include social value orientations, differences

MOTIVATION IN CONFLICT AND NEGOTIATION

251

in collectivist or individualistic values, need for affiliation, and Machiavellianism. Social value orientation refers to stable preferences for outcome distributions between oneself and another, unknown person (e.g., McClintock, 1977; Van Lange & Kuhlman, 1994). It is usually assessed through a decomposed game methodology (Kuhlman & Marshello, 1975). Using this methodology, about 50% of the participants in Western societies can be classified as prosocial (i.e., preferring equality and good outcomes for self and other), 40% as individualistic (i.e., preferring good outcomes for oneself while ignoring the other), and 5–10% is classified as competitive (i.e., preferring to gain higher outcomes than the other). Individualistic and competitive participants are often lumped together into one category labeled ‘‘pro-self’’ or ‘‘egoistic’’ (De Dreu & Van Lange, 1995). Social value orientation is a central feature of culture. Following the seminal work by Hofstede (1980) and Triandis (1989), a distinction has been made between collectivist and individualistic culture. The subscale of the IndCol, the predominant measure of cultural values at the level of the individual that assesses ‘‘vertical individualism’’ (Singelis, Triandis, Bhawuk, & Gelfand, 1998), is all about competition, where respondents indicate their agreement with statements such as ‘‘winning is everything.’’ Probst, Carnevale, and Triandis (1999) found positive correlations between the measure of vertical individualism and the measure of pro-self responding in the decomposed game method, and positive correlations between responses on the collectivism subscales of the IndCol and cooperative responding in decomposed games. Similarly, Hulbert, Correa da Silva, and Adegboyega (2001) found that collectivism was positively associated with a tendency to minimize differences between self and an interdependent other. Individualism was negatively correlated to the tendency to minimize differences. Individual differences variables that are linked to social motivation include the need for affiliation (McClelland, Koestner, & Weinberger, 1989), and the Big-Five concept dimension ‘‘agreeableness’’ (Hofstee, De Raad, & Goldberg, 1992). Individuals scoring high rather than low on agreeableness are rated by their peers as lower on competitiveness (Graziano, JensenCampbell, & Hair, 1996), and they achieve lower economic gain in singleissue negotiations (Barry & Friedman, 1998). Archival and laboratory research by Langner and Winter (2001) revealed a positive correlation between the need for affiliation and willingness to concede in a negotiation. Another individual differences variable that is related to social motivation is Machiavellianism. McHoskey (1999) examined the goals and motivational orientation associated with Machiavellianism. Results showed that Machiavellianism was associated with an emphasis on the extrinsic goal of financial success, and on a control motivational orientation in general. Additional results showed that Machiavellianism was positively associated with

252

DE DREU AND CARNEVALE

alienation and antisocial behavior, and negatively correlated with social interest and prosocial behavior. Wilson, Near, and Miller (1998) used a storytelling method to explore aspects of Machiavellianism. They found that individuals who scored high and low on Machiavellianism wrote stories that revealed the cooperative nature of low-Machs and the exploitative nature of high-Machs in particularly sharp focus. Participants who read these stories rejected high-Machs as social partners for most relationships, except when their exploitative skills could be directed against members of other groups. Likewise, Barnett and Thompson (1985) found a negative relationship between Machiavellianism, prosocial reasoning, and prosocial motivation in children. Several studies examined the effects of Machiavellianism in conflict and negotiation. Huber and Neale (1986) studied the negotiation performance and goal setting of novice negotiators. Subjects were randomly assigned to be either buyers or sellers and to attain either an easy, moderate, or difficult goal in a competitive market simulation. Participants assigned difficult goals were more profitable and set harder new goals than those assigned easier goals. Moreover, Machiavellianism had a powerful effect on self-set goals and performance such that high-Machs set higher goals and achieved greater individual profit than low-Machs. In a study on integrative bargaining, Fry (1985) manipulated visual access to assess its effect on the ability of high, low, and mixed Machiavellian bargaining pairs to discover mutually advantageous solutions on a bargaining task. Results showed that visual access interfered with the discovery of jointly profitable solutions in the high-low Mach pairs but had no effect on other dyad combinations. Fry concluded that the personality dynamics of high-low Mach dyads made them particularly vulnerable to low joint bargaining outcomes in face-toface negotiation. Finally, Gunnthorsdottir, McCabe, and Smith (2002) studied the behavior of participants in a two-person one-shot constituent game in which subjects face a choice between trust and distrust, and between reciprocation (trustworthiness) and defection. It was found that Machiavellianism predicted reciprocity. Over one-half of those who score low to average on Machiavellianism reciprocate trust, while high-Machs overwhelmingly defected when it was to their advantage to do so. In addition to individual differences, social motives may be cued by instructions from superiors, reinforcement schemes, or social relationships (e.g., Deutsch, 1973; Tjosvold, 1998). For example, Deutsch (1958) used instructions to induce prosocial and selfish motivation. In the prosocial motive condition, participants were instructed to be concerned about the other’s feelings and welfare, and to see the other as ‘‘partner’’ (cf. Burnham, McCabe, & Smith, 2000). In the selfish motive condition, participants were instructed to disregard the other and to do as well for themselves as possible. Other research used monetary incentives to induce social motives.

MOTIVATION IN CONFLICT AND NEGOTIATION

253

Participants in prosocial motive conditions are told that payment depends on how well they do as a dyad, whereas participants in the selfish motive conditions are told that payment depends on how well they do personally (see, e.g., Deutsch, 1973). Indirect ways to manipulate social motives are to have negotiators anticipate future interaction with their opponent or not (Ben-Yoav & Pruitt, 1984a,b), to emphasize shared versus different group membership (Kramer et al., 1993), or to have them negotiate with a friend versus a stranger (Fry, Firestone, & Williams, 1984). In a metaanalytical review of the studies on social motivation in integrative negotiation, De Dreu, Weingart, and Kwon (2000) contrasted effects on joint outcomes of individual differences (mainly social value orientation), incentives, instructions, and implicit cues (shared group membership, future interaction, friend vs. stranger). All four categories yielded positive and significant effect sizes, indicating that prosocial negotiators achieved higher joint outcomes than selfish negotiators. Importantly, the authors found no significant differences in effect sizes between the four classes. This suggests that for the effects on integrative agreement it does not matter where social motivation comes from.

B. STRATEGIC CHOICE Most research on social motivation has considered its effects on strategic choice. This research has been rooted in one of three dominant theoretical perspectives: Economic models of bargaining, the theory of cooperation and competition (Deutsch, 1949, 1973), and dual concern theory (Pruitt & Rubin, 1986). Traditional economic theories of bargaining assume that individuals try to maximize their individual outcome. In addition, according to this perspective, any apparent prosocial behavior is actually driven by underlying egoistic concerns. The individuals’ persistence in seeking maximum personal gain leads to the dilemma between impasse and seeking acceptable outcomes for all. Because the latter is considered more rational (i.e., some outcomes are better than no outcomes in the case of impasse), the most likely settlement is the solution that satisfies both individuals’ aspirations as much as possible. This solution is the one that fully exploits the available integrative potential. As such, traditional economic theories of bargaining predict that individuals reach integrative agreements regardless of their social motive. They explain failures to realize integrative potential in terms of bias such as fixed-pie perceptions (Schelling, 1960; Thompson, 1990). The theory of cooperation and competition sees social motives as the key to problem-solving behavior and integrative negotiation. It argues that selfish

254

DE DREU AND CARNEVALE

negotiators develop distrust, hostile attitudes, and negative interpersonal perceptions (Kramer & Carnevale, 2001). They use persuasive arguments, positional commitments, threats, bluffs, and coercive power to get their way. Prosocial individuals, in contrast, develop trust, have positive attitudes and perceptions, engage in constructive exchange of information, listen, and seek to understand one another’s perspective. As a result, prosocial motivated negotiators are more likely to uncover possibilities for trade-off and to realize integrative potential (Deutsch, 1973; Tjosvold, 1998). Some evidence for these ideas exists. In their meta-analysis, De Dreu, Weingart, et al. (2000) found that when negotiators had a prosocial rather than selfish motive, they engaged in more problem solving and information exchange, and in less contentious bargaining including the use of persuasive arguments, threats, and positional commitments. Cross-cultural research shows similar patterns for collectivist and individualistic cultural values (Gelfand & Realo, 1999; Gelfand & Christakopoulos, 1999). For example, Gelfand and Christakopoulos (1999) found that negotiators with a collectivist background exchanged more information and, therefore, developed more accurate perceptions of each other’s preferences and priorities than negotiators with an individualistic background. Consistent with the seminal work of Blake and Mouton (1964), Pruitt and Rubin (1986) proposed their dual concern theory. It postulates two kinds of concern, other-concern and self-concern, each ranging in strength from weak to strong. Other-concern is closely related to the concept of social motive, with selfish negotiators having weak other-concern and prosocial negotiators having strong other-concern (Pruitt, 1998). Selfconcern is closely related to ‘‘toughness’’ (Bartos, 1974), resistance to yielding (Kelley et al., 1967), and the negotiator’s intransigence to concession making (Druckman, 1994). As Kelley et al. (1967, p. 382) note: ‘‘This resistance is felt by each party at each point throughout the negotiation session and has implications for his concession-making propensities at each point.’’ In dual concern theory, high (versus low) resistance to yielding is usually operationalized by giving negotiators high (versus low) aspirations or limits below which they should not concede (e.g., Ben-Yoav & Pruitt, 1984a; Pruitt, 1981; Pruitt & Lewis, 1975; Siegel & Fouraker, 1960; Yukl, 1974). A meta-analytical review by Druckman (1994) further showed higher resistance to yielding when negotiators are subject to low (rather than high) time pressure (e.g., Yukl, Malone, Hayslip, & Pamin, 1976), or when they are accountable to constituents rather than to their opposing negotiator (Kramer et al., 1993). Pruitt (1998) reviewed evidence to suggest that negotiators also have high (rather than low) resistance to yielding when prospective outcomes from the negotiation are framed as losses rather than

MOTIVATION IN CONFLICT AND NEGOTIATION

255

gains (e.g., Bottom & Studt, 1993; Carnevale & De Dreu, 2002; De Dreu et al., 1994). Like the theory of cooperation and competition, dual concern theory predicts that when parties have a prosocial rather than a selfish motive they are likely to develop positive interpersonal attitudes and perceptions, to seek understanding of other’s point of view, and to make and reciprocate concessions. However, dual concern theory predicts that differences in behavior and outcomes depend on the negotiators’ level of resistance to yielding. When a prosocial motive is paired with low resistance to yielding, unilateral concession making dominates and parties either accept the other party’s demands or settle on an (obvious) fifty–fifty compromise. When a prosocial motive is paired with high resistance to yielding, however, parties face the dilemma of wanting good outcomes for the other, but not at their own expense. As a result, they concede slowly and engage in various kinds of problem solving (such as exchanging information) that promote the discovery and development of integrative solutions. In a narrative review of the evidence, Carnevale and Pruitt (1992, p. 540) concluded that ‘‘The dual-concern model fits data from (. . .) experimental studies.’’ Likewise, Thompson (1990) concluded that the prediction that high resistance to yielding coupled with high concern for the other party [i.e., a prosocial motive] leads to higher joint outcomes has generally been supported. The meta-analysis by De Dreu, Weingart, et al. (2000) mentioned earlier provided a quantitative evaluation of the theory. The authors concluded that with regard to problem-solving behavior, contentious behavior, and integrative agreements, dual concern theory makes good predictions—social motives affect negotiation behavior and outcomes, but more when resistance to yielding is high rather than low. This is important because accepting that Dual Concern Theory is valid suggests a need to revise traditional economic models of bargaining that assume negotiators are purely self-interested, as well as the theory of cooperation and competition that assumes resistance to yielding is irrelevant. Although the support for Dual Concern Theory is good, the theory fails to deal with the errors and biases in human information processing that affect negotiation and dispute resolution. Below, we will consider in more detail how social motives moderate errors and biases associated with ego defensiveness and naive realism, respectively.

C. NAIVE REALISM RECONSIDERED Research on social motives has traditionally considered strategic choice and outcomes from the negotiation. Recent studies shifted the emphasis

256

DE DREU AND CARNEVALE

toward cognitive processes. Several studies suggest that social motives not only influence strategic choice, but also the influence of naive realism. Naive realism is reflected in, among other things, fixed-pie perceptions. A study by Gelfand and Christakopoulou (1999) shows that fixed-pie perceptions are revised during negotiation, especially when negotiators come from a collectivist rather than individualistic culture. When negotiators were Greek (a collectivist culture), they exchanged more information about their preferences and priorities, and thus developed more accurate understanding of the task, than when negotiators were American (an individualist culture). In a similar vein, De Dreu, Koole et al. (2000) showed that revisions in fixed-pie perception during the negotiation were moderately correlated with (self-reported) social motives. The more prosocial motivation negotiators reported, the more they revised their fixed-pie perceptions during negotiation. Carnevale and De Dreu (2002) reported additional evidence that social motives moderate bias due to naive realism. They focused on optimistic overconfidence. In these studies, participants played the role of seller in a simulated bilateral negotiation; a computer controlled the behavior of the opposing negotiator. Half were given an individualistic orientation, and the other half were given a cooperative orientation. The position materials were identical in both conditions and the behavior of the opposing negotiator was held constant. Prior to the end of negotiation, they were asked to estimate the strength of their position relative to the other negotiator’s position, on a five-point scale, from ‘‘I have a much stronger case’’ (1) to ‘‘Both of our positions are equally strong’’ (3) to ‘‘The other negotiator has a much stronger case’’ (5). The negotiators with the individualistic motive rated their position as significantly stronger (M ¼ 2.54) than the opposing negotiator, relative to negotiators with the cooperative motive (M ¼ 3.03). Note that the negotiators with the cooperative motive were virtually at the center on this scale, at level 3, indicating they felt that both side’s positions were equally strong. This suggests that overconfidence may be less of a problem when negotiators adopt a cooperative motive than when they adopt an individualistic motive.

D. EGO DEFENSIVENESS RECONSIDERED There is evidence that social motivation moderates the impact of ego defensiveness. Camac (1992) studied how social value orientation influenced the recall of the values in a prisoner’s dilemma game. Participants were classified as prosocial or selfish, and they subsequently played a prisoner’s dilemma game. Afterward they were shown the dilemma again, with their

MOTIVATION IN CONFLICT AND NEGOTIATION

257

own and other’s payoffs being eliminated, and they were asked to provide as good an estimate of the payoffs as possible. Results showed that accuracy of recall was consistent with the individual’s social value orientation—people with a prosocial value orientation recalled the joint gain in four cells quite accurately, whereas people with a selfish orientation recalled their own gain in the four cells best. This suggests that social value orientation biases attention, encoding, and retrieval processes. A similar conclusion derives from a negotiation study by De Dreu and Boles (1998). They measured individuals’ social value orientation and classified individuals as prosocial or as selfish. Participants believed they would enter a negotiation and they prepared for it by reading the instructions for the standard integrative negotiation task discussed above. At the end of these instructions a list of decision heuristics was presented. The list contained a series of cooperative heuristics and a series of competitive heuristics. Examples of cooperative heuristics are ‘‘share and share alike’’ and ‘‘the other deserves the benefit of the doubt.’’ Examples of competitive heuristics are ‘‘never trust your opponent,’’ and ‘‘your gain is my loss.’’ Participants subsequently engaged in a set of unrelated tasks and then were asked to recall as many decision heuristics as they could. Results showed that prosocial negotiators recalled more cooperative than competitive heuristics, whereas selfish negotiators recalled more competitive than cooperative heuristics. The data reported by Camac (1992) and De Dreu and Boles (1998) dealt with social value orientations that were measured rather than manipulated. Therefore, causality cannot be established and effects may have been driven by unknown third variables. To address this, De Dreu (unpublished observations) conducted an experiment in which social motives were manipulated rather than measured. The methods, materials, and procedures were identical to those used by De Dreu and Boles, except that the measurement of social value orientations was replaced by instructions intended to manipulate social motives. Social motives were manipulated as in previous studies (e.g., Carnevale & Lawler, 1986; De Dreu & McCusker, 1997). One-half of the participants were informed that the more personal profit they would make in the upcoming negotiation, the higher would be their chances of winning a $50 bonus (selfish motive condition). The other half were informed that the more profit they and the other party would make in the upcoming negotiation, the higher would be their chances of winning a $50 bonus (prosocial motive condition). The number of cooperative and competitive heuristics correctly recalled, corrected for the total number of heuristics recalled, was analyzed (total recall did not differ as a function of social motivation). The significant interaction between social motive and type of heuristic recalled is presented

258

DE DREU AND CARNEVALE

Fig. 1. Recall of cooperative and competitive decision heuristics as a function of social motive.

in Figure 1. As can be seen, negotiators with a prosocial motive recalled more cooperative and less competitive heuristics, while negotiators with a selfish motive did the reverse. This result is very similar to that obtained by De Dreu and Boles, and rules out doubts about directionality. It clearly shows that social motivation biases encoding and retrieval processes. Another indication that social motives influence individual level cognition derives from a series of studies by Carnevale and Probst (1998). They found that expectations of conflict produced a ‘‘freezing’’ of cognitive schemas. The expectation of conflict reduced general problem-solving abilities, and it restricted categorizations. However, there was an interesting twist. In nonconflict settings, prosocial individuals, compared to competitive and individualistic people, had more inclusive cognitive categories; but competitive people were most likely to alter their thinking in response to the social context. When the social context required competition with others, competitors had low cognitive flexibility. But when competitors were in a context that required cooperation (a need to work with their partner in a between-group competition), they had the greatest cognitive flexibility. Moreover, the social context (cooperation or conflict) influenced how people processed information unrelated to the circumstances that defined the context. In other words, cognitive rigidity in conflict is not simply holding firm, or a reflection of the desire to ‘‘look tough’’ in negotiation. Rather, it suggests an unintentional alteration in underlying cognitive organization in cooperative and conflict situations. Van Kleef and De Dreu (2002) examined how social motives influence information search. In two experiments they studied the questions negotiators ask their opponent as a function of beliefs about the opponent’s personality (cooperative, competitive, unknown) and the negotiator’s own social value orientation (prosocial vs. selfish). In Experiment 1 participants

MOTIVATION IN CONFLICT AND NEGOTIATION

259

selected questions about other’s cooperation or other’s competition from a prewritten list, and in Experiment 2 participants generated questions themselves and these were coded as dealing with the other’s cooperative or competitive tendency. Social motives did not influence the amount of questions asked. However, negotiators engaged in confirmatory search, such that they asked about competition when the other was believed to be competitive and about cooperation when the other was believed to be cooperative. When other’s goals were unknown, negotiators relied on their own orientation—selfish negotiators asked about competition and prosocial negotiators asked about cooperation. The studies reviewed thus far strongly suggest that individuals in conflict and negotiation search for, encode, and retrieve information consistent with their prosocial or selfish motivation. Also, prosocial and selfish negotiators engage in confirmatory information search likely to support their initial assumptions and beliefs. All this suggests that social motive moderates ego defensiveness—selfish negotiators fall prey to motivational biases that strengthen their egocentric tendencies, whereas prosocial individuals appear geared toward equality, consensus, and joint gain. More direct evidence that social motive moderates ego defensiveness derives from a cross-cultural study by Gelfand and colleagues (2001). In three studies they investigated differences in egocentric perceptions of fairness in conflict and negotiation in the United States and Japan. Using a free recall methodology (Messick et al., 1985), Study 1 showed that individualists associated themselves with fair behaviors and others with unfair behaviors to a much greater extent than collectivists. In a study of naturally occurring conflicts, Study 2 illustrated that individualist disputants perceived that an objective third party would view their behavior as much more fair than their counterparts, and this bias was greatly attenuated in collectivist Japan. Finally, a laboratory study of negotiation showed that self-serving biases of fairness were much greater among individualist than collectivist negotiators, and demonstrated that such differences were related to lower negotiation outcomes among individualist negotiators. Thus, ego defensiveness and all it engenders in the case of conflict and negotiation appears to be less of an issue when prosocial and collectivist values dominate. Other evidence that social motive can moderate ego defensiveness comes from research by Carnevale and colleagues on the mere ownership and endowment effects. In these studies, mere ownership—the tendency to overvalue what is in one’s possession—was examined for participants with collectivist and with individualistic values. They replicated the standard individual ownership effect (Beggan, 1992) and endowment effect (Kahneman, Knetsch, & Thaler, 1990) among individualists, but not for collectivists. Collectivists showed a group endowment effect (an effect for

260

DE DREU AND CARNEVALE

property owned by the group), but not an individual endowment effect (see Carnevale, 1995).

E. CONCLUSIONS ABOUT SOCIAL MOTIVES Taken together, research on social motives in conflict and negotiation allows for four conclusions. First, social motives are rooted in individual differences or they are triggered by situational cues, but the origins of social motives do not moderate the effects social motives have in conflict and negotiation. Second, prosocial negotiators engage in more cooperative behavior and exchange more accurate information about preferences and priorities than selfish individuals. Third, naive realism and the biases it entails have less impact on negotiation processes and outcomes when disputants have a prosocial rather than selfish motivation. Fourth, and finally, ego defensiveness and the biases it entails have less impact on negotiation processes and outcomes when disputants have a prosocial rather than selfish motivation. These four conclusions are represented graphically in Figure 2. The top of the figure shows the antecedents to prosocial motivation: at the left are listed person-related, individual differences variables such as social value orientation, collectivist values, need for affiliation, and low levels of Machiavellianism. To the right, we have situation cues that trigger prosocial motivation: cooperative incentives, instructions to cooperate, a history of cooperative interaction, or the expectation of future cooperation, and a shared categorization. The figure further shows that prosocial motivation stimulates a problem-solving approach, including the sharing of information about preferences and priorities; as predicted by dual concern theory (Pruitt & Rubin, 1986), this effect of prosocial motivation on problem solving is strengthened by resistance to yielding. Finally, the figure shows the key notion that prosocial motivation, through its effect on problem solving and exchange of information, reduces the negative impact ego defensiveness and naive realism has on integrative negotiation.

V. Epistemic Motivation ‘‘Continuous effort—not strength or intelligence—is the key to unlocking our potential.’’ —Sir Winston Churchill

MOTIVATION IN CONFLICT AND NEGOTIATION

Person; Individual Differences Pro-social Values Collectivist Values High Need forAffiliation Low Machiavellianism

+

261

Situation Cues: Coop. Incentives Coop. Instructions Coop. Past or Future Shared Categorization

Pro-social Motivation + Resistance to Yielding

+

Problem Solving And Information Exchange -

-

Ego Defensiveness Integrative Negotiation Naive Realism Fig. 2. Antecedents and consequences of social motives for information processing and integrative negotiation.

Thus far we have discussed a variety of heuristics and biases that hinder constructive negotiation and dispute resolution, and we have reviewed evidence showing that social motivation drives the type of information people search and process, thus moderating the detrimental impact of ego defensiveness and naive realism. We have not considered the possibility that individuals may differ, across individuals and across situations, in the extent to which they process available information. Thus far, we have implicitly assumed that individuals process information available, and search for information consistent with the social motivation. However, according to dual process models of human information processing (Chaiken & Trope, 1999; Smith & DeCoster, 2001), individuals can choose from two alternative strategies for processing information. The first strategy is to solve logical problems, to evaluate persuasive arguments, or to form impressions of others through a quick, effortless, and heuristic processing of information that rests on well-learned prior associations. Alternatively, individuals may engage in more effortful, deliberate, and systematic processing that involves rule-based inferences. Thus, to deal with

262

DE DREU AND CARNEVALE

their cognitively and emotionally taxing situation, negotiators can use a ‘‘quick and dirty’’ strategy when processing information, arriving at usually reasonable approaches to the situation efficiently and effortlessly (Neale & Bazerman, 1991). Alternatively, they can try hard to think deeply about the situation, sometimes arriving at qualitatively different approaches than the ‘‘quick and dirty’’ strategy would have suggested. They may engage in deliberate and systematic processing of information about their opposing party’s preferences and priorities, and discover that agreements exist that integrate their own and the other party’s interests (De Dreu, Koole et al., 2000). The extent to which individuals search, encode, and retrieve information depends on their epistemic motivation—the desire to develop and hold accurate and well-informed conclusions about the world. In this section we first review personality characteristics and situations that give rise to low or high epistemic motivation. Subsequently, research is discussed to argue that epistemic motivation helps people overcome their cognitive limitations, and tempers the influence of naive realism in conflict and negotiation.

A. ORIGINS OF EPISTEMIC MOTIVATION Epistemic motivation relates to (lack of) need for cognitive closure, which is central in lay-epistemic theory developed by Kruglanski and associates (Kruglanski, 1989; Kruglanski & Ajzen, 1983; Kruglanski & Webster, 1996; Webster & Kruglanski, 1994). This theory argues that there exists a single dimension underlying the desire for different kinds of knowledge, termed ‘‘need for cognitive closure.’’ Individuals at the high need for closure end of the continuum are characterized by considerable cognitive impatience, leaping to judgment on the basis of inconclusive evidence and rigidity of thought. At the other end of the continuum, individuals with low need for closure may prefer to suspend judgment, engaging in extensive information search and generating multiple interpretations for known facts. Compared to their low need for closure counterparts, individuals high in need for closure display stronger primacy effects in impression formation, are more prone to the correspondence bias, rely more on existing stereotypic knowledge, and react more negatively to people opposing group consensus (Kruglanski & Webster, 1996, for a review). These findings offer converging evidence for the notion that individuals with high need for cognitive closure may be more likely to use cognitive heuristics in making judgments and decisions than individuals with low need for cognitive closure. The latter instead postpone judgment until they have processed as much information as possible, or until time and energy is depleted.

MOTIVATION IN CONFLICT AND NEGOTIATION

263

Epistemic motivation is also related to need for cognition (Petty & Cacioppo, 1986). Need for cognition refers to the intrinsic motivation for, and enjoyment of, effortful cognitive activities. Individuals with high need for cognition are more attentive to the quality of the arguments in a persuasive communication, and they are less likely than individuals with low need for cognition to be influenced by peripheral cues such as the attractiveness of a communicator (e.g., Petty & Cacioppo, 1986). In a recent study concerning dyadic decision making, members with high need for cognition were viewed by their discussion partners as being more effective persuaders, and as generating more valid arguments than their low need for cognition counterparts (Shestowsky, Wegener, & Fabrigar, 1998). Although need for cognitive closure and need for cognition are individual differences variables affecting epistemic motivation, situational cues may influence epistemic motivation as well. Research suggests that epistemic motivation increases when the task becomes more involving (Mayseless & Kruglanski, 1989), or when there is process accountability (Tetlock, 1992). Under process accountability, individuals expect to be observed and evaluated by others with unknown views about the process of judgement and decision making (Lerner & Tetlock, 1999; Tetlock, 1992). Simonson and Staw (1992) argued that individuals under process accountability tend to engage in preemptive self-criticism, leading to more evenhanded evaluation of decision alternatives and reduced need for self-justification. Indeed, a review of the relevant research suggests that ‘‘accountability attenuated bias on tasks to the extent that (a) suboptimal performance resulted from lack of self-critical attention to the judgmental process and (b) improvement required no special training in formal decision rules, only greater attention to the information provided’’ (Lerner & Tetlock, 1999, p. 263). In addition to task involvement and process accountability, epistemic motivation depends on the individual’s relationship with the other, including relative power and sense of interdependence. Fiske (1992) has argued that low power individuals (re)gain control over their situation by paying close and careful attention to their powerful other, so as to accurately predict other’s needs, desires, and possible actions. Several studies in the realm of person perception indeed support this view (e.g., Goodwin, Gubin, Fiske, & Yzerbyt, 2000). Gelfand and Christakopoulou (1999) suggest that interdependence can induce a person to be motivated to accurately predict other’s needs, desires, and possible actions, akin to low power individuals. Epistemic motivation also depends on the individual’s level of aspiration. When an individual has a high rather than low aspiration level more prospective outcomes fall below the person’s aspiration and are framed as losses. Framing outcomes as losses increases the tendency to think, as

264

DE DREU AND CARNEVALE

indicated by greater time needed for decision making (De Dreu, Emans, & Van de Vliert, 1992), and more systematic information processing (e.g., Dunegan, 1993). In a similar vein, Kimmel, Pruitt, Magenau, KonarGoldband, and Carnevale (1980) found that higher aspirations produced greater search for and provision of information including competitive statements and problem-solving activities. Thus, it appears that high aspirations induce more epistemic motivation than low aspirations. Finally, sometimes the context undermines epistemic motivation. For example, time pressure reduces epistemic motivation and increases the motivation to reach (cognitive) closure. Kruglanski and Freund (1983) reviewed evidence to suggest that time pressure intensifies the tendency to seek cognitive closure and they showed that individuals are more likely to fall prey to primacy effects, ethnic stereotyping, and numerical anchoring when time pressure is high rather than low. These results suggest that individuals in negotiation may be less likely to engage in systematic information processing when there is high rather than low time pressure because increased time pressure increases the need for cognitive closure. Similar effects may be expected when there is a high level of noise or when individuals become fatigued (for a review, see Kruglanski & Webster, 1996).

B. BIAS DUE TO COGNITIVE LIMITATIONS RECONSIDERED Above, we argued that epistemic motivation affects information processing in negotiation such that higher levels of epistemic motivation lead to more deliberate and more sophisticated information processing. Put differently, it can be predicted that the negotiator’s tendency to rely on cognitive heuristics is moderated by his or her epistemic motivation. Results reported by De Dreu, Koole, and Oldersma (1999) are consistent with this general prediction. These authors examined the influence of need for cognitive closure on anchoring-and-adjustment (Experiment 1) and stereotyping (Experiment 2) in negotiation. In Experiment 1, participants in the high (low) anchor condition were told, prior to negotiation, that some participants in earlier sessions had reached agreements that valued around 11,000 (3,000) points (with 14,000 being the maximum). Results showed that participants made more concessions in the case of low anchor values, but only when they had high need for cognitive closure. In Experiment 2, psychology students were told that their opponent majored in business or in theology. Psychology students hold more competitive beliefs about business majors than about religion majors (De Dreu, Yzerbyt, et al., 1995). Results showed, accordingly, that participants made smaller concessions when the

MOTIVATION IN CONFLICT AND NEGOTIATION

265

opponent was a business student, and this tendency was stronger when participants had high rather than low need for cognitive closure. Recent work by Galinsky and Mussweiler (2001) provides another indication that motivating negotiators to consider information carefully moderates their tendency to use heuristic cues in judgment and decision making. These authors conducted three experiments that explored the role of first offers, perspective-taking, and negotiator self-focus in determining distributive outcomes in a negotiation. Across three experiments, whichever party made the first offer obtained a better outcome, suggesting that the first offer in a negotiation provides an important anchor value. In addition, first offers were a strong predictor of final settlement prices. Interestingly, when the negotiator who did not make a first offer focused on information that was inconsistent with the implications of their opponent’s first offer, the advantageous effect of making the first offer was eliminated. Thinking about one’s opponent’s alternatives to the negotiation, one’s opponent’s reservation price, or one’s own target, all negated the effect of first offers on outcomes. Thus, anchoring processes in negotiation are eliminated when negotiators are stimulated to think carefully before proceeding with making offers and counteroffers. We have argued that time pressure and fatigue reduce epistemic motivation and increase reliance on heuristics and erroneous reasoning. De Dreu (2002) tested this idea. In Experiment 1, time pressure (high versus low) was crossed with information about the opponent’s group membership (business student vs. religion student). Individuals placed higher demands when they faced a business rather than a religion student, especially when time pressure was acute rather than mild. In Experiment 2, time pressure was either high or low, and prior to and immediately following the negotiation fixed-pie perceptions were measured. Results showed more accurate perceptions (i.e., less fixed-pie error) at the end of the negotiation and more integrative agreements when time pressure was mild rather than acute.

C. NAIVE REALISM RECONSIDERED The experiments so far examined the extent to which epistemic motivation moderates the reliance on cognitive heuristics provided at the outset of the negotiation. However, the study on time pressure by De Dreu (2002) also suggests that epistemic motivation may reduce the impact of naive realism, as exemplified by (revisions in) fixed-pie perceptions. A study by De Dreu, Koole et al. (2000) considered this possibility in more detail. It was predicted that epistemic motivation helps negotiators to

266

DE DREU AND CARNEVALE

release their fixed-pie perceptions. In two experiments, epistemic motivation was manipulated by holding negotiators accountable for the manner in which they negotiated or not (Lerner & Tetlock, 1999). Results showed that process accountability reduced fixed-pie perceptions during face-to-face negotiation, and produced more integrative agreements. Also, the accountable negotiators revised their fixed-pie perceptions even when information exchange was held constant; this indicates that the effects were due to the systematic processing of information rather than to a more elaborate search for new information. De Dreu and Van Kleef (2002) provide further evidence for this possibility. In this study, the influence of power on information search strategies was examined. Power was hypothesized to influence epistemic motivation, with low power negotiators having higher epistemic motivation than their high power counterparts (see also Fiske, 1992). High and low power negotiators were, prior to the negotiation, asked to generate questions they would like to ask their opponent, who was displayed as being cooperative or competitive. Results showed that negotiators engaged in less confirmatory search, as indicated by the number of diagnostic rather than leading questions they asked when participants had low power, and the other was seen as competitive. Finally, Kemmelmeier and Winter (2000) had participants read speeches given by Bill Clinton and Saddam Hussein shortly after Iraq military forces had attacked Kurd villages in northern Iraq in 1996. Participants were asked to evaluate these speeches in the role of a military officer or in the role of UN mediator. The idea was that participants in the role of a military officer would adopt a partisan mind-set and view the conflict from the U.S. perspective only. Participants in the role of UN mediator, in contrast, would adopt an accuracy mind-set, and view the conflict from a more distant perspective that allowed multiple views to come into play. Results showed that perceptual distortion was greater when participants had a partisan rather than accuracy mindset. Thus, this study shows that different mindsets influence perceptual distortion in conflict and, thereby, can reduce naive realism.

D. CONCLUSIONS ABOUT EPISTEMIC MOTIVATION The research on epistemic motivation in conflict and negotiation allows for three conclusions. First, epistemic motivation is rooted in individual differences or it is triggered by situational cues. Second, individuals with high epistemic motivation engage in more systematic processing of information and are less influenced by a variety of cognitive heuristics,

MOTIVATION IN CONFLICT AND NEGOTIATION

Person; Individual Differences: Low Need for Closure High Need for Cognition +

267

Situation Cues: Process Accountability Power Disadvantage High Task Involvement High Limit/Aspiration Mild Time Pressure Low Noise & Fatigue

Epistemic Motivation +

Systematic Information Processing and Search

-

-

Cognitive Limitations Integrative Negotiation Naive Realism Fig. 3. Antecedents and consequences of epistemic motives for information processing and integrative negotiation.

including inappropriate anchoring-and-adjustment and stereotyping. Third, negotiators with high epistemic motivation appear less subject to naive realism, because they engage in less confirmatory information search, and revise their fixed-pie assumption more readily. As Figure 3 shows, these processes are expected to result in accurate understanding of the negotiation task, which enhances the likelihood of integrative agreements. The top of the figure shows the antecedents to epistemic motivation: To the left we have listed person-related, individual differences variables such as low need for cognitive closure and high need for cognition. To the right, we have listed situation cues that trigger epistemic motivation: process accountability, a low power position, high task involvement, high limits and aspirations, low levels of time pressure, and low levels of fatigue. The figure further shows that epistemic motivation stimulates deliberate, effortful, and systematic processing of information that is or becomes available during the negotiation. Finally, the figure shows the key notion that epistemic motivation, through its effect on systematic

268

DE DREU AND CARNEVALE

information processing, reduces the negative impact cognitive limitations and naive realism on integrative negotiation.

VI. Motivated Information Processing Model of Conflict and Negotiation The work we have reviewed thus far reveals that motivational and cognitive processes often conspire against constructive conflict management and integrative negotiation. But it also reveals that sometimes motivation facilitates a rather constructive and cooperative interaction conducive to integrative agreement. In Figure 4 we summarize this main idea. As can be seen, ego defensiveness, naive realism, and reliance on cognitive heuristics are supposed to reduce the likelihood of integrative negotiation (Neale & Bazerman, 1991; Ross & Ward, 1995). The top part of the figure summarizes our thesis with regard to the effects of prosocial motivation.

Pro-social Motivation +

Resistance to Yielding

+

Information Exchange -

Ego Defensiveness -

Naive Realism

Cognitive Limitations

Integrative Negotiation -

Systematic information processing + Epistemic Motivation Fig. 4.

Motivated information processing model of negotiation.

MOTIVATION IN CONFLICT AND NEGOTIATION

269

In interaction with resistance to yielding, prosocial motivation increases the likelihood of a problem-solving approach and high levels of information exchange. A problem-solving approach, in turn, reduces the detrimental effects of ego defensiveness and of naive realism. The bottom part of the figure summarizes our thesis with regard to the effects of epistemic motivation. Because it stimulates the deliberate and systematic processing of information, epistemic motivation is expected to reduce the detrimental consequences of cognitive limitations and of naive realism. Because we have no empirical evidence so far, we do not hypothesize that prosocial motivation reduces the impact of cognitive limitations, or that epistemic motivation reduces the role ego defensiveness plays in integrative negotiation. From Figure 4 it follows that social and epistemic motives appear critical in determining whether conflict escalates into hostile and counterproductive interaction, or instead runs a relatively constructive course with integrative agreements being a likely result. In this section, we develop this idea further and introduce four archetypes of disputants—prosocial thinkers, prosocial misers, selfish thinkers, and selfish misers. Also, we discuss in more depth the interplay between personality and situation variables, and we consider the possibility that motives change during conflict interaction, and converge at the interpersonal level.

A. INTERPLAY AMONG MOTIVES: TOWARD FOUR ARCHETYPES OF DISPUTANTS IN CONFLICT Throughout our analysis and review, we assumed that social and epistemic motivation were independent. That is, a negotiator may be more or less motivated to engage in deliberate and deep processing of information as well as more or less motivated to seek good outcomes for themselves only, or for themselves as well as for the opposing party. The independence assumption holds up empirically. In three different samples, De Dreu et al. (1999, Study 3) examined the relationship between need for cognitive closure and social value orientation. Need for closure was assessed with the Webster and Kruglanski (1994) questionnaire, and social value orientation was assessed using the Kuhlman and Marshello (1975) decomposed game measure. Results showed that need for cognitive closure did not differ as a function of social value orientation in either sample. Comparing the means for prosocials and selfish participants in a meta-analysis, an overall effect size of Cohen’s d ¼ .21 was found. This effect size was small and not significant, and it also appeared that the effect size did not differ across samples.

270

DE DREU AND CARNEVALE

Similar conclusions were reached with regard to situation-based epistemic motivation. In two experiments, De Dreu, Koole et al. (2000) manipulated process accountability in a face-to-face integrative negotiation task. Manipulation checks confirmed that negotiators under process accountability had higher levels of epistemic motivation than negotiators not held accountable. In both experiments, process accountability did not influence self-reported social motivation. Thus, it appears safe to conclude that situation-based epistemic motivation is unrelated to measures of social motivation. To say that social and epistemic motivation are independent is not to say that the two have independent effects on strategic choice and the quality of agreements. In fact, the combination of social and epistemic motivation results in four conflict archetypes. These archetypes—the prosocial thinker, the prosocial miser, the selfish thinker, and the selfish miser—derive from the combination of prosocial versus selfish motivation, and high versus low epistemic motivation (see Figure 5). In analogy to the cognitive miser (Fiske & Taylor, 1991), the selfish miser relies on competitive heuristics such as ‘‘your gain is my loss’’ and ‘‘an eye for an eye (. . .).’’ The opposing individual is seen as an opponent who cannot be trusted. He or she engages in ‘‘black–white’’ thinking (Carnevale & Probst, 1998; Judd, 1978), and prominent behavioral options are threats, bluffs, and lies, and failure to reciprocate other’s concessions. The cooperative miser, in contrast, relies on cooperative heuristics such as ‘‘share and share alike’’ and ‘‘other deserves the benefit of the doubt.’’ The opposing negotiator is seen as a partner who deserves good outcomes as well. Prominent behavioral options are unconditional promises, large unilateral concessions, and search for easy fifty–fifty compromises. High epistemic motivation produces thinkers, but what they think about and what information they have to think about is largely determined by their social motivation. Thus, the selfish thinker provides, searches, and stores and retrieves information that helps him or her to further the goal of obtaining good outcomes for oneself, or better outcomes than the other side. The other negotiator is seen as an opponent, who cannot be trusted. Prominent behavioral options are heuristic trial-and-error—making concessions across issues that overall provides one with equal value (Pruitt & Lewis, 1975)—advancing persuasive arguments about why the other should give in, misrepresentation and deception, and so on. The prosocial thinker, in contrast, searches and stores and retrieves information that helps him or her to further the cooperative goal of reaching agreement. The other negotiator is seen as a partner who deserves to be trusted. Prominent behavioral options are providing and asking information about preferences and priorities, making concessions and proposals for settlement, and searching for mutually attractive solutions.

MOTIVATION IN CONFLICT AND NEGOTIATION

High

Selfish Thinker

Pro-social Thinker

Low

Selfish Miser

Pro-social Miser

271

Epistemic Motivation

Pro-social

Selfish Social Motivation

Fig. 5. Four archetypes of individuals in conflict as a function of social and epistemic motivation.

De Dreu, Beersma, Stroebe, and Euwema (2002) tested the interaction of social and epistemic motivation in three experiments crossing social motives (prosocial versus selfish) and epistemic motivation (high versus low). They argued that prosocial (selfish) negotiators tend toward information consistent with their cooperative (competitive) goals and expectations, and the impact of this selective tendency is stronger the more negotiators are motivated to engage in deep and thorough processing of information. This suggests the hypothesis that social motivation affects negotiator cognition and integrative agreements more under high rather than low levels of epistemic motivation. In the study by De Dreu et al. (2002), social motives were manipulated by describing the other party, in the instructions, as either ‘‘partner’’ (prosocial motive condition) or as ‘‘opponent’’ (selfish motive condition). Epistemic motivation was induced by making participants process accountable or not (see De Dreu, Koole et al., 2000). The first experiment by De Dreu et al. (2002) used a recall task previously used by De Dreu and Boles (1998). Results showed that the pattern of results displayed in Figure 1 was more pronounced under high rather than low levels of epistemic motivation. Subsequent experiments used the integrative negotiation task-methodology described in Section II, and results showed that prosocial negotiators engaged in more integrative negotiation, and reached more integrative agreements than did selfish negotiators, but only under high rather than low levels of epistemic motivation. These results together support the

272

DE DREU AND CARNEVALE

hypothesis—prosocial negotiators achieve more integrative agreements than selfish negotiators, but only under high levels of epistemic motivation. Put differently: Only prosocial thinkers are likely to attain high joint outcomes.

B. PERSON X SITUATION: HOW USEFUL ARE INDIVIDUAL DIFFERENCES? As indicated in Figures 2 and 3, some precursors to motivation are rooted in the person (e.g., need for cognition, need for affiliation), and some of them are due to the situation (e.g., power differences, cooperative incentives). Although this is consistent with the Lewinian notion that human behavior is a function of personality, the situation, and the interaction between the two, the place accorded to individual differences in the current analysis contrasts with a consensual tradition in the negotiation literature to negate and whenever possible downplay the importance of personality. For example, Thompson (1990, p. 520) states that ‘‘The scant number of clear relationships observed in this review is consistent with [the] conclusion that there are few significant relationships between personality and negotiation outcomes.’’ Similar conclusions were reached by Rubin and Brown (1975), Carnevale and Pruitt (1992), and more recently by Bazerman et al. (2000). In contrast to the rather pessimistic conclusions reached by reviewers of the negotiation literature, we have been quite successful in obtaining predicted effects of individual differences in social value orientation, in collectivism, or in need for cognitive closure. Perhaps in past research there has been a misfit between the personality variable measured and the dependent variable considered. For example, there is no strong argument as to why the need for cognitive closure should affect concession making, but there is a strong argument as to why it should affect information processing. Thus, although there may be few relationships between individual differences and negotiation, there may be many relationships between certain individual differences and certain aspects of negotiation. Individual differences are also more likely to affect cognition and behavior in weak situations—situations without clearly defined and established norms (Barry & Friedman, 1998; Kenrick & Funder, 1988). Many conflict and negotiation studies are conducted with MBA students. In most MBA classes there may be strong (competitive) beliefs and norms associated with negotiation (De Dreu & McCusker, 1997; Kasser & Ahuvia, 2002), something that may be less the case in the social psychological laboratory where students participate relatively anonymously.

MOTIVATION IN CONFLICT AND NEGOTIATION

273

C. SHIFTS IN MOTIVATION Motivational goals are not constant and change as the negotiation cycles through different phases (Olekalns, Smith, & Walsh, 1996). For example, social motivation may change when incentive structures are revised, when the likelihood of future interaction changes, or when one comes to like or dislike the opponent. Epistemic motivation may change when time pressure increases or when parties become fatigued. Also, epistemic motivation may change as a function of impasses in negotiation. Negotiators often begin with distributive behavior and, when a costly impasse looms, switch to more integrative strategies (Brett, Shapiro, & Lytle, 1998; Harinck et al., 2000; Walton & McKersie, 1965). Zartman (1991) refers to ripeness as the point in time at which conflict parties prefer negotiated agreement over the (conflict-laden) status quo. This work suggests that when parties impasse, they reflect upon their current (competitive) strategies, realize these lead nowhere, and that they need to better understand the task and their opponent’s preferences and priorities. One important source of change in motivation is the opponent’s behavior. In fact, it is quite likely that disputants with different levels of social or epistemic motivation converge toward each other in the course of conflict interaction. To reach agreement, parties need to understand each other and, therefore, they are motivated to converge on the same negotiation script (De Dreu et al., 1994; Richter & Kruglanski, 1999). Also, mental representations are intimately intertwined with, and reflected in both verbal and nonverbal communication (De Dreu et al., 1994), and emotions may ‘‘catch on’’ and are adopted by interdependent others through contagion-like processes (Hatfield, Cacioppo, & Rapson, 1992; Morris & Keltner, 2000). Finally, interdependent individuals have a strong tendency to reciprocate other’s behavior (Gouldner, 1960; Weingart, Thompson, Bazerman, & Carroll, 1990), and through self-perception this may alter previously held notions about the world. Research has revealed such convergence in goals and concomitant motives. A study by Pinkley and Northcraft (1994) showed that negotiators who start with a different representation of the conflict than their opponent converged in the course of the interaction toward one and the same conflict frame. Furthermore, the classic work by Kelley and Stahelski (1970) revealed that such convergence processes can be asymmetrical. That is, prosocially motivated individuals are more likely to adopt the selfish goals of their opponent than the other way around. Weingart, Brett, and Olekalns (2001) demonstrate, accordingly, that four-person groups containing one selfish negotiator over time become as selfish as groups containing two or

274

DE DREU AND CARNEVALE

more selfish negotiators. This suggests that selfish motivation is more likely to dominate and survive than prosocial motivation. Although asymmetry in adaptation of social motives has been wellsupported, we are unaware of research on convergence in epistemic motivation. In fact, in the course of a conflict, epistemic motivation may rise due to the need to avoid a costly impasse, but it may also decline due to increased time pressure, fatigue, and concomitant need for cognitive closure. Further, it may be that one’s high level of epistemic motivation and concomitant deep and systematic processing of information inspires the opponent to ‘‘play along,’’ but it may also reinforce the opponent’s tendency to act as a cognitive miser who needs not think too deeply about the situation. Research is needed to address this issue, in order to inform us whether a prosocial miser is likely to turn into a selfish miser, or into a selfish thinker.

VII. Other Motives and Other Models Social motives and epistemic motivation have been the focus of many studies, but they are not the only motives that influence strategic choice and information processing in conflict and negotiation. Examples of other motives are impression management concerns and a power motive. Also, the model developed in the previous section builds on (parts of) older models that have dominated the conflict and negotiation literature for a long time, and still do. In this section we review research on other motives in conflict and negotiation, and we compare the current motivational analysis to its two scientific parents—dual concern theory and behavioral decision theory.

A. IMPRESSION MOTIVATION AND POWER MOTIVATION In a recent chapter on persuasion in negotiation, Chaiken et al. (2001) considered impression motivation—the motivation to produce and maintain a certain impression for others, including the adversary (see also Wall, 1991). Although much research on impression motivation has been concerned with positive impressions, individuals in conflict and negotiation are sometimes motivated to convey a negative image, the image of someone not to be trifled with. This makes sense because, in some cases, perceived weakness of an adversary may lay the foundation for an attack or exploitation of an adversary (Holsti, 1962; Siegel & Fouraker, 1960). Indeed, negotiators do engage in behaviors to convey such a tough image. They look tough while competing (Carnevale, Pruitt, & Seilheimer, 1981),

MOTIVATION IN CONFLICT AND NEGOTIATION

275

retaliate when attacked (Felson, 1978), and they ‘‘cut their nose to save their face’’ (Brown, 1977; Tjosvold & Sun, 2000). Like social and epistemic motivation, impression motivation is rooted in the individual as well as in the situation. For example, individuals high in self-monitoring are more concerned with image than individuals low in selfmonitoring. Jordan and Roloff (1997) classified negotiators as being high or low in self-monitoring, and found that negotiators scoring high on selfmonitoring engaged more in planning of impression management (e.g., ‘‘be friendly so he’ll think I’m giving him a good deal’’). Also, there is evidence that impression motivation increases when there is accountability to constituents—the need to account to constituents the outcomes one obtained from the negotiation (note the differences with process accountability, where negotiators do not know what is expected from them; Tetlock, 1992). Negotiators under outcome accountability tend to be more competitive with their opponent, presumably because they want to impress their constituents by winning the negotiation (e.g., Ben-Yoav & Pruitt, 1984a,b; Carnevale et al., 1981; Wall, 1977). Similarly, constituent surveillance tends to increase negotiator competitiveness, apparently because negotiators make greater effort to impress their constituents (e.g., Carnevale, Pruitt, & Britton, 1979). Copeland (1994) and Jones (1986) have argued that individuals with a power disadvantage have higher impression motivation than individuals with a power advantage. The idea is that low power individuals may seek situational control by managing positive impressions of themselves (see also Goodwin et al., 2000). Earlier we reviewed research suggesting that power differences influence epistemic motivation. Thus, negotiators with low power have, compared to their powerful opponents, higher epistemic motivation as well as higher impression motivation (De Dreu & Van Kleef, 2002). Although impression motivation is considered a core motive in conflict and negotiation, research has focused on strategic choice and ignored information processing effects. An exception is the study by Jordan and Roloff (1997) mentioned earlier. These authors obtained positive correlations between self-monitoring and the variety of strategies and tactics negotiators thought of prior to the negotiation. It appeared that selfmonitoring and concomitant impression motivation made negotiators creative and flexible in their strategic thinking. Consistent with this, Ohbuchi and Fukushima (1997) found self-monitoring to be associated with more integrative negotiation, albeit only when there was mild (rather than acute) time pressure. Integrative behavior can be seen as requiring substantial cognitive flexibility. Conflict and negotiation research has also considered the antecedents and the consequences of power motive. Results suggest that power motive can

276

DE DREU AND CARNEVALE

affect negotiation behavior in important ways. Individuals with a high power motive have a strong concern for impact, prestige, and reputation (McClelland et al., 1989). Research has shown, for example, that negotiators with a high rather than low power motive adopt a more exploitative negotiation style, and use physical and verbal aggression more frequently (Winter, 1996). Langner and Winter (2001) coded various statesmen for power motive on the basis of historical documents (e.g., speeches, minutes from meetings, interviews) and related this to conciliatory behavior in the crises with which these historical documents were concerned. Results showed that political parties led by statesmen with a high power motive were less likely to make conciliatory moves than were those led by statesmen with a low power motive.

B. OTHER MODELS Reflecting the fact that there is more than social and epistemic motivation, other models have been proposed. In fact, many studies have centered on dual concern theory (Pruitt & Rubin, 1986; Rubin et al., 1994). Other work has considered judgment and decision making in light of behavioral decision theory (Neale & Bazerman, 1991). Dual concern theory (Pruitt & Rubin, 1986) has been introduced when we discussed strategic choice as a function of social motives. It predicts strategic choice in negotiation to be a function of the interaction between social motivation and resistance to yielding—prosocial motivation leads to problem solving and high levels of information exchange, especially when negotiators have high rather than low resistance to yielding. Although dual concern theory has received good empirical support, it has been criticized for not providing useful advice (Bazerman et al., 2000; but see De Dreu, Weingart et al., 2000), and for being simplistic and unable to account for conglomerated conflict behavior—the cooccurrence of several strategies such as contending and problem solving at the same time (Van de Vliert, 1997). The current analysis accepts dual concern theory as it is, but adds several critical elements. Dual concern theory makes no predictions about the cognitive consequences of social motives. The current analysis has clarified that social motives bias information processing and search so that information consistent with one’s motive is encoded and retrieved better than information inconsistent with one’s motive. One interesting issue is that several, but not all, of the variables considered as antecedents to resistance to yielding in dual concern theory are considered antecedents to epistemic motivation in the current analysis. For example, time pressure reduces resistance to concession making

MOTIVATION IN CONFLICT AND NEGOTIATION

277

(Carnevale & Lawler, 1986) and epistemic motivation (De Dreu, 2002; Kruglanski & Freund, 1983). Likewise, high limits and aspirations are considered to raise resistance (Ben-Yoav & Pruitt, 1984a,b), and they may increase epistemic motivation as well. The question is whether effects of time pressure and aspirations on integrative agreements are mediated by resistance to yielding, by epistemic motivation, or by both. Some variables were argued to increase resistance to yielding and to reduce epistemic motivation. An example is power—a power disadvantage reduces resistance to yielding (Bacharach & Lawler, 1981), but increases epistemic motivation (Fiske, 1992). The question is whether the potentially beneficial effects of epistemic motivation on information processing may be offset by the fact that reduced resistance to concession making leads one to make large and unilateral concessions, thus foregoing integrative agreements. Key facets of behavioral decision theory (Neale & Bazerman, 1991) have been discussed in Section III. In brief, the theory argues that negotiators are bounded in their rationality, have limited cognitive abilities, and therefore are likely to rely on cognitive heuristics that prevent them from reaching integrative agreements. The current motivational analysis can be seen as an extension and elaboration of several of its basic principles. First, behavioral decision theory has been criticized for not taking into account the social context within which conflict and negotiation take place (Carnevale & Pruitt, 1992; Kramer & Messick, 1995; Weingart et al., 1996). The current focus on social motivation provides a way to introduce the social context into the analysis of cognitive processes. Second, following Kruglanski and Ajzen (1983; Ross, 1977; Ross & Ward, 1995) we distinguished between types of bias, and clarified that some of the suboptimal negotiation processes are due to limited cognitive capacity and the need to simplify the complexities of the world around us (Neale & Bazerman, 1991). Some of them are due to basic tendencies, including ego defensiveness and naive realism (Ross & Ward, 1995). Furthermore, our analysis revealed two critical moderators—social motivation and epistemic motivation. Compared to a selfish motivation, a prosocial motivation reduces the impact of naive realism and ego defensiveness. Compared to low epistemic motivation, high epistemic motivation reduces the impact of cognitive heuristics, as well as naive realism and the biases to which it gives rise.

VIII. Conclusion In reviewing the social psychological literatures on conflict and negotiation, Carnevale and Pruitt (1992) made a distinction between a

278

DE DREU AND CARNEVALE

motivation and strategy perspective and a cognitive processes perspective. The motivation and strategy perspective dominated research between 1960 and the mid-1980s. It was assumed that individuals have difficulty reaching (integrative) agreement because features of the situation, such as power differences, and motivational concerns such as the desire to beat the opponent, conspire against cooperative problem-solving and integrative negotiation (Druckman, 1994). The cognitive processes perspective has been the dominant perspective since the mid-1980s. It is assumed that the negotiators’ inclination to base judgments and decisions on faulty beliefs and inappropriate sources of information, such as the framing of the negotiation outcomes, or momentarily accessible anchor values, prohibits negotiators from reaching integrative agreements. The motivation and strategy perspective and the cognitive processes perspective have long continued to exist and develop in isolation (De Dreu, Koole et al., 2000; Kramer & Messick, 1995; O’Connor, 1997). This is unfortunate because motivation and cognition are difficult to separate (Tetlock & Levi, 1982), and because human behavior is better understood when motivation and cognition are considered together rather than in isolation (Higgins & Kruglanski, 2001). By focusing on social and epistemic motivation, we have taken a first step toward integrating the various research literatures on conflict and negotiation. Our motivational analysis has several other advantages. First, it allows us to understand the influence of many different personality variables and situational features on cognitive and behavioral processes in conflict and negotiation. That is, the influence of individual differences in social value orientation or collectivist values, and the influence of situational features such as incentive structures or the expectation of future interaction, all can be understood in terms of the difference between selfish and prosocial motivation. Likewise, the influence of individual differences in need for cognitive closure or need for cognition, and situational variation in time pressure or power differences, all can be understood in terms of high versus low epistemic motivation. Obviously, there is no simple one-to-one relationship between person and situation variables on the one hand, and social or epistemic motivation on the other. Not all of the effects of time pressure, power differences, or process accountability can be explained in terms of epistemic motivation, and not all the effects of need for affiliation, cooperative future interaction, and shared group membership can be understood in terms of social motives. Nevertheless, the current motivational analysis allows us to reduce the number of predictor variables considerably and to develop a rather parsimonious model of cognition and behavior in conflict and negotiation.

MOTIVATION IN CONFLICT AND NEGOTIATION

279

Second, the combination of social and epistemic motivation yields four archetypes of individuals in conflict, and these archetypes may allow disputants, partisans, and third parties assisting in the conflict process to quickly classify disputants’ behavior. Such diagnosis may foster adequate intervention and help parties to redirect the conflict process to run a more constructive course. Third, the current motivational analysis easily connects to the study of affect and emotion in conflict and negotiation. Our discussion thus far, and much of the research on conflict and negotiation, has ignored the role of emotions. However, emotions are key and we echo comments made by Barry (1999): ‘‘as an impetus for and byproduct of social conflict, emotion is potentially central to understanding how individuals think about, behave within, and respond to bargaining situations’’ (p. 94). Emotions involve physiological reactions, action tendencies, and subjective experience (Lazarus, 1991), and they can be considered as an antecedent to, and a by-product of motivation (Frijda, 1993). For example, fear produces the desire to leave the situation, and happiness produces the desire to preserve the situation. Also, receiving lower outcomes than one’s opponent frustrates one’s desire to come out ahead and, hence, produces dissatisfaction (Gillespie, Brett, & Weingart, 2001; Thompson, Valley, & Kramer, 1995). The connection between motivation and affect was clearly seen in the classic conference paper by Kuhlman and Carnevale (1984), which reported that people with a cooperative social value orientation were more likely to display happy facial expressions, and people with a competitive social value orientation were more likely to display angry facial expressions. A related effect was observed by Lanzetta and Englis (1989), that conflict can elicit a counterempathy response, where a smile on an opponent’s face elicits a grimace and vice versa. Studies considering affect and emotion as triggers of behavior have focused on positive versus negative moods and emotions. Although positive affect, empathy, and happiness all appear to induce a prosocial motivation, negative affect, anger, and frustration seem to make people more egoistic and selfcentered. Thus, positive affect increases concession making (Baron, 1990), joint gains (Allred, Mallozzi, Matsui, & Raia, 1997; Carnevale & Isen, 1986), and preferences for cooperation (Baron, Fortin, Frei, Hauver, & Shack, 1990; Forgas, 1998), and negative affect reduces joint gains (Allred, 2000; Allred et al., 1997), promotes the rejection of ultimatum offers (Pillutla & Murnighan, 1996), and increases the use of competitive strategies (Forgas, 1998). In a similar vein, it may be that surprise enhances epistemic motivation (e.g., Baker & Petty, 1994), whereas fear induced by the opponent’s angry messages appears to reduce epistemic motivation (Van Kleef, De Dreu, & Manstead, 2002). Figures 2 and 3 could easily be modified to include affect as a parameter.

280

DE DREU AND CARNEVALE

Most of the research on conflict and negotiation has, in the past 20 years, considered cognitive processes, yet ignored the motivational underpinnings and affective consequences of conflict and negotiation. The motivated information processing perspective developed here redresses this one-sided focus. Based on extensive evidence, we have argued that although ego defensiveness and naive realism constitute important psychological barriers to constructive negotiation and dispute resolution, social and epistemic motivation alone and in combination moderate the impact of bias due to cognitive limitations, ego defensiveness, and naive realism. Individuals in conflict and negotiation can be characterized as motivated information processors and, depending on their social and epistemic motives, behave as prosocial or selfish misers, or as prosocial or selfish thinkers. We agree with Sir Winston Churchill when he noted that continuous effort—not strength or intelligence—is the key to unlocking our potential. But the course this unlocked potential takes us depends very much on whether we have a competitive, selfish orientation towards the other party, or instead consider the consequences of our actions for other’s outcomes as well.

Acknowledgments Preparation of this manuscript was supported by a grant from the Netherlands Foundation for Scientific Research (NWO Grant No. 490-22-173) awarded to the first author, and NSF Grant No. SBR-9210536 awarded to the second author. For their help and comments on a previous draft of this article, we thank Bianca Beersma, Don Choi, Wolfgang Steinel, Gerben Van Kleef, Ching Wan, and Mark Zanna.

References Allred, K. G. (2000). Anger and retaliation in conflict: The role of attribution. In M. Deutsch & P. T. Coleman (Eds.), The handbook of conflict resolution: Theory and practice (pp. 236–255). San Francisco: Jossey-Bass. Allred, K. G., Mallozzi, J. S., Matsui, F., & Raia, C. P. (1997). The influence of anger and compassion on negotiation performance. Organizational Behavior and Human Decision Processes, 70, 175–187. Amato, P. R., & Booth, A. (2001). The legacy of parents’ marital discord: Consequences for children’s marital quality. Journal of Personality and Social Psychology, 81, 627–638. Axelrod, R. (1984). The evolution of cooperation. New York: Basic Books. Bacharach, S. B., & Lawler, E. J. (1981). Bargaining: Power, tactics and outcomes. Greenwich, CT: JAI Press.

MOTIVATION IN CONFLICT AND NEGOTIATION

281

Baker, S. M., & Petty, R. E. (1994). Majority and minority influence: Source-position imbalance as a determinant of message scrutinity. Journal of Personality and Social Psychology, 67, 5–19. Barnett, M. A., & Thompson, S. (1985). The role of perspective taking and empathy in children’s Machiavellianism, prosocial behavior, and motive for helping. Journal of Genetic Psychology, 146, 295–305. Baron, R. A. (1990). Environmentally induced positive affect: Its impact on self-efficacy, task performance, negotiation, and conflict. Journal of Applied Social Psychology, 20, 368–384. Baron, R. A., Fortin, S. P., Frei, R. L., Hauver, L. A., & Shack, M. L. (1990). Reducing organizational conflict: The potential role of inducing positive affect. International Journal of Conflict Management, 1, 133–152. Barry, B. (1999). The tactical use of emotion in negotiation. Research on Negotiation in Organizations, 4, 93–121. Barry, B., & Friedman, R. A. (1998). Bargainer characteristics in distributive and integrative negotiation. Journal of Personality and Social Psychology, 74, 345–359. Bartos, O. J. (1974). Process and outcome in negotiation. New York: Columbia University Press. Batson, C. D. (1998). Altruism and prosocial behavior. In D. Gilbert, S. T. Fiske & G. Lindzey (Eds.), The handbook of social psychology (4th ed., Vol. 2, pp. 282–316). Boston, MA: McGraw-Hill. Batson, C. D., & Shaw, L. L. (1991). Evidence for altruism: Toward a pluralism of prosocial motives. Psychological Inquiry, 2, 107–122. Bazerman, M. H., Curhan, J. R., Moore, D. A., & Valley, K. L. (2000). Negotiation. Annual Review of Psychology, 51, 279–314. Bazerman, M. H., Magliozzi, T., & Neale, M. A. (1985). Integrative bargaining in a competitive market. Organizational Behavior and Human Decision Processes, 35, 294–313. Bazerman, M. H., & Neale, M. (1983). Heuristics in negotiation: Limitations to effective dispute resolution. In M. H. Bazerman & R. J. Lewicki (Eds.), Negotiating in organizations, (pp. 51–67). Beverly Hills, CA: Sage. Beersma, B., & De Dreu, C. K. W. (2002). Integrative and distributive negotiation in small groups: Effects of task structure, decision rule, and social motive. Organizational Behavior and Human Decision Processes, 87, 227–252. Beggan, J. K. (1992). On the social nature of nonsocial perception: The mere ownership effect. Journal of Personality and Social Psychology, 62, 229–237. Ben-Yoav, O., & Pruitt, D. (1984a). Resistance to yielding and the expectation of cooperative future interaction in negotiation. Journal of Experimental Social Psychology, 34, 323–335. Ben-Yoav, O., & Pruitt, D. G. (1984b). Accountability to constituents: A two-edged sword. Organizational Behavior and Human Performance, 34, 283–295. Berg, J., Dickhaut, J., & McCabe, K. (1995). Trust, reciprocity, and social history. Games and Economic Behavior, 10, 122–142. Berton, P., and Kimura, H., & Zartman, I. W. (Eds.) (1999). International negotiation: Actors, structure/process, values. New York: St. Martin’s Press. Blake, R., & Mouton, J. S. (1964). The managerial grid. Houston, TX: Gulf. Bottom, W. P. (1998). Negotiator risk: Sources of uncertainty and the impact of reference points on negotiated agreements. Organizational Behavior and Human Decision Processes, 76, 89–112. Bottom, W. P., & Paese, P. W. (1997). False consensus, stereotypic cues, and the perception of integrative potential in negotiation. Journal of Applied Social Psychology, 27, 1919–1940. Bottom, W. P., & Studt, A. (1993). Framing and distributive aspects of integrative negotiation. Organizational Behavior and Human Decision Processes, 56, 459–474.

282

DE DREU AND CARNEVALE

Brett, J. M., Shapiro, D. L., & Lytle, A. L. (1998). Breaking the bonds of reciprocity in negotiations. Academy of Management Journal, 41, 410–424. Brown, B. R. (1977). Face-saving and face-restoration in negotiation. In D. Druckman (Ed.), Negotiations, (pp. 275–300). Beverly Hills, CA: Sage. Burnham, T., McCabe, K., & Smith, V. L. (2000). Friend-or-foe intentionality priming in an extensive form trust game. Journal of Economic Behavior & Organization, 43, 57–73. Camac, C. (1992). Information preferences in a two-person social dilemma. In W. Liebrand, D. Messick & H. Wilke (Eds.), Social dilemmas: Theoretical issues and research findings, (pp. 147–161). Berlin: Springer-Verlag. Caporael, L. R., Dawes, R. M., Orbell, J. M., & Van de Kragt, A. J. C. (1989). Selfishness examined: Cooperation in the absence of egoistic incentives. Behavioral and Brain Sciences, 12, 683–739. Carnevale, P. J. (1995). Property, culture, and negotiation. In R. Kramer & D. M. Messick (Eds.), Negotiation as a social process. Newbury Park, CA: Sage. Carnevale, P. J., & De Dreu, C. K. W. (2002). Motivational orientation moderates cognitive bias effects in negotiation. Submitted. Carnevale, P. J., & Isen, A. (1986). The influence of positive affect and visual access on the discovery of integrative solutions in bilateral negotiation. Organizational Behavior and Human Decision Processes. 37, 1–13. Carnevale, P. J., & Lawler, E. J. (1986). Time pressure and the development of integrative agreements in bilateral negotiations. Journal of Conflict Resolution, 30, 636–659. Carnevale, P. J., & Probst, T. M. (1998). Social values and social conflict in creative problem solving and categorization. Journal of Personality and Social Psychology, 74, 1300–1309. Carnevale, P. J., & Pruitt, D. G. (1992). Negotiation and mediation. Annual Review of Psychology, 43, 531–582. Carnevale, P. J., & Pruitt, D. G. (2003). Negotiation in social conflict. (2nd ed.). Buckingham, UK: Open University Press. Carnevale, P. J. D., Pruitt, D. G., & Britton, S. D. (1979). Looking tough: The negotiator under constituent surveillance. Personality and Social Psychology Bulletin, 5, 118–121. Carnevale, P. J., Pruitt, D. G., & Seilheimer, S. (1981). Looking and competing: Accountability and visual access in integrative bargaining. Journal of Personality and Social Psychology, 40, 111–120. Chaiken, S., Gruenfeld, D. H., & Judd, C. M. (2000). Persuasion in negotiations and conflict situations. In M. Deutsch & P. Coleman (Eds.). Conflict resolution, New York. Chaiken, S. & Trope, Y. (Eds.). (1999). Dual-process theories in social psychology. New York: Guilford. Clyman, D. R. (1995). Measures of joint performance in dyadic mixed-motive negotiations. Organizational Behavior and Human Decision Processes, 64, 38–48. Copeland, J. T. (1994). Prophecies of power: Motivational implications of social power for behavioral confirmation. Journal of Personality and Social Psychology, 67, 264–277. Cross, J. G. (1977). Negotiation as a learning process. Journal of Conflict Resolution, 21, 581–606. de Callieres, F. (1716, reissued in 1963). On the manner of negotiating with princes. (Trans. A. F. White). Notre Dame, IN: Notre Dame University Press. De Dreu, C. K. W. (2002). Time pressure and motivated closing of the mind in negotiation. Organizational Behavior and Human Decision Processes. Submitted. De Dreu, C. K. W., Beersma, B., Stroebe, K., & Euwema, M. (2002). Motivated information processing in negotiation. In preparation.

MOTIVATION IN CONFLICT AND NEGOTIATION

283

De Dreu, C. K. W., & Boles, T. (1998). Share and share alike or winner take all? Impact of social value orientation on the choice and recall of decision heuristics in negotiation. Organizational Behavior and Human Decision Processes, 76, 253–267. De Dreu, C. K. W., Carnevale, P. J. D., Emans, B. J. M., & Van de Vliert, E. (1994). Effects of gain loss frames in negotiation: Loss aversion, mismatching, and frame adoption. Organizational Behavior and Human Decision Processes, 60, 90–107. De Dreu, C. K. W., Emans, B. J., & Van de Vliert, E. (1992). Frames of reference and cooperative social decision-making. European Journal of Social Psychology, 22, 297–302. De Dreu, C. K. W., Giebels, E., & Van de Vliert, E. (1998). Social motives and trust in integrative negotiation: Disruptive effects of punitive capability. Journal of Applied Psychology, 83, 408–422. De Dreu, C. K. W., Koole, S., & Oldersma, F. L. (1999). On the seizing and freezing of negotiator inferences: Need for cognitive closure moderates the use of heuristics in negotiation. Personality and Social Psychology Bulletin, 25, 348–362. De Dreu, C. K. W., Koole, S., & Steinel, W. (2000). Unfixing the fixed-pie: A motivated information processing of integrative negotiation. Journal of Personality and Social Psychology, 79, 975–987. De Dreu, C. K. W., & McCusker, C. (1997). Gain-loss frames in two-person social dilemmas: A transformational analysis. Journal of Personality and Social Psychology, 72, 1093–1106. De Dreu, C. K. W., Nauta, A., & Van de Vliert, E. (1995). Self-serving evaluation of conflict behavior and escalation of the dispute. Journal of Applied Social Psychology, 25, 2049–2066. De Dreu, C. K. W., & Van Kleef, G. A. (2002). The influence of power on information search, impression formation and concession making in negotiation. Journal of Experimental Social Psychology. Submitted. De Dreu, C. K. W., & Van Lange, P. A. M. (1995). The impact of social value orientations on negotiator cognition and behavior. Personality and Social Psychology Bulletin, 21, 1178–1188. De Dreu, C. K. W., Weingart, L. R., & Kwon, S. (2000). Influence of social motives on integrative negotiation: A meta-analytical review and test of two theories. Journal of Personality and Social Psychology, 78, 889–905. De Dreu, C. K. W., Yzerbyt, V., & Leyens, J-Ph. (1995). Dilution of stereotype-based cooperation in mixed-motive interdependence. Journal of Experimental Social Psychology, 21, 575–593. Deutsch, M. (1949). A theory of cooperation and competition. Human Relations, 2, 199–231. Deutsch, M. (1958). Trust and suspicion. Journal of Conflict Resolution, 2, 265–279. Deutsch, M. (1973). The resolution of conflict: Constructive and destructive processes. New Haven, CT: Yale University Press. Deutsch, M. (1982). Interdependence and psychological orientation. In V. Derlega & J. Gezelak (Eds.), cooperation and helping behavior (pp. 15–42). Cambridge: Cambridge University Press. Druckman, D. (1994). Determinants of compromising behavior in negotiation: A metaanalysis. Journal of Conflict Resolution, 38, 507–556. Druckman, D., & Zechmeister, K. (1973). Conflict of interest and value dissensus: Propositions in the sociology of conflict. Human Relations, 26, 449–466. Dunegan, K. J. (1993). Framing, cognitive modes, and image theory: Toward an understanding of a glass half full. Journal of Applied Psychology, 78, 491–503. Eisenberg, N., & Miller, P. A. (1987). The relation of empathy to prosocial and related behaviors. Psychological Bulletin, 101, 91–119. Felson, R. B. (1978). Aggression as impression management. Social Psychology, 41, 205–213.

284

DE DREU AND CARNEVALE

Fisher, R., & Ury, W. (1981). Getting to yes: Negotiating agreement without giving in. Boston, MA: Houghton Mifflin. Fiske, S. T. (1992). Controlling other people: The impact of power on stereotyping. American Psychologist, 48, 621–628. Fiske, S. T., & Neuberg, S. L. (1990). A continuum model of impression formation: From category-based to individuating processes as a function of information, motivation, and attention. In L. Berkowitz (Ed.). Advances in experimental social psychology (Vol. 23, pp. 1–108). New York: Academic Press. Fiske, S. T., & Taylor, S. E. (1991). Social cognition. New York: McGraw-Hill. Forgas, J. P. (1998). On feeling good and getting your way: Mood effects on negotiator cognition and behavior. Journal of Personality and Social Psychology, 74, 565–577. Frijda, N. H. (1993). Moods, emotion episodes, and emotions. In M. Lewis & J. M. Haviland (Eds.), Handbook of emotions (pp. 381–403). New York: Guilford. Froman, L. A., & Cohen, M. D. (1970). Compromise and logroll: Comparing the efficiency of two bargaining processes. Behavioral Science, 30, 180–183. Fry, W. R. (1985). The effect of dyad machiavellianism and visual access on integrative bargaining outcomes. Personality and Social Psychology Bulletin, 11, 51–62. Fry, W. R., Firestone, I. J., & Williams, D. L. (1983). Negotiation process and outcome of stranger dyads and dating couples: Do lovers lose? Basic and Applied Social Psychology, 4, 1–16. Galinsky, A. D., & Mussweiler, T. (2001). First offers as anchors: The role of perspective-taking and negotiator focus. Journal of Personality and Social Psychology, 81, 657–669. Gelfand, M. J., & Christakopoulou, S. (1999). Culture and negotiator cognition: Judgment accuracy and negotiation processes in individualistic and collectivistic cultures. Organizational Behavior and Human Decision Processes, 79, 248–269. Gelfand, M. J., Higgins, M., Nishii, L., Raver, J., Dominguez, A., Yamaguchi, S., Murakami, F., & Toyama, M. (2001). Culture and egocentric biases of fairness in conflict and negotiation. Journal of Applied Psychology, 86, 1059–1074. Gelfand, M. J., & Realo, A. (1999). Individualism-collectivism and accountability in intergroup negotiations. Journal of Applied Psychology, 84, 721–736. Gillespie, J. J., Brett, J. M., & Weingart, L. M. (2000). Interdependence, social motives, and outcome satisfaction in multiparty negotiation. European Journal of Social Psychology, 30, 779–797. Goodwin, S. A., Gubin, A., Fiske, S. T., & Yzerbyt, V. Y. (2000). Power can bias impression processes: Stereotyping subordinates by default and by design. Group Processes and Intergroup Relations, 3, 227–256. Gouldner, A. W. (1960). The norm of reciprocity: A preliminary statement. American Sociological Review, 25, 161–178. Graziano, W. G., Jensen-Campbell, L. A., & Hair, E. C. (1996). Perceiving interpersonal conflict and reacting to it: The case for agreeableness. Journal of Personality and Social Psychology, 70, 820–835. Greenwald, A. G. (1980). The totalitarian ego: Fabrication and revision of personal history. American Psychologist, 35, 603–618. Gunnthorsdottir, A., McCabe, K., & Smith, V. (2002). Using the Machiavellianism instrument to predict trustworthiness in a bargaining game. Journal of Economic Psychology, 23, 49–66. Guth, W., Schmittberger, R., & Schwarze, B. (1982). An experimental analysis of ultimatum bargaining. Journal of Economic Behavior and Organization, 3, 367–388. Hammond, K. R. (1965). New directions in research on conflict resolution. Journal of Social Issues, 21, 44–66.

MOTIVATION IN CONFLICT AND NEGOTIATION

285

Harinck, F., De Dreu, C. K. W., & Van Vianen, A. E. M. (2000). The impact of issue content on fixed-pie perceptions, problem solving and integrative outcomes in negotiation. Organizational Behavior and Human Decision Processes, 81, 329–358. Hatfield, E., Cacioppo, J. T., & Rapson, R. L. (1994). Emotional contagion. New York: Cambridge University Press. Higgins, E. T., & Kruglanski, A. W. (2001). Motivational science: The nature and functions of wanting. In E. T. Higgins & A. W. Kruglanski (Eds.), Motivational science: Social and personality perspectives (pp. 1–20). New York: Psychology Press. Hofstede, G. (1980). Culture’s consequences. London: Sage. Hofstee, W. K. B., De Raad, B., & Goldberg, L. R. (1992). Integration of the big five and circumplex approaches to trait structure. Journal of Personality and Social Psychology, 63, 146–163. Holsti, O. R. (1962). The belief system and national images: A case study. Journal of Conflict Resolution, 6, 244–252. Huber, V. L., & Neale, M. A. (1986). Effects of cognitive heuristics and goals on negotiator performance and subsequent goal setting. Organizational Behavior and Human Decision Processes, 38, 342–365. Hulbert, L. G., Correa da Silva, M. L., & Adegboyega, G. (2001). Cooperation in social dilemmas and allocentrism: A social values approach. European Journal of Social Psychology, 31, 641–658. Jehn, K. (1995). A multimethod examination of the benefits and detriments of intragroup conflict. Administrative Science Quarterly, 40, 256–282. Jervis, R. (1976). Perception and misperception in international politics. Princeton, NJ: Princeton University Press. Jones, E. E. (1986). Interpreting interpersonal behavior: The effects of expectancies. Science, 234, 41–46. Jordan, J. M., & Roloff, M. E. (1997). Planning skills and negotiator goal accomplishment. Communication Research, 24, 31–63. Judd, C. M. (1978). Cognitive effects of attitude conflict resolution. Journal of Conflict Resolution, 22, 483–498. Kahneman, D., Knetsch, J., & Thaler, R. (1990). Experimental tests of the endowment effect and the Coase theorem. Journal of Political Economy, 98, 1325–1348. Kahneman, D., & Tversky, A. (1973). On the psychology of prediction. Psychological Review, 80, 237–251. Kahneman, D., & Tversky, A. (1979). Prospect theory: An analysis of decision under risk. Econometrica, 47, 263–291. Kasser, T., & Ahuvia, A. (2002). Materialistic values and well-being in business students. European Journal of Social Psychology, 32, 137–146. Kelley, H. H. (1966). A classroom study of the dilemmas in interpersonal negotiations. In K. Archibald (Ed.), Strategic interaction and conflict: Original papers and discussion, (pp. 67–85). Berkeley, CA: Institute of International Studies. Kelley, H. H., Beckman, L. L., & Fischer, C. S. (1967). Negotiation the division of reward under incomplete information. Journal of Experimental Social Psychology, 3, 361–389. Kelley, H. H., & Schenitzki, D. P. (1972). Bargaining. In C. McClintock (Ed.), Experimental social psychology (pp. 298–337). New York: Holt, Rinehart, & Winston. Kelley, H. H., & Stahelski, A. J. (1970). Social interaction basis of cooperators’ and competitors’ beliefs about others. Journal of Personality and Social Psychology, 16, 66–91.

286

DE DREU AND CARNEVALE

Kelley, H. H., & Thibaut, J. W. (1978). Interpersonal relations: A theory of interdependence. New York: Wiley. Keltner, D., & Robinson, R. J. (1993). Imagined ideological differences in conflict escalation and resolution. International Journal of Conflict Management, 4, 249–262. Kemmelmeier, M., & Winter, D. G. (2000). Putting threat into perspective: Experimental studies on perceptual distortion in international conflict. Personality and Social Psychology Bulletin, 26, 795–809. Kemp, K. E., & Smith, W. P. (1994). Information exchange, toughness, and integrative bargaining: The roles of explicit cues and perspective-taking. International Journal of Conflict Management, 5, 5–21. Kenrick, D. T., & Funder, D. C. (1988). Profiting from controversy: Lessons learned from the person  situation debate. American Psychologist, 43, 23–34. Kimmel, M. J., Pruitt, D. G., Magenau, J. M., Konar-Goldband, E., & Carnevale, P. J. D. (1980). Effects of trust, aspiration, and gender on negotiation tactics. Journal of Personality and Social Psychology, 38, 9–22. Kramer, R. M., & Carnevale, P. J. (2001). Trust and intergroup negotiation. In R. Brown & S. Gaertner (Eds.), Blackwell handbook of social psychology, Vol 4: Intergroup relations (pp. 431–450). Oxford, UK: Blackwell Publishers. Kramer, R. M. & Messick, D. M. (Eds.), (1995). Negotiation as a social process. London: Sage. Kramer, R. M., Pommerenke, P., & Newton, E. (1993). The social context of negotiation. Journal of Conflict Resolution, 37, 633–654. Kray, L. J., Thompson, L., & Galinsky, A. (2001). Battle of the sexes: Gender stereotype confirmation and reactance in negotiations. Journal of Personality and Social Psychology, 80, 942–958. Kristensen, H., & Ga¨ rling, T. (1997). The effects of anchor points and reference points on negotiation process and outcome. Organizational Behavior and Human Decision Processes, 71, 85–94. Kruglanski, A. W. (1989). The psychology of being ‘‘right’’: The problem of accuracy in social perception and cognition. Psychological Bulletin, 106, 395–409. Kruglanski, A. W., & Ajzen, I. (1983). Bias and error in human judgment. European Journal of Social Psychology, 13, 1–44. Kruglanski, A. W., & Freund, T. (1983). The freezing and unfreezing of lay-inferences: Effects on impressional primacy, ethnic stereotyping, and numerical anchoring. Journal of Experimental Social Psychology, 19, 448–468. Kruglanski, A. W., & Webster, D. M. (1996). Motivated closing of the mind: ‘‘Seizing and ‘‘freezing.’’ Psychological Review, 103, 263–283. Kuhlman, D. M., & Carnevale, P. J. (1984). Differences in nonverbal demeanor between cooperators, individualists, and competitors. Paper presented at the 92nd Annual Convention of the American Psychological Association, Toronto, Ontario, Canada. Kuhlman, D. M., & Marshello, A. (1975). Individual differences in game motivation as moderators of preprogrammed strategic effects in prisoner’s dilemma. Journal of Personality and Social Psychology, 32, 922–931. Langner, C. A., & Winter, D. G. (2001). The motivational basis of concessions and compromise: Archival and laboratory studies. Journal of Personality and Social Psychology, 81, 711–727. Lanzetta, J. T., & Englis, B. G. (1989). Expectations of cooperation and competition and their effects on observers’ vicarious emotional responses. Journal of Personality and Social Psychology, 56, 543–554. Lazarus, R. S. (1991). Emotion and adaptation. Oxford, UK: Oxford University Press.

MOTIVATION IN CONFLICT AND NEGOTIATION

287

Lerner, J. S., & Tetlock, P. E. (1999). Accounting for the effects of accountability. Psychological Bulletin, 125, 255–275. Levine, J. M., & Thompson, L. (1996). Conflict in groups. In T. Higgins & A. W. Kruglanski (Eds.), Social psychology: Handbook of basic principles (pp. 745–776). New York: Guilford. Luce, R. D., & Raiffa, H. (1957). Games and decisions. New York: Wiley. MacCrimmon, K. R., & Messick, D. (1976). A framework for social motives. Behavioral Science, 21, 86–100. Mayseless, O., & Kruglanski, A. W. (1987). What makes you so sure? Effects of epistemic motivations on judgmental confidence. Organizational Behavior and Human Decision Processes, 39, 162–183. McClelland, D. C., Koestner, R., & Weinberger, J. (1989). How do self-attributed and implicit motives differ? Psychological Review, 96, 690–702. McClintock, C. (1977). Social motives in settings of outcome interdependence. In D. Druckman (Ed.), Negotiations: Social psychological perspective, (pp. 49–77). Beverly Hills, CA: Sage. McHoskey, J. W. (1999). Machiavellianism, intrinsic versus extrinsic goals, and social interest: A self-determination theory analysis. Motivation and Emotion, 23, 267–283. Messick, D. M., Bloom, S., Boldizar, J. P., & Samuelson, C. D. (1985). Why we are fairer than others. Journal of Experimental Social Psychology, 21, 480–500. Messick, D. M., & McClintock, C. (1968). Motivational bases of choice in experimental games. Journal of Experimental Social Psychology, 4, 1–25. Morris, M. W., & Keltner, D. (2000). How emotions work: An analysis of the social functions of emotional expression in negotiations. Research in Organizational Behavior, 22, 1–50. Morris, M. W., Larrick, R. P., & Su, S. K. (1999). Misperceiving negotiation counterparts: When situationally determined bargaining behaviors are attributed to personality traits. Journal of Personality and Social Psychology, 77, 52–67. Nash, J. F. (1954). The bargaining problem. Econometrica, 18, 155–162. Neale, M. A., & Bazerman, M. H. (1985). The effects of framing and negotiator overconfidence on bargaining behaviors and outcomes. Academy of Management Journal, 28, 34–49. Neale, M. A., & Bazerman, M. H. (1991). Rationality and cognition in negotiation. New York: Free Press. Northcraft, G. B., & Neale, M. A. (1987). Experts, amateurs, and real estate: An anchoringand-adjustment perspective on property pricing decisions. Organizational Behavior and Human Decision Processes, 39, 84–97. O’Connor, K. (1997). Motives and cognitions in negotiation: A theoretical integration and empirical test. International Journal of Conflict Management, 8, 114–131. Ohbuchi, K-I., & Fukushima, O. (1997). Personality and interpersonal conflict: Aggressiveness, self-monitoring, and situational variables. International Journal of Conflict Management, 8, 99–113. Ohbutsu, Y., & Kameda, T. (1998). The function of equality heuristic in distributive bargaining: Negotiated allocation of costs and benefits in a demand revelation context. Journal of Experimental Social Psychology, 34, 90–108. Olekalns, M., Smith, P. L., & Walsh, T. (1996). The process of negotiating: Strategy and timing as predictors of outcomes. Organizational Behavior and Human Decision Processes, 68, 68–77. Paese, P. W., & Yonker, R. D. (2001). Toward a better understanding of egocentric fairness judgments in negotiation. International Journal of Conflict Management, 12, 97–186. Parks, C. D., Henager, R. F., & Scamahorn, S. D. (1996). Trust and reactions to messages of intent in social dilemmas. Journal of Conflict Resolution, 40, 134–151.

288

DE DREU AND CARNEVALE

Petty, R. E., & Cacioppo, J. T. (1986). The elaboration likelihood model of persuasion. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 19, pp. 123–205). New York: Academic Press. Pillutla, M. M., & Murnighan, J. K. (1996). Unfairness, anger, and spite: Emotional rejections of ultimatum offers. Organizational Behavior and Human Decision Processes, 68, 208–224. Pinkley, R. L., Griffith, T. L., & Northcraft, G. B. (1995). ‘‘Fixed Pie’’ a` la Mode: Information availability, information processing, and the negotiation of suboptimal agreements. Organizational Behavior and Human Decision Processes, 62, 101–112. Pinkley, R. L., & Northcraft, G. B. (1994). Conflict frames of reference: Implications for dispute processes and outcomes. Academy of Management Journal, 37, 193–205. Polzer, J. T., Mannix, E. A., & Neale, M. A. (1998). Interest alignment and coalitions in multiparty negotiation. Academy of Management Journal, 41, 42–54. Probst, T., Carnevale, P. J., & Triandis, H. C. (1999). Cultural values in intergroup and singlegroup social dilemmas. Organizational Behavior and Human Decision Processes, 77, 171–191. Pruitt, D. G. (1967). Reward structure and cooperation: The decomposed prisoner’s dilemma. Journal of Personality and Social Psychology, 7, 21–27. Pruitt, D. G. (1981). Negotiation behavior. New York: Academic Press. Pruitt, D. G. (1998). Social conflict. In D. Gilbert, S. T. Fiske & G. Lindzey (Eds.), Handbook of social psychology (4th ed., Vol. 2, 89–150). New York: McGraw-Hill. Pruitt, D. G., & Drews, J. L. (1969). The effect of time pressure, time elapsed, and the opponent’s concession rate on behavior in negotiation. Journal of Experimental Social Psychology, 5, 43–60. Pruitt, D. G., & Kimmel, M. J. (1977). Twenty years of experimental gaming: Critique, synthesis, and suggestions for the future. Annual Review of Psychology, 28, 363–392. Pruitt, D. G., & Lewis, S. A. (1975). Development of integrative solutions in bilateral negotiation. Journal of Personality and Social Psychology, 31, 621–633. Pruitt, D. G., & Rubin, J. Z. (1986). Social conflict: escalation, stalemate, and settlement. New York: McGraw-Hill. Raiffa, H. (1982). The art and science of negotiation. Cambridge, MA: Belknap. Rapoport, A. (1960). Fights, games and debates. Ann Arbor: University of Michigan Press. Rettinger, D. A., & Hastie, R. (2001). Content effects on decision making. Organizational Behavior and Human Decision Processes, 85, 336–359. Richter, L., & Kruglanski, A. W. (1999). Motivated search for common ground: Need for closure effects on audience design in interpersonal communication. Personality and Social Psychology Bulletin, 25, 1101–1114. Robinson, R. J., Keltner, D., Ward, A., & Ross, L. (1995). Actual versus assumed differences in construal: ‘‘Naı¨ve realism’’ in intergroup perception and conflict. Journal of Personality and Social Psychology, 68, 404–417. Ross, L. (1977). The intuitive psychologist and his shortcomings. In L. Berkowitz (Ed.) Advances in experimental social psychology (Vol. 10). New York: Academic Press. Ross, L., & Stillinger, C. (1991). Barriers to conflict resolution. Negotiation Journal November, 389–404. Ross, L., & Ward, A. (1995). Psychological barriers to dispute resolution. In M. Zanna (Ed.), Advances in experimental social psychology, (Vol. 27, pp. 255–304). San Diego, CA: Academic Press. Ross, W. H., & Conlon, D. E. (2000). Hybrid forms of third party dispute resolution: Theoretical implications of combining mediation and arbitration. Academy of Management Review, 25, 416–427.

MOTIVATION IN CONFLICT AND NEGOTIATION

289

Roth, A. E. (1994). Bargaining. In J. Kagel & A. E. Roth (Eds.), Handbook of experimental economics (pp. 253–348). Princeton, NJ: Princeton University Press. Rothbart, M., & Hallmark, W. (1988). In-group and out-group differences in the perceived efficacy of coercion and conciliation in resolving social conflict. Journal of Personality and Social Psychology, 55, 248–257. Rubin, J., & Brown, R. (1975). Social psychology of bargaining and negotiation. New York: Academic Press. Rubin, J., Pruitt, D. G., & Kim, S. (1994). Social conflict: Escalation, stalemate, and settlement. New York: McGraw-Hill. Rubinstein, A. (1999). Experience from a course in game theory: Pre-and postclass problem sets as a didactic device. Games and Economic Behavior, 28, 155–170. Schelling, T. C. (1960). The strategy of conflict. Boston, MA: Harvard University Press. Shalev, J. (2002). Loss aversion and bargaining. Theory and Decision, 52, 210–232. Shestowsky, D., Wegener, D. T., & Fabrigar, L. R. (1998). Need for cognition and interpersonal influence: Individual differences in impact on dyadic decisions. Journal of Personality and Social Psychology, 74, 1317–1328. Siegel, A. E., & Fouraker, L. E. (1960). Bargaining and group decision making. New York: McGraw-Hill. Simon, H. A. (1957). A behavioral model of rational choice. Quarterly Journal of Economics, 69, 99–118. Simonson, I., & Staw, B. M. (1992). Deescalation strategies: A comparison of techniques for reducing commitment to losing courses of action. Journal of Applied Psychology, 77, 419–426. Singelis, T. M., Triandis, H. C., Bhawuk, D. P. S., & Gelfand, M. J. (1995). Horizontal and vertical dimensions of individualism and collectivism: A theoretical and measurement refinement. Cross-Cultural Research, 29, 240–275. Smith, E. R., & DeCoster, J. (2000). Dual process models in social and cognitive psychology: Conceptual integration and links to underlying memory systems. Personality and Social Psychology Review, 4, 108–131. Smith, V. L. (2001). From old issues to new directions in experimental psychology and economics. Behavioral and Brain Sciences, 24, 428–429. Snyder, M. (1992). Motivational foundations of behavioral confirmation. In M. Zanna (Ed.). Advances in experimental social psychology (Vol. 25, pp. 67–114). New York: Academic Press. Susskind, L., McKearnan, S. & Thomas-Larmer, J.(Eds.), (1999). The consensus building handbook: A comprehensive guide to reaching agreement. Thousand Oaks, CA: Sage. Taylor, S. E., & Brown, J. D. (1988). Illusion and well-being: A social-psychological perspective on mental health. Psychological Bulletin, 103, 193–210. Tetlock, P. E. (1992). The impact of accountability on judgment and choice: Toward a social contingency model. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 25, pp. 331–376). New York: Academic Press. Tetlock, P. E., & Levi, A. (1982). Attribution bias: On the inconclusiveness of the cognitionmotivation debate. Journal of Experimental Social Psychology, 18, 68–88. Thompson, L. L (1990). Negotiation behavior and outcomes: Empirical evidence and theoretical issues. Psychological Bulletin, 108, 515–532. Thompson, L. L. (1991). Information exchange in negotiation. Journal of Experimental Social Psychology, 26, 82–90. Thompson, L. L. (1995). ‘‘They saw a negotiation:’’ Partisanship and involvement. Journal of Personality and Social Psychology, 68, 839–853.

290

DE DREU AND CARNEVALE

Thompson, L. L. (1998). The mind and heart of the negotiator. Upper Saddle River, NJ: Prentice-Hall. Thompson, L. L., & Hastie, R. (1990). Social perception in negotiation. Organizational Behavior and Human Decision Processes, 47, 98–123. Thompson, L. L., & Hebrec, D. (1996). Lose-lose agreements in interdependent decision making. Psychological Bulletin, 120, 396–409. Thompson, L. L., & Loewenstein, G. (1992). Egocentric interpretations of fairness in interpersonal conflict. Organizational Behavior and Human Decision Processes, 51, 176–197. Thompson, L. L., Valley, K. L., & Kramer, R. M. (1995). The bittersweet feeling of success: An examination of social perception in negotiation. Journal of Experimental Social Psychology, 21, 1–26. Tjosvold, D. (1998). Cooperative and competitive goal approach to conflict: Accomplishments and challenges. Applied Psychology, 46, 285–342. Tjosvold, D., & Sun, H. (2000). Social face in conflict: Effects of affronts to person and position in China. Group Dynamics: Theory, Research, and Practice, 4, 259–271. Triandis, H. C. (1989). The self and social behavior in different cultural contexts. Psychological Review, 96, 506–520. Tripp, T., & Sondak, H. (1992). An evaluation of dependent variables in experimental negotiation: Effects of decision rule, agenda, and aspiration. Organizational Behavior and Human Decision Processes, 51, 273–295. Tyler, T., Huo, Y. J., & Lind, E. A. (1999). The two psychologies of conflict resolution: Differing antecedents of pre-experience choices and post-experience evaluations. Group Processes and Intergroup Relations, 2, 99–118. Van Boven, L., Dunning, D., & Loewenstein, G. (2000). Egocentric empathy gaps between owners and buyers: Misperceptions of the endowment effect. Journal of Personality and Social Psychology, 79, 66–76. Van de Vliert, E. (1997). Complex interpersonal conflict behavior. London: Psychology Press. Van Dyne, L., & Le Pine, J. A. (1998). Helping and voice extra-role behaviors: Evidence of construct and predictive validity. Academy of Management Journal, 41, 108–119. Van Kleef, G. A., De Dreu, C. K. W. (2002). Social value orientation, information search, and impression formation in negotiation. International Journal of Conflict Management. Van Kleef, G. A., De Dreu, C. K. W., & Manstead, A. S. R. (2002 June). Interpersonal effects of opponent’s emotions in negotiation. Paper presented at the annual meeting of the International Association for Conflict Management, Park City, UT, 13, 59–77. Van Lange, P. A. M. (1999). The pursuit of joint outcomes and equality in outcomes: An integrative model of social value orientations. Journal of Personality and Social Psychology, 77, 337–349. Van Lange, P. A. M., & Kuhlman, M. D. (1994). Social value orientations and impressions of partner’s honesty and intelligence: A test of might versus morality effect. Journal of Personality and Social Psychology, 67, 126–141. Van Lange, P. A. M., Otten, W., De Bruin, E. M. N., & Joireman, J. A. (1997). Development of prosocial, individualistic, and competitive orientations: Theory and preliminary evidence. Journal of Personality and Social Psychology, 73, 733–746. Von Neumann, J., & Morgenstern, O (1947). Theory of games and economic behavior. Princeton, NJ: Princeton University Press. Wall, J. A. (1977). Intergroup bargaining: Effects of opposing constituents’ stance, opposing representative’s bargaining, and representatives’ locus of control. Journal of Conflict Resolution, 21, 459–474. Wall, J. A. (1991). Impression management in negotiations. In R. A. Giacalone & P. Rosenfeld (Eds.), Applied impression management, London: Sage.

MOTIVATION IN CONFLICT AND NEGOTIATION

291

Walton, R. E., & McKersie, R. (1965). A behavioral theory of labor negotiations: An analysis of a social interaction system. New York: McGraw-Hill. Webster, D., & Kruglanski, A. W. (1994). Individual differences in need for cognitive closure. Journal of Personality and Social Psychology, 67, 1049–1062. Weingart, L. R., Bennett, R. J., & Brett, J. M. (1993). The impact of consideration of issues and motivational orientation on group negotiation process and outcome. Journal of Applied Psychology, 78, 504–517. Weingart, L. R., Brett, J. M., & Olekalns, M. (2001). Mixed motivational orientations in negotiating groups: Convergence and reaching agreement. Paper presented at the Society for Industrial and Organizational Psychology 13th Annual Conference, Dallas, TX. Weingart, L. R., Hyder, E., & Prietula, M. J. (1996). Knowledge matters: The effect of tactical descriptions on negotiation behavior and outcome. Journal of Personality and Social Psychology, 70, 1205–1217. Weingart, L. R., Thompson, L., Bazerman, M. H., & Carroll, J. S. (1990). Tactical behavior and negotiation outcomes. International Journal of Conflict Management, 1, 7–31. Wilson, D., Near, D. C., & Miller, R. R. (1998). Individual differences in Machiavellianism as a mix of cooperative and exploitative strategies. Evolution and Human Behavior, 19, 203–212. Winter, D. G. (1996). Personality: Analysis and interpretation of lives. New York: McGraw-Hill. Yukl, G. A. (1974). The effects of situational variables and opponent concessions on a bargainer’s perception, aspirations, and concessions. Journal of Personality and Social Psychology, 29, 227–236. Yukl, G. A., Malone, M. P., Hayslip, B., & Pamin, T. A. (1976). The effects of time pressure and issue settlement order on integrative bargaining. Sociometry, 39, 277–281. Zartman, I. W. (1991). Regional conflict resolution. In V. A. Kremenyuk (Ed.), International negotiation (pp. 302–314). San Francisco: Jossey-Bass.

REGULATORY MODE: LOCOMOTION AND ASSESSMENT AS DISTINCT ORIENTATIONS

E. Tory Higgins Arie W. Kruglanski Antonio Pierro

The everyday conception of self-regulation is that individuals decide what they want that they don’t currently have, figure out what they need to do to get what they want, and then do it. This conception captures two key functions in self-regulation. The first is individuals assessing both the different goals to pursue and the different means to pursue them, and the second is individuals locomoting or ‘‘moving’’ from their current state in pursuit of some alternative goal pursuit state. Some version of these assessment and locomotion functions is contained in all major models of selfregulation. In classic control theory terms, for example, the purpose of regulation is to minimize the discrepancy between a current state and a desired end state, and self-regulation involves both moving toward the desired end state and determining the extent of discrepancy between the current state and the desired end state (e.g., Miller, Galanter, & Pribram, 1960; Powers, 1973; Wiener, 1948; for recent descriptions, see Carver & Scheier, 1990; Kuhl, 1985). The Rubicon model of self-regulation distinguishes between the preactional phases of deliberation and implementation and the action initiation and maintenance phase (e.g., Gollwitzer, 1990; Gollwitizer, Heckhausen, & Stellar, 1990; Heckhausen & Gollwitzer, 1987). In these models of self-regulation, as in the everyday conception, assessment and locomotion are conceived as working together as parts of the same goal pursuit orientation. They are integrated tools of goal attainment. Self-regulatory theories generally consider the functions of assessment and locomotion to be inseparable parts of the whole of 293 ADVANCES IN EXPERIMENTAL SOCIAL PSYCHOLOGY, VOL. 35

Copyright 2003 Elsevier Science (USA). All rights reserved. 0065-2601/03 $35.00

294

HIGGINS et al.

self-regulation and treat them as functionally interdependent. Thus, as motivational intensity increases, both locomotion and assessment should increase in intensity, yielding a positive correlation between them. Although we agree that such positive correlations do occur, we propose that locomotion and assessment can also function as independent orientations or modes that can receive differential emphasis by different people and in different situations. It is certainly useful to consider how locomotion and assessment are combined as parts of goal pursuit. It is also useful, however, to treat locomotion and assessment as distinct motivational orientations that have their own independent effects on motivation more generally. We propose that locomotion and assessment need to be studied as separate motivational principles in their own right. It our purpose in this chapter to demonstrate how the nature and consequences of goal pursuit vary depending on which of these orientations predominates.

I. Locomotion and Assessment as Independent Modes In its most general sense, to ‘‘regulate’’ means to govern or direct attention, resources, or action. Psychologists have paid considerable attention to the determinants and consequences of regulation, especially to the control aspects of attaining goals or meeting standards (see Carver & Scheier, 1990; Gollwitzer, 1990; Higgins, 1989; Kuhl, 1985, 1986; Mischel, 1974, 1981). In classic models of self-regulation, locomotion and assessment are regulatory means for attaining a desired end state. This chapter introduces regulatory mode theory that considers the locomotion and assessment modes to be both more general and more independent than described in classic theories of self-regulation (see also Kruglanski, Thompson, Higgins, Atash, Pierro, Shah, & Spiegel, 2000). Though locomotion and assessment may be thought of as interdependent ‘‘ingredients’’ of the larger self-regulatory system, we shall consider each individually. Specifically, we assume that the nature and the consequences of locomotion versus assessment are to some extent independent of each other. From an individual difference perspective, this independence would be reflected in the fact that not only are there individuals who are high in locomotion and low in assessment and individuals who are high in assessment and low in locomotion, but there are also individuals who are relatively high in both locomotion and assessment and other individuals who are relatively low in both. Even for individuals who are high in both, however, one or the other regulatory mode can be emphasized at any point in time. Specific activities can also emphasize one or the other mode. In this sense,

LOCOMOTION AND ASSESSMENT REGULATORY MODES

295

locomotion and assessment can compete for an individual’s resources (i.e., his or her time, energy, or attention), and whichever orientation is emphasized at any point will be the predominant influence on the nature and consequences of a goal pursuit. This chapter is concerned with what happens to judgement and decision making and to activity engagement and performance when individuals emphasize locomotion or assessment in their goal pursuits.

A. LOCOMOTION Our nature is movement; absolute stillness is death. —Pascal Fools are aye fond o’ flittin’, and wise men o’ sittin’. —J. Ray

In regulatory mode theory, the term ‘‘locomotion’’ does not refer only to movement from a current state toward a valued or desired end state. ‘‘Locomotion’’ refers to ‘‘moving from place to place’’ (Webster’s Ninth New Collegiate Dictionary, 1989, p. 701). According to field theory (see Deutsch, 1968; Lewin, 1951), a locomotion is any change of position of any region within the life space, typically a change of state from a current position toward or away from some position. Thus, locomotion can be away from, as well as toward, some end state, and the end state need not be a valued or desired end state. In keeping with this general view on locomotion, regulatory mode theory proposes that locomotion constitutes the aspect of self-regulation that is concerned with movement from state to state. In keeping with Lewin’s (1951) conceptualization, this change of state takes place psychologically and not necessarily behaviorally. Actual physical movement need not occur. As thus defined, the essential nature of locomotion from a motivational perspective involves simply moving away from a current state with no particular destination or direction in mind. As the intent to locomote gets translated into the specifics of actual movement, the direction of the motion will become determined. But the specific destination might still not be determined. Moreover, the direction that does get determined may do so without a great deal of involvment or deliberation on the individual’s part. For example, a high locomotor might engage in whichever activity is currently most accessible, irrespective of its particular direction. It is enough to ‘‘just do it.’’ The initial intent could simply be to move away from the current state— the intent to do ‘‘something, anything, other than the same thing.’’ The basic

296

HIGGINS et al.

orientation is for change, to do something different. Thus, the initial intent need not include any representation of direction or destination. The locomotion orientation is itself a dimension—it varies from high to low. From an individual difference perspective, some people are high locomotors, some are low locomotors, and some are in between. The quotations at the beginning of this section illustrate the viewpoint of a high locomotor (‘‘Our nature is movement; absolute stillness is death’’) and the viewpoint of a low locomotor (‘‘Fools are aye fond o’ flittin’, and wise men o’ sittin’ ’’). Situations can also vary in whether they are likely to induce a high locomotion orientation or a low locomotion orientation. Locomotion, of course, is involved in classic goal pursuit given that it involves a motivation to move from a current state to a different desired end state, and in this case the initial intent does represent a direction and destination. Thus, all goal pursuit involves locomotion. But in contrast to classic self-regulatory models, regulatory mode theory does not assume that locomotion always involves goal pursuit in the classic sense of having a destination in mind, a destination that one is moving toward. For locomotion, it is enough to be moving away from one’s current state because such movement would involve a change of state. Indeed, from the perspective of regulatory mode theory, a high locomotor can be motivated to move not only as a means to arrive at some desired destination (i.e., exogenous) but also as an end in itself (i.e., endogenous). Change of state is its own reward. In this chapter we are interested in the case in which the primary motivation is simply to initiate and maintain locomotion as well as when the motivation is to decrease distance from some desired end state (i.e., approach) or to increase distance from some undesired end state (i.e., avoid). Locomotion motivation involves a desire to move away from the current state even when the current state is inherently positive. It is not that a person is motivated to avoid an inherently negative state. Rather, it is the condition of immobility that a person is motivated to change. Moreover, even when the move is toward a desired end state, this goal may be the (epiphenomenal) consequence rather than the cause of the locomotion motivation. It is the initiation and maintenance of locomotion that are desired. It is a failure to initiate or maintain locomotion that is undesired. As Goethe said, ‘‘He who moves not forward goes backward!’’

B. ASSESSMENT We might remind ourselves that criticism is as inevitable as breathing. —T. S. Eliot

LOCOMOTION AND ASSESSMENT REGULATORY MODES

297

Comparisons are odorous. —Shakespeare

Regulatory mode theory is also concerned with ‘‘assessment’’ in the most general sense. As in classic control theories (see, for example, Carver & Scheier, 1990; Kuhl, 1985), the assessment function includes measuring the discrepancy between the current state and some desired end state, as well as the rate of progress in reducing the discrepancy. But other kinds of assessment are included as well. Most generally, ‘‘assessment’’ refers to determing the rate, amount, size, value, or importance of something, to appraise critically for the purpose of understanding or interpreting, or as a guide in taking action (Webster’s Ninth New Collegiate Dictionary, 1989, pp. 109 and 426). Thus, the value or importance of both the current state and the end state can be independently assessed, as well as the value or utility of the means used to move toward or away from that end state. Assessment constitutes the comparative aspect of self-regulation that includes critically evaluating entities or states, such as goals or means, in relation to alternatives in order to judge relative quality (e.g., judging the quality of something by considering both its merits and demerits in comparison with an alternative). More generally, assessment involves measuring, interpreting, or evaluating something through comparing one thing to another. In keeping with this general view on assessment, regulatory mode theory proposes that assessment constitutes the aspect of selfregulation that is concerned with making comparisons. In classic control theories (for an overview, see Carver & Scheier, 1981, 1990), assessment tends to be equated with monitoring the relation between the current state and the desired (or undesired) end state, which provides the self-regulator with feedback about ‘‘How am I doing?’’ Comparing the current state to the desired end state does involve assessment. However, feedback about how one is doing is necessary for both locomotion and assessment to function effectively and efficiently. A locomotion orientation needs feedback about whether a change of state has occurred and whether it is being maintained. An assessment orientation needs feedback about whether a comparison process has been initiated and whether it needs to continue. Thus, monitoring as feedback is important for both locomotion and assessment and needs to be treated as a separate function. As for locomotion, the assessment orientation is also a dimension—it too varies from high to low. From an individual difference perspective, again some people are high assessors, some are low assessors, and some are in between. The quotations at the beginning of this section illustrate the viewpoint of a high assessor (‘‘We might remind ourselves that criticism is as inevitable as breathing.’’) and the viewpoint of a low assessor

298

HIGGINS et al.

(‘‘Comparisons are odorous.’’). And once again, situations can also vary in whether they are likely to induce a high assessment orientation or a low assessment orientation. ‘‘It is critical vision alone which can mitigate the unimpeded operation of the automatic’’ (Marshall McLuhan). In sum, both individuals and situations can vary in the extent to which locomotion or assessment is emphasized. Because both locomotion and assessment are dimensions, there are many possible combinations of different levels of locomotion and assessment that can be emphasized by different individuals or in different situations. There are some stereotypic cases with which we are all familiar, however. The lone cowboy drifter in Western movies represents a case of high locomotion/low assessment. Mercenaries might be a real life example of this type. When high assessment is added to high locomotion in the movies, a secret agent is born, such as James Bond who has discriminating taste in addition to a love for adventure. International spies might be a real life example of this type. Other cultural icons of individuals high in both locomotion and assessment would be movie portrayals of historical figures such as Napoleon or Ghandi. A case of higher assessment with lower locomotion would be classic movie detectives who are crime solvers rather than action figures, such as Sherlock Holmes or Poirot. Then there are the comedic characters in the movies who are high in neither assessment nor locomotion. A classic comedic character like Lou Costello, or the recent comedic characters portrayed by Adam Sandler or Ben Stiller, have things happen to them rather than making things happen. Occasionally, relatively pure types are portrayed together in the same story. In Winnie-the-Pooh, for example, Tiger is high in locomotion, Eeyore is high in assessment, Christopher Robin is high in both, and lovable Pooh is high in neither. There are also activities that are stereotypically associated with the different types, such as skipping rope as locomotion, birdwatching as assessment, playing bridge or poker as both, and lying on the beach as neither. We should note that the independence of locomotion and assessment permits many possible effects to be revealed. One possibility is the classic case of locomotion and assessment working together in an interdependent manner such that both need to be present for some outcome to occur. The nature of such interdependency can take many forms. For example, as we discuss later, high achievement often requires both a correct choice (high assessment) and a commitment to following through on the choice (high locomotion). There can also be the case where an outcome varies as a function of level of locomotion but not level of assessment, and the reverse case is also possible. Finally, there is the case in which an outcome is influenced by both locomotion and assessment but their effects are in opposite directions. We will present evidence in this chapter for each of these cases.

LOCOMOTION AND ASSESSMENT REGULATORY MODES

299

II. Discriminating Locomotion and Assessment from Each Other Compared to classic theories of self-regulation, regulatory mode theory is concerned with the locomotion and assessment functions more generally and as independent from one another. A key difference between regulatory mode theory and control theories of self-regulation is that control theories consider the primary function of regulation to be minimizing the discrepancy between the current state and some desired end state (or increasing the discrepancy from undesired end states). Locomotion and assessment are treated as the means to serve this function. Locomotion is typically treated as the primary means, with assessment serving a subsidiary role of providing feedback about the progress of the locomotion. In contrast, regulatory mode theory proposes that either the locomotion or the assessment function can receive primary emphasis as an end in itself. They have equal status as distinct regulatory orientations. The term ‘‘mode’’ in regulatory mode theory is intended to capture this notion that locomotion and assessment have equal regulatory status, with either function being capable of receiving primary emphasis. ‘‘Mode’’ refers to a particular form or variety of something, a particular functioning arrangement (in this case, distinct motivational orientations), and a customary or preferred way of doing something (Webster’s Ninth New Collegiate Dictionary, 1989, p. 762). Regulatory mode theory proposes that individual and situational variables, both momentary and chronic, can cause either the locomotion mode or the assessment mode to become the customary or the preferred orientation to goal pursuit. To provide an initial flavor for how locomotion and assessment are different, let us begin with a recent study by Camacho and Higgins (2001a) on cultural maxims or proverbs. The participants in the study had all previously filled out the Regulatory Mode Questionnaire, which measures chronic individual differences in strength of locomotion orientation and strength of assessment orientation (see Kruglanski et al., 2000). Table I shows the locomotion and assessment items in this questionnaire. For both scales, respondents indicate the extent to which they endorse each item by responding to a six-point Likert scale ranging from 1 (strongly disagree) to 6 (strongly agree). Across different samples, the coefficient alphas for each scale range from .76 to .86, the temporal stability (test–retest correlations) for each scale is around r ¼ .75 on average, and the intercorrelation of the two scales is only slightly positive (typically r < .15). Embedded among the locomotion and assessment items is a set of six ‘‘faking’’ items designed to tap self-promotional or socially desirable tendencies in participants’ responding (e.g., ‘‘I have never been late for work

300

HIGGINS et al. TABLE I Items Comprising the Locomotion and Assessment Scalesa

1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12.

Locomotion Items I don’t mind doing things even if they involve extra effort. I am a ‘‘workaholic.’’ I feel excited just before I am about to reach a goal. I enjoy actively doing things, more than just watching and observing. I am a ‘‘doer.’’ When I finish one project, I often wait awhile before getting started on a new one. (–) When I decide to do something, I can’t wait to get started. By the time I accomplish a task, I already have the next one in mind. I am a ‘‘low energy’’ person. (–) Most of the time my thoughts are occupied with the task I wish to accomplish. When I get started on something, I usually persevere until I finish it. I am a ‘‘go-getter.’’ Assessment Items I never evaluate my social interactions with others after they occur. (–) I spend a great deal of time taking inventory of my positive and negative characteristics. I like evaluating other people’s plans. I often compare myself with other people. I don’t spend much time thinking about ways others could improve themselves. (–) I often critique work done by myself or others. I often feel that I am being evaluated by others. I am a critical person. I am very self-critical and self-conscious about what I am saying. I often think that other people’s choices and decisions are wrong. I rarely analyze the conversations I have had with others after they occur. (–) When I meet a new person I usually evaluate how well he or she is doing on various dimensions (e.g., looks, achievements, social status, clothes).

a In our validation research, the locomotion and assessment items were interspersed in a single questionnaire along with six faking items. They were not presented as separate sets. This ordering may be obtained by request from the authors.

or for an appointment’’). Across many samples, neither higher locomotion nor higher assessment has been found to be associated with faking. (For a full review of the psychometric construction and testing of the scales, see Kruglanski et al., 2000.) In addition to the data from the United States and Italy described in Kruglanski et al. (2000), recent studies in Germany, Spain, Israel, South Korea, and Japan have provided cross-cultural validation for the structure of the locomotion and assessment scales (Higgins, Pierro, & Kruglanski, 2002). The participants in the Camacho and Higgins (2001a) study also answered a second questionnaire in which they were asked to indicate how much they agreed or disagreed with each of a set of maxims or proverbs. To identify

LOCOMOTION AND ASSESSMENT REGULATORY MODES

301

TABLE II Agreement with Proverbs as a Function of Locomotion and Assessment Proverbs with which high locomotors agreed (+) or disagreed () (+) ‘‘If at first you don’t succeed, try, try again’’ (+) ‘‘Never put off till tomorrow what you can do today’’ (+) ‘‘A stitch in time saves nine’’ (+) ‘‘Too many cooks spoil the broth’’ () ‘‘He that fights and runs away may live to fight another day’’ Proverbs with which high assessors agreed (+) (+) ‘‘The road to hell is paved with good intentions’’ (+) ‘‘Let the buyer beware’’ (+) ‘‘Think on the end before you begin’’ (+) ‘‘There is a scorpion under every stone’’ (+) ‘‘It is easier to descend than to ascend’’

unique relations between either locomotion or assessment and each proverb that controlled for the alternative mode, simultaneous multiple regressions were conducted that included both the locomotion index and the assessment index as predictors of agreement with each proverb. Table II shows the proverbs with which high locomotors uniquely agreed [or disagreed], and the proverbs with which high assessors uniquely agreed. What Table II illustrates is that high locomotors agreed with proverbs that imply that forward movement is good and obstacles to forward movement are bad. In contrast, high assessors agree with proverbs that imply that consideration of all possibilities—negative as well as positive—is good, and insufficient attention to all possible alternatives is bad. A recent study by Pierro, Higgins, and Kruglanski (2002) also provides a flavor of the difference between locomotion and assessment orientations. The participants in this study completed a measure of their film preferences for different types of film, including action/adventure films, horror films, science fiction/fantasy films, and dramatic films. The study found that as locomotion mode (but not assessment mode) increased, preference for action/adventure films and comedy films increased. In contrast, as assessment mode (but not locomotion mode) increased, preference for dramatic films increased. Action/adventure films and comedy films share the property of having quick and sudden changes in what is happening to the central characters—either what they make happen (e.g., action heros) or what happens to them (e.g., slapstick comedians). In contrast, dramatic films have a slower development in understanding and evaluating what the central characters are like. If locomotion and assessment are different motivational orientations, then they should have different relations to other psychological variables.

302

HIGGINS et al.

Using the Regulatory Mode Questionnaire, Kruglanski et al. (2000) identified several specific characteristics that have distinct relations to chronic individual differences in locomotion orientation and chronic individual differences in assessment orientation. We next review these distinct relations.

A. SELF-EVALUATIVE CONCERNS Assessment involves making comparisons. An especially important comparison in self-regulation is comparing the self to some standard, where the standard could be another person or some valued end state. Thus, higher assessment should be associated with stronger self-evaluative concerns. Selfevaluative concerns should be less relevant to the locomotion orientation. 1. Self-Consciousness and Social Comparison Kruglanski et al. (2000) measured self-consciousness with Fenigstein, Scheier, and Buss’s (1975) scales of public and private self-consciousness. Public self-consciousness refers to a ‘‘general awareness of self as a social object,’’ and private self-consciousness refers to ‘‘attending to one’s inner thoughts and feelings’’ (p. 523). Comparison with standards was measured with the need for the social comparison subscale of Hill’s (1987) Interpersonal Orientation Scale. As predicted, assessment was strongly and positively associated with public self-consciousness, private selfconsciousness, and need for social comparison. In contrast, locomotion had little or no relation to these variables. As expected, the relation between each of these measures of self-evaluative concerns and assessment was significantly greater than the relation to locomotion. 2. Sensitivity to Social Appraisal and Social Norms Related to public self-consciousness is a sensitivity to others’ expectations for one’s behavior. One aspect of this sensitivity is concern with others’ appraisal of one’s actions, and another aspect is being influenced by social norms. In a recent study with German participants, Higgins, Pierro, and Kruglanski (2002) measured participants’ sensitivity to criticism, shyness, and fear of being negatively evaluated in social situations using Kuhl and Kazen’s (1997) measure of the self-critical and insecure personality style. The study found that higher assessment was positively associated with sensitivity to social criticism, whereas higher locomotion was negatively associated. Such sensitivity to social criticism should in turn increase anxiety

LOCOMOTION AND ASSESSMENT REGULATORY MODES

303

in social interactions. Consistent with this prediction, Higgins et al. (2002) found in additional studies with both American and Italian participants that higher assessment was positively related to feeling anxious in social interactions, as measured by the Interaction Anxiety Scale (Leary, 1983). In contrast, higher locomotion had a negative relation to interaction anxiety. Individuals with a high assessment orientation want to do what is best, what is right. Doing the right thing includes being sensitive to social norms. Success and failure to meet social norms are other aspects of comparing oneself to standards. Thus, one would expect individuals with a high assessment orientation to be more influenced by social norms. A recent study by Pierro, Mannetti, Higgins, and Kruglanski (2002) investigated this possibility as part of a study testing Fishbein and Ajzen’s theory of reasoned action and theory of planned behavior (Ajzen, 1988, 1991; Ajzen & Fishbein, 1980; Fishbein & Ajzen, 1975). These theories postulate that overt behavior is a function of a person’s intentions, which in turn depend on that person’s attitude toward the behavior, his or her perceived behavioral control, and his or her subjective norms. We investigated whether these postulated relations are influenced by individuals’ locomotion and assessment orientations in a longitudinal study on frequency of attendance at gym exercise classes. The participants consisted of people who registered for gym classes in Rome, Italy. At the moment of registration at the gym course, the participants filled out a questionnaire containing measures of the study variables: locomotion mode, assessment mode, attitudes toward ‘‘regularly attending gym classes in the next 6 months’’ (assessed through semantic differential scales for ‘‘useful,’’ ‘‘advantageous,’’ ‘‘wise,’’ etc.), perceived behavioral control (e.g., ‘‘for me regularly attending gym classes in the next 6 months is . . . 1 ¼ difficult to 7 ¼ easy’’), and subjective norms. The subjective norms were measured with two items: ‘‘Most people who are important to me would approve that I continue regularly attending gym classes in the next 6 months’’ and ‘‘My closest friends would approve that I continue regularly attending gym classes in the next 6 months.’’ The dependent measure was behavioral intentions, which was measured by asking participants to express their intentions to regularly attend gym classes in the next 6 months. As predicted by the Theory of Reasoned Action, intentions to exercise regularly were predicted by both attitudes and perceived behavioral control. They were also predicted by an interaction of assessment and subjective norms. As predicted, this interaction reflected the fact that intentions to exercise were increased by social norms favoring exercise when individuals had a higher assessment orientation. This finding is consistent with the notion that higher assessment is related to greater

304

HIGGINS et al.

concerns with normative standards and doing what is ‘‘right.’’ Locomotion in this study did not influence intentions. Locomotion did influence the relation between intentions and overt behavior, however, as we discuss later in the section on achievement performance.

3. Self-Appraisal Vulnerability As we have seen, higher assessment is associated with sensitivity to social appraisal by others. Higher assessment should also be associated with sensitivity to self-appraisal. Because of their concerns with comparing themselves to others and to standards, high assessors are more likely than low assessors to evaluate themselves continuously in relation to standards. Some of these standards are valued end states, such as personal hopes and aspirations (ideals). When individuals evaluate themselves in relation to such standards, this can make accessible any discrepancies they possess (see, for example, Duval & Wicklund, 1972; Higgins, 1987). Continuous self-evaluation is also likely to involve a wide range of standards, including very high and very low standards. Thus, even individuals doing quite well are likely to compare themselves sometimes to standards that make them feel like relative failures. Continous selfevaluation, then, is a vulnerability factor to experiencing failures. Combined with their greater sensitivity to social criticism described earlier, this would make high assessors vulnerable to negative affect, lower self-esteem, and lower optimism. As expressed by the British writer Jan Morris, ‘‘The command of comparisons that develops with age is, while no doubt good for the mind, somewhat numbing for the emotions.’’ It is also likely to produce emotional instability from experiencing both successes and failures relative to low and high standards of comparison. Once again, because selfevaluative concerns are less relevant to locomotion, one would not expect high (versus low) locomotors to show these effects. Kruglanski et al. (2000) measured negative affect with two measures of social anxiety (Fenigstein et al., 1975; Leary, 1983) and a measure of depression (Radloff, 1977). They measured self-esteem with Rosenberg’s (1965) Self-Esteem Scale, and optimisim with the Life Orientation Test— Revised (Scheier, Carver, & Bridges, 1994). As predicted, higher assessment had significant positive correlations with both social anxiety and depression, and significant negative correlations with self-esteem and optimism. Locomotion had significant relations in the opposite direction. A recent study by Higgins et al. (2002) with German participants also found that higher assessment was negatively related to self-esteem on Heatherton and Polivy’s (1991) measure and was positively related to stress load (e.g.,

LOCOMOTION AND ASSESSMENT REGULATORY MODES

305

experiencing difficulties in one’s life) on Kuhl and Fuhrmann’s (1998) measure. What underlies these different relations? An important aspect of selfregulation is people’s representations of their own or others’ hopes or wishes for them (i.e., Ideal self-guides) or people’s representations of their own or others’ beliefs about their duties or obligations (i.e., Ought self-guides). These self-guides (Higgins, 1987, 1990) function as both goals toward which people are motivated to move and as standards in relation to which current states are assessed. Some modes of socialization that involve frequent and consistent intervention and discipline increase the likelihood that strong selfguides will be acquired. People who acquire strong self-guides are more likely to possess both high locomotion motivation to move toward their selfguides as goals and high assessment motivation to evaluate their current states in relation to their self-guides as standards. Indeed, in a study comparing ‘‘first and only’’ borns to ‘‘later’’ borns, where the former have stronger self-guides than the latter, Newman, Higgins, and Vookles (1992) found that ‘‘first and only’’ borns had smaller self-discrepancies to their ideal and ought self-guides (reflecting their stronger positive locomotion toward their goals), while at the same time emotionally suffering more from whatever discrepancies remained (reflecting their stronger negative assessment of any failure to meet their standards). It is possible, of course, for individuals with strong self-guides to differ in the strength of their locomotion and assessment modes for other reasons. Everything else being equal, a difference in strength of locomotion versus assessment would produce differences in self-esteem and emotional vulnerability. Stronger locomotion would increase the sense of personal success in making progress toward one’s goals, whereas stronger assessment would increase the sense of personal failure in not fully meeting one’s standards. Thus, one would expect that individuals with a strong locomotion orientation would generally have higher self-esteem than individuals with a strong assessment orientation. Also contributing to the positive relation between locomotion and high self-esteem and optimism would be the fact that a positive outlook facilitates locomotion toward goals (e.g., Taylor & Brown, 1988). In addition, high (versus low) locomotors are more concerned with change per se rather than comparing themselves to high standards, which would make them less likely to suffer from a sense of failure. It should also be noted that there can be a cost to high locomotors being motivated to have a positive outlook on their performance and being less self-critical. In a recent study by Camacho and Higgins (2001b), participants were asked to think of a time in their past when they failed to do the right thing, did not do what would have been the moral thing to do. They were

306

HIGGINS et al.

then asked how they felt about failing to perform this moral action. On measures of how ‘‘guilty’’ they felt and how ‘‘ashamed’’ they felt, higher locomotion (but not higher assessment) was associated with feeling less guilty and less ashamed for past moral inactions. (Feelings of guilt or shame did not vary as a function of either locomotion or assessment when the participants were asked how they felt about committing an immoral action in the past.) We have also seen that on Kuhl and Kazen’s (1997) measure of sensitivity to social criticism, higher locomotion is associated with being less sensitive. 4. Emotional Instability As mentioned above, continuous self-evaluation is likely to involve a wide range of standards, including very high and very low standards. It is likely to produce emotional instability from experiencing both successes and failures relative to low and high standards of comparison. Kruglanski et al. (2000) measured emotional instability by using the neuroticism factor (emotional instability) from the NEO-Five Factor Inventory (Costa & McCrae, 1992) and the emotional stability factor from the Sixteen Personality Factor (PF) Questionnaire (Cattell, Cattell, & Cattell, 1993). As predicted, Kruglanski et al. (2000) found for both an American sample of participants and an Italian sample of participants that higher assessment was positively related to emotional instability whereas higher locomotion had the opposite relation. This finding was recently replicated with both a sample of Japanese participants and a sample of German participants (Higgins et al., 2002). 5. Discomfort with Ambiguity Despite the fact that making comparisons to high as well as low standards has the potential cost of increasing negative affect and emotional instability, it will be done by individuals with a strong assessment orientation because it is the comparison process itself that matters in that orientation and not simply the outcomes of the process. There is a functional benefit, of course, in the increased accuracy that can come from a full and exhaustive evaluation process. Individuals with a strong assessment orientation want to be accurate. They want to find the best alternative, to ‘‘do the right thing.’’ To do so, they cannot afford ambiguity. They need to have confidence in and certainty about their judgments. Thus, one would expect assessment to be positively associated with ‘‘discomfort with ambiguity.’’ Kruglanski et al. (2000) included among its measures the Discomfort With Ambiguity subscale of Webster and Kruglanski’s (1994) Need For Cognitive Closure

LOCOMOTION AND ASSESSMENT REGULATORY MODES

307

Scale. As predicted, higher assessment was strongly associated with higher discomfort with ambiguity. In contrast, discomfort with ambiguity was not related to locomotion. This finding has been replicated with a sample of Italian participants (Higgins et al., 2002). 6. Performance (versus Learning) Goal Orientation Individuals with greater self-evaluative concerns are also more likely to want to prove their competence, to prove that they can meet some standard of success and do better in comparison to others. In the literature on achievement motivation, this has been referred to as having a performance orientation, and it is contrasted with a mastery or learning orientation that involves the development of skills (see, for example, Dweck, 1991). Higher assessment should be associated with emphasizing performance rather than mastery. If anything, one might expect the opposite to be true for higher locomotion because mastery or learning provides an opportunity for continued movement and change. Kruglanski et al. (2000) measured these goal orientations with the Goal Orientation Scale (GOS; Button, Mathieu, & Zajac, 1996) that includes items tapping both performance orientation (e.g., ‘‘I prefer to do things that I can do well rather than things that I do poorly’’) and mastery orientation (e.g., ‘‘The opportunity to do challenging work is important to me’’). The analyses controlled for the positive correlation between performance and mastery goals. As predicted, higher assessment was found to be uniquely related to having a performance but not a mastery orientation, whereas higher locomotion was found to be uniquely related to having a mastery but not a performance orientation. A recent study by Taylor and Higgins (2002) with MA organizational students replicated these results. They found that performance orientation had a strong positive correlation with assessment (and slightly negative with locomotion), and learning orientation had a strong positive correlation with locomotion (and slightly negative with assessment). They also found that higher assessment was positively related to believing that people cannot change the kind of person they are (with higher locomotion having a nonsignificant negative relation to this belief). This pattern of findings demonstrates that assessment is distinct from locomotion by having higher self-evaluative concerns that produce sensitivity to social appraisal and social norms, affective vulnerabilities, emotional instability, discomfort with ambiguity, and a performance orientation. All self-regulatory orientations have trade-offs. High assessors are oriented to compare themselves to standards, which means they will compare themselves to high standards as well as to low standards. They can even perform a within-self comparison in which they compare their

308

HIGGINS et al.

own ‘‘average’’ qualities to their own strongest qualities. In general, this selfevaluative style has the potential cost of increasing failure experiences from self-comparisons to high standards. By comparing to both high and low standards, it also has the potential cost of increasing emotional instability. The benefit part of the trade-off for high assessment would include being willing to consider alternative possibilities. Indeed, Kruglanski et al. (2000) found that higher assessment was positively associated with the trait of reflectiveness or openness to experience as measured by both the NEO-Five Factor Inventory (Costa & McCrae, 1992) and the Sixteen Personality Factor Questionnaire (Cattell et al., 1993). In contrast, locomotion had no relation to this trait. Another possible benefit is that high assessors might be more realistic and more accurate in their self-evaluations than low assessors, and they might have a better idea of areas for self-improvement. These possibilities need to be examined in future research.

B. ACTIVITY FLOW CONCERNS People in a strong locomotion orientation do not like to stand still, to do nothing. They like to move along, to flow. Locomotion involves initiating and maintaining movement, changing from state to state. To initiate movement or change, it is important to be decisive in decision making because decisiveness allows action to proceed. Thus, higher locomotion should be associated with higher decisiveness. If anything, one might expect higher assessment to be associated with lower decisiveness because a more extensive comparison of alternatives and relating them to valued standards would complicate and prolong decision making, which would impede decisiveness. 1. Decisiveness Kruglanski et al. (2000) used several measures of decisiveness: (1) the Decision-Related Action versus State Orientation subscale of Kuhl’s (1985) Action—Control Scale (i.e., the Action—Decision subscale), which measures commitment to prompt action; (2) Thompson, Naccarato, and Parker’s (1989) Personal Fear of Invalidity Scale, which measures individual differences in the indecisiveness associated with evaluation apprehension; (3) Dickman’s (1990) Functional Impulsivity Scale, which measures the tendency to spend less time on preactional decision processes before taking action when to do so is functional; and (4) the Decisiveness subscale of Webster and Kruglanski’s (1994) Need for Cognitive Closure (NFCC) Scale, which measures the decisiveness of respondents’ judgments and

LOCOMOTION AND ASSESSMENT REGULATORY MODES

309

choices (e.g., ‘‘I usually make important decisions quickly and confidently’’). As predicted, for both a sample of American participants and a sample of Italian participants, the pattern of results clearly showed that higher locomotion was associated with more decisiveness and higher assessment was associated with less decisiveness. This pattern of results has been recently replicated with a new sample of Italian participants (Higgins et al., 2002).

2. Flow Locomotion involves not only the initiation of movement or change but also its maintenance. Thus, locomotion needs not only decisiveness to proceed with action but also commitment to maintain action and deal with interruptions and obstacles. Higher locomotion, then, should be associated with greater attention, persistence, and vitality when engaging activities, with increased level of experiential involvement, i.e., with greater flow (e.g., Csikszentmihalyi, 1975). Higher assessment on the other hand should be, if anything, associated with lesser flow because stopping to evaluate and reevaluate alternatives during activity engagement could disrupt smooth task flow. Kruglanski et al. (2000) used several measures of flow: (1) attentional control (e.g., ‘‘I tend to be quite wrapped up and interested in whatever I am doing,’’ ‘‘No matter how hard I try to concentrate, thoughts unrelated to my work always creep in’’ [reverse-scored]; (Huba, Aneshensel, & Singer, 1981); (2) conscientiousness (e.g., ‘‘I’m pretty good about pacing myself so as to get things done on time,’’ ‘‘I work hard to accomplish my goals’’; Costa & McCrae, 1992); and (3) vitality (e.g., ‘‘I feel alive and vital’’; Ryan & Frederick, 1997). As predicted, higher locomotion had a strong positive association with each of these measures, whereas higher assessment had little relation or a somewhat negative relation to them. The recent study by Higgins et al. (2002) with German participants also found that higher locomotion (but not assessment) had a positive association with Kuhl and Fuhrmann’s (1998) measure of self-regulatory competence and self-maintenance, which has properties such as maintaining control over one’s attention, motivation, emotion, planning, and decisions during task engagement. The same study found that higher locomotion had a negative association with Kuhl and Fuhrmann’s (1998) measure of volition inhibition, which has properties such as being distractible, ruminating, postponing, rigid, overcontrolled, and unable to get going. In contrast, higher assessment had a positive association with these properties of volition inhibition.

310

HIGGINS et al.

This latter finding relates to the self-regulatory distinction between behavioral activation versus behavioral inhibition. Decisiveness and commitment to initiating and maintaining movement or change involve taking action rather than inaction. One would expect, therefore, that higher locomotion (compared to higher assessment) would be associated more positively with the behavioral activation system (BAS) than the behavioral inhibition system (BIS) (see Carver & White, 1994). The results of a recent study with Italian participants (Higgins et al., 2002) supports this conclusion. Higher locomotion (but not higher assessment) was positively associated with BAS, especially with the Drive subscale that measures motivation to pursue things and willingness to take risks in doing so. In contrast, locomotion, if anything, had a mildly negative correlation with BIS, whereas assessment had a moderate positive association with BIS consistent with a wait and check-it-out attitude. 3. Intrinsic (versus Extrinsic) Another indicator of flow is the experience that people have when they engage in an activity for its own sake rather than as just a means to an end, or when they feel autonomous rather than controlled while engaging in an activity. This distinction is generally referred to as intrinsic versus extrinsic activity engagement (see Deci & Ryan, 1985; Kruglanski, 1975; Sansone & Harackiewicz, 1996). For the reasons described earlier, higher locomotion should be associated more with intrinsic than extrinsic engagement, whereas, if anything, the reverse should be true for higher assessment. Kruglanski et al. (2000) used a couple of measures of intrinsic and extrinsic motivational orientation: (1) the Work Preference Inventory (WPI; Amabile, Hill, Hennessy, & Tighe, 1994) and (2) the five subscales of Academic Motivation Scale (AMS) of Vallerand et al. (1992). Kruglanski et al. (2000) found, as predicted, that higher locomotion was associated with greater intrinsic than extrinsic motivation, whereas higher assessment was associated with greater extrinsic than intrinsic motivation. Pierro, Kruglanski, and Higgins (2002a) recently replicated these distinct relations with three different samples from diverse organizational settings in Italy. In two additional studies, Pierro et al. (2002a) used Sheldon and Elliot’s (1999) self-concordance measure of intrinsic motivation and their measure of effort investment during goal pursuit (with items such as ‘‘I intend to devote all my energies to pursuing these goals?’’). Both studies found that higher locomotion, but not higher assessment, was related to higher intrinsic motivation, which in turn predicted higher effort investment during goal pursuit. In addition to this mediated effect, higher locomotion also had a direct positive relation to higher effort investment that was

LOCOMOTION AND ASSESSMENT REGULATORY MODES

311

independent of intrinsic motivation. These results suggest that higher locomotion can enhance flow during goal pursuit both by increasing intrinsic motivation and by increasing commitment to action maintenance independent of intrinsic motivation. 4. Activity Involvement In another program of research, Pierro, Kruglanski, and Higgins (2002b) examined the relation between locomotion and flow in another way. They hypothesized that activity involvement might contribute to the relation between locomotion and effort investment. They again tested this hypothesis in an organizational setting in Italy. The first set of studies tested the relation between locomotion and effort investment, where effort investment (see Brown & Leigh, 1996) was measured in terms of workintensity items (e.g., ‘‘When there is a job to be done, I devote all my energy to getting it done’’) and time commitment items (e.g., ‘‘I put in more hours throughout the year than most of our salespeople do’’). For three different samples of participants (state employees, firemen, bank clerks), higher locomotion was positively related to higher effort investment, conceptually replicating the findings of Pierro et al. (2002a) described earlier with a different measure of effort investment. Job involvement can be defined as psychological identification with one’s job (Kanungo, 1982). It was measured in our study using Lodhal and Kejner’s (1965) Job Involvement Scale, which is an eight-item instrument with statements such as ‘‘I am very much personally involved in my job’’ or ‘‘The most important things that happen to me involve my work.’’ The study found, as predicted, that higher locomotion was related to higher job involvement, which in turn predicted higher effort investment. In addition to this mediated effect, higher locomotion also had a direct positive relation to higher effort investment that was independent of job involvement. Analogous to the intrinsic motivation findings of Pierro et al. (2002a) discussed above, these results suggest that higher locomotion can enhance flow during goal pursuit both by increasing activity involvement (e.g., job involvement) and by increasing commitment to goal pursuit independent of activity involvement per se. 5. Openness to Change These studies demonstrate that individuals with a high locomotion orientation have high flow concerns. Do these concerns also translate into higher locomotion being associated with greater openness to change, i.e., ‘‘going with the flow’’? This question was addressed in a set of four studies

312

HIGGINS et al.

by Kruglanski, Pierro, Mannetti, Livi, Higgins, and Capozza (2002) that were conducted in various organizations in Italy and employed crosssectional as well as longitudinal designs. The participants in the first study were nurses in a hospital in Rome. These nurses had been recently subjected to several role changes related to new rules about the exercise of their profession. The participants in the second study were workers in a Postal Service Center who had recently been subjected to far-reaching changes because of a privatization of this sector, including organizational shifts that substantially modified old habits and conventions of being postal workers. Both studies used a measure of coping with organizational change (Judge, Thoresen, Pucik, & Welbourr, 1999) that considered both reactance to change (e.g., ‘‘The changes in this organization cause me stress,’’ item reversed) and adaptation to change (e.g., ‘‘When dramatic changes happen in this organization, I feel I can handle them with ease’’). Both studies found that workers with a higher locomotion orientation reported being better able to cope with change. A third study on municipal workers who were undergoing a major reorganization of the administration found this same relation on a new measure of coping with change in which the participants were asked to characterize their reactions to the organizational change on items reflecting reactance to change (e.g., insecure, worried, overwhelmed, pessimist) and items reflecting acceptance of change (involved, active, committed, enterprising). To examine the relation between locomotion and coping in a more predictive manner, the fourth study used a longitudinal design. This study also measured participants’ expectancies regarding change and the actual shift in work attitudes over time from the organizational change. Like the second study, postal workers were the participants. The first phase of the study collected measures of locomotion orientation and expectancies about change (e.g., expectancies about quality of the work, the physical environment, relations with co-workers, career opportunities, etc.). Higher locomotion was related to more positive expectancies about change. The second phase of the study took place approximately 3 months later. Coping with organizational change was measured as in the first and second studies. In both phases, positive work attitudes were measured by an organizational commitment scale (Meyer, Allen, & Smith, 1993) with items such as ‘‘I would be very happy to spend the rest of my career with this organization’’ and ‘‘I really feel as if this organization’s problems are my own,’’ and by Lodahl and Kejner’s (1965) job involvement scale with items such as ‘‘I am very much involved personally in my work.’’ Controlling for positive expectancy about change (which did not itself predict coping with change), higher locomotion was found to predict coping

LOCOMOTION AND ASSESSMENT REGULATORY MODES

313

with change. Higher locomotion was also found to predict a positive shift in organizational commitment and a positive shift in job involvement. In sum, this pattern of findings from multiple studies with many different types of measures, settings, and population samples demonstrates that locomotion, as distinct from assessment, has high activity flow concerns that produce greater decisiveness, attentional control, vitality, conscientiousness, and openness to change, as well as activity engagement with higher intrinsic motivation, effort investment, behavioral activation, maintenance, and involvement. Once again, all self-regulatory orientations have trade-offs. Because locomotion is concerned with change per se, it is possible for resources to be committed to the pursuit of a goal that is not worth the resources allocated to it. It would be like ‘‘much ado about nothing.’’ This potential downside of high locomotion also needs to be investigated in future research. There is, in fact, evidence (which we discuss later) that high locomotion alone is not sufficient to produce high achievement. It needs to be combined with high assessment.

III. Distinguishing Regulatory Modes from Other Orientations The previous section has reviewed evidence supporting the proposal of regulatory mode theory that locomotion and assessment are distinct motivational orientations. Regulatory mode theory is not the only model of self-regulation that distinguishes between different aspects of selfregulation. There are other motivational orientations that have at least a family resemblance to locomotion and assessment. Thus, it is important not only to distinguish locomotion and assessment from each other, but also to distinguish them from other orientations that have been described in the literature. Let us begin with a couple of models that also propose a conceptual distinction in terms of one or another function of self-regulation receiving greater emphasis.

A. OTHER MOTIVATIONAL ORIENTATIONS 1. Action versus State Orientations Kuhl (1985) distinguishes between action and state orientations receiving differential self-regulatory emphasis. According to Kuhl (1985), a fully developed intentional action structure includes a represented relation between a present state and a desired future state, action alternatives that

314

HIGGINS et al.

may transform the present state into the future state, and the commitment of the actor to perform the intended action under specified situational conditions. Individuals have an action orientation when their attention is focused on such a fully developed intentional action structure. In contrast, individuals have a state orientation when they perseverate on a particular state, such as a present state, a past state, or a future state, or when there is the absence of any coherent cognition, as in absentmindedness. When individuals have a state orientation, the intentional action structure is represented only a little if at all. Thus, individuals with a state orientation are absent an action orientation. The distinction between action versus state orientations is quite different from the distinction between locomotion versus assessment functions. The distinction between locomotion and assessment functions is not concerned with the extent to which a fully developed intentional action structure is represented. It is the role of commitment in Kuhl’s (1985) model that creates some relation between his and our distinction. According to Kuhl (1985), a state orientation fails to represent the relational component representing intended action that underlies commitment. This relational component is represented in an action orientation, however. Thus, Kuhl (1985) proposes that commitment to taking action is likely to be higher in the action than the state orientation. Actions should be initiated sooner, therefore, when individuals are in an action versus state orientation. Because locomotion is concerned with initiating and maintaining movement and change, higher locomotion should increase prompt action initiation. In contrast, higher assessment should, if anything, decrease prompt action initiation because the concern is with making extensive comparisons that could slow down action initiation. Thus, one might expect action versus state orientation (as measured by the Action—Decision subscale of Kuhl’s (1985) Action-Control Scale) to have a positive association with locomotion, and possibly a negative association with assessment. As described earlier, Kruglanski et al. (2000) did, indeed, find for the Action—Decision subscale of Kuhl’s (1985) Action—Control Scale (which measures commitment to prompt action) that higher locomotion had a moderately strong positive association with this subscale, whereas higher assessment had a weaker (but significant) negative relation. 2. Deliberation versus Implementation The notion that distinct self-regulatory functions can receive differential emphasis is also found in the Rubicon model of action phases in goaloriented behavior (see, e.g., Gollwitzer, 1990; Gollwitzer, Heckhausen, & Stellar, 1990; Heckhausen & Gollwitzer, 1987). This model was inspired by

LOCOMOTION AND ASSESSMENT REGULATORY MODES

315

Lewin’s distinction between goal setting and goal striving (e.g., Lewin, Dembo, Festinger, & Sears, 1944). The model takes a temporal perspective on the course of action and proposes that different self-regulatory phenomena are associated with different phases of self-regulation. The model distinguishes between (1) the deliberative function of establishing and committing to a preference or wish to be fulfilled, which forms a goal intention, and (2) the implemental function of planning and committing oneself to a particular course of action for wish fulfillment, which forms behavioral intentions. Mind-sets can emphasize either the deliberative or implemental functions. It is tempting to think of deliberation as having greater assessment concerns and implementation as having greater locomotion concerns. Indeed, there may be some differences in strength of association that could be explored in future research. What needs to be highlighted, however, is that both locomotion and assessment concerns play a role in both deliberation and implementation. Deliberation involves comparing and evaluating alternative goals, and implementation involves comparing and evaluating alternative means or plans. Thus, there are assessment concerns in each phase. Each phase needs to be initiated and progress maintained in order to reach a goal intention (deliberation) or to reach behavioral intentions (implementation). Thus, there are also locomotion concerns in each phase. It should be noted that the Rubicon model of action phases also distinguishes between the preactional phases of deliberation and implementation and the actual action phase of action initiation and effort exertion. Rather than assessment versus locomotion having differential associations with deliberation and implementation, it is possible that assessment concerns are stronger in preactional than actional phases whereas the reverse is true for locomotion concerns. This would also be worth investigating. It is important to note once again, however, that assessment and locomotion concerns, even if differentially weighted, would both be involved in every phase. Indeed, the associations between regulatory modes and Rubicon phases could be fruitfully examined from an opposite perspective. How would each phase in the Rubicon model play out differently when either locomotion or assessment concerns are emphasized? When deliberating, for example, one could be concerned with critically comparing alternative goals to select the best one or the right one, or one could be more concerned with completing deliberation promptly in order to move on quickly through implementation to actual action initiation. What is clear is that the functional distinctions in the Rubicon model are different than the regulatory mode theory distinction between assessment and locomotion functions, and, indeed, the fact that

316

HIGGINS et al.

they are different raises interesting research questions about the impact of these different functions on one another. In addition to distinguishing regulatory mode theory from other models of self-regulation, it is also important to show that locomotion and assessment as motivational orientations are themselves distinct from other orientations that have been described in the literature. Because these other orientations have a family resemblance to one or the other regulatory mode, it would not be surprising to find a positive relation in some cases. However, we would expect any relation to be only moderate if locomotion and assessment are, indeed, distinct orientations. We have already discussed one such relation—the relation to Kuhl’s (1985) distinction between action versus state orientation. As discussed earlier, one might expect action (versus state) orientation to have a positive association with locomotion because they both relate to commitment to prompt action, and, indeed, Kruglanski et al. (2000), did find that higher locomotion had a moderately strong positive association with Kuhl’s (1985) Action—Decision subscale that measures such commitment. (In contrast, higher assessment had a smaller but reliable negative relation.) It should also be noted, however, that this positive association was around r ¼ .40 across the different samples, and thus was still in the moderate range. 3. Need to Evaluate One might also expect a positive association between assessment and Jarvis and Petty’s (1996) Need to Evaluate Scale. The Need to Evaluate Scale measures individuals’ orientation to have strong opinions about everything (even when not personally involved), to have extreme likes and dislikes or preferences, and to pay attention to the goodness or badness of everything. This need to evaluate involves that aspect of assessment that has to do with comparing something to some qualitiative standard of goodness or badness. Thus, there should be some positive association between the need to evaluate and assessment. The positive association between the need to evaluate and assessment should be only moderate, however, because there are also important differences between these orientations. First, assessment is broader than the need to evaluate because it can involve quantitative as well as qualitative comparisons, and comparisons among things along nonevaluative dimensions, such as comparing people’s hair color, age, height, and so on. Second, the fact that an assessment orientation is concerned with making comparisons does not imply that strong or extreme judgments necessarily will be made. By comparing some target of evaluation to a broad range of positive and negative alternatives or to different standards, individuals with

LOCOMOTION AND ASSESSMENT REGULATORY MODES

317

an assessment orientation could identify both the strengths and the weaknesses of the target. This could both increase the variability of high assessors’ evaluative judgments and reduce their overall extremity. In fact, there is some evidence for these effects that we will describe later. It should be noted that such moderation does not mean that high assessors have fuzzy or wishy-washy opinions or preferences. As discussed earlier, high assessment is associated with disliking ambiguity. Individuals with a high assessment orientation want to have definite judgments, but not necessarily extreme ones. Indeed, the tendency under high assessment to carry out a full and exhaustive comparison process prior to any final judgment could increase confidence and certainty in that judgment, even when the judgment is moderate. Under high assessment, therefore, strong ‘‘moderate’’ judgments or attitudes are possible. But such confidence and certainty would derive from the exhaustive comparison process that produced the judgment rather than from any extremity of the judgment per se. Such confidence and certainty should not be confused with rigidity, however. As we also discussed earlier, higher assessment is positively associated with reflectiveness or openness to new experience. It would be useful in future research to test systematically the relation between higher assessment and increased variability, moderation, confidence, and reflectiveness in judgments. In sum, one would expect a positive, but moderately positive, association between Jarvis and Petty’s (1996) Need to Evaluate Scale and assessment. Consistent with this prediction, Kruglanski et al. (2000) found across different samples a positive association between these orientations around r ¼ .3. Also as expected, the Need to Evaluate had a significantly stronger association with assessment than with locomotion. This basic pattern of findings was recently replicated with a sample of Italian participants (Higgins et al., 2002). 4. Achievement Orientation In classic models of goal pursuit, locomotion is concerned with goal striving. One might expect, therefore, that higher locomotion orientation would be positively associated with an orientation toward achievement. Consistent with this prediction, Kruglanski et al. (2000) found that higher locomotion was positively associated with Jackson’s (1974) PRF measure of achievement orientation (e.g., ‘‘I enjoy difficult work’’), negatively associated with Herman’s (1990) measure of fear-of-failure (e.g., ‘‘When I start doing poorly on a task, I feel like giving up’’), and positively associated with the Type A behavior pattern that relates to personality characteristics such as ‘‘hard driving’’ and ‘‘speed/impatience.’’ These moderately positive

318

HIGGINS et al.

associations were all significantly stronger for higher locomotion than for higher assessment. Indeed, higher assessment actually had a small but significant positive association with fear-of-failure, which is probably related to the self-evaluative concerns of high assessment. 5. Extroversion When one thinks of people such as high locomotors who, among other things, strive for change and seek out opportunities to engage in activities, one naturally thinks of extraverts. Extraverts are talkative, friendly, bold, and spontaneous. A locomotion orientation is certainly broader than these personal characteristics, but one might expect there to be a positive association between higher locomotion and extraversion. Consistent with this prediction, Kruglanski et al. (2000) found for both American and Italian participants that higher locomotion had a moderate positive association with extraversion as measured by both the NEO-Five Factor Inventory (Costa & McCrae, 1992) and the Sixteen Personality Factor Questionnaire (Cattell et al., 1993), whereas assessment had no relation to extraversion. This pattern of results was replicated in studies by Higgins et al. (2002) with both German and Japanese participants. To further demonstrate that locomotion and assessment are distinct and discriminant motivational orientations, it is important to show that there are variables to which neither locomotion nor assessment is related. Of course, there are potentially a host of such variables. Kruglanski et al. (2000) checked on a few basic ones. They found that neither locomotion nor assessment was associated with participants’ self-reported political orientation (1 ¼ strongly liberal, to 6 ¼ strongly conservative). Neither was associated with the 16-item measure of social dominance orientation of Pratto et al. (1994), the tendency to support antiegalitarianism, or groupbased dominance in social relations. Neither was associated with a measure of ingroup favoritism (attitude toward one’s own racial–ethnic group minus mean attitude toward other racial–ethnic groups, e.g., ‘‘Asians,’’ ‘‘Blacks,’’ ‘‘Hispanics,’’ ‘‘Whites’’; see Shah, Kruglanski, & Thompson, 1998). Finally, and especially important, neither locomotion nor assessment had a significant association with age or with gender.

IV. Consequences of Distinct Locomotion and Assessment Modes The previous sections have provided evidence that the self-regulatory modes of locomotion and assessment are distinct from one another and

LOCOMOTION AND ASSESSMENT REGULATORY MODES

319

distinct from other self-regulatory orientations. This section considers the consequences of these distinct modes for judgment and decision making and for activity engagement and performance.

A. JUDGMENT AND DECISION MAKING Locomotion constitutes the aspect of self-regulation that is concerned with movement from state to state, whereas assessment constitutes the aspect of self-regulation that is concerned with making comparisons. These different self-regulatory orientations can influence the process of decision making in various ways, including preferences for different strategies of decision making and differential weighting of factors used in making decisions. 1. Strategic Preferences Two common strategies for making a choice among alternatives are relevant to locomotion and assessment orientations (see Avnet & Higgins, in press). The ‘‘progressive elimination’’ strategy looks at the first evaluative attribute and eliminates whichever brand has the worst value for that attribute. This elimination process is repeated for the second evaluative attribute, and so on until a single option remains. This strategy is a variation of the classic elimination-by-aspects strategy that eliminates options that do not meet a minimum cutoff value for the most important attribute. This elimination process is repeated for the second most important attribute, with processing continuing until a single option remains (Tversky, 1972; Bettman, Luce, & Payne, 1998). Because how the final choice is made involves changing the state of alternative possibilities at the end of each phase (i.e., eliminating the worst alternative), this progressive elimination strategy suits a locomotion orientation. Because more and more of the original alternatives are no longer part of the comparison process as the decision making proceeds, this strategy does not suit an assessment orientation. The ‘‘full comparison’’ strategy considers all of the alternatives and all of the attribute values for each alternative. A choice is made by making comparisons among all of the alternatives for all of the attributes, and then choosing the brand that has the best attributes overall. This strategy is a variation of the classic equal-weight strategy (Bettman et al., 1998). In the equal-weight strategy, a choice is made according to the highest value achieved. This value is obtained for each alternative by summing all of the attribute values for that option, and the alternative with the highest value is selected. Because how the final choice is made involves making as many

320

HIGGINS et al.

comparisons as possible, this full comparison strategy suits an assessment orientation. Because all of the alternatives remain as possibilities until the final choice phase (i.e., no change of state), this strategy does not suit a locomotion orientation. Avnet and Higgins (in press) experimentally induced either a locomotion or an assessment orientation and randomly assigned participants to make their choice using either a full comparison strategy or a progressive elimination strategy. They combined regulatory fit theory (Higgins, 2000) and regulatory mode theory to make their predictions. According to regulatory fit theory, people experience a regulatory fit when they use a means of goal pursuit that fits their regulatory orientation, and this regulatory fit increases the value of what they are doing (see Higgins, 2000). In addition, the value experience that derives from how a decision is made can transfer to outcome value. As a result, people will assign higher value to an object that was chosen with higher regulatory fit. Thus, value can be transferred from ‘‘how’’ a choice is made to ‘‘what’’ was chosen. For the reasons described above, Avnet and Higgins (in press) predicted that participants with an assessment orientation would have a higher regulatory fit with the full comparison strategy than the progressive elimination strategy, and that participants with a locomotion orientation would have a higher regulatory fit with the progressive elimination strategy than with the full comparison strategy. The experiment used an ‘‘unrelated studies’’ paradigm in which the first study manipulated regulatory mode and the second, supposedly unrelated, study manipulated strategic decision style and measured monetary value of the choice. When participants arrived at the study, they first completed a ‘‘Behavior Over Time’’ task that induced either a locomotion orientation or an assessment orientation. In both regulatory mode conditions, the participants were told that the study was ostensibly about how people recall their past behaviors over time. They were asked to recall three different types of behavior they had used successfully in the past and to write a short example for each type of behavior. In the locomotion mode condition, the participants were requested to give a short example for each of three types of behavior that were items taken from the locomotion scale of Kruglanski et al. (2000), such as ‘‘Think back to the times when you decided to do something and you couldn’t wait to get started.’’ In the assessment mode condition, the participants were requested to give a short example for each of three types of behavior that were items taken from the assessment scale of Kruglanski et al. (2000), such as ‘‘Think back to the times when you compared yourself with other people.’’ After the experimental induction of regulatory mode, the participants were given a decision-making task in which they chose among different

LOCOMOTION AND ASSESSMENT REGULATORY MODES

321

brands of reading lights. They were assigned either a full comparison strategy or a progressive elimination strategy by which to choose the brand they preferred. Prior to the experimental study, the different brands of reading lights were examined to identify one that was clearly superior to the others. The purpose of the study was not to test which light would be chosen, but rather to test the perceived monetary value of the same chosen light as a function of regulatory fit and value transfer. Thus, the light alternatives were selected so that one brand would be preferred by all participants. This was, indeed, the case as all participants in the study chose the same brand as their favorite. After making their choice, the participants were asked whether they would like to purchase the product they chose and, if so, how much they were willing to pay to purchase it. Avnet and Higgins (in press) predicted that because of value transfer from regulatory fit to outcome value, the participants would offer more money to buy the light when the strategy they used to make their choice fit their orientation (i.e., assessment orientation/ full comparison; locomotion orientation/progressive elimination/) than when it did not fit (i.e., assessment orientation/progressive elimination; locomotion orientation/full comparison). As shown in Fig. 1, the study found no main effect of either regulatory mode per se or choice strategy per se on money offered for the chosen book-light. However, as predicted, there was a significant interaction between the two variables, with participants in the fit conditions offering over 40% more to buy the book-light than participants in the nonfit conditions. This study provides strong evidence for distinct decision strategies being preferred when individuals are in a locomotion mode versus an assessment mode. The study also demonstrates that either a locomotion or assessment orientation can be experimentally induced. Thus, locomotion and assessment orientations are best thought of as psychological states that can vary across both persons (as chronic predispositions) and situations (see Higgins, 2000). 2. Strategic Preference as Target of Others’ Social Influence People not only have self-regulatory strategies that they use in their own goal pursuits. They also have social-regulatory strategies that they use to influence the goal pursuits of others. French and Raven (1959) in their ground-breaking analysis of social power (see also Raven & Kruglanski, 1969; Raven, 1992, 1993) distinguished among several specific power bases, including coercive power, legitimate power, expert power, and referent power. Coercive and legitimate power constitute bases of social influence in which compliance is demanded of others via the invocation of strictly

322

HIGGINS et al.

Value of book-light (in $)

7 6 5 4 3 2 1 0

Assessment Progressive elimination Full comparison

Locomotion

Fig. 1. Value from fit.

enforcable rules or through the threat of painful consequences contingent on the failure to comply. By contrast, expert and referent power constitute bases of social influence in which others are essentially free to decide whether to accept the advice or counsel of the influencer. Conceptually related to these two types of power bases is the distinction between autocratic/directive versus democratic/participative leadership styles (see, for example, Bass, 1990; Lewin, Lippit, & White, 1938). Kruglanski, Pierro, and Higgins (2002) investigated the relation between these two basic types of leadership styles and regulatory mode. To avoid associations with extraneous content of political labels (e.g., ‘‘autocratic’’ versus ‘‘democratic’’) or labels with evaluative connotations (e.g., ‘‘strong’’ versus ‘‘soft’’), they referred to the ‘‘strong’’/‘‘autocratic’’ type of leadership style as ‘‘forceful’’ and the ‘‘soft’’/‘‘democratic’’ type of leadership style as ‘‘advisory.’’ ‘‘Forceful’’ captures the demanding, directive, and coercive nature of the ‘‘strong’’/‘‘autocratic’’ type of leadership while remaining more neutral in political content and evaluative tone. To ‘‘force’’ means to press, drive, or compel. ‘‘Advisory’’ captures the counselling, consultative, and participative nature of the ‘‘soft’’/‘‘democratic’’ type of leadership style while also remaining more neutral in political content and evaluative tone. To ‘‘advise’’ means to recommend, counsel, consult, and ‘‘advised’’ means considered and thought out. The ‘‘forceful’’ and ‘‘advisory’’ leadership styles represent different ways of influencing others. In organizational settings, supervisors use these different strategies to influence the goal pursuits of their subordinates. Kruglanski et al. (2002) were concerned with how these different leadership strategies of supervisors affect the job satisfaction of the subordinates who are the target of the influence strategies. Supervisors affect the day-to-day

LOCOMOTION AND ASSESSMENT REGULATORY MODES

323

activities of their subordinates through the use of different influence strategies. These influence strategies may provide a better fit for the selfregulatory orientations of some subordinates than others. Kruglanski et al. (2002) hypothesized that locomotion and assessment would have a better fit, respectively, with the ‘‘forceful’’ and the ‘‘advisory’’ strategies. A leadership style involving ‘‘forceful’’ strategic influence is demanding, directive, and coercive. This style compels and presses for movement and change and ensures against disruption. Individuals with higher locomotion orientation should have a preference for this leadership style because they want to initiate and maintain movement from state to state. They should desire strong situational forces that press for or compel movement and change without disruption. A leadership style involving ‘‘advisory’’ strategic influence, on the other hand, involves counselling, consultation, and participation. Advice about alternative possibilities is given. Options to be considered and mulled over are offered. Standards and critical evaluation are provided. Individuals with higher assessment orientation should have a preference for this leadership style because they want to make comparisons among alternatives and critically evaluate past and future actions. Thus, they should desire a strategic influence that allows them to consider, compare, and critically evaluate different future options or past actions. The first set of predictions of Kruglanski et al. (2002) was that preference for a ‘‘forceful’’ leadership style would be greater as individuals’ locomotion orientation was greater, and preference for an ‘‘advisory’’ leadership style would be greater as individuals’ assessment orientation was greater. These predictions were tested in three different studies that included different geographic locations, diverse employee populations (including bank clerks, firemen, and employees of an investment firm), and two different measures of the ‘‘forceful’’ and ‘‘advisory’’ strategic influence styles [namely, Hinkin and Schriesheim’s (1989) scale of social power bases, and the Path-Goal Leadership Questionnaire (Indvik, 1985, 1988; Northouse, 1997)]. The first two studies found that individuals with a stronger locomotion orientation preferred a leader who used ‘‘coercive’’ power and ‘‘legitimate’’ power as ways to influence subordinates—both ‘‘forceful’’ influence strategies. This preference for forceful or demanding strategies is natural because high locomotors want to initiate and maintain movement from state to state. These studies also found that individuals with a stronger assessment orientation preferred a leader who used ‘‘expert’’ power and ‘‘referent’’ power as ways to influence subordinates—both ‘‘advisory’’ influence strategies. This preference for advisory or counselling strategies is also natural because high assessors want to consider different options and relate past and future actions to critical standards. The third study generalized these findings. It tested the ‘‘forceful’’ style by studying ‘‘directive’’

324

HIGGINS et al.

leadership and the ‘‘advisory’’ style by studying ‘‘participative’’ leadership. Like the first two studies, the third study found that individuals with a stronger locomotion orienation preferred the forceful ‘‘directive’’ leadership style, and individuals with a stronger assessment orientation preferred the advisory ‘‘participative’’ leadership style. Once these distinct preferences were empirically established, Kruglanski et al. (2002) turned to the issue of how the different leadership strategies of supervisors affect the job satisfaction of the subordinates who are the target of the influence strategies. As discussed above, when there is a fit between individuals’ regulatory orientation and the strategic nature of an activity, the value of the activity increases. In the context of organizational activities, the value of work activities is reflected in employees’ job satisfaction. Thus, the second set of predictions of Kruglanski et al. (2002) was that subordinates’ job satisfaction would be higher when there was a greater fit between their regulatory mode orientation and the leadership style of their supervisor (high locomotion/‘‘forceful’’ style; high assessment/‘‘advisory’’ style). These predictions were tested in a fourth study that involved yet another new population (members of the police force). As predicted, the study found that job satisfaction was greater when subordinates with higher locomotion orientation had supervisors who used a more forceful ‘‘directive’’ style and when subordinates with higher assessment orientation had supervisors who used a more advisory ‘‘participative’’ style.

3. Emphasizing Expectancy versus Value in Decisions Expectancy-value (or subjective utility) models have been historically the most influential models regarding the evaluative processes underlying decision making. A basic assumption shared by these models is that both expectancy and value influence goal commitment (Lewin et al., 1944; Tolman, 1955; Vroom, 1964; for a review, see Feather, 1982). Regulatory mode should moderate whether it is expectancy or value that receives greater emphasis. Individuals with higher locomotion orientation like flow and dislike barriers to flow. For locomotion, it is goal attainability, ease of advancement, that is critical (see Kruglanski, 1996). High expectancy suggests smooth sailing whereas low expectancy suggests a bumpy ride. Thus, as locomotion increases there should be increasing concern with whether expectancy of goal attainment is high or low. Indeed, it could be that this emphasis on wanting (and seeking out) high expectancy goal pursuits contributes to the strong association found by Kruglanski et al. (2000) between high locomotion and high optimism.

LOCOMOTION AND ASSESSMENT REGULATORY MODES

325

Although individuals with a high assessment orientation can certainly compare alternatives along an expectancy dimension, their primary concern would naturally be to compare alternatives along a value dimension. Value comparisons are critical to making the ‘‘right’’ or ‘‘best’’ choice, to making decisions on the basis of merit, relative worthiness, and excellence. Thus, as assessment increases there should be increasing concern with whether value of goal attainment is high or low. In contrast, individuals with high locomotion, like people in general, would prefer high to low value goal pursuits, but because their distinctive orientation concerns flow rather than outcomes per se, their natural preference for higher value goal pursuits need not be greater than individuals with low locomotion. In a study by Kruglanski et al. (2000), the participants were asked to name spontaneously five personal attributes they wanted to attain. Participants listed attribute goals such as being ‘‘educated,’’ ‘‘strong,’’ ‘‘fit,’’ ‘‘knowledgeable,’’ or ‘‘outgoing.’’ After listing all five attribute goals, participants were asked to list all the different means they might employ to attain each goal. Participants then rated the likelihood they would attain each goal, as well as its perceived value. Both types of ratings were made on seven-point scales anchored with 1 ‘‘not at all’’ (likely/valuable) and 7 ‘‘extremely’’ (likely/valuable). If high locomotors are concerned with expectancy or attainability, they should generally report having attribute goals with high expectancy of attainment. If high assessors are concerned with value or worth, they should generally report having attribute goals with high value if attained. Total attainment expectancy was calculated by summing up the expectancy ratings for each of the attributes. This measure indexed the degree to which one pursues goals whose attainment has high expectancy. Total attainment value was calculated by summing up the value ratings for each of the attributes listed. This total indexed the degree to which one pursues goals whose attainment has high value. Separate regression analyses were carried out to examine how differences in locomotion and assessment scores related to participants’ pursuit of high value goals and high expectancy goals. To control for any positive association between high value and high expectancy (see, for example, Atkinson, 1964), each analysis controlled for the alternative commitment factor. Kruglanski et al. (2000) found, as predicted, that participants’ assessment scores had a strong positive relation to value, whereas their locomotion scores had no significant relation to value. In contrast to having no significant relation to value, participants’ locomotion scores did have a strong positive relation to expectancy, as predicted. Also as expected, participants’ assessment scores were not positively related to expectancy. Indeed, they had a significant negative relation. This significant negative

326

HIGGINS et al.

relation is reminiscent of the lower optimism of higher assessors found by Kruglanski et al. (2000) that we described above. This lower optimism could be produced by high assessors, more than other people, considering all alternative outcomes, including more pessimistic scenarios. 4. Speed and Number in Generating Decision Alternatives An important aspect of decision making in goal pursuit is generating alternative means to attain a goal. Two features of this process of generation are the speed of generation and the number of alternatives generated. Speed of generating alternative means should be higher for individuals with higher locomotion for the reasons discussed earlier when predicting their higher urgency to begin. Urgency to begin is not a concern of individuals with higher assessment. Rather, high assessors are concerned with making comparisons and thus they would want to consider all reasonable alternative means. The number of means generated should be greater, therefore, for individuals with higher assessment. In contrast, what matters to high locomotors is having means that permit quick initiation and smooth maintenance of goal pursuit rather than the number of means per se. As part of the same study of listing personal goal attributes that we described above, Kruglanski et al. (2000) instructed participants, after they had reported their expectancies and values for the different goal attributes, to type in as quickly as possible the best means for attaining each of the attributes. (The order of attributes was randomly generated by the computer for each participant. In addition, for use later as a covariate to control for individual differences in response speed, participants first typed in as quickly as possible responses to a number of control questions, such as their first name.) The number of means listed for each attribute was summed across the five attributes to measure total number of means. The average time taken to type in the best means for each of the attributes was summed across the five attributes to measure the speed with which participants listed their best means. It should be noted that total means number and choice speed were not significantly correlated in the study. As predicted, higher assessment was positively related to the number of means generated, but higher locomotion was not. Also as predicted, higher locomotion was positively related to choice speed, but higher assessment was not. 5. Heterogeneity versus Homogeneity of Judgment When people judge themselves or other people they can make few distinctions or many distinctions. When evaluating themselves, for example,

LOCOMOTION AND ASSESSMENT REGULATORY MODES

327

they can judge their personal attributes as all being quite positive or they can judge that they have both strengths and weaknesses. As another example, when determining the closeness of their relationships with different people, they can judge that they are quite close to all of their significant others or they can judge that they are closer to some people than others. Individuals in assessment mode are inclined to make comparisons. For self-evaluation, this would mean comparing oneself to different people and comparing one’s own attributes to one another. Because there is both variability across people and variability within a person in actual strengths and weaknesses, this kind of comparison process should increase heterogeneity of judgment among high assessors. In contrast to assessment mode, individuals in locomotion mode are inclined to move and initiate action. A general positivity or leniency bias in judging oneself and others would have the advantage of increasing motivation to approach and engage social and nonsocial activities. This would increase positive homogeneity of judgment. This difference between assessment mode and locomotion mode would have two effects. First, high assessors should have more variance in their judgments than low assessors, whereas high locomotors should have less variance in their judgments than low locomotors. Second, high assessors should be less positive in their judgments than low assessors, whereas high locomotors should be more positive in their judgments than low locomotors. These predictions were tested in a couple of studies by Higgins et al. (2002). German participants in one study were given Heatherton and Polivy’s (1991) social self-esteem scale (‘‘I feel inferior to others at this moment’’ [reverse scored]). As predicted and shown in Table III, the variance of the social self-esteem scores was higher for high than low assessors, but was lower for high than low locomotors. Also as predicted and shown in Table III, the level of social self-esteem scores was lower for high than low assessors, but was higher for high than low locomotors. Italian participants in another study completed the Social Network Inventory (SNI) developed by Treadwell, Leach, and Stein (1993). This inventory measures individual networks defined as the smallest number of significant persons that influences one’s life. It measures the closeness the participants feel toward the significant others, as well as the participants’ perception of the closeness felt by the significant others toward him or her. The instructions include three steps: (1) to list the names of significant persons that affect one’s life; (2) to describe one’s relationship to each (e.g., brother, friend, etc.); and (3) to rate each person on two scales (seven-point Likert scales, with 1 representing ‘‘distance’’ and 7 indicating ‘‘closeness’’) listed beside each name. The first scale (Criterion 1) measures how close the

328

HIGGINS et al.

TABLE III Social Self-Esteem Level and Variance as a Function of Locomotion and Assessment

Assessment high

Social self-esteem

t test

Test of variance homogeneity

M

SD

M

SD

t

Sig.

F

Sig.

3.81

1.01

4.34

.80

4.32

.001

4.70

.03

Locomotion high

Social self-esteem

Assessment low

Locomotion low

M

SD

M

SD

4.37

.73

3.77

1.01

t test t

Sig.

4.91 .001

Test of variance homogeneity F

Sig.

8.30

.004

participant feels he or she is to the person. The second scale (Criterion 2) involves a role reversal and asks the participant to place him or herself in the role of the other person and estimate how close that individual thinks he or she is to the participant. The analysis for the evaluative judgments across significant others was performed only on the first three significant persons listed by each participant because 97% of the participants listed three significant persons. (Because it was not 100%, the number listed was included as a covariate in the analysis.) There were two measures of variance in evaluative judgments. One measure was the variance between the participants in their overall closeness score (averaged across the three significant others). The second measure was the variance within each participant’s closeness scores between their three significant others. For each of these measures of variance and for both Criterion 1 (me judging them) and Criterion 2 (them judging me), the study found, as predicted, that the variance of the closeness scores was greater for high than low assessors. If anything, the closeness scores tended to be nonsignificantly smaller overall for high than low locomotors. Also as predicted, the level of the closeness scores was lower for high than low assessors, but were, if anything, higher (nonsignificantly) for high than low locomotors. Combined with the social self-esteem findings, there was strong and consistent evidence across these two studies that high assessors have more variance and less positivity in their judgments than low assessors. If anything, there was an opposite pattern for locomotion mode, with high locomotors having less variance and more positivity in their judgments than low locomotors, although this pattern was significant only for the social selfesteem measure.

LOCOMOTION AND ASSESSMENT REGULATORY MODES

329

B. ACTIVITY ENGAGEMENT AND PERFORMANCE Individuals in a high locomotion orientation are concerned with initiating and maintaining activity engagement. They want to ‘‘get on with it.’’ In contrast, individuals in a high assessment orientation are concerned with making comparisons among alternative activities in order to select the correct one to engage. They want to ‘‘make the right choice.’’ This is a basic distinction between the locomotion and assessment modes, and it was examined in a couple of studies by Klem, Higgins, and Kruglanski (1996). 1. Getting Started on Something and Waiting to Do the ‘‘Right’’ Thing When people have a choice between doing several different activities, sometimes they don’t wait to think much about it and simply do one of them. At other times, however, they wait for information that would tell them which of their choices is the ‘‘right’’ one or the ‘‘best’’ one to do. Individuals who have a high locomotion orientation prefer to do something, almost anything, rather than nothing at all. Waiting and doing nothing is highly unpleasant. Thus, if all of the alternatives are reasonable, they should prefer a random activity that they can begin immediately rather than simply waiting and doing nothing. Individuals high in assessment orientation, however, are not interested simply in doing something. They want to compare the alternatives and decide which one is the right one or the best one to do. Thus, they should prefer to wait for information that will allow comparisons to be made rather than randomly jumping into whatever activity happens to come up. Indeed, they should be willing to wait quite awhile rather than lose the opportunity to make such comparisons. In a study by Klem et al. (1996), the participants were recruited for a supposed ‘‘nationwide study on the utility of 9 different tasks for evaluating creativity.’’ They were told that each task examined creativity in a different way that some students found very interesting and others found somewhat boring. They were told that because there were such wide individual differences in preferences for different tasks, we were providing background information on all of the tasks so that they would have the opportunity to choose the task that best suited them. The participants were then informed that because we were having problems with the software that provided this background information, it would take 5 minutes or more for the information to appear on the screen. But they didn’t have to wait to start one of the tasks. They were told that by pressing any number between 1 and 9, they could begin one of the nine tasks immediately without waiting for the background information. If they chose to wait, and later became tired of waiting, they were told that they could press a number between 1 and 9 at

330

HIGGINS et al.

any time to begin one of the tasks. The background information never appeared on the screen and the waiting time period ended after 10 minutes. There were two dependent measures: (1) whether a participant chose to wait for the background information or chose to begin a task immediately without waiting; and (2) for those who chose to wait for the background information, the length of time they waited before either they gave up and chose a number or the time period ended. The study found two independent effects. As predicted, participants with higher locomotion were more likely to choose a number immediately and not wait to begin. Also as predicted, among those who decided to wait for the background information, participants with higher assessment waited a longer time before either they gave up and chose a number or the time period ended. A second study by Klem et al. (1996) extended this study. The participants in this study also believed that they were choosing to do one among five creativity tasks (instead of nine tasks). In this study, the participants were given one example each of three types of background information they could look at for each of the different tasks. The quality of this information was experimentally manipulated. For half of the participants, the examples revealed that the background information was of low quality, such as information about who had developed the task, where in the nation the task had been developed, and the number of questions included in the task. For the other half of the participants, the examples revealed that the background information was of high quality, such as information about how other students from their university had previously evaluated the task, the type of person who would enjoy the task (e.g., ‘‘People who like tactical games such as Chess enjoy task 2 the most’’), and what they would actually be doing in the task. Pilot testing was used to select low quality information that, on a scale from 1 (‘‘not at all useful’’) to 7 (‘‘extremely useful’’), received an average of less than ‘‘2,’’ and high quality information that received an average of more than ‘‘6.5.’’ There was no overlap in the low-quality versus high-quality ratings. Unlike the first study, there was no delay in this study before the background information became available. For this reason, the vast majority of participants did look at some of the background information. The participants were told that when they asked for information about a task, they could look at one, two, or all three of the different types of information available for that task. At any point, they could decide to begin one of the tasks by choosing its number. There were two main dependent measures: (1) the number of tasks considered prior to making a final choice (which varied between 1 and 5); and (2) the number of total sets of information investigated prior to making a final choice (which varied from 1 to 15). Because the number of total sets of information investigated tends

LOCOMOTION AND ASSESSMENT REGULATORY MODES

331

to increase with an increase in the number of tasks considered, measure (1) was included as a covariate in the analysis for measure (2). For the same reasons described earlier, we predicted that individuals with a higher locomotion orientation would consider less different tasks before beginning one of them. We also predicted that individuals with a higher assessment orientation would investigate more sets of background information about the different tasks before choosing a task. Both of these predictions were confirmed. The experimental manipulation of quality of background information permitted a third prediction as well. Individuals with a high assessment orientation want to make comparative evaluations in order to make the right or best decision. Given this goal, they should want to make evaluative comparisons more when the background information is of high quality than low quality. This predicts an interaction of assessment and quality of background information, such that the greater investigation of the background information by individuals with higher assessment should be especially true for the high-quality information. This prediction was also confirmed. 2. Time to Completion and Looking Out for Errors According to regulatory mode theory, locomotion and assessment should also relate to different aspects of task performance. Time to completion is an aspect of all tasks and looking out for errors is an aspect of many. As we have seen, higher locomotion is associated with task flow, with decisiveness, attentional control, and persistence. One would expect, then, that individuals with a high locomotion orientation would go swiftly through tasks, producing a fast completion time. There is no obvious relation between assessment and completion time, but assessment should relate to looking out for errors. An error is a discrepancy from some standard. Individuals with a high assessment orientation make comparisons, including comparing a target to a standard. Thus, higher assessment should be positively associated with looking out and finding errors. These predictions were tested in a ‘‘proofreading’’ study by Kruglanski et al. (2000). The participants were told that the study was about ease of reading and detection of spelling errors as a function of the different types of print font used in a passage. Participants were given a red pen and two fivepage booklets with identical passages printed in different fonts. They were then asked to check one of the booklets, labeled Sample Copy, against the other, labeled Master Copy. Participants were told to circle on the Sample Copy any differences found between it and the Master Copy. Such differences included the appearance of different words, numbers, or dates, different punctuation marks, and actual grammatical or spelling errors

332

HIGGINS et al.

found only in the Sample Copy. There were a total of 63 differences between the two passages. All participants were allowed as much time as needed to complete their booklet. As predicted, the study found that higher locomotion (but not higher assessment) was positively associated with quicker completion time, and higher assessment (but not higher locomotion) was positively associated with the total number of errors circled. 3. Achievement Performance We described earlier how achievement motivation has a stronger relation to locomotion than to assessment. This does not imply, however, that actual achievement depends only on having a high locomotion orientation. Successful goal pursuit, like successful self-regulation generally, requires that both locomotion and assessment be functioning properly. To achieve, it is certainly necessary to initiate and maintain movement toward the goal. But it is also important to choose good strategies, prioritize, and plan one’s time and effort well, especially for difficult and complex endeavors that unfold over a long period. Thus, we expected that a combination of high locomotion and high assessment would predict high achievement. Kruglanski et al. (2000) tested this prediction in two studies, one involving achievement in university and one involving achievement in military training. The undergraduates in the university study had completed either three, five, or seven semesters of course work. Blind to individual students’ names, each of their grade point averages (GPAs) was obtained. A multiple regression analysis was conducted to predict the students’ cumulative GPA scores from their locomotion scores, their assessment scores, and the interaction between these scores. The analysis controlled for Scholastic Aptitude Test (SAT) score, which was a strong positive predictor of GPA. There was a significant positive effect of locomotion overall which was qualified by a locomotion  assessment interaction. As predicted, the locomotion effect on GPA increased as a positive function of assessment. For participants scoring below the median on assessment, locomotion was actually unrelated to GPA. For participants above the median on assessment, however, locomotion was a positive predictor of GPA. Thus, achievement from high locomotion depended on there being high assessment as well. The participants in the military training study were male applicants to an elite combat training unit in the U.S. Army who began participation just prior to onset of the training period. All of them were already in military service. This program is highly selective and the training is extremely demanding, and the majority do not complete the training. The dependent

LOCOMOTION AND ASSESSMENT REGULATORY MODES

333

measure of the study was successful completion of the training course. The study controlled for other predictor variables recommended by U.S. Army researchers who administered the precourse testing battery, including scores from a general technical survey and from a spatial abilities test, whether applicants were commissioned officers or enlisted soldiers, and whether applicants were or were not ‘‘Ranger-qualified.’’ A logistic regression analysis found that higher locomotion scores predicted a higher likelihood of success, but this was qualified by a locomotion  assessment interaction. As predicted, the impact of locomotion depended on the level of assessment. Among soldiers scoring below the median in assessment, locomotion had no significant impact on program completion. Among soldiers scoring above the median in assessment, locomotion was a strong predictor of success. These results for soldiers’ military training achievement replicate those found for college students’ academic achievement. A recent study by Pierro et al. (2002a) asked employees of a computer firm in Rome, Italy to evaluate their perceived goal attainment (see Brunstein, Schultheiss, & Grassmann, 1998) with respect to advancement (‘‘I have made a great deal of progress in my attempts to advance this goal’’), success (‘‘I have quite a lot of success in my pursuit of this goal’’), and failure (‘‘Many of my efforts to realize this goal have failed’’ [reverse scored]). Conceptually replicating the results for soldiers’ and students achievement, the study found that the employees who were high in both locomotion and assessment had the highest perceived goal attainment. High locomotion is necessary to succeed at endeavors that require persistence over a long period, but it is not sufficient. High assessment is also important because considering alternatives, which includes managing one’s time and effort, enhances performance on such endeavors. This interaction between locomotion and assessment is telling. One might expect that individuals who describe themselves on a questionnaire as being a persistent, high-energy go-getter and doer (high locomotors) are likely to be high achievers. Indeed, one might argue that high achievers look at their success and then infer that they are the kind of person who would agree with high locomotion items on the questionnaire, that is, a Bem-like selfperception effect (see Bem, 1967). What the interaction between locomotion and assessment reveals is that being a high locomotor, in fact, is not enough by itself to be a high achiever. Instead, high achievers are individuals who are high in both locomotion and assessment. Earlier we described a study by Pierro, Mannetti, Higgins, and Kruglanksi (2002) that found that intentions to exercise were increased by social norms favoring exercise when individuals had a higher assessment orientation. The same study also examined how these intentions translated to the actual behavior of frequently and regularly going to the gym for

334

HIGGINS et al.

exercise. This study found that actual exercise was a function of the interaction between intentions and locomotion, such that positive intentions to exercise led to actually going to the gym to exercise when individuals had a high locomotion orientation. It is well-known that people do not always carry out their intentions (e.g., Zanna, Higgins, & Herman, 1982). Because higher locomotion relates to higher commitment to initiate and maintain actions in pursuit of a chosen goal, higher locomotion increases the likelihood that a goal-related intention will be carried out. The results of the Pierro et al. (2002) study show in more detail how a combination of high assessment, which increases positive intentions through following social norms, and high locomotion, which increases commitment to carry out behaviors inspired by positive intentions, can together produce high achievement. 4. Activity Affordances and Effort Investment The research we have described thus far has examined variability in individuals’ locomotion and assessment orientations that was either chronic or experimentally induced. These studies concerned locomotion and assessment as states of individuals, as something associated with persons. However, locomotion and assessment can also be considered as something associated with activities themselves. That is, for people in general, there are some activities that naturally afford locomotion and other activities that naturally afford assessment. Taylor and Higgins (2002) conducted a study on motivations associated with doing different kinds of common activities. In an initial phase of the study, one sample of participants generated the different activities. In the next phase of the study, a different sample of participants was asked their reason for doing each activity, as follows: (1) Locomotion—my reason for doing this activity is because it involves action or movement away from the current situation. It satisfies my need for change, to do something, anything different, regardless of what I am currently doing. In the extreme case of locomotion, ‘‘I would rather do anything than nothing at all.’’ (2) Assessment—my reason for doing this activity is because it involves evaluating, measuring, or interpreting information. It satisfies my need for critically appraising and evaluating something in order to be certain I am doing it correctly. In the extreme case of assessment, ‘‘I would rather do nothing than the wrong thing.’’ The participants were also explicity told: ‘‘As these two reasons for doing activities are independent, an activity can be thought of as involving one or both of these reasons.’’ For each activity, the participants responded by choosing among the following reasons for doing it: ‘‘just locomotion,’’ ‘‘primarily locomotion but also assessment,’’

LOCOMOTION AND ASSESSMENT REGULATORY MODES

335

TABLE IV Situational Affordances for Locomotion and Assessment

Activities

Just and primarily locomotion (%)

Just and primarily assessment (%)

Both equal (%)

Playing sports Exercise Dancing Video games Partying Academic activities Financial duties Thinking Cultural event News Correspondence Meditation Going places Hygiene Creating art Friends Performing

68 84 87 86 70 0 7 0 10 0 23 16 25 31 26 48 42

10 6 0 2 8 87 93 88 69 93 67 69 10 42 37 20 37

22 10 13 12 21 13 0 12 21 7 11 16 65 27 38 32 21

equally locomotion and assessment, ‘‘primarily assessment but also locomotion,’’ and ‘‘just assessment.’’ Taylor and Higgins (2002) found that for many activities there was a strong consensus (by a Chi-square analysis, all ps < .001) as to whether it was just or primarily locomotion, just or primarily assessment, or equally locomotion and assessment. As shown in Table IV, just or primarily locomotion activities were ‘‘playing sports’’ (defined as ‘‘engaging in competitive physical games’’), ‘‘exercise’’ (‘‘activities that strengthen muscles and increase your heart rate’’), ‘‘dancing’’ (‘‘dancing, moving to music’’), ‘‘video games’’ (‘‘playing games on the computer or video game system’’), and ‘‘partying’’ (‘‘going to parties and bars’’). Just or primarily assessment activities were ‘‘academic activities’’ (‘‘required learning and course work’’), ‘‘financial duties’’ (‘‘managing money, paying bills, balancing your checkbook’’), ‘‘thinking’’ (problem solving, idea generation), ‘‘cultural event’’ (‘‘attending cultural events such as theatre, opera, ballet . . . ’’), and ‘‘news’’ (‘‘watching the news or reading the newspaper’’), ‘‘correspondence’’ (‘‘writing letters or e-mail’’), and ‘‘mediation’’ (‘‘letting your mind focus in thought’’). An activity that was consensually considered to be equally locomotion and assessment was ‘‘going places’’ (‘‘traveling, exploring, or going places in general’’).

336

HIGGINS et al.

There were also some activities for which there was no consensus, with approximately the same percentage of participants rating it as just or predominantly locomotion, just or predominantly assessment, and equally locomotion and assessment. These nonconsensual activities were ‘‘hygiene’’ (‘‘daily activities for personal cleanliness’’), ‘‘creating art’’ (‘‘making sculpture, drawing, taking photographs’’), ‘‘friends’’ (‘‘hanging out with friends and acquaintances’’), ‘‘performing’’ (‘‘performing in an artistic, musical, or theatrical sense’’), ‘‘TV’’ (‘‘watching television’’), ‘‘relaxing’’ (‘‘activities that relax you or put you in a calm state’’), ‘‘music’’ (‘‘listening to music’’), and ‘‘internet’’ [‘‘surfing the web (not including e-mail)’’]. In a follow-up study, Taylor and Higgins (2002) gave a new sample of participants the financial duties, academic activity, playing sports, and exercise activities, and for each activity asked them ‘‘how much effort would you put into this activity?’’ They were asked to imagine engaging in the activity for different motivational reasons—either ‘‘just for locomotion’’ or ‘‘just for assessment.’’ Taylor and Higgins (2002) predicted that individuals generally would invest more effort in an activity when the reason for engaging in that activity fit its (consensual) regulatory mode affordance. As predicted, for the two activities with locomotion affordances (i.e., ‘‘playing sports’’ and ‘‘exercise’’), participants said they would put more effort into the activity when the reason for engaging in the activity fit the activity affordance (‘‘just locomotion’’) than when it did not fit (‘‘just assessment’’). Also as predicted, for the two activities with assessment affordances (i.e., ‘‘financial duties’’ and ‘‘academic activity’’), participants said they would put more effort into the activity when the reason for engaging in the activity fit the activity affordance (‘‘just assessment’’) than when it did not fit (‘‘just locomotion’’). The participants also varied in the strength of their locomotion and assessment orientations as measured by the Regulatory Mode Questionnaire (Kruglanski et al., 2000). Taylor and Higgins (2002) predicted that individuals would invest more effort in an activity whose regulatory mode affordance fit their own personal regulatory mode orientation. As predicted, for the two activities with locomotion affordances, participants with a stronger locomotion orientation (but not a stronger assessment orientation) said they would put more effort into the activity. Also as predicted, for the two activities with assessment affordances, participants with a stronger assessment orientation (but not a stronger locomotion orientation) said they would put more effort into the activity. Together, the results of these studies suggest that both persons and activities can be associated with locomotion and assessment, and a regulatory mode fit between the affordance of the activity and either individuals’ current reason for engaging the activity or their chronic orientation can increase their motivation to engage the activity.

LOCOMOTION AND ASSESSMENT REGULATORY MODES

337

V. Conclusions This chapter has reviewed evidence for distinguishing between an assessment mode concerned with making comparisons and a locomotion mode concerned with movement from state to state. Assessment has been found to be distinct from locomotion by having higher self-evaluative concerns that produce greater sensitivity to social appraisal and social norms, affective vulnerabilities, emotional instability, discomfort with ambiguity, and a performance orientation, as well as greater reflectiveness. Locomotion has been found to be distinct from assessment by having higher activity flow concerns that produce greater decisiveness, attentional control, vitality, conscientiousness, and openness to change, as well as activity engagement with higher intrinsic motivation, effort investment, behavioral activation, and involvement. Higher locomotion and higher assessment were also shown to have distinct effects on judgment and decision making, including different preferences regarding decision strategies (‘‘progressive elimination’’ versus ‘‘full comparison,’’ respectively) and leadership styles (‘‘forceful’’ versus ‘‘advisory,’’ respectively), different emphases in the decision process (expectancy versus value; speed versus number, respectively), and different self-evaluative judgment styles (homogeneity versus heterogeneity, respectively). Other studies found that higher locomotion and higher assessment had differential effects on activity engagement and performance, including, respectively, distinct preferences for getting started on something versus waiting to do the ‘‘right’’ thing and for fast completion time versus looking for errors. Our review highlights the fact that there are trade-offs to each regulatory mode. Each mode has its own strengths and weaknesses, its own benefits and costs. Indeed, evidence was presented to show that high achievement performance often depends on individuals emphasizing both locomotion and assessment. In addition, effort investment depends not just on which mode is emphasized but also on whether the mode that is emphasized fits the type of activity engagement because activities themselves vary in being more related to locomotion (e.g., exercise) or assessment (e.g., financial duties). Similarly, fit between individuals’ chronic regulatory mode and the strategy used to make a decision was found to create a value experience that increased the value of the outcome of the decision (e.g., the monetary value of a chosen object), and, in an organizational context, fit between subordinates’ chronic regulatory mode and their supervisor’s leadership style was found to increase job satisfaction. Future research needs to examine the interrelations between locomotion and assessment in more

338

HIGGINS et al.

detail to determine which combinations of mode emphasis under which conditions are most effective and efficient. The distinction between locomotion and assessment modes is very general because it is not tied to any specific goal content, sphere of life, or type of need satisfaction. Rather, all self- regulation involves some combination of these two modes functioning at different levels of emphasis. Because these two modes are independent, one mode or the other or both or neither can be emphasized at any given time. In addition, different emphases in regulatory mode are not personality differences per se. Regulatory mode emphasis is a state of an individual that can vary across situations, across activites, and across time. It can be studied in terms of individual differences in chronic tendencies to emphasize each mode, but it can also be studied in terms of situational differences or temporal phase differences in mode emphasis. There can be developmental differences, organizational differences, and cultural differences in regulatory mode in addition to personality differences. This chapter has described studies on chronic individual differences and momentary situational differences in regulatory mode, as well as organizational differences (i.e., leadership styles) that vary in their fit with regulatory mode. We have not as yet studied developmental differences. It is reasonable to postulate that, in general, there is a relative emphasis on locomotion in early childhood (e.g., toddlers) and that assessment increases in emphasis during this period. It is also reasonable to postulate that even early in development both temperament and socialization styles influence whether a child emphasizes locomotion or assessment. This is an important area for future research. We have begun to study cross-cultural differences in regulatory mode. It should first be noted, as mentioned earlier, that there is strong cross-cultural validation for the structure of the locomotion and assessment scales (Higgins et al., 2002). Moreover, as is evident from our discussion of research conducted in different countries, the discriminant associations of locomotion with particular psychological phenomena and assessment with other psychological phenomena have been found in Italy, Germany, and Japan in addition to the United States. Thus, locomotion and assessment modes function similarly across cultures. What varies across cultures is the emphasis given to the different modes. Research by Higgins et al. (2002) has found that in Italy and Spain approximately 80% of the participants have higher locomotion than assessment scores (predominant locomotors) and 20% have higher assessment than locomotion scores (predominant assessors). In the United States and Germany there is more of an even split between the percentage of predominant locomotors and predominant assessors (about 60% to 40%, respectively). In direct contrast to these countries, approximately 70% of the

LOCOMOTION AND ASSESSMENT REGULATORY MODES

339

participants in Japan and South Korea are predominant assessors and 30% are predominant locomotors. In tests of homogeneity across countries for the difference between locomotion and assessment scores, these three sets of countries fall out as being distinct from one another. What are the cultural antecedents and consequences of a predominant locomotion emphasis versus a predominant assessment emphasis versus a balanced emphasis among the members of a culture? This is a fascinating question for future research. Another fascinating issue are the consequences of a match versus a mismatch between individuals’ personal regulatory mode emphasis and the regulatory mode emphasis of their surrounding culture. What does it mean to be a Japanese high locomotor or an Italian high assessor? The same question could be asked of organizations that have a particular regulatory mode emphasis that does or does not match the surrounding culture. Our exploration of regulatory mode has just begun. We now know that there is definitely a set of phenomena associated with regulatory mode and we can turn to understanding more about when, how, and why they occur (see Zanna & Fazio, 1982). Having established the psychological distinction between locomotion and assessment modes, we can now consider its more general implications for personal well-being, interpersonal relationships, intragroup and intergroup processes, as well as economic and political behaviors. We believe that the distinction between locomotion and assessment modes is basic and that it has broad implications for selfregulation and social regulation. Like all fundamental psychological principles, there are trade-offs of each mode. For both theoretical and practical reasons, it is the exploration of these trade-offs and how they can be handled that will enrich motivation science.

References Ajzen, I. (1988). Attitudes, personality and behavior. Buckingham, UK: Open University Press. Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human Decision Processes, 50, 179–211. Ajzen, I., & Fishbein, M. (1980). Understanding attitudes and predicting social behavior. Englewood Cliffs, NJ: Prentice Hall. Amabile, T. M., Hill, K. G., Hennessy, B. A., & Tighe, E. M. (1994). The Work Preference Inventory: Assessing intrinsic and extrinsic motivational orientations. Journal of Personality and Social Psychology, 66, 950–967. Avnet, T., & Higgins, E. T. (in press). Locomotion, assessment, and regulatory fit: Value transfer from ‘‘how’’ to ‘‘what.’’ Journal of Experimental Social Psychology. Bass, B. M. (1990). Bass & Stogdill’s handbook of leadership: Theory, research, & managerial applications (3rd ed.). New York: Free Press.

340

HIGGINS et al.

Bem, D. J. (1967). Self-perception: An alternative interpretation of cognitive dissonance phenomena. Psychological Review, 74, 183–200. Bettman, J. R., Luce, M. F., & Payne, J. W. (1998). Constructive consumer choice processes. Journal of Consumer Research, 25, 187–217. Brown, S. P., & Leigh, T. W. (1996). A new look at psychological climate and its relationship to job involvement, effort, and performance. Journal of Applied Psychology, 81, 358–368. Brunstein, J. C., Schultheiss, O. C., & Grassmann, R. (1998). Personal goal and emotional wellbeing: The moderating role of motive dispositions. Journal of Personality and Social Psychology, 75, 494–508. Button, S. B., Mathieu, J. E., & Zajac, D. M. (1996). Goal orientation in organizational research: A conceptual and empirical foundation. Organizational Behavior and Human Decision Processes, 67, 26–48. Camacho, J., & Higgins, E. T. (2001a). Agreement with proverbs as a function of locomotion and assessment. Unpublished manuscript, Columbia University Camacho, J., & Higgins, E. T. (2001b). Feelings of guilt and shame for past moral transgressions as a function of locomotion and assessment. Unpublished manuscript, Columbia University Carver, C. S., & Scheier, M. F. (1981). Attention and self-regulation: A control-theory approach to human behavior. New York: Springer-Verlag. Carver, C. S., & Scheier, M. F. (1990). Origins and functions of positive and negative affect: A control process view. Psychological Review, 97, 19–35. Carver, C. S., & White, T. L. (1994). Behavioral inhibition, behavioral activation, and affective responses to impending reward and punishment: The BIS/BAS scales. Journal of Personality and Social Psychology, 67, 319–333. Cattell, R. B., Cattell, A. K., & Cattell, H. E. (1993). Sixteen Personality Factor Questionnaire (5th ed.). Champaign, IL: Institute for Personality and Ability Testing. Costa, P. T., , Jr., & McCrae, R. R. (1992). Revised NEO Personality Inventory (NEO-PI-R) and NEO Five Factor Inventory (NEO-FFI) professional manual. Odessa, FL: Psychological Assessment Resources. Csikszentmihalyi, M. (1975). Beyond boredom and anxiety. San Francisco: Jossey-Bass. Deci, E. L., & Ryan, R. M. (1985). Intrinsic motivation and self-determination in human behavior. New York: Plenum Press. Deutsch, M. (1968). Field theory. In G. Lindzey & E. Aronson (Eds.), Handbook of social psychology. Reading, MA: Addison-Wesley. Dickman, S. J. (1990). Functional and dysfunctional impulsivity: Personality and cognitive correlates. Journal of Personality and Social Psychology, 58, 95–102. Duval, S., & Wicklund, R. A. (1972). A theory of objective self-awareness. New York: Academic Press. Dweck, C. S. (1991). Self-theories and goals: Their role in motivation, personality and development. In R. Dienstbier (Ed.), Nebraska symposium on motivation (Vol. 38, pp. 199–235). Lincoln, NE: University of Nebraska Press. Feather, N. T. (1982). Actions in relation to expected consequences: An overview of a research program. In N. T. Feather (Ed.), Expectations and actions: Expectancy-value models in psychology (pp. 53–95). Hillsdale, NJ: Erlbaum. Fenigstein, A., Scheier, M. F., & Buss, A. H. (1975). Public and private self-consciousness: Assessment and theory. Journal of Consulting and Clinical Psychology, 43, 522–527. Fishbein, M., & Ajzen, I. (1975). Belief, attitude, intention, and behavior: An introduction to theory and research. Reading, MA: Addison-Wesley. French, J. R. P., Jr., & Raven, B. H. (1959). The bases of social power. In D. Cartwright (Ed.), Studies in social power (pp. 150–167). Ann Arbor, MI: University of Michigan Press.

LOCOMOTION AND ASSESSMENT REGULATORY MODES

341

Gollwitzer, P. M. (1990). Action phases and mind-sets. In E. T. Higgins & R. M. Sorrentino (Eds.), Handbook of motivation and cognition: Foundations of social behavior (Vol. 2, pp. 53–92). Gollwitzer, P. M., Heckhausen, H., & Stellar, B. (1990). Deliberative vs. implemental mind-sets: Cognitive tuning toward congruous thoughts and information. Journal of Personality and Social Psychology, 59, 1119–1127. Heatherton, T. F., & Polivy, J. (1991). Development and validation of a scale for measuring state self-esteem. Journal of Personality and Social Psychology, 60, 895–910. Heckhausen, H., & Gollwitzer, P. M. (1987). Thought contents and cognitive functioning in motivational versus volitional states of mind. Motivation and Emotion, 11, 101–120. Herman, W. E. (1990). Fear of failure as a distinctive personality trait measure of test anxiety. Journal of Research and Development in Education, 23, 180–185. Higgins, E. T. (1987). Self-discrepancy: A theory relating self and affect. Psychological Review, 94, 319–340. Higgins, E. T. (1989). Knowledge accessibility and activation: Subjectivity and suffering from unconscious sources. In J. S. Uleman & J. A. Bargh (Eds.), Unintended thought (pp. 75–115). New York: Guilford. Higgins, E. T. (1990). Personality, social psychology, and person-situation relations: Standards and knowledge activation as a common language. In L. A. Pervin (Ed.), Handbook of personality (pp. 301–338). New York: Guilford. Higgins, E. T. (2000). Does personality provide unique explanations for behavior?: Personality as cross-person variability in general principles. European Journal of Personality, 14, 391–406. Higgins, E. T., Pierro, A., & Kruglanski, A. W. (2002). Cross-cultural validation of the psychological distinction between locomotion and assessment modes. Unpublished manuscript, Columbia University Hill, C. A. (1987). Affiliation motivation: People who need people . . . but in different ways. Journal of Personality and Social Psychology, 52, 1008–1018. Hinkin, T. R., & Schriesheim, C. A. (1989). Development and application of new scales to measure the French and Raven (1959) bases of power. Journal of Applied Psychology, 74, 561–587. Huba, G. J., Aneshensel, C. S., & Singer, J. L. (1981). Development of scales for three secondorder factors of inner experience. Multivariate Behavioral Research, 16, 181–206. Indvik, J. (1985). A path-goal theory investigation of superior subordinate relationships. Unpublished doctoral dissertation, University of Wisconsin, Madison. Indvik, J. (1988). A more complete testing of path-goal theory. Paper presented at Academy of Management, Anaheim, CA. Jackson, D. N. (1974). Personality Research Form Manual. Goshen, NY: Research Psychologists Press. Jarvis, W. B. G., & Petty, R. E. (1996). The need to evaluate. Journal of Personality and Social Psychology, 70, 172–194. Judge, T. A., Thoresen, C. J., Pucik, V., & Welbourne, T. M. (1999). Managerial coping with organizational change: A dispositional perspective. Journal of Applied Psychology, 84, 107–122. Kanungo, R. N. (1982). Measurement of job and work involvement. Journal of Applied Psychology, 67, 341–349. Klem, A., Higgins, E. T., & Kruglanski, A. W. (1996). Getting started on something versus waiting to do the ‘‘right’’ thing: Locomotion and assessment as distinct regulatory modes. Unpublished manuscript, Columbia University

342

HIGGINS et al.

Kruglanski, A. W. (1975). The endogenous-exogenous partition in attribution theory. Psychological Review, 82, 387–406. Kruglanski, A. W. (1996). Goals as knowledge structures. In P. M. Gollwitzer & J. A. Bargh (Eds.), The psychology of action: Linking cognition and motivation to behavior (pp. 599–619). New York: Guilford. Kruglanski, A. W., Pierro, A., & Higgins, E. T. (2002). Regulatory mode and preferred leadership styles: How fit increases job satisfaction. Unpublished manuscript, University of Maryland at College Park. Kruglanski, A. W., Pierro, A., Mannetti, L., Livi, S., Higgins, E. T., & Capozza, D. (2002). ‘‘On the move’’ or ‘‘staying put’’: Locomotion, need for closure and reactions to organizational change. Unpublished manuscript, University of Maryland at College Park. Kruglanski, A. W., Thompson, E. P., Higgins, E. T., Atash, M. N., Pierro, A., Shah, J. Y., & Spiegel, S. (2000). To ‘‘do the right thing’’ or to ‘‘just do it’’: Locomotion and assessment as distinct self-regulatory imperatives. Journal of Personality and Social Psychology, 79, 793–815. Kuhl, J. (1985). Volitional mediation of cognition-behavior consistency: Self-regulatory processes and action versus state orientation. In J. Kuhl & J. Beckman (Eds.), Action control: From cognition to behavior (pp. 101–128). Berlin: Springer-Verlag. Kuhl, J. (1986). Motivation and information processing: A new look at decision making, dynamic change, and action control. In R. M. Sorrentino & E. T. Higgins (Eds.), Handbook of motivation and cognition: Fundations of social behavior (pp. 404–434). New York: Guilford. Kuhl, J., & Fuhrmann (1998). Decomposing self-regulation and self-control: The volitional components inventory. In J. Heckhausen & C. S. Dweck (Eds.), Motivations and selfregulation across the life span (pp. 15–49). New York: Cambridge University Press. Kuhl, J., & Kazen, M. (1997). Personlichkeits-, Stil- und Storungsinentar. Gottingen: Hogrefe. Leary, M. R. (1983). Social anxiousness: The construct and its measurement. Journal of Personality Assessment, 47, 66–75. Lewin, K. (1951). Field theory in social science. New York: Harper. Lewin, K., Dembo, T., Festinger, L., & Sears, P. S. (1944). Level of aspiration. In J. McHunt (Ed.), Personality and the behavior disorders (Vol. 1, pp. 333–378). New York: Ronald Press. Lewin, K., & Lippit, R. (1938). An experimental approach to the study of autocracy and democracy: A preliminary note. Sociometry, 1, 292–300. Lodhal, T. M., & Kejner, M. (1965). The definition and measurement of job involvement. Journal of Applied Psychology, 49, 24–33. Meyer, J. P., Allen, N. J., & Smith, C. A. (1993). Commitment to organizations and occupations: Extension and test of a three-component conceptualization. Journal of Applied Psychology, 78, 538–551. Mischel, W. (1974). Processes in delay of gratification. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 7, pp. 249–292). New York: Academic Press. Mischel, W. (1981). Metacognition and the rules of delay. In J. H. Flavell & L. Ross (Eds.), Social cognitive development: Frontiers and possible futures (pp. 240–271). New York: Cambridge University Press. Miller, G. A., Galanter, E., & Pribram, K. (1960). Plans and the structure of behavior. New York: Holt. Newman, L. S., Higgins, E. T., & Vookles, J. (1992). Self-guide strength and emotional vulnerability: Birth order as a moderator of self-affect relations. Personality and Social Psychology Bulletin, 18, 402–411. Northouse, P. G. (1997). Leadership. theory and practice. Thousand Oaks, CA: Sage.

LOCOMOTION AND ASSESSMENT REGULATORY MODES

343

Pierro, A., Higgins, E. T., & Kruglanski, A. W. (2002). Regulatory mode and film preferences. Unpublished manuscript, University of Rome ‘‘La Sapienza.’’ Pierro, A., Kruglanski, A. W., & Higgins, E. T. (2002a). Regulatory mode and the joys of doing: Effects of ‘‘locomotion’’ and ‘‘assessment’’ on intrinsic and extrinsic task-motivation in organizations. Unpublished manuscript, University of Rome ‘‘La Sapienza.’’ Pierro, A., Kruglanski, A. W., & Higgins, E. T. (2002b). Progress takes work: Effects of the locomotion dimension on job involvement, effort investment and task performance in organizations. Unpublished manuscript, University of Rome ‘‘La Sapienza.’’ Pierro, A., Mannetti, L., Higgins, E. T., & Kruglanski, A. W. (2002). Moderating role of regulatory mode on relationships between theory of planned behavior’ variables. Unpublished manuscript, University of Rome ‘‘La Sapienza.’’ Powers, W. T. (1973). Behavior: The control of perception. Chicago: Aldine. Pratto, F., Sidanius, J., Stallworth, L. M., & Malle, B. F. (1994). Social dominance orientation: A personality variable predicting social and political attitudes. Journal of Personality and Social Psychology, 67, 741–763. Radloff, L. S. (1977). The CES-D scale: A self-report depression scale for research in the general population. Applied Psychological Measurement, 1, 385–401. Raven, B. H. (1992). A power/interaction model of interpersonal influence: French and Raven thirty years later. Journal of Social Behavior and Personality, 7, 217–244. Raven, B. H. (1993). The bases of power: Origins and recent developments. Journal of Social Issues, 49(4), 227–251. Raven, B. H., & Kruglanski, A. W. (1970). Conflict and power. In P. Swingle (Ed.), The structure of conflict (pp. 69–109). New York: Academic Press. Rosenberg, M. (1965). Society and the adolescent self-image. Princeton, NJ: Princeton University Press. Ryan, R. M., & Frederick, C. (1997). On energy, personality, and health: Subjective vitality as a dynamic reflection of well-being. Journal of Personality, 65, 529–565. Sansone, C., & Harackiewicz, J. M. (1996). ‘‘I don’t feel like it’’: The function of interest in selfregulation. In L. Martin & A. Tesser (Eds.), Striving and feeling: Interactions between goals and affect (pp. 203–228). Mahwah, NJ: Erlbaum. Scheier, M. F., Carver, C. S., & Bridges, M. W. (1994). Distinguishing optimism from neuroticism (and trait anxiety, self-mastery, and self-esteem): A reevaluation of the Life Orientation Test. Journal of Personality and Social Psychology, 67, 1063–1078. Shah, J. Y., Kruglanski, A. W., & Thompson, E. P. (1998). Membership has its (epistemic) rewards: Need for closure effects on intergroup favoritism. Journal of Personality and Social Psychology, 75, 383–393. Sheldon, K. M., & Elliot, A. J. (1999). Goal striving, need satisfaction, and longitudinal wellbeing: The self-concordance model. Journal of Personality and Social Psychology, 76, 482–497. Taylor, A., & Higgins, E. T. (2002). Regulatory mode and activity orientations. Unpublished manuscript, Columbia University. Taylor, S. E., & Brown, J. D. (1988). Illusion and well-being: A social psychological perspective on mental health. Psychological Bulletin, 103, 193–210. Thompson, M. M., Naccarato, M. E., & Parker, K. E. (1989, June). Assessing cognitive need: The development of the Personal Need or Structure and Personal Fear of Invalidity scales. Paper presented at the annual meeting of the Canadian Psychological Association, Halifax, Nova Scotia. Tolman, E. C. (1955). Principles of performance. Psychological Review, 62, 315–326. Treadwell, T. W., Leach, E., & Stein, S. (1993). The social networks inventory. A diagnostic instrument measuring interpersonal relationships. Small Group Research, 24, 155–178.

344

HIGGINS et al.

Tversky, A. (1972). Elimination by aspects: A theory of choice. Psychological Review, 79, 281–299. Vallerand, R. J., Pelletier, L. G., Blais, M. R., Briere, N. M., Senecal, C., & Vallieres, E. F. (1992). The academic motivation scale: A measure of intrinsic, extrinsic, and amotivation in education. Educational and Psychological Measurement, 52, 1003–1017. Vroom, V. H. (1964). Work and motivation. New York: Wiley. Webster’s Ninth New Collegiate Dictionary (1989). Springfield, MA: Merriam-Webster. Webster, D. M., & Kruglanski, A. W. (1994). Individual differences in need for cognitive closure. Journal of Personality and Social Psychology, 67, 1049–1062. Wiener, N. (1948). Cybernetics. New York: Wiley. Zanna, M. P., & Fazio, R. H. (1982). The attitude-behavior relation: Moving toward a third generation of research. In M. P. Zanna, E. T. Higgins, & Herman (Eds.), Consistency in social behavior: The Ontario Symposium (Vol. 2, pp. 283–301). Hillsdale, NJ: Erlbaum.

AFFECTIVE FORECASTING

Timothy D. Wilson Daniel T. Gilbert

It must be troubling for the god who loves you To ponder how much happier you’d be today Had you been able to glimpse your many futures. —Dennis (2001, p. 72)

Foreseeing the future is one of the most appealing of all psychic powers. Who has not dreamed of making millions by predicting which new offering on Wall Street will be the next Microsoft and whether the Red Sox or Phillies will win the World Series? Seeing into the future would bring many advantages other than fattening our wallets, such as eliminating all decisionmaking angst. Rather than worrying about whether we are best suited for a career as a lawyer or an interior designer, whether we should marry Sam or Harry, or whether we should buy our neighbor’s 1992 Volvo, we could simply glance into our crystal balls and see how these various options would pan out. People do not have crystal balls, of course (at least not accurate ones) and thus must prognosticate as best they can, based on what they know in the present. There is a great deal of research on how people make predictions about the future, including decision making under uncertainty (e.g., Kahneman & Tversky, 2000; Gilovich, Griffin, & Kahneman, 2002; Nisbett & Ross, 1980), the accuracy of people’s predictions about their future behavior (e.g., Osberg & Shrauger, 1986; Sherman, 1980; Wilson & LaFleur, 1995), the effects of temporal perspective on prediction (e.g., Trope & Liberman, in press), and optimistic biases in self-prediction (Armor & Taylor, 1998; Taylor & Brown, 1988; Weinstein, 1980). Until recently virtually all research on prediction has concerned people’s ability to anticipate the occurrence of future external events (e.g., ‘‘will the 345 ADVANCES IN EXPERIMENTAL SOCIAL PSYCHOLOGY, VOL. 35

Copyright 2003 Elsevier Science (USA). All rights reserved. 0065-2601/03 $35.00

346

WILSON AND GILBERT

price of Microsoft stock go up or down?’’) or their own behavior (e.g., ‘‘am I likely to get divorced in the next 10 years?’’). A crucial form of prediction has been overlooked, namely people’s ability to forecast their own feelings. What people really want to know about the future, we maintain, is what their level of happiness and well-being will be, and many questions about future events and behaviors are really proxies for questions about these affective states. People want to be able to predict whether they will get married or divorced or have children because they believe that such life events are crucial determinants of their happiness. They want to know the future price of Microsoft stock so that they can make money, which they believe will increase their happiness. The pursuit of happiness is one of the most fundamental of all human motives, and if people had crystal balls in good working order, they would peer into them most often to try to achieve that goal. How successful would people be at achieving happiness? This crucial question depends on whether people can predict accurately which events will make them happy, by how much, and for how long. That is, predictions about future events are good proxies for predictions about feelings only if people know for sure how much happiness these events will bring and how long that happiness will last. Will a cool drink be more refreshing than another helping of guacamole? Will it be more enjoyable to go to La Traviata at the Met or watch a soap opera on TV? Will marrying Sam bring more or less happiness, in the long run, than living life as a single parent? In recent years several researchers have begun to investigate affective forecasting, namely people’s predictions about their future feelings (e.g., Baron, 1992; Gilbert, Driver-Linn, & Wilson, 2002; Gilbert & Wilson, 2000; Kahneman, 1994; Kahneman & Snell, 1990, Linville & Fischer, 1991; Loewenstein & Frederick, 1997; Loewenstein, Nagin, & Paternoster, 1997; Loewenstein & Prelec, 1993; Loewenstein & Schkade, 1999; Mellers, 2000; Mellers & McGraw, 2001; Snell, Gibbs, & Varey, 1995; Wilson, Centerbar, & Gilbert, in press; Zeelenberg, van Dijk, Manstead, & van der Pligt, 2000). Most of the early work in this area measured people’s forecasts but not their actual emotional responses, and thus did not assess forecasting accuracy. Increasingly researchers have assessed accuracy by measuring both predicted and experienced emotional responses, permitting the systematic study of errors in affective forecasting. I. Types of Affective Forecasts and Errors Affective forecasts can be broken down into four components: predictions about the valence of one’s future feelings, the specific emotions that will be experienced, the intensity of the emotions, and their duration. People can be

AFFECTIVE FORECASTING

347

accurate or inaccurate in predicting each of these facets of emotional experience.

A. PREDICTING VALENCE Life would be difficult indeed if people were mistaken about the valence of future events. People would leave the house wearing down coats on hot summer days, thinking that bundling up would make them comfortable. Customers in music stores would have no idea what type of music to shop for and once they got home they would often discover that they hated the recording that they thought they would love. People would fail to anticipate that a full body massage is a pleasant experience whereas a full body beating is not, or that lemonade with sugar tastes better than lemonade without. We do not mean to imply that people always make correct predictions about valence. When experiencing something for the first time, such as the Thunderbolt Roller Coaster at an amusement park, people might discover that what they thought would be thrilling is more like a trip through the underworld. In general, however, people make accurate predictions about which side of the neutral point their emotional experiences will fall, especially if they have had experience in that domain. In one study, for example, Wilson, Wheatley, Kurtz, Dunn, and Gilbert (2002) staged a simulated dating game, in which college students competed with a same-sex student for a hypothetical date with an opposite-sex student. Experiencers were randomly assigned to win or lose the date, after which they rated their mood. Forecasters estimated what their mood would be if they won or lost the date. Without exception, all forecasters estimated that they would be in a better mood if they won than if they lost, and indeed, experiencers who won were, on average, in a better mood than experiencers who lost. Forecasters overestimated how positive or negative they would feel—a point we will return to shortly—but for present purposes, we note that they were accurate about the valence of winning versus losing.

B. PREDICTING SPECIFIC EMOTIONS Even if people correctly predict the valence of their feelings (e.g., that they will feel negatively), they still need to identify the specific emotions they will experience (e.g., disgust, anger, fear, or a blend of all three). They probably are able to do so much of the time; people have a pretty good idea of what will disgust them and do not mistakenly think that these events will cause anger or fear instead. People know that watching a comedy is more likely to

348

WILSON AND GILBERT

produce mirth than pride, and that watching one’s favorite team win a championship is more likely to result in joy than fits of laughter. Robinson and Clore (2001), for example, gave people written descriptions of a series of emotion-provoking pictures and asked them to predict on 20 emotion scales how the actual pictures would make them feel. Not surprisingly, the forecasters were generally correct about which emotion they would experience, compared to people who actually saw the pictures. They correctly estimated, for example, that a snarling wolf would trigger more fear than a picture of kissing lovers would, and that a photo of a filthy toilet would trigger more disgust than would the snarling wolf. Emotions can occur in complex blends, of course, and people might fail to anticipate the precise nature of the mix they will experience. We suspect that this is especially true for events that produce a combination of positive and negative emotions. People often view the future in a simplistic manner, assuming that events will cause primarily good or bad feelings, rather than a rich mixture of both. When imagining their graduation day, for example, college students might focus on the feelings of joy and pride they are likely to experience, overlooking the fact that they are likely also to feel sadness about leaving their friends and apprehension about the future (Larsen, McGraw, & Cacioppo, 2001). As compelling as this idea sounds (at least to us), there is not yet much evidence for it. In the Wilson, Wheatley, Kurtz, Dunn, and Gilbert (2002) dating game study, for example, forecasters predicted how positive/negative, happy, sad, frustrated, calm, and agitated they would feel if they won or lost the date, and experiencers rated their actual feelings on these same scales. There was no evidence that forecasters’ predictions were less complex than experiencers’ reports of their actual experiences; factor analyses generally revealed the same number of factors for predictions as experiences. Similarly, Robinson and Clore (2001) performed cluster analyses of people’s predicted emotional reactions to the descriptions of emotion-provoking pictures, and found that the structure of people’s predicted reactions matched closely the structure of people’s actual reactions to the pictures. Liberman, Sagristano, and Trope (2002), on the other hand, found that people have overly simplistic, schema-driven views of their reactions to emotional events when thinking about the distant future (as opposed to the near future). When imagining what a good day would be like tomorrow, for example, participants described a series of events that would be mostly positive but that would be tempered by some negative occurrences (e.g., ‘‘I’ll have fun going out with my friends but I also have a test in my chemistry class’’). When imagining what a good day will be like in a year, participants described a more uniform set of positive events, untempered by anything negative (e.g., ‘‘I’ll have fun going out with my friends and we’ll hear some

AFFECTIVE FORECASTING

349

good music’’). These findings suggest that people’s forecasts might be more realistic for events that will happen soon but overly simplistic for events far in the future. People’s predictions also might be overly simplistic when they anticipate events that cause more ambivalence than the ones we or Robinson and Clore (2001) examined. Winning or losing a dating game, for most college students, produces straightforward feelings of joy or disappointment, and a picture of a snarling wolf probably causes little ambivalence. Other events, such as graduating from college, produce a more complex blend of emotions, and people might not be very good at predicting the precise mixture that will occur. Further, people might misconstrue the nature of complex social situations when imagining them in advance, and thereby mispredict the dominant emotion that they will experience. Woodzicka and LaFrance (2001) described a job interview to women and asked them how they would feel if they were asked a specific set of sexually harassing questions. The women predicted that their predominant emotional reaction would be anger and that they would experience little fear. When women were actually interviewed for a job and asked the harassing questions, however, their predominant emotional reaction was fear and relatively few reported anger. Failing to appreciate the nuances of complex social situations is, we suspect, the major cause of mispredicting which emotions people will experience, a point we will return to later.

C. PREDICTING INTENSITY AND DURATION Often people predict correctly the valence of their emotional reactions (‘‘I’ll feel good if I get the job’’) and correctly predict the specific emotions they will experience (e.g., joy). Even when achieving such accuracy, however, it is important for people to predict what the initial intensity of the reaction will be (how much joy they will experience) and the duration of that emotion (how long they will feel this way). It is useful to know that we will feel happy on our first day at a new job, but better to know how happy and how long this feeling will last, before committing ourselves to a lifetime of work as a tax attorney. It is helpful to know that it will be painful to end a long-term relationship, but better to know how painful and whether the pain will last half a second or half a decade. As we will soon document, numerous studies have found that people overestimate the impact of future events on their emotional lives. We initially referred to this as a durability bias, defined as the tendency to overestimate the duration of one’s future emotional reactions (Gilbert,

350

WILSON AND GILBERT

Pinel, Wilson, Blumberg, & Wheatley, 1998). It is now clear that this term can be misleading, because overestimating the impact of future emotional events can involve a misprediction of several different aspects of emotional experience. Figure. 1 depicts an experienced emotional reaction to an event and people’s hypothetical predictions about their reactions. To predict correctly how they will feel some time after the event occurs, people would need to know the acceleration of their initial emotional reaction (the rate of increase in the emotional reaction after the event occurs), the peak level of intensity of their reaction, and the rate of deceleration (i.e., the rate of the return to baseline). The dotted line in Fig. 1 represents a prediction error on all three components: People overestimate the rate of acceleration, overestimate the peak level of intensity, and underestimate the rate of deceleration. Many other forms of error are possible as well; for example, people could make correct predictions about the rate of acceleration and deceleration, but underestimate the peak intensity. Or, they could be correct about the peak intensity but underestimate the rate of deceleration. Further, emotional reactions could occur in more complex patterns than depicted in Fig. 1; for example, they could oscillate in response to environmental reminders of the event. To make a correct prediction, people would need to anticipate these complex response patterns. When people’s emotions are measured at only one point in time following an emotional event, it is impossible to tease apart the precise nature of any error they have committed. If predicted and experienced emotions were measured only at Time 3 in Fig. 1, for example, it would be clear that people made a prediction error, estimating that their emotional reaction at that point would be more intense than it in fact was. Whether this error was due to an overestimation of the initial intensity of the experience, an error about how quickly the reaction would dissipate, or some other error about the

Predicted Experienced Emotional intensity

Time 2 Time 3

Time 1 (Emotional event)

Time 4

Time Fig. 1.

The hypothetical time course of predicted and experienced emotion.

AFFECTIVE FORECASTING

351

time course of their emotions would be unknown. Wilson, Wheatley, Meyers, Gilbert, and Axsom (2000), for example, found that college sports fans overestimated how happy they would be the day after their favorite team won a football game. Fans might have correctly predicted exactly how happy they were immediately after the game, but have been wrong about how long this feeling would last. Alternatively, they might have been accurate about the rate of deceleration of their emotional reactions to games, but overestimated how happy they were initially. Or, they might have made both errors, underestimating intensity and overestimating deceleration. We will refer to mispredictions of this sort as an impact bias, defined as the tendency to overestimate the enduring impact that future events will have on our emotional reactions (Gilbert, Driver-Linn, & Wilson, 2002). This term is intentionally broad, covering a number of more specific errors (e.g., overestimating initial intensity and overestimating duration). It would be quite interesting, of course, to investigate the temporal structure of people’s emotional reactions, compare this to their predictions, and isolate the various components of the impact bias. In some studies researchers have measured affect at more than one point in time, allowing some insight into its time course. However, there are several reasons why it is difficult to focus on the precise time course of affective forecasts and emotional reactions. First, the measurement problems associated with the time course of affective response are not trivial (Larsen & Fredrickson, 1999). It is difficult enough to ask people to predict and report their emotional reactions at one or two points in time, but to have them make the number of predictions required to estimate the velocity and acceleration of their responses is challenging (Hsee & Abelson, 1991). Second, the impact bias is an important phenomenon regardless of which elements compose it in any particular instance. Whether people overestimate how good or bad they will feel, overestimate how quickly those feelings will arise, or underestimate how quickly they will dissipate, the important point is that they overestimate how powerfully the event will impact their emotional lives. Put differently, what matters to us is whether an emotion is more or less intense than predicted at some point after an emotional event has occurred, and in most cases, it does not matter whether this misprediction is due to a mistaken estimate of acceleration, peak magnitude, or deceleration. We are interested, of course, in the psychological processes that make people’s emotions different from what they expected at different time points after the event (from a few seconds to a few years or more). We have focused on a number of the mechanisms responsible for the impact bias and other kinds of prediction errors, which we will review shortly.

352

WILSON AND GILBERT

D. MEASUREMENT AND VALIDITY Most studies of affective forecasting use self-report scales, such as ones that ask people to report how happy they are now or (or will be) ‘‘in general these days.’’ These measures have been found to have good psychometric properties (Andrews & Robinson, 1991; Fordyce, 1988). When researchers have included more extensive measures (e.g., the ‘‘Satisfaction with Life Scale’’; Diener, Emmons, Larsen, & Griffin, 1985; the ‘‘Affectometer’’; Kammann & Flett, 1983), they have found healthy correlations between one-item measures of happiness and multiitem measures (Gilbert et al., 1998). Other studies have included measures of more specific hedonic states, using such instruments as the Positive and Negative Affect Schedules (PANAS; Watson, Clark, & Tellegen, 1988); the Multiple Affect Adjective Check List (MAACL; Zuckerman & Lubin, 1965), or emotion words specific to an individual study. Self-report scales such as these have a proven track record of reliability and validity and allow a close comparison between predicted and actual ratings on the same scales. Although it is possible that these indices are susceptible to self-presentational concerns, perspective shifts, and demand characteristics, there are a number of reasons to doubt that these sources of bias are a problem in most affective forecasting studies. For example, many of the findings we will encounter cannot be explained by the desire to put one’s best foot forward, such as the fact that people often report feeling less positively than they predicted they would. If people were trying to present themselves in a positive light, it is unclear why they would admit to feeling negatively, more so than they had predicted. Further evidence for the validity of self-report measures of affect comes from studies that used behavioral as well as self-report indices. A version of the Wilson, Wheatley, Kurtz, Dunn, and Gilbert (2002) dating game study, for example, included both types of measures. Forecasters predicted how they would feel if they lost the dating game, on standard mood scales. Experiencers found out that they had, in fact, lost the dating game and then rated their mood on the same scales. Forecasters predicted that they would be significantly less happy after a loss than experiencers actually were. The important point for present purposes is that Wilson et al. (2002) also included a behavioral forecasting measure. All participants were given the opportunity to take a mood-enhancing herbal drug, ostensibly as part of a pilot program to ensure that participants left psychology studies in the same frame of mind as when they first arrived. Forecasters chose the dosage of drug they would want to take if they found out that they were not chosen for the date, whereas experiencers found out that they were not chosen and then selected the dosage of drug they wanted to take. Consistent with the results

AFFECTIVE FORECASTING

353

on the mood scales, forecasters selected a significantly higher dosage of the drug than did experiencers. In addition to increasing our faith in the validity of self-report scales, results such as these demonstrate that affective forecasts are often translated into social behaviors of considerable interest. People base important decisions on affective forecasts, such as the types of consumer goods they want to purchase in the future (Gilbert, Gill, & Wilson, 2002; Read, Loewenstein, & Kalyanaraman, 1999; Read & van Leeuwen, 1998; Simonson, 1990) and which gambles they prefer to take (Mellers & McGraw, 2001; Mellers, Schwartz, & Ritov, 1999; Read et al., 1999). We assume that many other important life decisions are based on affective forecasts as well, such as whom to marry and what career to pursue. More broadly, happiness and subjective well being are highly valued states in virtually all cultures that predict many positive outcomes in life, such as marital satisfaction, job success, good health, and longevity (Danner, Snowdon, & Friesen, 2001; Diener, 2000). Thus, it would seem to be important for people to be able to predict their future levels of happiness.

E. THE IMPACT BIAS The most prevalent error found in research on affective forecasting is the impact bias, whereby people overestimate the impact of future events on their emotional reactions (Buehler & McFarland, 2001; Crawford, McConnell, Lewis, & Sherman, 2002; Gilbert, Brown, Pinel, & Wilson, 2000; Gilbert et al., 1998; Mellers & McGraw, 2001; Mitchell, Thompson, Peterson, & Cronk, 1997; Rachman, 1994; Rachman & Arntz, 1991; Robinson & Clore, 2001; Schkade & Kahneman, 1997; Schmidt, Jacquin, & Telch, 1994; Sieff, Dawes, & Loewenstein, 1999; van Hout & Emmelkamp, 1994; Wilson et al., 2000, in press). The impact bias has been found in a variety of populations (e.g., college students, professors, sports fans, dieters, vacationers, snake phobics, people taking medical tests), with a wide range of emotional events (e.g., romantic breakups, personal insults, sports victories, electoral defeats, parachute jumps, failures to lose weight, reading tragic stories, and learning the results of pregnancy and HIV tests). Although the impact bias is by far the most common finding in the affective forecasting literature, a few studies have found an underprediction of the impact of one’s future affective reactions. There are also cases of the misprediction of specific emotions. We will note these exceptions as we discuss the process by which people make affective forecasts and possible sources of error.

354

WILSON AND GILBERT

II. The Process of Affective Forecasting and Sources of Error Figure 2 depicts the process of affective forecasting (shown in the boxes). The left of the dotted line represents the time at which forecasts are made, whereas the right of the dotted line represents people’s future experiences at the predicted point in time. As seen in the figure there are several sources of error (shown in the circles), which we will discuss in turn. A. CONSTRUAL When people think about how they will feel when a future event occurs, they first must bring to mind a representation of that event. If people have experienced the event many times before (e.g., commuting to work), they can form such a representation effortlessly by recalling a prototype or exemplar of it. When people think about events that they have not experienced before, such as the birth of a child, getting married, or attending a party at the house of a new acquaintance, they need to construct a representation of what the event is likely to entail. The first source of error is the misconstrual problem, whereby people mistakenly imagine the wrong event. When asked how she will feel at the birth of her first child, a woman might imagine a trouble-free, natural delivery followed by a quiet period of intimate bonding with the baby. What happens instead is 24 hours of painful labor, a Cesarean section, and intrusive visits from in-laws armed with video cameras. People do not have PRESENT (Affective Forecasts) Correction For Unique Influences

Recall, Affective Theories Assessed Affective Reaction

Representation Of Event

Construal

Framing Effects

Fig. 2.

FUTURE (Affective Experiences) Unique Influences

Affective Forecast

Initial Affective Experience

Expectation Effects

Affective Experience Over Time

Sense Making Processes (Ordinization, Psychological Immune System)

Sources of influence and bias on affective forecasts.

AFFECTIVE FORECASTING

355

crystal balls and the future often unfolds in ways they do not expect, producing discrepancies between their predicted and actual feelings. Because such unexpected outcomes are common, one might think that people would be reticent about making confident predictions about future behavior and feelings. ‘‘I’m not sure how I’ll feel when my baby is born,’’ an expectant mother might say, ‘‘because the birth could unfold in any number of ways.’’ Although people know that the future is uncertain, they seem to think they can predict it better than they can. Griffin and Ross (1991) reviewed substantial evidence that people do not appreciate how much their views of the future (and the present) are construals rather than representations of objective reality. Thus, by failing to appreciate the fact that a future event may not occur in precisely the way they imagine, people are prone to errors in their forecasts about how they will feel. Most research in this area has been concerned with the way in which misconstruals lead to errors in prediction about how people will behave. Social psychologists, for example, have documented many instances in which people underestimate the power of a social situation and thus make incorrect behavioral predictions, such as the failure to appreciate how much people will obey an authority figure in the Milgram (1974) studies or will fail to help a bystander in the Latane´ and Darley (1970) studies (Ross & Nisbett, 1991). Clearly, if people’s misconstruals of how a situation will influence their behavior are off to a large degree (e.g., they assume they will help a person in need but in fact do not), their predictions about how they will feel in these situations will be inaccurate. Few of these studies, however, asked people directly to make affective forecasts. As noted earlier an exception is a study by Woodzicka and LaFrance (2001), who asked women to predict how they would react if they were asked sexually harassing questions during a job interview and compared these predictions to the actual reactions of women who really were asked the sexually harassing questions during an interview. The forecasters’ predictions were surprisingly at odds with the experiences of the women who participated in the real interview, in part because they imagined a different situation than the one faced by the experiencers. The forecasters imagined a situation in which it would be easy to confront the interviewer and where their primary emotional reaction would be anger. In the real interview the women were suddenly faced by a confusing and surprising interaction in which their primary emotional reaction was fear. Sixty-eight percent of forecasters said they would refuse to answer at least one of the three questions, whereas every experiencer answered every question. Twenty-eight percent of the forecasters said that they would confront the interviewer or leave; none of the experiencers did so. Woodzicka and LaFrance (2001) note that women are often unfairly blamed for failing to confront harassment,

356

WILSON AND GILBERT

precisely because of this kind of misconstrual error. People imagine a situation in which it is easy to confront a harasser, failing to appreciate that the situation will be a more complex one in which they experience intimidation, confusion, and fear. 1. Types of Forecasting Errors Caused by Misconstrual Misconstruals of future situations produce the greatest latitude of affective forecasting errors, because there is no limit to how far off people’s construal of the situation can be. If a prospective mother completely misconstrues what the birth of her child will be like, for example, she could be wrong about the specific emotions she will experience and the valence, intensity, and duration of her emotional reactions. There is no reason to assume that misconstrual leads more to overprediction than underprediction of the intensity of affective reactions. If an expectant mother imagines a serene, peaceful birth but has a painful Cesarean section, she will have overestimated how positive she will feel. However, she could just as easily have imagined a painful Cesarean section and experienced a serene, peaceful birth, thereby underestimating how positive she will feel. Misconstrual surely leads to many prediction errors, but it does not explain the prevalence of the impact bias.

B. FRAMING EFFECTS People’s representations of events also depend on the way in which people frame them, such as the particular attributes of the events that happen to capture people’s attention (e.g., Kahneman & Tversky, 1979, 1984). People are prone to an isolation effect, for example, which Kahneman and Tversky (1979) defined as the case in which people ‘‘disregard components that the alternatives share, and focus on the components that distinguish them’’ (p. 271; see also Hodges, 1997; Houston, Sherman, & Baker, 1991; Tversky, 1972). When making affective forecasts people often think about how they will feel under alternative scenarios, such as how happy they will be if they vacation at the beach versus the mountains, or if they purchase a home on Elm Street versus Main Street. The isolation effect suggests that when comparing alternative future events people focus too much on features that differentiate the alternatives and too little on features they share, even if the shared features will influence their future happiness. To test this hypothesis, Dunn, Wilson, and Gilbert (in press) asked firstyear college students to forecast what their overall level of happiness would

AFFECTIVE FORECASTING

357

be the following year if they lived in various dormitories (‘‘houses’’) at their university, soon before they were randomly assigned to one of the houses. Because the task was framed as a comparison between the different houses, people were expected to focus more on ways in which the houses differed (e.g., physical features such as their beauty and location) and less on factors that the houses would have in common, such as the quality of people’s social life (e.g., relationship with their roommates, who would move with the students to whichever house they were assigned), even though social relationships have been found to have a big impact on people’s subjective well-being (e.g., Diener & Seligman, 2002). Consistent with this hypothesis, people’s forecasts were much more a function of their ratings of physical features than social features, but when contacted a year later, people’s happiness was much more a function of social features than physical features. The net effect was a strong impact bias; by focusing too much on a variable that distinguished between the houses but was uncorrelated with their future happiness, people overestimated the effect of house assignment on their eventual well-being. As seen in Table I, for example, people predicted that they would be much less happy if they were assigned to undesirable than desirable houses, but they were wrong; those living in undesirable houses were nearly identical in their overall happiness to those living in desirable houses. In a second study Dunn et al. (in press) manipulated people’s framing of the task by asking some participants to think about factors that were the same about the houses and things that were different, in counterbalanced order. A recency effect was found, whereby people who thought last about what would be the same across houses based their forecasts more on social features (that were constant across houses). Other participants, replicating Study 1, based their forecasts more on physical factors that distinguished between houses. In short, the way in which people frame the

TABLE I Predicted versus Actual Overall Happiness for College Students Randomly Assigned to Undesirable and Desirable Housesa

Undesirable houses Desirable houses a

Predicted

Actual

Significance of difference

3.43 (1.50) 5.96 (.85)

5.37 (1.16) 5.45 (.92)

t(52) = 10.71, p < .0001 t(55) = 3.04, p < 0.01

Predicted and actual happiness were rated on seven-point scales with endpoints labeled 1 = unhappy and 7 = happy. Standard deviation are in parentheses. Adapted from Dunn et al. (in press).

358

WILSON AND GILBERT

forecasting task can influence the features of an event they use to formulate their forecasts. 1. Types of Forecasting Errors Caused by Framing Effects Framing effects will produce an impact bias if, as in the Dunn et al. (in press) studies, people focus their attention on features that they think will influence their emotional states but that actually will be of little importance. In principle, framing effects could increase the accuracy of forecasts, if they focused people’s attention on factors that really will influence their later happiness. Suppose, for example, that Dunn et al. (in press) had framed the forecasting task differently, asking students to predict how happy they would be in a year if they were living with different groups of roommates in the same house, rather than in different houses with the same group of roommates. The isolation effect would focus their attention on the effect of social relationships with roommates (which would be variable in this task) rather than on the physical features of the house (which would be constant in this task). Because social relationships are important for people’s happiness (Diener & Seligman, 2002; Myers, 1999), framing the task in this way would be likely to increase the accuracy of people’s affective forecasts.

C. RECALL AND AFFECTIVE THEORIES Suppose that people avoid misconstrual and framing effects and bring to mind an accurate representation of the future event. As seen in Fig. 2, the next step in the process is to figure out how one feels about that event (Loewenstein, O’Donoghue, & Rabin, 1999; Robinson & Clore, 2002). If people have experienced the event before, one way they can assess their feelings about it is simply to recall how they felt in the past. When imagining how they feel about colonoscopies, for example, people could try to recall how they felt when they had one 3 years ago, and assume that they will feel this way again. It is well known, however, that memory for past affective experiences is poor. Robinson and Clore (2002) note that emotional experiences are not stored in memory in a form that can be retrieved directly later. People can remember that a colonoscopy was painful, but the pain itself is not stored in memory and then recalled in its original form. If emotional experiences could be retrieved in their original form, there would be no need to go to the trouble to recreate positive experiences such as vacations or roller coaster

AFFECTIVE FORECASTING

359

rides; we could relive them by recalling and ‘‘replaying’’ our past reactions to these events (Robinson & Clore, 2002). Instead of replaying past emotions, people often recall the details of an experience and have emotional reactions to these memories (e.g., someone who feels happy when thinking about his or her vacation in Paris). Because people’s recall of their past experiences can be biased, however, there is no guarantee that the feelings evoked by their memories are the same as the feelings they originally experienced. Kahneman and his colleagues, for example, found that whereas on-line emotional experience is quite sensitive to time (e.g., whether a colonoscopy lasts for 20 or 30 minutes), retrospective emotion reports are quite insensitive to time. Instead, retrospective reports are heavily influenced by the peak intensity of the experience and the intensity of the emotional experience when it ended (e.g., the highest amount of pain during the colonoscopy and the amount of pain at its conclusion; see Ariely, 1998; Fredrickson & Kahneman, 1993; Kahneman, Fredrickson, Schreiber, & Redelmeier, 1993; Varey & Kahneman, 1992). Thus, to the extent that people predict their future online experiences (how they will feel during a colonoscopy or their next vacation in Paris) from their recall of their past experiences, systematic errors are likely to occur. As time goes by, people’s memory for the episodic details of an experience fades and they must rely more on their theories about how the event will make them feel, rather than on the details of the actual experience (Robinson & Clore, 2002). This is also true of events people have never before experienced; they must rely on their affective theories rather than their past experiences (e.g., ‘‘A colonoscopy will be horrible’’; ‘‘My wedding day will be full of joy’’). Every culture has rich theories about the causes of emotions and people develop idiosyncratic theories based on their past experiences (Hochschild, 1979; Nisbett & Wilson, 1977; Wilson & Stone, 1985). These theories are part of the semantic and conceptual knowledge that people use when making predictions about and recalling their emotional experiences and can be at variance with people’s actual experiences. Wilson, Laser, and Stone (1982), for example, found that people relied on shared and idiosyncratic theories when assessing the predictors of their daily mood, and although some of these theories were correct (e.g., the relationship between the quality of their interpersonal relationships and their moods), others were not (e.g., the relationship between the amount of sleep they had gotten the previous night and their moods). McFarland, Ross, and DeCourville (1989) found that many women held the theory that they were in worse moods when they were menstruating, and recalled being in worse moods during their periods. When asked to rate their mood on a daily basis for several weeks, however,

360

WILSON AND GILBERT

these women were in no worse a mood when they were menstruating than when they were not. 1. Types of Forecasting Errors Caused by Poor Recall and Inaccurate Theories As discussed, people’s recall of their emotional experiences is biased in systematic ways; they weight the peaks and endings of their past experiences more than their duration, which will lead to prediction errors (e.g., that a 30 minute colonoscopy will be no worse than a 20 minute one). Incorrect theories can lead to errors in predicting specific emotions as well as their intensity and duration; people might predict that a lack of sleep will produce prolonged grumpiness and irritation, for example, when in fact it does not cause these emotions at all. There is no reason to assume that inaccurate theories are biased in the direction of overprediction. Sometimes people underestimate their emotional reactions because of faulty theories, as in the case of a man who believes that sad movies have little effect on him, only to find tears trickling down his cheek as the hero and heroine say goodbye and the camera fades to black.

D. CORRECTION FOR UNIQUE INFLUENCES There is another problem with gauging our affective reaction to an event and deciding how likely we are to have the same reaction in the future. The circumstances under which people make affective forecasts are almost always different from the circumstances under which they will actually experience an event, and people must subtract out several potential sources of bias on their current assessments of their feelings (labeled ‘‘correction for unique influences’’ on the left side of Fig. 2; see Elster & Loewenstein, 1992; Gilbert, Gill, & Wilson, 2002; Loewenstein, Weber, Hsee, & Welch, 2001). As noted by Kahneman and Tversky (1984), ‘‘Some factors that affect decisions do not have a comparable impact on the experience of outcomes’’ (p. 350). Suppose, for example, that a person is suffering from a cold when trying to decide whether to accept an invitation to a party the following month, and her current negative feelings taint her assessment of how she will feel at the party. Loewenstein et al. (1999) have referred to this phenomenon as a projection bias, defined as the tendency for people to ‘‘underappreciate the effects of changes in their states, and hence falsely project their current preferences . . . onto their future preferences’’ (p. 1; see also Loewenstein, 1996). The projection bias is an instance of mental contamination, whereby people’s judgments, emotions, or behaviors are influenced in unwanted ways

AFFECTIVE FORECASTING

361

(Wilson & Brekke, 1994). In the present context, people attempt to come up with an unbiased estimate of what their affective state will be in the future, but their assessment is contaminated by unique influences on their current well-known affective state. It is well known that it is difficult to decontaminate one’s own biased judgments. In order to do so people would have to be aware that their judgment is biased, be motivated to correct the bias, be aware of the precise direction and magnitude of the bias, and be able to correct their responses accordingly (Gilbert, 2002; Martin & Stapel, 1998; Tversky & Kahneman, 1974; Wegener & Petty, 1997; Wilson & Brekke, 1994; Wilson, Centerbar, & Brekke, 2002). When thinking about how much they will enjoy a party next month, people might not realize that their estimate is biased by the fact that they have a cold, thereby failing to satisfy the first criterion for correction. Even if they recognize that they are biased, people might not know by how much. And, even if they know by how much, they might not have the resources or motivation to correct their judgments sufficiently (Gilbert, 2002). Several studies have shown, for example, that when people shop for food to be consumed later, they are influenced by their current state of hunger (Gilbert, Gill, & Wilson, 2002; Nisbett & Kanouse, 1969; Read & van Leeuwen, 1998). Shoppers who have not eaten for several hours think, ‘‘Surely I will want several bags of corn chips and a couple of cartons of ice cream next week,’’ failing to adjust for the fact that they will often be full during the week and not experiencing the same cravings. Gilbert, Gill, and Wilson (2002) found that this failure to adjust was exacerbated when people were under cognitive load. Hungry or not hungry participants predicted how much they would enjoy eating spaghetti with meat sauce either the next morning or the next evening. People seem to have considered how appetizing spaghetti sounded to them right then (‘‘Mm, spaghetti!’’) and then adjusted for the time they would eat it the following day (‘‘But it wouldn’t be all that great for breakfast’’). When they were not under cognitive load people adjusted to some extent but were still influenced by their current hunger; hungry participants predicted that they would enjoy the spaghetti more the next morning than nonhungry participants did. When they were under cognitive load people found it even more difficult to adjust; hungry participants said that they would enjoy spaghetti in the morning a great deal, as much as they would enjoy it in the evening. 1. Types of Forecasting Errors Caused by Inadequate Correction for Unique Influences Inadequate correction can lead to a number of different kinds of forecasting errors, depending on the direction and strength of the influence

362

WILSON AND GILBERT

and the degree to which people correct for it. If people have a cold when thinking about next month’s party, and do not correct for this fact, they might be wrong about the valence of their reaction (‘‘I’ll have a terrible time’’), or at least the intensity and duration of their affect (‘‘It will be OK but I’ll won’t want to stay very long’’). If people are in an extreme positive state when making the prediction, however, they might make forecasting errors in the opposite direction, overestimating how enjoyable the party will be. There is no reason to assume that inadequate correction is a major source of the impact bias.

E. EXPECTATION EFFECTS Expectation effects occur when people’s affective forecasts change their actual emotional experience. People who watch a movie with the expectation that it will be one of the best they have ever seen might have a different experience than people who see the same movie with no expectations. Assimilation occurs when people who expect to like the movie enjoy it more; contrast occurs when people who expect to like the movie enjoy it less, if it turns out not to live up to their expectations. Wilson and Klaaren (1994) reviewed the conditions under which assimilation and contrast are likely to occur in the realm of affective expectations. Assimilation is observed when people’s expectations are not too discrepant from the experience (e.g., the movie is not quite as good as people expected) and people rapidly assimilate the experience to their expectations. Wilson, Lisle, Kraft, and Wetzel (1989), for example, showed participants a series of six cartoons, the first three of which were relatively funny and the last three of which were not. When people looked at the cartoons with no expectations about how funny they would be they noticed the discrepancy and rated the first three as significantly funnier than the last three. When people were told that previous participants had found all six cartoons to be very funny they showed clear evidence of assimilation. They found the last three cartoons to be as funny as the first three, and significantly funnier than did people with no expectations. This was true not only of their self-reported ratings of the cartoons, which might have been susceptible to demand characteristics. People’s facial expressions were videotaped without their prior knowledge, and those who expected the cartoons to be funny showed significantly more facial mirth while watching the final three cartoons than did people in the no expectations condition. People with no expectations spent more time looking at the last three cartoons than the first three, apparently because they noticed the change in how funny they were. People who expected the cartoons to be funny looked

AFFECTIVE FORECASTING

363

at the first and last three cartoons for about the same amount of time, apparently because they did not notice the change in funniness. Armed with the expectation that the last three cartoons would be funny, these people looked at them for significantly less time and found them to be significantly funnier, compared to people with no expectations (see also Kirsch, 1995; Klaaren, Hodges, & Wilson, 1994). Research on affective expectations has rarely found evidence of contrast effects, whereby people evaluate a stimulus in a direction away from the expectations, after having those expectations disconfirmed. An exception is a study by Geers and Lassiter (1999) in which people were given the expectation that a film would be funny when in fact it was not. In addition, participants were instructed to segment the film into units, by pressing a button whenever a meaningful action occurred (Newtson, 1973). In the gross unitization condition people were instructed to segment the film into the ‘‘largest actions that are meaningful to you,’’ whereas in the fine unitization condition people were instructed to segment the film into the ‘‘smallest actions that are meaningful to you’’ (Geers & Lassiter, 1999, p. 404). When people segmented the film into gross units, those who expected it to be funny rated it as funnier than those who did not, replicating the Wilson et al. (1989) study. When people segmented the film into fine units they were more likely to notice that it was not as funny as they expected, resulting in contrast: People who expected the film to be funny rated it as less funny than those who did not. 1. Types of Forecasting Errors Caused by Expectation Effects Affective expectations can increase the accuracy of forecasts to the extent that assimilation occurs. If people expect to like a movie, and that expectation causes them to like it more, their expectation will be confirmed—not because the movie was objectively enjoyable but because of a self-fulfilling prophecy. Such an effect is similar to Sherman’s (1980) demonstration of self-erasing prediction errors, whereby the act of predicting how they would behave in the future increased the likelihood that people would behave in that manner. Affective expectations will reduce the accuracy of a forecast if they lead to contrast effects, whereby people recognize a discrepancy and contrast the experience to the expectation (as in the Geers & Lassiter, 1999 study). For example, people who expect to like a film but recognize that it is disappointingly dull will like it less than people who had no expectations. It should be noted, however, that studies on affective expectations differ in important ways from most studies of affective forecasting, chiefly in the extent to which people’s expectations are accessible at the time of the

364

WILSON AND GILBERT

affective experience. Many forecasting studies use between-subjects designs, whereby some people predict how they will feel in a specific situation and these predictions are compared to the actual reactions of experiencers who were never asked to make forecasts. Studies on the effects of affective expectations, in contrast, deliberately make people’s affective expectations salient at the time they experience the stimulus. In the Wilson et al. (1989) and Geers and Lassiter (1999) studies, for example, people were given information about how much other people liked cartoons or a film right before seeing the cartoons or the film. The kinds of assimilation and contrast effects found in the literature on affective expectations (e.g., Wilson & Klaaren, 1992) are probably more likely to occur when people’s expectations are highly accessible at the time of the actual experience. Such effects are less likely to occur when people have never explicitly made an affective forecast, as is in studies of affective forecasting that use between-subjects designs, or when people have made forecasts at a time far removed from the actual experience (e.g., Wilson et al., 2000). This prediction is consistent with our informal observation that when within-subject designs are used to study the accuracy of affective forecasts, and people’s forecasts are measured right before they experience the emotional event, their forecasts can contaminate their actual experience. It is not always clear whether this contamination is due to demand characteristics, whereby people feel compelled to report an emotional reaction that is consistent with how they just predicted they would feel, or a genuine affective expectation effect, whereby people assimilate or contrast an experience to their expectation. In any case, it is clear that many of the errors in affective forecasts documented in the literature are unlikely to be due to expectation effects, because these studies often use betweenparticipants designs in which people’s expectations about how they would feel are not highly accessible at the time of the actual emotional experience.

F. UNIQUE INFLUENCES ON ACTUAL EMOTIONAL EXPERIENCE I: HOT/COLD INTRAPERSONAL EMPATHY GAPS Just as there are unique influences on people’s assessments of their emotional reactions to an event when making an affective forecast, so are there unique influences on their actual emotional experiences—unique in the sense that they influence people’s emotions but not their forecasts, leading to discrepancies between the two. Earlier, we saw cases in which people failed to recognize influences on their emotional or motivational states at

AFFECTIVE FORECASTING

365

the time of the forecast (e.g., anger, hunger) that would not be in force in the future. There are also cases in which people fail to anticipate influences on their emotional or motivational states in the future that are not impinging on them when they make their forecasts. Just as people who shop when hungry purchase too many junk foods, people who shop when they are full may purchase too few, underestimating how much they will want a bag of corn chips when watching television late the next night. In Loewenstein, Prelec, and Shatto’s (1998) terms, people in temporary ‘‘hot’’ emotional states have difficulty anticipating what they will want later, when in a ‘‘colder’’ state, and people in ‘‘cold’’ emotional states have difficulty anticipating what they will want later when in a ‘‘hot’’ state. In a clever demonstration of the latter phenomenon, visitors to a museum were given an 11-item trivia quiz and asked to choose as compensation a candy bar or the answers to the questions. Most participants who made their choice before taking the test, when they were in a relatively ‘‘cold’’ state (i.e., their curiosity about the answers to the questions was not yet piqued), chose the candy bar (79%). Most participants who made their choice after taking the test, when they were in a relatively ‘‘hot’’ state (i.e., their curiosity was piqued), preferred the answers (only 40% chose the candy bar). A third group of participants was asked, before taking the test, to predict which compensation they would want after taking it. Consistent with the idea that people in ‘‘cold’’ states have difficulty imagining what they will prefer in ‘‘hot’’ states, 62% said they would prefer the candy bar (see also Loewenstein & Adler, 1995). Loewenstein (1996, 2001) notes that the failure for people in ‘‘cold’’ states to anticipate what it will be like to be in a ‘‘hot’’ state explains a wide array of important phenomena, including drug addiction (e.g., addicts’ underestimation of the cravings they will experience when drug deprived), pain management (e.g., women who decide in advance of child birth not to use anesthesia, only to reverse that decision while giving birth), and risky sexual behavior (e.g., people at risk for AIDS who assume in advance that they will use condoms, failing to take into account the ‘‘heat of the moment,’’ especially when they are intoxicated; MacDonald, MacDonald, Zanna, & Fong, 2000). People who understood better how they are likely to feel and behave when in ‘‘hot’’ states would be in a better position to anticipate and prevent risky behaviors. 1. Types of Forecasting Errors Caused by Hot/cold Intrapersonal Empathy Gaps Hot/cold empathy gaps can lead to either an underprediction or overprediction of the intensity of affective states, depending on the direction

366

WILSON AND GILBERT

of the gap. If people in hot states are predicting how they will feel in cold states, they will overestimate intensity (e.g., people who are hungry overestimate how hungry they will be the next day). If people in cold states are predicting how they will feel in hot states, they will underestimate intensity, as in the Loewenstein et al. (1998) studies just reviewed.

G. UNIQUE INFLUENCES ON ACTUAL EMOTIONAL EXPERIENCE II: FOCALISM In addition to failing to anticipate unique influences on their emotional reactions to an event, people often fail to anticipate the extent to which unrelated events will influence their thoughts and emotions, a tendency that Wilson et al. (2000) termed focalism (Schkade & Kahneman, 1998, independently used the term focusing illusion). As noted by Wladyslaw Tatarkiewicz (1962/1976), ‘‘The pleasures and pains, joys and sufferings, which people actually experience, often fall short of what they had anticipated . . . . In anticipating a coming event we have it alone in mind, and make no provision for other occurrences’’ (p. 111). When considering how their emotional lives will be influenced by a future occurrence, such as the outcome of an election or sporting event, people tend to think of their lives in a vacuum, focusing on that occurrence alone (‘‘I’ll be thrilled for days if Finkleberry wins the election’’). Events do not occur in a vacuum of course, but in the rich context of many other events in people’s lives. By neglecting to consider how much these other events will capture their attention and influence their emotions, people overestimate the impact of the focal event. Wilson et al. (2000) demonstrated focalism and a corrective for it. In one study college football fans predicted, 2 months before an upcoming football game, how happy they would be if their team won. They predicted what their overall level of happiness would be immediately after the game and on each of the succeeding 3 days. The day after the game (which their team did, in fact, win), participants rated their overall happiness. As seen in the control condition in Fig. 3, participants showed a strong impact bias. People predicted that they would be above their baseline level of happiness right after the game and on each of the next 3 days; in fact, their actual level of happiness was no different from their baseline by the day after the game. Before making their predictions people in a second condition were asked, ostensibly as part of another study, to imagine a specific day in the future— which happened to be the Monday after the upcoming football game. They estimated the number of hours they would spend on 10 everyday activities, such as going to class, socializing with friends, studying, and eating. In

AFFECTIVE FORECASTING

367

3

Predicted-Baseline Happiness

Control, Predicted 2

1

Diary, Predicted

0 Control, Actual Diary, Actual −1 Saturday Sunday Monday Tuesday (Game Day)

Fig. 3. Predicted versus actual happiness after a victory by one’s college football team. Adapted from Wilson et al. (2000).

addition they were given a sheet of paper with 24 blanks corresponding to the hours of that same day, and asked to write in what they thought they would be doing each hour. Wilson et al. hypothesized that when people rated how happy they would be after the football game, the ‘‘prospective diary’’ they had filled out would remind them that the game would not occur in a vacuum, and that their lives would be full of other events that would occupy their thoughts and influence their feelings. Consistent with this prediction, people in the diary condition predicted that the game would impact their happiness significantly less than did people in the control condition (see Fig. 3). There was evidence that the effects of the diary manipulation on predicted happiness were mediated by how much people believed they would think about the focal events. In the absence of the diary manipulation, people believed that the focal event (e.g., the football game) would dominate their thoughts on subsequent days and thus have a large impact on their happiness. People in the diary condition, who first imagined how full a typical day is with other activities, believed that they would not think about the focal event as much and thus predicted (correctly) that the event would not influence their happiness for very long. Mediation analyses revealed that the effects of the diary manipulation on affective forecasts was mediated at least in part by how much people believed they would think about the focal event in the future.

368

WILSON AND GILBERT

The focalism hypothesis is relevant to a possible artifactual explanation of the impact bias that concerns the way in which people interpret the forecasting and actual happiness questions. When asked to predict their future happiness, people’s attention is drawn to the focal event (e.g., the football game), which might have led them to interpret the question to mean, ‘‘How happy I will be in the future when I am thinking about the game?’’ Later, people are often asked to rate their overall level of happiness without reference to the focal event. Thus, they might have reported something different (their general level of happiness) than they predicted (their happiness when thinking about the game). Perhaps the impact bias is due to the fact that people interpret the forecasting and actual happiness questions to mean different things. On the one hand, this explanation is not an artifact but the point of the focalism hypothesis: When people make affective forecasts they exaggerate how much the event will be focal in their thoughts and thus overestimate how much it will influence their happiness. On the other hand, the explanation would be an artifact if people know that they will not think about the event very often and know that their overall level of happiness will not be affected for very long, but simply misinterpret the questions they are asked. There is a considerable amount of evidence, however, that rules out this artifactual explanation, discussed in detail by Wilson et al. (2000). In brief, the impact bias has been found even when people are reminded of the focal event at the time they rate their actual happiness. It has also been found when the people are asked to predict how they will feel 10 minutes in the future and then report how they feel 10 minutes later (e.g., Study 6 by Gilbert et al., 1998, which we will discuss later). Given the short interval of time, it is reasonable to assume that people were thinking about the focal event both at the time they made their predictions and the time at which they rated their actual affect. Finally, Wilson et al. (2000) conducted a study in which people predicted how often they would be in good and bad moods after their favorite team won a football game. If people misinterpret forecasting questions to mean, ‘‘how happy will I be when I’m thinking about the event,’’ and people believe that future events will therefore influence them only sporadically, then they might do better at predicting the frequency of their good and bad moods after the event occurs. In fact, people predicted that they would be in good moods significantly more often than they in fact were and that they would be in bad moods significantly less often than they in fact were. The impact bias appears not to be due to an artifact of the way in which prediction questions and actual happiness questions are asked.

AFFECTIVE FORECASTING

369

1. Types of Forecasting Errors Caused by Focalism Focalism is the first mechanism we have discussed that produces forecasting errors in only one direction, namely the impact bias, whereby people overestimate the enduring impact of a future event on their emotional reactions. By underestimating the impact of other events on their thoughts and feelings, people by definition overestimate the impact of the focal event. This error can occur in predictions about the initial intensity of an affective experience as well as its duration, to the extent that people neglect to take into account the extent to which other events will moderate intensity and duration (see Fig. 2).

H. SENSE MAKING PROCESSES The last (and arguably most important) source of affective forecasting errors involves psychological processes that temper people’s emotional reactions. Major life events can surely affect our emotional lives for a very long time. Catastrophic events such as the death of a loved one, divorce, or sexual assault can have enduring effects on our emotions, as can positive events such as finding Mr. or Ms. Right, the birth of a child, or winning huge sums of money. Invariably, however, people’s emotional reactions to life events become less intense with time, a phenomenon we have called emotional evanescence (Wilson, Gilbert, & Centerbar, in press). As noted by Adam Smith (1759/1853), ‘‘The mind of every man, in a longer or shorter time, returns to its natural and usual state of tranquillity. In prosperity, after a certain time, it falls back to that state; in adversity, after a certain time, it rises up to it’’ (p. 149). A major source of the impact bias, we suggest, is that people fail to anticipate the extent to which they will transform events psychologically in ways that ameliorate their impact. Although everyone knows that pains and pleasures recede with time, we believe that the reasons for emotional evanescence have been underappreciated by psychologists (in their theories of emotion) and lay people (in their understanding of their own emotional lives). Once people experience an emotional reaction to a life event, what causes the emotion to lessen in intensity? Emotions are not like radioactive isotopes with short half-lives that naturally decay over time. There must be psychological processes that are responsible for the decay of emotional reactions. Wilson et al. (in press) reviewed previous explanations of emotional evanescence, such as adaptation-level theories. These theories argue that people constantly compare their experiences to similar events in the past, and that their emotional reactions depend on how the current experience

370

WILSON AND GILBERT

compares to this comparison level (e.g., Brickman & Campbell, 1971; Helson, 1964; Parducci, 1995; Tversky & Kahneman, 1979). When a person travels from Minneapolis to Hawaii in December, for example, a partly cloudy, 70-degree day in Honolulu is likely to cause a good deal of pleasure, because the person is comparing it to the 20-degree, snowy day that was left behind in Minnesota. But over the course of the visit 80-degree sunny days become the standard of comparison, such that a partly cloudy 70-degree day triggers disappointment. As people become used to pleasurable events, they become the standard of comparison and it takes an even more pleasurable event to make them happy. The same principle applies to negative events. A person who travels from Hawaii to Minneapolis might find a 50-degree March day to be intolerable, because it is being compared to a beautiful, Hawaiian, 80-degree, sun-filled days. As time goes by, however, the average weather in Minneapolis becomes the comparison level, such that a 50-degree March day seems quite lovely (Schkade & Kahneman, 1998). Changes in adaptation level might explain why emotional reactions to external events are short lived. Like sunny Hawaiian days to a Minnesotan, novel events change people’s comparison level, dampening their reactions to similar events in the future. One problem with adaptation level theories, however, is that they do not specify the comparison point that people will use at any given point in time. How do people categorize an event in a way that defines the appropriate comparison level? When visiting Hawaii, do people compare the weather to the previous day in Minneapolis, to a running average of the weather they experienced the past two weeks, or to their vacation in Hawaii last year? People have to categorize the event (e.g., ‘‘weather in Hawaii,’’ ‘‘weather during the past 2 weeks’’), and then decide which events in that category are the most appropriate standard of comparison (e.g., the average weather they have experienced during all Hawaiian vacations, the best day they ever had in Hawaii, the worst day they ever had in Hawaii). The process by which people compare new experiences to old ones is undoubtedly complex, determined by such things as the accessibility in memory of previous experiences and the similarity of the present complex to previous ones (see Eiser, 1990, and Frederick & Loewenstein, 1999, for a discussion of these issues). Further, adaptation level theories cannot explain why people’s reactions to a single, newly experienced event taper off, because they focus on people’s reactions over time to repeated events. For example, the theories can explain why doing well in college courses will change students’ adaptation level, such that they enjoy their tenth ‘‘A’’ less than the first. The theories cannot explain why people’s enjoyment of the first ‘‘A’’ wears off relatively quickly, because the comparison level remains the same for that experience.

AFFECTIVE FORECASTING

371

1. The Emotional Implications of Human Sense Making Our basic argument is that people make sense of their worlds in a way that speeds recovery from emotional events, and that this sense-making process is largely automatic and nonconscious. Humans inexorably explain and understand events that were initially surprising and unpredictable, and this process lowers the intensity of emotional reactions to the events. Humans beings are adept at orienting to important, novel events in their environment and then transforming these events psychologically in order to understand them better. As noted by Roese and Olson (1996), ‘‘People perceive the occurrence of an outcome and are compelled to make sense of it’’ (p. 297). Although such sense-making processes are well known, we believe their ramifications for people’s emotional experiences have not been fully appreciated. Sense-making processes and emotional evanescence can be broken down into four basic steps. First, people orient to unexpected but relevant information in their environment. Such an orienting response has been found in many species (e.g., Anderson, 1994; Cheal, Johnson, Ellingboe, & Skupny, 1984; Hilgetag, Lomber, & Payne, 2001; Vinogradova, 2001; Wechsler, 1992) and in human infants in the first few months of life (e.g., Quinn, Eimas, & Tarr, 2001; Fantz, 1964). It has been found on a wide variety of cognitive tasks and with various measures of brain activity (e.g., Donchin, 1981; Enns, Austen, Di Lollo, Rauschenberger, & Yantis, 2001; Johnston & Schwarting, 1997; Hamann, Ely, Hoffman, & Kilts, 2002; Daffner et al., 2000; Kimmel, Van Olst, & Orlebeke, 1979; Spencer, Dien, & Donchin, 2001). It is not novelty per se that attracts human attention and information processing. If it were, people could not walk down the street without stopping every second to examine parts of their visual fields that they had never seen before, such as the face of every stranger, novel patterns of clouds, and the fact that a gopher burrowed a hole in a new part of the adjacent park. Instead, people engage in selective attention, in which they screen out information that is irrelevant to their processing goals, unless that information is so ‘‘newsworthy’’ (i.e., important to some other goal) that it suddenly attracts attention. The park groundskeeper would be much more likely to notice the new gopher hole, even if he or she had been thinking about which trees needed pruning. Second, people have more intense emotional reactions to unexpected, relevant information than to other events. The groundskeeper will react more strongly, for example, to a gopher hole that is unexpected and suddenly pops into attention than to one that was there yesterday. The more discrepant new information is from people’s existing knowledge structures, the more intense their emotional reactions to it. Sometimes events occur that

372

WILSON AND GILBERT

are not relevant to our current goals or focus of attention; they appear out of the blue. In Ortony, Clore, and Collins’ (1988) terms, these events are characterized by ‘‘unexpectedness,’’ and ‘‘unexpectedness is positively correlated with the intensity of emotion’’ (Ortony et al., 1988, p. 64). Other times an event is not completely unexpected but is of low perceived probability; Ortony et al. (1988) refer to such occurrences as violations of perceived likelihood. Surprising occurrences such as these also produce intense emotional reactions. Mellers and McGraw (2001), for example, found that people who won $8 on a gambling task were happier when the odds of winning were 20% than when the odds were 80%. Similarly, people who lost $8 were unhappier when the odds of losing were 20% versus 80% (see also Coughlan & Connolly, 2001). Third, once an unexpected event occurs and people have a relatively intense emotional reaction, they attempt to make sense of the event, quickly and automatically. Unlike almost all other species, humans possess the ability to perform sophisticated cognitive operations on their representation of the stimulus; indeed, Mandler (1975) equates attention with an analysis of the meaning of a stimulus or event; the two are indistinguishable: ‘‘Any input to the cognitive-interpretive system is subjected to an analysis of its relation to existing structures’’ (p. 26). As Gilovich (1991) noted, ‘‘We are predisposed to see order, pattern, and meaning in the world, and we find randomness, chaos, and meaninglessness unsatisfying. Human nature abhors a lack of predictability and the absence of meaning’’ (p. 9). The way in which humans transform the world into a predictable place is well documented. Piaget (1952; Piaget & Inhelder, 1969) discussed how children assimilate new events to existing knowledge structures, or, if that is not possible, alter their knowledge structures to accommodate the new information. People are also skilled explainers of their social worlds, making quick attributions about the causes of their own and other people’s behavior (Gilbert, 1991; Heider, 1958; Jones & Davis, 1965; Kelley, 1967). As Heider (1958) noted, a person makes causal attributions ‘‘not only because of intellectual curiosity, but also because such attribution allows him to understand his world, to predict and control events involving himself and others’’ (p. 146). Holyoak and Simon found that when making decisions, such as about a legal case, people inexorably show coherence shifts, whereby complex, contradictory, information is transformed into an internally consistent, coherent viewpoint (Holyoak & Simon, 1999; Simon, Pham, Le, & Holyoak, 2001). If people feel that they cannot control, predict, or understand their environments, they are at risk for severe motivational and cognitive deficits, such as depression (Abramson, Seligman, & Teasdale, 1978; Langer & Rodin, 1976; Pittman, 1998; Schulz, 1976; Seligman, 1975; Taylor & Brown, 1988; Thompson, Armstrong, & Thomas, 1998).

AFFECTIVE FORECASTING

373

Fourth, when people make sense of an event it no longer seems surprising or unexpected, and as a result they think about it less and it produces a less intense emotional reaction. The process of sense making ‘‘ordinizes’’ events in a way that robs them of their emotional power (Wilson et al., in press). Emotional evanescence may also be adaptive in and of itself, independent of the human proclivity to make sense of the environment (Wilson et al., in press). A number of theorists have argued that a basic function of an emotion is to signal people about dangers and opportunities in their environments (e.g., Damasio, 1994; LeDoux, 1996). Rather than having to stop and analyze every situation consciously and deliberately, people have quick emotional reactions that tell them whether to approach or avoid a stimulus. To serve this signaling function, it is important that emotional reactions to previous events not last too long; for emotional signals to get through, people’s systems must not be swamped by responses to past events. If we are still euphoric or depressed over something that happened yesterday, we will be less sensitive to new dangers or opportunities that are occurring today. One reason for this is that intense emotional reactions interfere with higher order cognitive processing, making it difficult to think clearly. Although moderately positive states enhance creative problem solving (Fredrickson, 1998; Isen, 1993), extreme states—in either a positive or negative direction— are likely to reduce attention and impede processing (Easterbrook, 1959). Another reason that it is to people’s advantage to recover quickly from emotional reactions is the conservation of energy. To the extent that emotions are accompanied by physiological arousal, our bodies can maintain high levels of arousal for only so long. Imagine, for example, how it feels to experience a great surge of joy and excitement, such as the day you were married, the day your child was born, or when you learned that you had won a coveted prize. Now imagine that you felt that way for a week. As wonderful as such experiences are, it would be exhausting to maintain them for very long. It would be dangerous, or even fatal, for our heart rate and blood pressure to remain elevated for a prolonged period of time. To protect our health, there must be mechanisms that counteract perturbations to our emotional system. Wilson et al. (in press) and Wilson (2002) discussed different mechanisms that foster emotional evanescence, such as those posited by opponent process theory (Solomon, 1980). Perhaps the most powerful way, we suggest, is by invoking the kinds of sense-making processes we have already discussed. The human sense maker may have evolved in part because it serves the important, adaptive function of emotional evanescence. Sense making can take several forms, with the common feature that it reduces discrepancies between people’s schemas and unexpected, relevant

374

WILSON AND GILBERT

events. As mentioned it can involve assimilation, accommodation, and causal attribution. Another example is the hindsight bias, whereby people transform an event psychologically after it occurs to make it seem more predictable than it really was. Historical upheavals, improbable outcomes of sporting events, and sudden break ups of relationships all seem like things that we should have anticipated, in retrospect (Carli, 1999; Fischhoff, 1975; Hawkins & Hastie, 1990; Roese & Olson, 1996; Wasserman, Lempert, & Hastie, 1991). Although the hindsight bias is well-known, its consequences for emotional evanescence have been not been examined systematically. As we have seen, events that are unexpected have more emotional impact than ones that are easy to explain and understand. Explaining events in a way that make them seem predictable, then, should lower the intensity of people’s emotional reactions to them. Further, this process occurs automatically and nonconsciously (e.g., Pohl & Hell, 1996). Indeed, if people were fully aware of their post hoc sense making, they would not commit the hindsight bias. In fact, most sense-making processes, such as causal attributions, whereby people strive to understand and explain each others’ behavior, require little or no mental effort, are unintentional, and occur outside of awareness (Gilbert, 1998; Gilbert, Pelham, & Krull, 1988). This fact is crucial to our understanding of forecasting errors. 2. Failures to Anticipate Sense-Making Processes: Ordinization Neglect in Reaction to Positive Events A major source of the impact bias, we suggest, is people’s failure to anticipate how much they will transform events psychologically in ways that reduce their emotional power. In other words, people do not appreciate the extent to which they will ‘‘ordinize’’ an event by engaging in the kinds of sense-making processes we have discussed, leading to ordinization neglect. ‘‘I will be thrilled if I receive tenure,’’ an assistant professor thinks, ‘‘and I’ll be happy for years to come.’’ When assistant professors first learn that their tenure has been approved they probably are thrilled, particularly if their case was not a sure thing. Inevitably they begin to make sense of their achievement, however, as they assimilate it into their knowledge structures or create new schemas. They find themselves thinking less and less about their tenure as time goes by, and sooner rather than later, it recedes into the background of their busy lives. The knowledge that ‘‘I am an associate professor with tenure’’ becomes ordinary knowledge with little ‘‘zing’’ to it. Consequently, the tenure decision does not lead to the lasting happiness that people anticipated. As noted by Nathaniel Hawthorne in his short story Rappaccini’s Daughter, ‘‘How often is it the case, that, when impossibilities

AFFECTIVE FORECASTING

375

have come to pass, and dreams have condensed their misty substance into tangible realities, we find ourselves calm. . .amid circumstances which it would have been a delirium of joy to anticipate!’’ (1846/1937, p. 1055). Consistent with this reasoning, Gilbert et al. (1998) found that a positive tenure decision did not cause the lasting happiness that untenured professors anticipated. Assistant professors at a large university predicted what their general level of happiness would be in the 5 years after receiving tenure, and these forecasts were compared to the actual level of happiness of professors who had been granted tenure at that university in the previous 5 years. The assistant professors predicted that they would be significantly happier than were the professors who had successfully achieved tenure, suggesting that they were overestimating the impact that a positive decision would have. Other studies have examined more directly whether impact biases to positive events are due to ordinization neglect. The strategy in these studies was to manipulate the ease with which people could ordinize a positive emotional event by making sense of it, with the prediction that people’s emotional reactions would last the longest in the conditions in which sense making was most difficult. People who forecasted their emotional reactions were expected to be insensitive to this manipulation of sense making. That is, they were expected to fail to realize that their emotional reactions would fade the quickest in the conditions in which sense making was the easiest, due to ordinization neglect. Centerbar, Wilson, and Gilbert (2002), for example, gave students positive social feedback, thereby improving their mood, and then manipulated how easily the students could make sense of the feedback. When a student arrived the experimenter took his or her picture, scanned it into a computer, and entered it into an instant messaging program to be sent to five students at other universities, ostensibly as part of a study of impression formation over the internet. The participant then saw the pictures of the other students appear on the screen and exchanged information with these students about their values, interests, and backgrounds. There was in fact only one real participant in each session; the information about the other five (two of the same gender as the participant, three of the opposite gender) was preprogrammed into the computer. After people exchanged information, they asked the students to choose the one opposite sex member of the group who they thought would make their ‘‘best potential friend’’ and to write a paragraph explaining why. Participants were told that their choice and explanations would be sent to the other group members, and that they would learn who the opposite sex group members chose and read their explanations. After

376

WILSON AND GILBERT

sending their choice and explanation, participants waited while the other students’ choices and explanations appeared on the screen one at a time, ostensibly as they were sent by the students. In the experiencer conditions participants learned that all three of the opposite sex students chose them as their best potential friend. The paragraphs explaining the reasons for these choices were uniformly positive, though distinct in their details. To manipulate how easily people could make sense of this positive feedback, experiencers were randomly assigned to one of two conditions. In the revealed condition people were told which of the other students had written each paragraph explaining their choice. When a paragraph appeared on the screen, the picture and name of the author of that paragraph was displayed next to it. In the anonymous condition participants were told that in order to maintain confidentiality, the authors of each paragraph would not be shown. These participants received the same positive feedback (that all three opposite sex students had chosen them) and read the same paragraphs explaining why; the only difference was that they did not know which of the students had written which paragraph. Participants completed mood scales immediately after receiving the feedback, completed a filler task of word search puzzles for 15 minutes, and then completed the mood scales again. Centerbar et al. (2002) predicted that participants in both the revealed and anonymous condition would be very happy right after receiving the positive feedback; the surprisingly positive news should produce a highly favorable response. Over the next several minutes, however, people in the revealed condition should find it easier to make sense of the feedback, given that they knew who said what (‘‘It is no surprise that Sarah liked my values; I remember that she answered the questions similarly to me’’). The specific student that was said to author each paragraph was counterbalanced, but regardless of who said what, participants could probably find reasons why. As a consequence of this sense making, the positive reactions of people in the revealed condition were expected to fade over time, demonstrating emotional evanescence. People in the anonymous condition could not as easily make sense of why each person wrote what they did, given that the authorship of the paragraphs was unknown. Consequently their positive reactions were expected to fade more slowly, showing less emotional evanescence. Participants in a control condition took part in the study up to the point at which they expected to receive the feedback about who the other students chose as their best potential friend, at which point they rated their mood. Forecasters also participated up to the point at which they expected to receive the feedback from the other students. These students were then asked to imagine that they had been chosen by everyone as their best

377

AFFECTIVE FORECASTING

potential friend and to read the paragraphs explaining why (these paragraphs were identical to the ones experiencers received). They predicted what their mood would be right away and 15 minutes later, if they knew who had authored each paragraph and if they did not. Finally, choosers were given the actual feedback and paragraphs to read and were allowed to choose whether they wanted to know which person had authored each paragraph. The computer program said that some previous participants had preferred to know and others had not, and instructed the participant to click on one button if they wanted to learn the author of each paragraph and another if they did not. As predicted, experiencers in both the revealed and anonymous conditions were significantly happier than control participants, right after learning that they had been chosen as everyone’s best potential friend, but did not differ from each other (see means in Table II). That is, the feedback made people in the revealed and anonymous conditions equally happy at first. As predicted, however, experiencers’ positive mood decreased more rapidly in the revealed than the anonymous condition. The drop in positive mood in the revealed condition was significantly larger than the drop in the anonymous condition. Table II also shows forecasters’ predictions about what their mood would be in the different conditions. People were generally accurate about how happy they would be in the revealed condition; their predicted happiness at Times 1 and 2 correspond closely to the experiencers’ moods at these points in time. People were inaccurate about how happy they would be in the anonymous condition. In fact, their predicted mood was lower than people’s predicted mood in the revealed condition, especially at Time 2, which was the opposite of what was found for experiencers. People’s theory seems to have been that not knowing who authored which paragraph would spoil their mood when in fact it enhanced it (at Time 2). Consistent with this TABLE II Mean Positive Affect in Internet Impression Formation Studya Experiencers

Forecasters

Time

Control

Revealed

Anonymous

Revealed

Anonymous

Time 1 Time 2

6.73 (.74)

7.76 (.86) 6.39 (1.29)

7.98 (.79) 7.06 (1.44)

7.63 (1.12) 6.56 (1.13)

7.29 (1.00) 5.91 (.99)

a

Means are the average of several mood scales rated on nine-point scales, with higher numbers reflecting a more positive mood. Standard deviations are in parentheses. From Centerbar, Wilson, and Gilbert (2002).

378

WILSON AND GILBERT

interpretation, 100% of the people allowed to choose whether to find out who authored which paragraph elected to do so; that is, they placed themselves in the condition that forecasters believed would make them happy the longest, but which in fact made experiencers less happy. Our preferred explanation of these results is that experiencers in the revealed condition were better able to make sense of (‘‘ordinize’’) the positive feedback they received, and by so doing reduced its emotional impact. The feedback stayed alive longer in the anonymous condition precisely because participants could not as easily make sense of it and stop thinking about it. An alternative explanation is that at least some people in the revealed condition were disappointed by knowing who had authored which paragraph, to the extent that they liked one person more than the others, preferred one of the paragraphs to the others, and were disappointed that their favorite person did not give them their favorite feedback. To avoid this possibility we carefully pretested the descriptions of the three opposite sex students and the paragraphs, such that most people did not strongly prefer one of the students or paragraphs. Further, when asked to rate how pleased they were with the feedback they received from the three students, people in the revealed condition were just as pleased overall as people in the anonymous condition. Contrary to the disappointment hypothesis, people in the revealed condition were significantly more pleased with the feedback they received from their favorite person (the one they had selected as their best potential friend) than were people in the anonymous condition. Nonetheless, it is possible that the revealed and anonymous conditions differed in some way other than the ease of making sense of the feedback. Wilson, Kermer, and Gilbert (2002) performed another study that attempted to make a positive event as similar as possible except that it was easier to explain under some conditions. Students who were studying alone in a university library were approached by a research assistant who handed them a card with a dollar coin attached to it (a United States Sacagawea dollar), said ‘‘Hi, this is for you, have a nice day,’’ and walked away. There was written information on the cards that varied slightly across conditions. In the uncertain condition three facts were listed, namely ‘‘The Smile Society,’’ ‘‘A Student/Community Secular Alliance,’’ and ‘‘We Like to Promote Random Acts of Kindness.’’ These facts were expected to be somewhat puzzling and difficult to make sense of. In the certain condition the information on the cards was identical, except that two question were added to which the facts provided an answer. At the top of the card, in a different color ink, was the question, ‘‘Who Are We?’’ followed by ‘‘The Smile Society’’ and ‘‘A Student/Community Secular Alliance.’’ Following this, also in a different colored ink, was the question, ‘‘Why are We Doing

379

AFFECTIVE FORECASTING TABLE III Mean Positive Affect in Library Coin Studya Experiencers

Forecasters

Control

certain

uncertain

certain

uncertain

5.1 (1.7)

4.9 (0.9)

6.7 (1.2)

7.3 (0.9)

6.8 (0.9)

a Means are the average of mood ratings on 9-point scales; higher numbers indicate more positive moods. Standard deviations are in parentheses. From Wilson, Kermer, and Gilbert (2002).

This?’’ followed by ‘‘We Like to Promote Random Acts of Kindness.’’ Thus, the information (the answers) was the same in both conditions, but the question-and-answer format in the certain question was expected to be easier to make sense of. Rather than puzzling over what the information meant, people might view it as answers to reasonable-sounding questions, and thereby find it easier to continue with what they were doing. To see if people in the two conditions had different affective reactions to the cards, a different research assistant approached people 5 minutes later and asked if they would complete a brief survey for their psychology class. The survey contained a few filler questions about their study habits followed by standard mood scales. We also included a control condition in which people received the survey without having first been approached and given a dollar coin. As seen in Table III, the results were largely as predicted. Five minutes after receiving the coin, participants in the certain condition were no happier than control participants who had not been given a dollar. We assume that they were happy to receive the dollar at first but that this happiness had faded over the next 5 minutes (though of course we do not know for sure what their immediate affective reaction was because we were unable to measure it). As predicted, people in the uncertain condition were in a significantly better mood at the 5 minute point than people in either the control or certain condition. Table III also shows the predictions made by forecasters who were approached in the library, asked to imagine that they were given one of the cards with the dollar on it, and predict what their mood would be 5 minutes later. Forecasters in the uncertain condition made quite accurate predictions about how happy they would be. Forecasters made inaccurate predictions in the certain condition, however (the one with the question-and-answer format on the card). They predicted that they would be in a better mood than did forecasters in the uncertain condition, when in fact experiencers in this condition were in a significantly less positive mood. Like the

380

WILSON AND GILBERT

forecasters in the Centerbar et al. (2002) internet impression formation study, people seem to base their forecasts on the theory that certainty (in this case, the ease of making sense of the questions and answers) would be more pleasing and mood enhancing, when in fact it was less pleasing and mood enhancing. 3. Types of Forecasting Errors Caused by Ordinization Neglect Like the focalism bias, ordinization neglect produces errors in one direction only. Because people underestimate the speed with which they make sense of novel events they overestimate the duration of their emotional reactions, leading to the impact bias.

I. THE PSYCHOLOGICAL IMMUNE SYSTEM: MAKING SENSE OF NEGATIVE EVENTS So far, we have explored ordinization neglect in response to positive events, such as receiving positive social feedback or an unexpected monetary gift. Does ordinization neglect occur when people predict how they will respond to negative events? People possess powerful psychological defenses that serve to ameliorate the impact of negative information. These defenses are so pervasive and effective that they can be thought of as a psychological immune system that detects and neutralizes events that challenge people’s sense of well-being (Gilbert et al., 1998). Social and clinical psychology have documented many such psychological defenses, including psychoanalytic defense mechanisms, dissonance reduction, self-affirmation, motivated reasoning, self-deception, positive illusions, and terror management (e.g., Aronson, 1968; Dunning, 1999; Festinger, 1977; Folkman, 1984; S. Freud, 1924/1968; A. Freud, 1966; Greenwald, 1980; Kunda, 1990; Pyszczynski, Greenberg, & Solomon, 1997; Steele, 1988; Taylor, 1989; Tesser, 2000; Vaillant, 2000). There is, of course, a great deal of human suffering in the world, and the psychological immune system can only do so much to ameliorate this suffering. Our psychological pain and suffering would be a lot worse, however, if we did not possess potent psychological defenses that hasten our recovery from them. The psychological immune system can be thought of as a special case of the kind of human sense making we have already discussed. When any novel important event occurs, cognitive processes are triggered to make sense of it. If that event is negative and challenges people’s sense of well-being, the psychological immune system turbo charges the sense-making process, giving it extra force and direction. People are motivated to make sense of

AFFECTIVE FORECASTING

381

any novel event, but are especially motivated to interpret negative events in ways that minimize their impact. A feature that the psychological immune system shares with other sensemaking processes is that it occurs largely outside of awareness. In fact, as noted by Gilbert et al. (1998), psychological defenses are more effective by operating behind the mental scenes. If people recognized the extent to which they transformed events psychologically in order to make themselves feel better, these transformations would not be nearly as compelling. When Bob learns that Sarah has left him, for example, he gradually reduces the pain by emphasizing Sarah’s flaws and deciding that she was not right for him after all. His rationalization works best if Bob does not recognize it as such. It is difficult to accomplish such a rationalization deliberately and consciously; it would not be very effective for Bob to say, ‘‘As of 2 pm today I will stop loving Sarah by focusing on the fact that she always leaves empty water bottles and gum wrappers on the floor of my car.’’ Rather, his psychological defenses are mobilized automatically and nonconsciously. Sarah begins to appears objectively different to Bob, and he does not realize that it was not she who changed but his construals of her.

1. Failures to Anticipate Sense-Making Processes: Immune Neglect in Reaction to Negative Events A major source of the impact bias in response to negative events is people’s failure to anticipate how much their psychological immune systems will hasten their recovery, a phenomenon that Gilbert et al. (1998) termed immune neglect (which, as noted above, can be viewed as a special case of ordinization neglect). Before Sarah broke up with him, Bob would likely have predicted that it would take him months, if not years, to recover. He would be underestimating the extent to which a break-up would trigger defensive psychological processes that would speed his recovery. Just as people fail to appreciate the extent to which they will ‘‘ordinize’’ positive events by making sense of them, they fail to appreciate the extent to which they will ‘‘defang’’ negative events by rationalizing, reconstruing, or minimizing them. Gilbert et al. (1998) found support for this hypothesis in several studies. The logic of these studies was to manipulate the ease with which people could rationalize a negative emotional event, with the prediction that they would recover the quickest in the conditions in which rationalization was easiest. People who forecasted their emotional reactions were expected to be insensitive to this manipulation of rationalization. That is, they were expected to fail to realize that they would recover most quickly from the

382

WILSON AND GILBERT TABLE IV Mean Happiness in Job Interview Studya Experiencers

Forecasters

Time

Unfair (n = 20)

Fair (n = 19)

Unfair (n = 20)

Fair (n = 19)

Time 1 Time 2


.40 (1.19) .00 (1.12)


.68 (1.34)
1.26 (1.97)


2.10 (1.68)
1.90 (2.02)


2.11 (1.94)
2.00 (1.45)

a

Means are changes from a baseline measure of happiness. Standard deviations are in parentheses. From Gilbert et al. (1998, Study 6).

negative events when rationalization was the easiest, due to immune neglect. In one study, for example, college students interviewed for a desirable job in which they would be paid to sample consumer products (Gilbert et al., 1998). Participants in the unfair decision condition were interviewed by a lone business student who asked questions that were only marginally related to the job, such as ‘‘Why did you pick your major?’’ Participants in the fair decision condition were interviewed by a panel of three business students who asked highly relevant questions. Further, participants were told that they would fail to get the job only if the three interviewers unanimously voted against them. All participants then predicted how happy they would be immediately and 10 minutes after learning they had and had not received the job. Next, all participants learned that they did not, in fact, get the job, and rated their actual happiness then and 10 minutes later. Gilbert et al. (1998) hypothesized that people in both conditions would be unhappy when they first learned that they did not get the job but that those in the unfair condition would recover more quickly, by finding it easier to rationalize this negative outcome. That is, people in the unfair condition could rationalize the impact of the decision by blaming it on the capricious interviewer, whereas people in the fair condition could not rationalize the outcome as easily, given the unanimity of the interviewers and the fairness of their questions. Consequently, the negative impact of the event should last longer in the fair decision condition. This is precisely what happened, as seen in Table IV under ‘‘experiencers.’’ Right after learning that they had not received the job people in both the fair and unfair condition were less happy than they had been at the beginning of the study, and did not differ significantly from each other (the cell entries are their happiness ratings minus this baseline measure). After 10 minutes people in the unfair condition were back to their baseline

AFFECTIVE FORECASTING

383

level of happiness, whereas people in the fair condition had become even more unhappy; the two groups differed significantly at this point in time. Consistent with the immune neglect hypothesis, people did not anticipate that they would react differently in the fair and unfair conditions. As seen on the right side of Table IV, forecasters predicted that they would be unhappy if they did not get the job to an equal degree in both conditions, and that they would still be unhappy 10 minutes later. These results are consistent with the hypothesis that people in the unfair condition failed to anticipate the ease with which they would rationalize a failure to get the job by blaming the capricious interviewer. Gilbert et al. (1998) found similar results in several other studies. In one, for example, people predicted that they would be equally unhappy if they received negative personality feedback from two clinical psychologists who had examined their test results or from a computer program that had analyzed their test results. In fact, people who received negative feedback from the computer were not as unhappy as people who received it from the clinicians, presumably because they found it easier to rationalize by questioning the validity of the feedback (‘‘what does a bunch of circuit boards and computer chips know anyway?’’).

J. SUMMARY We documented a number of sources of error on affective forecasts, including misconstruing the nature of the future event, errors in recall of past emotional experiences, faulty affective theories, failures to correct for unique influences on forecasts, and framing. Additional sources of error stem from people’s failure to take into account, when making affective forecasts, factors that will influence their later emotions. A failure to anticipate expectation effects can increase or decrease the accuracy of people’s forecasts, depending on whether people’s expectations lead to assimilation (people change their actual reactions to conform to their forecasts) or contrast (people’s expectations lead to an even larger discrepancy between the forecast and the emotional experience). Hot/cold intrapersonal empathy gaps can lead to under- or overprediction, depending on the direction of the gap. People forecasting while in hot states (e.g., grocery shopping while hungry) overestimate how they will feel in the future when in a cold state (e.g., how much they will want to eat corn chips after a large meal later in the week). People forecasting while in cold states (e.g., grocery shopping when full) tend to underestimate how they will feel when in hot states in the future (e.g., how much they will want to eat corn chips while watching television late at night). Although each of these factors can

384

WILSON AND GILBERT

produce errors in forecasting, none leads systematically to an overestimation of the impact of future events on one’s affective reactions. Focalism and the underestimation of sense-making processes (ordinization neglect and immune neglect) lead uniquely to the impact bias, whereby people overestimate the enduring impact that future events will have on their emotional reactions. Focalism, whereby people fail to anticipate the extent to which unrelated events will influence thoughts and emotions, causes people to overestimate the impact of the focal event, such as the outcome of an upcoming football game. Perhaps the most pervasive influence on affective forecasting is the failure to anticipate one’s own ability to make sense of the world in ways that minimize its emotional impact. We documented ordinization neglect for positive events and immune neglect for negative events, both of which produced the impact bias. The fact that the impact bias is by far the most common error found in affective forecasting research is testimony, perhaps, to the pervasiveness of people’s tendency to fail to anticipate their own sense-making processes.

III. Dynamics of Sense-Making Processes The failure to appreciate the workings of one’s own sense-making processes has several intriguing consequences, including the promotion of beliefs in external agents and several types of failures to maximize one’s own happiness.

A. THE ILLUSION OF EXTERNAL AGENCY Things often work out for the best, or so it seems. It might be difficult to decide between a job in Grand Rapids and one in Plattsburgh, and after moving to Grand Rapids, we might worry that we made the wrong choice. After settling in and making a life for ourselves, however, we find that we are quite happy and thank our lucky stars that we made the right decision. In fact, the more we think about it, the more it seems that most of our difficult decisions turned out for the best, as if there were a ‘‘guiding hand’’ shepherding us in the right direction whenever we had tough choices to make. Although we cannot say definitely whether there is a guiding hand helping people in need, we can say that immune neglect fosters such a belief. In the above example, people may have overlooked the fact that it was their

AFFECTIVE FORECASTING

385

psychological immune systems that created the belief that their chosen lives were superior to the unchosen ones, as has been demonstrated in many studies of postdecisional dissonance (e.g., Brehm, 1956). Because people are largely unaware that their internal dynamics promote such positive change, they look outward for an explanation. Gilbert et al. (2000) demonstrated such an illusion of external agency in a study in which participants were told that they would play a ‘‘self-disclosure game’’ with another student. After performing an unrelated visual detection task, participants read four autobiographical statements of other students and either rated how much they would like each one as a partner (the committed group) or did not rate their preferences (the uncommitted group). The autobiographies were then placed in folders and the participants randomly chose one to determine which person would be their partner. Through experimental sleight-of-hand, all participants chose a partner who was not their first choice, after which they rated how much they liked this person. Gilbert et al. (2000) assumed that it would be easier for uncommitted participants to reduce dissonance by increasing their liking of the chosen partner, because they had not publicly stated earlier that they did not like this person. This was indeed the case; people in the uncommitted condition rated their partner significantly more favorably than people in the committed condition did. How did they explain their lucky choice of such a wonderful partner? At this point the experimenter said that the study was over and explained that it was actually a test of the effectiveness of subliminal messages. The visual detection task they had completed earlier, participants were told, actually included subliminal primes designed to increase the likelihood that they would pick the folder with the most favorable partner. In fact, there were no subliminal messages in the earlier task. As predicted, though, when asked to rate how effective the subliminal primes had been, people in the uncommitted condition believed that they had been influenced more by the primes than people in the committed condition did. Why? By failing to realize that it was their own psychological immune systems that had created the belief that their choice of partner was optimal, they attributed this ‘‘lucky choice’’ to the influence of subliminal messages that were not, in fact, present. According to national surveys, most people believe in a divine being. Eighty-one percent of Americans, for example, reported that they had felt God’s presence (Gallup & Castelli, 1989). Again, we certainly cannot speak to the accuracy of such beliefs. We can suggest, however, that immune neglect might have contributed to them, just as it contributed to a belief in subliminal influence in the Gilbert et al. (2000) study. Gilbert and Rimsky

386

WILSON AND GILBERT

(1999), for example, found a correlation between the strength of people’s psychological immune system and their belief in God. Participants filled out a survey in return for a bag of one of two brands of chocolate chip cookies; to determine which brand they would get, participants picked a folded slip of paper labeled ‘‘Brand A’’ or ‘‘Brand B.’’ They then tasted cookies of each brand and rated how much they liked them. Unbeknownst to participants, the two cookies were identical. The extent to which they said they preferred the brand they had chosen was used as a measure of their propensity to reduce dissonance; that is, the more they exaggerated their liking for their chosen alternative, the more they had engaged in a rationalization process. Gilbert and Rimsky (1999) hypothesized that people who are most prone to reduce dissonance in this manner might be most susceptible to the illusion of external agency. For example, if they consistently find that their difficult choices turn out for the best, they might misattribute such dissonance reduction to a supreme being. Consistent with this hypothesis, the more people exaggerated their liking for their chosen brand of cookie, the more likely they were to report on a subsequent survey that they believed in God, r ¼ .31, p < .05. This is a correlational finding, of course, and it is not certain whether it was people’s propensity to reduce dissonance that caused their belief in a higher power. It is possible, for example, that it was people’s belief in God that made them more likely to reduce dissonance. The results are consistent with the Gilbert et al. (2000) laboratory studies, though, that found experimental evidence for a tendency to misattribute one’s own ability to ‘‘make the best’’ of a choice to an external agent.

B. KEEPING ONE’S OPTIONS OPEN—AT AN AFFECTIVE COST Another consequence of immune neglect is that people might arrange their choices in ways that inadvertently make it difficult for them to reduce dissonance. Imagine that Professor Jones has been offered an attractive job at a university and decides to accept. Her current university offers her the opportunity to take a leave of absence rather than resigning, in case she discovers that she does not like her new position and wants to return. Should she take her university up on this offer and keep her options open? Why wouldn’t people want to give themselves the opportunity to change their minds? What if Professor Jones discovers that she hates living in a small town and pines for the big city life she left behind? People often pay extra for the ability to revoke a decision, such as those who shop for clothes at an expensive boutique with a liberal return policy rather than a discount store in which all sales are final. Indeed, it is often rational to try out

387

AFFECTIVE FORECASTING TABLE V Mean Change in Rank of Rating of Postersa Experiencers

Forecasters

Unchangeable

Changeable

Unchangeable

Changeable

.71


.07


.44


.33

a Means are changes the rank assigned to the poster participants chose to take home. From Gilbert and Ebert (2002, Study 2a).

something before committing oneself to it. Extended courtships are usually a better idea, for example, than marrying someone after the first date. If people have experience with the options and know what they entail, however, it might be best to make an irrevocable choice, given that postdecisional dissonance reduction is most likely to occur after a decision is final (e.g., Frey, 1981; Lowe & Steiner, 1968). The cost to keeping one’s options open is that it prevents people from making the best of their situation psychologically by reducing postdecisional dissonance. Suppose that Professor Jones discovers that she is expected to do a considerable amount of undergraduate advising at her new institution, which she finds onerous. If she has given up the option of returning to her home institution she will be motivated to reduce her dissonance by deciding, perhaps, that she always wanted to spend more time with undergraduates. If she still has the option of returning home, however, she is more likely to resent the time that the advising takes away from her scholarly activities. Because she can still decide to revoke her decision she needs to keep an accurate tally of the pros and cons of her new job, and the extra advising goes into the debit column. To test the hypothesis that keeping one’s options open can have an affective cost, Gilbert and Ebert (2002) asked people to rate their liking for nine art posters and then told them they could have either their third- or fourth-highest ranked poster to take home. Participants in the changeable condition learned that if at any point in the next month they wanted to change their minds, they could exchange the poster they had chosen for the other one, whereas participants in the unchangeable condition learned that their choice was final. As seen in Table V, people in the unchangeable condition subsequently ranked their choice of poster more favorably than people in the changeable condition did. They ‘‘made the best’’ of their choice, given that they could not change their minds. Forecasters asked to predict how they would rate the posters under changeable or unchangeable conditions did not anticipate this effect of irrevocability; they predicted that they would like their chosen poster

388

WILSON AND GILBERT

equally under changeable or unchangeable conditions. Finally, an additional group of participants read a description of the study and chose whether they would prefer to be in the changeable or unchangeable condition. Most participants chose the changeable condition, just as in everyday life people often prefer to retain the ability to change their mind later. In the Gilbert and Ebert study, however, this was the condition that resulted in less satisfaction with one’s choice. If our goal is to maximize our satisfaction with a chosen alternative, it may be best not to keep our options open.

C. MINOR INSULTS, LASTING PAINS Suppose that homeowners were asked whether they would rather have something minor go wrong with their house, such as a screen door that develops loose hinges and does not close properly, or something major, such as a furnace that stops working in the middle of winter. This seems like an absurd question; who would rather experience a major problem than a minor one? The question is not so silly, however, when we take into account people’s motivation to fix problems of different degrees of seriousness. When the furnace conks out in the middle of winter, homeowners call the heating and plumbing contractor immediately, and it is fixed as soon as possible. People are less motivated to fix minor problems; loose hinges, leaky faucets, and broken light switches can annoy people for months or years, because they never get around to repairing them. Ironically the net amount of discomfort caused by a minor problem might exceed that caused by a major problem. The same paradox might apply to psychological traumas. Suppose we asked people whether they would rather be insulted by a close friend or a stranger. Again, this seems like an absurd question; an insult from a friend will surely hurt more than an insult from someone we don’t know and will never see again. Suppose, however, that people are more motivated to ‘‘repair’’ the insult from the friend, by invoking their psychological defenses (e.g., ‘‘Sue didn’t really mean it when she said my new haircut makes me look 5 years older; she must be having a bad day’’). If they are able to defend against the insult from the friend successfully, the amount of pain it causes will be less than that caused by the insult from the stranger, which, precisely because it is less threatening, does not trigger psychological defenses to the same degree. To test this hypothesis Gilbert, Lieberman, Morewedge, and Wilson (in press) placed participants in a situation in which they received negative personality feedback from another participant. Half of the participants

AFFECTIVE FORECASTING

389

expected to meet and interact with the person who gave them the feedback whereas the other half believed they would never meet the person. When asked to predict how they would feel under these circumstances, forecasters said that they would feel worse if they received the negative feedback from the person they expected to meet. This prediction makes eminent sense; surely it would feel worse to be insulted by someone who we expect to meet and work with in a few minutes than by a stranger who we think we will never meet. As hypothesized, however, forecasters’ predictions were wrong. People who actually received the negative feedback felt better when it came from the person they expected to meet than from the person they would never see. Presumably the insult from the never-to-be-seen stranger was too minor an annoyance to trigger the psychological immune system. It was strong enough to evoke some negative affect, but was not so important that people were motivated to repair it psychologically. The insult from the person people were about to meet was presumably more threatening, triggering a successful attempt to ameliorate its impact (‘‘he was probably just pulling the experimenter’s leg when he rated me so negatively; I bet he didn’t mean it’’). Minor insults can cause more lasting pains, if they do not trigger attempts to defend against them.

D. THE TIMING OF RATIONALIZATION PROCESSES Sometimes people know in advance that a negative event is likely to occur and it is to their advantage to get a head start on dealing with it psychologically. If Jane is certain that she will be laid off from her job next month, for example, she might prepare herself by devaluing her company and reconstruing it as a terrible place to work. Preemptive rationalization is risky, though, because if the negative event does not, in fact, occur, people have rained on their own parades. Suppose that Jane is wrong and that instead of being laid off, she is promoted to vice president. If she has convinced herself that the company is a poorly run, top-heavy behemoth that is sure to go bankrupt, she will get less pleasure from her achievement. There is a delicate trade-off between softening the blow of a future negative event by rationalizing it in advance and reserving judgment in case the negative event does not occur. Wilson, Wheatley, Kurtz, Dunn, and Gilbert (2002) found that people are quite skilled at making this trade-off. As discussed earlier, participants took part in a simulated ‘‘dating game,’’ in which they believed that they and a same-sexed student were competing for a hypothetical date with an opposite-sex student. The purpose of the study was ostensibly to test a computer program that analyzed information about the students and tried

390

WILSON AND GILBERT

to predict which of the two ‘‘contestants’’ the ‘‘date’’ would prefer. The contestant learned that the computer had assigned them a low or high probability of being chosen, and this probability was either moderate or extreme: The computer said that their probability of being chosen was 1.5% (extreme-low), 12% (moderate-low), 88% (moderate-high), or 98.5% (extreme-high). Before learning the date’s actual choice, participants had the opportunity to examine positive and negative information about him or her, after which they learned the date’s choice and rated their mood. When given moderate probabilities (12% or 88%), people seemed to reserve judgment and avoid preemptive rationalization. Before learning the date’s decision, for example, people in the 12% and 88% conditions did not differ in how long they looked at negative versus positive information. But, as soon as they learned whether the date had actually chosen them or the other contestant, they succeeded in rapidly reconstruing information about him or her in a direction consistent with the decision. Those who lost rated the date more negatively than those who won, and were least likely to recall positive information about him or her—regardless of whether they initially expected to win or lose. Thus, rather than run the risk of rationalizing too soon, and being wrong about what the outcome would be, people with moderate expectations waited until the date’s actual choice was known and then selectively ‘‘spun’’ the information to make themselves feel better. The more people who lost reconstrued the date in a negative direction, for example, the better their mood. And, consistent with our prior work on immune neglect, forecasters did not anticipate that they would be able to rationalize a negative outcome so readily; they predicted that they would feel worse than experiencers reported feeling. People given extreme probabilities (1.5% or 98.5%) engaged in a different strategy. Rather than reserving judgment, these participants engaged in preemptive rationalization whereby they began to reconstrue the date in anticipation of a loss or a win, before knowing his or her actual choice (Pyszczynski, 1982). People in the 1.5% condition, for example, spent the most time looking at negative information about the date relative to positive information. People in the 98.5% condition, in contrast, formed the most positive impressions of the date prior to learning his or her decision. Interestingly, they did so at a cost, if their expectations were violated. Participants in the 98.5% condition who found out they had lost the dating game were least successful at rationalizing this negative outcome after the fact, and were relatively unhappy afterward. By engaging in preemptive rationalization (e.g., focusing on the date’s positive qualities) they seem to have made it more difficult for themselves to rationalize when they found out that the date did not choose them. In sum, people generally made good decisions about when to prepare for

AFFECTIVE FORECASTING

391

affective events in advance by rationalizing and when to reserve judgment. In the unlikely case that an extreme expectation is violated, however (e.g., losing a date that was 98.5% certain), there is a cost to preemptive rationalization.

E. THE PLEASURES OF UNCERTAINTY We have just seen several intriguing consequences of the dynamics of the psychological immune system. We turn now to some implications of sensemaking processes and emotional evanescence more generally, particularly in response to positive events. As seen earlier, one implication is that positive emotional reactions can be prolonged by interfering with people’s sense-making abilities. Centerbar, Wilson, and Gilbert (2002) found that participants who had been chosen as the ‘‘best potential friend’’ by oppositesex students were happiest the longest when they did not know which student had written the paragraphs explaining their choice. Wilson, Kermer, and Gilbert (2002) found that students who were given a dollar unexpectedly were happiest the longest when it was most difficult to explain why they had received the money. Both of these studies manipulated people’s ability to make sense of a positive event after it occurred. There are many times in life in which there is uncertainty about the nature of a positive event in advance, such as when people anticipate receiving a birthday present but do not know what it will be. Do people prefer to delay such uncertainty, or resolve it soon? Loewenstein (1987) noted that the anticipation of a positive event is itself pleasurable and has utility; for example, when asked when they would want to receive a kiss from their favorite movie star, participants preferred to wait a few days than receive the kiss right away, presumably to allow themselves the pleasure of anticipation. In these cases, however, people knew the exact nature of the event and thus knew what to savor. When there is uncertainty about what the event will be—imagine, for example, that people were told that they would get to kiss their favorite movie star or get a free ride in a hot air balloon—people generally prefer to resolve the uncertainty sooner rather than later. Lovallo and Kahneman (2000) found that people were willing to pay $9, on average, to know the outcome of a positive gamble (e.g., whether they won $1100 or $100) sooner (the next day) rather than later (in 2 weeks). Loewenstein (1994) suggested that ‘‘curiosity is always aversive’’ and that ‘‘the process of satisfying curiosity is itself pleasurable’’ (p. 90).

392

WILSON AND GILBERT

The current conception of sense-making processes suggests a different hypothesis. As long as the valence of the outcome is known to be positive, people might be better off delaying the resolution of uncertainty. Imagine that people knew that they would win one of two attractive gifts, such as a camera or a compact disc player. As soon as they learn which gift they have won they begin to make sense of it; if they win the camera, for example, they alter their schemas accordingly, think about what they will do with it, and move on to think about other things. If people do not know which gift they will win this sense making is kept in check, and people might continue to think about the gifts and derive pleasure from these thoughts. Wilson, Kurtz, and Gilbert (2002) tested this hypothesis by asking students to participate in a consumer attitudes study, in return for entering them in a lottery in which they could win attractive prizes (such as a camera, portable compact disc player, and a blender). The lottery was held in two stages. First, the students drew a slip of paper from a fish bowl to see if they had won a prize, ostensibly with a one-in-five chance of winning. All participants in fact won this stage of the lottery. Second, students wrote the names of their two most preferred prizes on slips of paper and randomly selected one, to find out which prize they had won. Participants in the certain condition did the second drawing right away and thus found out immediately what they had won. They were told that they would receive the prize when they returned for a second session held about 2 hours later. Participants in the uncertain condition were told that the second drawing, to determine which of their two favorite prizes they would win, would be held when they returned for the second session. Participants in both conditions were relatively happy at the first session, after having found out that they would win a prize. There was no difference between the certain and uncertain conditions at this point; even though those in the latter condition did not yet know which prize they had won, they were as happy as people in the certain condition. By the end of the second session people in the certain condition had become less happy; presumably they had had ample time to ‘‘make sense’’ of which prize they had won. People’s happiness increased slightly over time in the uncertain condition, resulting in a significant Condition  Time interaction. Presumably thoughts about the prizes stayed alive more in the uncertain condition, given that they could not yet ‘‘resolve’’ what they had won. We have no doubt that uncertainty is often an aversive state, especially when the valence of an outcome is unknown (e.g., whether an article will be accepted or rejected, or whether an investment has gained or lost money). As long as people know that the event is positive, however, a degree of uncertainty about it appears to prolong the pleasure people derive from it.

AFFECTIVE FORECASTING

393

IV. Why Don’t People Learn from Experience about the Impact Bias? Given that the impact bias appears to be quite pervasive, why don’t people learn about and correct for this bias over time? It might seem that as people experience life’s successes and setbacks, they should come to recognize that their emotional reactions do not last as long as they expected. After suffering through the loss of several loved ones, for example, do people learn that ‘‘this, too, shall pass?’’ After achieving numerous professional milestones, do they take note of the fact that their pride and elation did not last as long as they expected, and correct their future forecasts accordingly? It may not be as easy as it seems to learn about emotional evanescence. For example, there is evidence that people’s memory for their past emotional reactions is poor, which would limit their ability to learn from these reactions (Christianson & Safer, 1996; Robinson & Clore, 2002). People often have to reconstruct how they felt in the past by consulting their current thoughts, feelings, and theories (Bem & McConnell, 1970; Gilovich & Medvec, 1995; Goethals & Reckman, 1973; Holmberg & Holmes, 1994; Ross, 1989; Ross & Newby-Clark, 1998; Wilson, Houston, & Meyers, 1998). Therefore, people’s recall of the intensity and duration of past emotions might be subject to the same biases as predictions about future happiness. Under some conditions people might overestimate the impact that past events had on their emotions, committing a retrospective impact bias and preventing them from learning from their forecasting errors. For example, people might be prone to the focalism bias when trying to recall how they felt after a specific event in the past. Just as people tend to think of a future event in a vacuum, neglecting to adjust for the many other events that are likely to occupy their thoughts and influence their happiness, so might people think of past events in a vacuum. Baseball fans might overestimate how happy they were after their favorite team won the World Series, for example, because they think about that event alone and neglect to adjust for the fact that at the time they were busy at work, occupied by family matters, and thinking about other sports. On the other hand, we have seen that novel, emotion-evoking events come to seem more ordinary over time, as people make sense of them. When the Cardinals win the World Series it seems wonderful and unexpected, but as time goes by and people make sense of it, the Cardinal’s victory comes to seem less unexpected and amazing. When people think back to how they felt, they might be imagining an event (the inevitable Cardinal victory) that was different from the event they experienced at the time (the amazing, unexpected Cardinal victory).

394

WILSON AND GILBERT

One process, then, would seem to produce a retrospective impact bias (focalism), whereas another would seem to inhibit it (sense making). The net result might be an impact bias in retrospect that is smaller in magnitude than one that occurs in prospect. Before an event occurs both focalism and a failure to consider one’s sense-making processes produce a large impact bias (e.g., ‘‘If the Cardinals win the World Series I’ll be thrilled for days because all I will think about will be this momentous event’’). When people try to recall how they felt after the event, focalism again occurs but the event itself seems less momentous (e.g., ‘‘It was nice when the Cardinals won and their victory dominated my thoughts, but they were destined to win it all that year’’). We are aware of five studies that have tested for a retrospective impact bias and all found results that are consistent with this reasoning, at least for positive events. In three studies, Mitchell et al. (1997) assessed people’s predicted enjoyment, actual enjoyment, and recalled enjoyment of vacation trips (e.g., a trip to Europe). Wilson, Meyers, and Gilbert (2002) asked Republicans to predict how they would feel after the 2000 presidential election was decided, how they actually felt after the election was settled in George W. Bush’s favor, and 4 months later to recall how they had felt right after the election. In a second study Wilson, Meyers, and Gilbert (2002) asked forecasters to predict how they would feel if they did very well on a test of social aptitude. Experiencers took the test and were told that they had done very well, and then returned 1–3 weeks later and tried to recall how they had felt right after getting this positive feedback. The results were remarkably consistent across the five studies. Participants showed a strong impact bias in prospect, predicting that they would have stronger emotional reactions than they in fact did. In all five studies participants also committed a retrospective impact bias, whereby after the event they overestimated how strong their reaction had been. But, in each case the retrospective impact bias was of a smaller magnitude than the prospective impact bias. In the Wilson, Meyers, and Gilbert (2002) election study, for example, Republicans’ forecasts before the election were significantly greater than their recalled level of happiness after the election, though both were significantly higher than their actual level of happiness. These results are consistent with our reasoning that biases such as focalism can produce a retrospective impact bias, but that the magnitude of this bias is moderated by the fact that due to their sense-making processes, people recall an event that seems more ordinary than it did in prospect. Even if people are able to recall correctly how they felt after an event they still might not learn from experience when making forecasts about similar events in the future. It is not enough to be able to recall how one felt in the past, people must also exert the effort to compare the future event (e.g., how

AFFECTIVE FORECASTING

395

they will feel if their favorite baseball team wins the World Series) with relevant past events (e.g., how they felt the last time their team won the World Series). If people do not go to the trouble of searching their memory for past events to guide their predictions, but rather think about the future event in isolation, they obviously will not learn from their past experiences. Buehler and McFarland (2001), for example, found that people did not make very accurate forecasts about their reactions to future events, unless they were induced to think about similar events in the past. Finally, when people do consult their memories of past events, they need to decide which events are most applicable to the future event they are thinking about (Higgins, 1996; Koehler, 1996). In a study by Wilson, Meyers, and Gilbert (2001), for example, people who learned that they had done quite well on a test of social aptitude were not as happy as they had anticipated they would be. Fifteen minutes later, these participants predicted how happy they would be after learning they had done equally well on several tests, including a different form of the same test they had just taken. They predicted that they would be quite happy, significantly more so than the level of happiness they had just experienced after doing well on the test, and as happy as control participants who had not received the earlier positive feedback—in short, they did not learn at all from their prior experience. Although the evidence was indirect, Wilson, Meyers, and Gilbert (2001) assumed that the participants who had received the positive feedback had accurate recall of their prior feelings, given that they had received positive feedback on the test 15 minutes earlier. Presumably they also knew which of their prior experiences were most applicable to their forecasts, given that they were asked to predict how they would feel after doing well on a similar test in the future. The participants seem to have failed to meet the mental effort criterion, in that they apparently did not go to the trouble of consulting their earlier experiences when making their forecasts about their future ones. Perhaps if they were making forecasts that were more important to them than in these studies they would have gone to more of an effort to consult their prior experiences. At least at times, however, it is not sufficient to be able to recall how one felt in the past after similar events—people have to consult their memories and use this information when making forecasts about future events. So far we have limited our discussion of learning from experience to positive events, such as doing well on a test or seeing one’s favorite candidate win an election. As discussed earlier, when negative events occur people’s sense-making processes are turbo charged by the psychological immune system, because people are motivated to reconstrue or rationalize the event in a way that makes them feel better about it.

396

WILSON AND GILBERT

Because of this increased tendency to reconstrue negative events, people might view the events in more neutral terms after they occur (e.g., after they have been ‘‘defanged’’ by their psychological immune system), and thus be less likely to commit a retrospective impact bias. For example, forecasters in a study by Wilson, Meyers, and Gilbert (2002) predicted that they would be quite unhappy if they did poorly on a test of social aptitude. When experiencers found out they had done poorly they engaged in considerable rationalization, reporting that the test was unfair and was measuring something trivial (compared to people who had done well, who reported that the test was quite fair, etc.). Consequently, participants did not feel nearly as badly as forecasters had predicted. A few weeks later, when the participants attempted to recall how they had felt right after doing poorly on the test, they were still rationalizing; they continued to rate the test as unfair and unimportant. Consequently, they recalled correctly that they had not been very unhappy when they found out that they had done poorly. In other words, because they were viewing the test through the lens of their rationalization (‘‘the test was trivial and meaningless’’), they concluded, correctly, that they must not have been upset at doing poorly. Participants in this study were also asked to estimate how happy they would have predicted they would be right after taking the test, if they had been asked prior to taking it. If those who did poorly were viewing the test through the lens of their rationalization, then they should underestimate how badly they would have expected to feel in advance. Recall that forecasters had predicted that they would be very unhappy if they did poorly on the test. Those who really did do poorly, however, said that they would have predicted in advance that doing poorly would not bother them very much. Why? Forecasters (who had not taken the test) were imagining a devastating failure on a fair test of an important trait, whereas those who had done poorly were imagining an unfair test that assessed a trivial trait. The upshot is that the people who did poorly had not really learned from experience that negative events sometimes do not make them feel as badly as they would have predicted. Instead, they came to view the test as less negative, and assumed (wrongly) that they would have predicted in advance that it would have little impact on them. Indeed, Wilson, Meyers, and Gilbert (2001) found that when these people were asked to predict how they would feel if they did poorly on different tests that they had not rationalized, they said that they would feel as negatively as did participants who had not done poorly. Thus, there was no evidence that participants had learned the general lesson that they are equipped with a powerful psychological immune system that will enable them to recover quickly from future negative events of all kinds.

AFFECTIVE FORECASTING

397

A. THE IMPACT BIAS OVER THE LIFE SPAN The studies of learning from experience that we have discussed concerned single events in people’s lives, such as presidential elections, vacations, and performance on tests. The question remains whether people learn over their life spans about emotional evanescence. Are older people less prone to the impact bias than younger people, as a result of having noticed that they recovered more quickly than expected from many life events? To find out, Wilson, Gilbert, and Salthouse (2001) asked a sample of people from ages 20 to 91 to read descriptions of 12 events and estimate (1) how easy it was for them to imagine that the events influenced them a set amount (e.g., 2 points on a happiness scale), (2) how they would feel right after the events occurred, and (3) how long it would take for their general happiness to return to the level it was before the events occurred. The events included both large, impactful ones, such as the death of a relative and winning a million dollars in the lottery, and small, commonplace ones, such as having a pleasant telephone chat with a friend and learning that a party they were looking forward to was canceled due to inclement weather. There was no relationship between people’s age and the ease with which they could imagine the events having an impact on them, suggesting that the events we chose were applicable to people of all ages. Nor were there any relationships between age and how people said they would feel right after the events; people of all ages said they would be very sad after the death of a loved one, for example, and very happy if they won a million dollars. There was, however, a relation between age and how long people said it would take for the emotional impact of the events to wear off. As seen in Fig. 4, there was a slight increase in the predicted duration of one’s emotional reactions between ages 20 and 60, though this relationship was not strong; r ¼ .17, df ¼ 129, p ¼ .06. After the age of 60 the trend reversed, such that the older the participants the less time they said it would take them to recover from emotion events, r ¼
.32, df ¼ 64, p ¼ .008. The quadratic trend shown in Fig. 4 was obtained when people predicted their reaction to both positive and negative events. A trivial explanation of the downward slope after age 60 is that the older people are the less time they have left to experience the impact of major life events, thus they make shorter predictions. However, the negative correlation after age 60 was present for both major and minor life events (such as how long they thought it would take for the emotional impact of a social gathering at a friend’s house to wear off). Further, the correlation remained when we truncated all predictions at one month, to minimize the possibility that younger people were predicting that the events would last for longer than the expected life span of the older people.

398

WILSON AND GILBERT

Log of Predicted Time to Recover 7 1 month 6 1 week 5 4 1 day 3 2

1 hour

1 0

20s 30s 40s 50s 60s 70s 80s & 90s Age

Fig. 4. Relationship of people’s age to their predicted duration of emotional reactions to life events.

We confess that we do not have a ready-made explanation for the negative correlation after the age of 60, other than the possibility that it takes until that age for people to begin to learn from their many emotional experiences that they return to their emotional baseline relatively quickly. A limitation of the Wilson, Gilbert, and Salthouse (2001) study is that it examined predicted emotions only; there was no way of knowing how accurate these predictions were. In this light it is interesting to consider the results of a study of age and actual emotional experience by Carstensen, Pasupathi, Mayr, and Nesselroade (2000). People from ages 18 to 94 rated the degree to which they were feeling several different emotions at five, randomly chosen times a day for 1 week. There was a change in actual emotional experience that occurred at about the same time point (age 60) as the change we found in predicted emotional experience. When people were feeling more negatively than usual at one time point, Carstensen et al. computed the likelihood that they had ‘‘recovered’’ by the next time point and were feeling less negatively than usual. Between the ages of 18 and 60 there was a positive correlation between this recovery index and age; the older the participant, the more likely they were to have recovered from a negative emotional experience. However, from age 60 onward this trend

AFFECTIVE FORECASTING

399

reversed slightly; the older people were, the longer it took them to recover from negative experiences (though the correlation was nonsignificant for people over 60). It is striking that the age at which people seemed to become ‘‘wise’’ in the Wilson, Gilbert, and Salthouse (2001) study, predicting that emotional reactions to life events would not influence them for very long, is precisely the age at which negative emotional experiences began to increase in duration in the Carstensen et al. (2000) study. There were many differences between the two studies, of course, and it will take further research on adults across the life span to unravel the accuracy of people’s affective forecasts. The available evidence suggests a curious disconnect between forecasts and experience from 60 onward that deserves a closer look.

V. Is the Impact Bias Functional? The pervasiveness of the impact bias, and the difficulty people seem to have in learning about it from experience, raises the question of whether the bias is functional in some way. It is easy to imagine that at times it is, such as for negative events over which people can exert some control. Overestimating the intensity and duration of negative outcomes can serve a selfregulatory function, motivating people to work hard in the present to prevent these outcomes from occurring in the future, such as a person who thinks, ‘‘If I do poorly on my statistics exam I will feel terrible for weeks; I better skip the party and go to the review session’’ (Mischel, Cantor, & Feldman, 1996). Research on defensive pessimism, for example, suggests that some people benefit from exaggerating the chances that negative consequences will occur, because it motivates them to prepare for the worst (e.g., Norem, 2001; Norem & Cantor, 1986; Rachman, 1994; Sanna, 1996). It is not to people’s advantage to think, ‘‘I guess I’ll stay in bed this morning and skip work; because after all, if I get fired my psychological immune system will succeed in making me feel OK about it.’’ Exaggerating the intensity and duration of reactions to positive events can also serve as a motivator, leading people to work harder to obtain these outcomes. Further, there is utility in anticipating positive events (Loewenstein, 1987), and people might thus enjoy exaggerating the pleasure they will experience in the future (e.g., ‘‘the concert will be a real peak experience’’). Sometimes people have no control over future outcomes, however, such as the possibility that the large corporation that employs them is about to go bankrupt or that the college to which they have already applied will turn

400

WILSON AND GILBERT

them down. It does not seem as functional to overestimate the intensity and duration of one’s negative emotional reactions when people cannot do anything to influence the outcome, especially given that doing so is likely to cause unnecessary worry and anxiety. People need to prepare for uncontrollable negative outcomes by engaging in proactive coping (Aspinwall, 1997; Taylor & Pham, 1996), such as by looking for another job or working hard on applications to other colleges, but it does not seem beneficial to exaggerate how badly one will feel if an uncontrollable event happens. Even when events are controllable, the case could be made that people would be better off if they knew exactly how much pleasure or pain the events would cause, so that they could make wise decisions about how much effort to exert. Exaggerating the impact of events such as a job promotion might serve to increase people’s motivation to work toward it, but why should people work hard toward something that will not bring as much gratification as they think? Wouldn’t it be to people’s advantage to have a better idea of the intensity and duration of the pleasure that they would derive from different events? Consider people who are thinking of buying an expensive consumer item such as a television set or new car. Part of the decision involves an affective forecast; people are willing to pay more for an item if they think it will cause lasting intense pleasure than if they think it will make them happy for only an hour or a day. If people commit the impact bias and overestimate how much pleasure a television set or new car will bring, they are paying more for these items than they should. Further, the mechanisms that produce the impact bias, such as people’s lack of appreciation of their sense-making abilities (including immune neglect), come with a cost. As we have seen, people sometimes arrange their environments in ways that lead to less happiness, such as keeping their options open when they would be better off making a final decision (Gilbert & Ebert, 2002) and reducing their uncertainty about pleasurable events (Centerbar, Wilson, & Gilbert, 2002). Whether people are better off committing the impact bias is thus an open question. There has been very little research that has addressed this question directly, by, for example, examining the relationship between mental health indices and the propensity to commit the impact bias, as has been done with other kinds of illusions and biases (Taylor & Brown, 1988). The only evidence of which we are aware comes from the study of aging by Wilson, Gilbert, and Salthouse (2001). In this study the length of time that people said it would take them to recover from emotional events was positively correlated with a measure of depression (the Center for Epidemiologic Studies Depression Scale), r ¼ .30 for positive events, r ¼ .22 for negative events, ps < .005. It is unclear, of course, whether long predicted recovery times increased people’s risk for depression, depression increased the

AFFECTIVE FORECASTING

401

predicted amount of recovery time, or some third variable was positively correlated with both predicted recovery time and depression. Nonetheless these positive correlations are inconsistent with the hypothesis that the impact bias is functional.

VI. Summary and Conclusions Many important decisions are based on affective forecasts. Should we get married? Have a family? Pursue a career as musician or teacher? Go out to dinner or stay home and watch a made-for-television movie? Decisions such as these—indeed, virtually all important decisions—are based on predictions about how the different options will make us feel. People are skilled at predicting the valence of their future emotional reactions; if the choice is between going out to dinner or watching dogs be euthanized at the local animal shelter, most people would know which activity would be fun and which would not. People are also skilled at guessing the specific kinds of emotional reactions they will have, such as predicting whether an insult from a co-worker is more likely to cause anger, fear, or disgust. Predictions about valence and emotional specificity are not perfect; people sometimes underestimate the emotional complexity of an event or the predominant emotion they will experience, especially if they badly misconstrue what key elements of the situation will be like. In general, however, there is a good deal of accuracy to predictions about valence and specific emotions. People are less accurate at predicting the intensity and duration of their future emotional reactions and we reviewed a number of mechanisms that can result in over- or underprediction. By far the most common error is the impact bias, the tendency to overestimate the enduring impact that future events will have on our emotional reactions. One reason for the impact bias is focalism, whereby people fail to anticipate the extent to which unrelated events will influence their thoughts and emotions, and thus overestimate the impact of the event they are considering. Perhaps the most prevalent cause of the impact bias is ordinization neglect, whereby people fail to anticipate the extent to which they will ‘‘ordinize’’ an event after it occurs, by automatically making sense of it. People are consummate sense makers who transform novel, emotion-producing events into ones that seem ordinary and mundane, through the processes of assimilation, accommodation, and explanation. Doing so reduces the emotional power of an event. By failing to take into account how rapidly such sense-making processes will occur, people overestimate the intensity and duration of their future emotional reactions.

402

WILSON AND GILBERT

People are especially motivated to ‘‘defang’’ negative events in ways that help them recover from the negative emotions they produce. They are equipped with a psychological immune system that often succeeds in rationalizing or reconstruing an event in a way that ameliorates its impact. Because the psychological immune system operates largely outside of conscious purview, people do not take it into account when predicting their future emotions, an error called immune neglect. People are thus especially likely to overestimate the emotional impact of negative events. We documented several consequences of immune neglect, including the belief that a benevolent external agent is controlling one’s behavior and the tendency to arrange our environments in ways that are not optimal for maximizing our happiness. We also examined evidence that people do not readily learn about and correct for the impact bias, which suggests that it may be functional in some way (such as motivating people to work to avoid negative events). We are not yet convinced that the impact bias is an altogether good thing, however, and believe that people would be better off if they could glimpse their many futures and predict well what their emotional reactions are likely to be.

Acknowledgments The writing of this article and much of the research we report was supported by a grant from the National Institute of Mental Health (RO1-MH56075). We thank Sara Algoe, Kevin Carlsmith, David Centerbar, and Robyn Mallett for comments on an earlier draft.

References Abramson, L. Y., Seligman, M. E. P., & Teasdale, J. D. (1978). Learned helplessness in humans: Critique and reformulation. Journal of Abnormal Psychology, 87, 49–74. Anderson, B. (1994). The volume of the cerebellar molecular layer predicts attention to novelty in rats. Brain Research, 641, 160–162. Andrews, F. M., & Robinson, J. P. (1991). Measures of subjective well-being. In J. P. Robinson, P. R. Shaver, & L. S. Wrightsman (Eds.), Measures of personality and social psychological attitudes (Vol. 1, pp. 61–114). San Diego, CA: Academic Press. Ariely, D. (1998). Combining experiences over time: The effects of duration, intensity changes and on-line measurements on retrospective pain evaluations. Journal of Behavioral Decision Making, 11, 19–45. Armor, D. A., & Taylor, S. E. (1998). Situation optimism: Specific outcome expectancies and self-regulation. In M. Zanna (Ed.), Advances in experimental social psychology (Vol. 30, pp. 309–379). New York: Academic Press.

AFFECTIVE FORECASTING

403

Aronson, E. (1968). Dissonance theory: Progress and problems. In R. P. Abelson, E. Aronson, W. J. McGuire, T. M. Newcomb, M. J. Rosenberg, & P. H. Tannenbaum (Eds.), Theories of cognitive consistency: A sourcebook (pp. 5–27). Chicago: Rand McNally. Aspinwall, L. G. (1997). Where planning meets coping: Proactive coping and the detection and management of potential stressors. In S. L. Friedman & E. K. Scholnick (Eds.), The developmental psychology of planning: Why, how, and when do we plan? (pp. 285–320). Mahwah, NJ: Erlbaum. Baron, J. (1992). The effects of normative beliefs on anticipated emotions. Journal of Personality and Social Psychology, 63, 320–330. Bem, D. J., & McConnell, H. K. (1970). Testing the self-perception explanation of dissonance phenomena: On salience of premanipulation attitudes. Journal of Personality and Social Psychology, 14, 23–31. Brehm, J. W. (1956). Post-decision changes in desirability of alternatives. Journal of Abnormal and Social Psychology, 52, 384–389. Brickman, P., & Campbell, D. T. (1971). Hedonic relativism and planning the good society. In M. H. Appley (Ed.), Adaptation-level theory (pp. 287–305). New York: Academic Press. Buehler, R., & McFarland, C. (2001). Intensity bias in affective forecasting: The role of temporal focus. Personality and Social Psychology Bulletin, 27, 1480–1493. Carli, L. L. (1999). Cognitive reconstruction, hindsight, and reactions to victims and perpetrators. Personality and Social Psychology Bulletin, 25, 966–979. Carstensen, L. L., Pasupathi, M., Mayr, U., & Nesselroade, J. R. (2000). Emotional experience in everyday life across the adult life span. Journal of Personality and Social Psychology, 79, 644–655. Centerbar, D. B., Wilson, T. D., & Brekke, N. (2002). Mental contamination and the debiasing problem. In T. Gilovich, D. W. Griffin, & D. Kahneman (Eds.), The psychology of judgment: Heuristics and biases (pp. 185–200). New York: Cambridge University Press. Cheal, M., Johnson, M. R., Ellingboe, J., & Skupny, A. S. (1984). Preservation of attention to conspecific odors and novel objects in castrated gerbils. Physiology and Behavior, 33, 563–570. Christianson, S. A., & Safer, M. A. (1996). Emotional events and emotions in autobiographical memory. In D. Rubin (Ed.), Remembering our past: Studies in autobiographical memory (pp. 218–243). New York: Cambridge University Press. Coughlan, R., & Connolly, T. (2001). Predicting affective responses to unexpected outcomes. Organizational Behavior and Human Decision Processes, 85, 211–225. Crawford, M., T., McConnell, A. R., Lewis, A. C., & Sherman, S. J. (2002). Reactance, compliance, and anticipated regret. Journal of Experimental Social Psychology, 38, 56–63. Daffner, K. R., Mesulam, M. M., Scinto, L. F. M., Acar, D., Calvo, V., Raust, R., Chabrerie, A., Kennedy, B., & Holcomb, P. (2000). The central role of the prefrontal cortex in directing attention to novel events. Brain, 123, 927–939. Damasio, A. (1994). Descarte’s error: Emotion, reason, and the human brain. New York: Grosset/Putnam. Danner, D. D., Snowdon, D. A., & Friesen, W. V. (2001). Positive emotions in early life and longevity: Findings from the nun study. Journal of Personality and Social Psychology, 80, 804–813. Dennis, C. (2001). Practical gods. New York: Penguin. Diener, E. (2000). Subjective well-being: The science of happiness, and a proposal for a national index. American Psychologist, 55, 34–43. Diener, E., Emmons, R. A., Larsen, R. J., & Griffin, S. (1985). The Satisfaction With Life Scale. Journal of Personality Assessment, 49, 71–75.

404

WILSON AND GILBERT

Diener, E., & Seligman, M. E. P. (2002). Very happy people. Psychological Science, 13, 81–84. Donchin, E. (1981). Surprise! . . . Surprise? Psychophysiology, 18, 493–513. Dunn, E. W., Wilson, T. D., & Gilbert, D. T. (in press). Location, location, location: The misprediction of satisfaction in housing lotteries. Personality and Social Psychology Bulletin. Dunning, D. (1999). A newer look: Motivated social cognition and the schematic representation of social concepts. Psychological Inquiry, 10, 1–11. Easterbrook, J. A. (1959). The effect of emotion on cue utilization and the organization of behavior. Psychological Review, 66, 183–201. Eiser, J. R. (1990). Social judgment. Buckingham, UK: Open University Press. Elster, J., & Loewenstein, G. F. (1992). Utility from memory and anticipation. In G. F. Loewenstein & J. Elster (Eds.), Choice over time (pp. 213–234). New York: Russell Sage Foundation. Enns, J. T., Austen, E. L., Di Lollo, V., Rauschenberger, R., & Yantis, S. (2001). New objects dominate luminance transients in setting attentional priority. Journal of Experimental Psychology: Human Perception and Performance, 27, 1287–1302. Fantz, J. F. (1964). Visual experience in infants: Decreased attention to familiar patterns relative to novel ones. Science, 146, 668–670. Festinger, L. (1957). A theory of cognitive dissonance. Palo Alto, CA: Stanford University Press. Fischhoff, B. (1975). Hindsight6¼foresight: The effect of outcome knowledge on judgment under uncertainty. Journal of Experimental Psychology: Human Perception and Performance, 1, 288–299. Folkman, S. (1984). Personal control and stress and coping processes: A theoretical analysis. Journal of Personality and Social Psychology, 46, 839–852. Fordyce, M. W. (1988). A review of research on the happiness measures: A sixty second index of happiness and mental health. Social Indicators Research, 20, 355–381. Frederick, S., & Loewenstein, G. (1999). Hedonic adaptation. In D. Kahneman, E. Diener, & N. Schwarz (Eds.), Well-being: The foundations of hedonic psychology (pp. 302–329). New York: Russell Sage Foundation. Fredrickson, B. L. (1998). What good are positive emotions? Review of General Psychology, 2, 300–319. Fredrickson, B. L., & Kahneman, D. (1993). Duration neglect in retrospective evaluations of affective episodes. Journal of Personality and Social Psychology, 65, 45–55. Freud, A. (1966). The ego and the mechanisms of defense. New York: International Universities Press. Freud, S. (1968). A general introduction to psychoanalysis (J. Riviere, Trans.). New York: Washington Square Press (originally published 1924). Frey, D. (1981). Reversible and irreversible decisions: Preferences for consonant information as a function of attractiveness of decision alternatives. Personality and Social Psychology Bulleting, 7, 621–626. Gallup, G., Jr, & Catelli, J. (1989). The people’s religion: American faith in the 90’s. New York: Macmillan. Geers, A. L., & Lassiter, G. D. (1999). Affective expectations and information gain: Evidence for assimilation and contrast effects in affective experience. Journal of Experimental Social Psychology, 35, 394–413. Gilbert, D. T. (1991). How mental systems believe. American Psychologist, 46, 107–119. Gilbert, D. T. (1998). Ordinary personology. In D. Gilbert, S. Fiske, & G. Lindzey (Eds.), The handbook of social psychology (4th ed., Vol. 2, pp. 89–150). New York: Random House.

AFFECTIVE FORECASTING

405

Gilbert, D. T. (2002). Inferential correction. In T. Gilovich, D. W. Griffin, & D. Kahneman (Eds.), The psychology of judgment: Heuristics and biases (pp. 167–184). Cambridge: Cambridge University Press. Gilbert, D. T., Brown, R. P., Pinel, E. C., & Wilson, T. D. (2000). The illusion of external agency. Journal of Personality and Social Psychology, 79, 690–700. Gilbert, D. T., Driver-Linn, E., & Wilson, T. D. (2002). The trouble with Vronsky: Impact bias in the forecasting of future affective states. In L. Feldman-Barrett & P. Salovey (Eds.), The wisdom of feeling (pp. 114–143). New York: Guilford. Gilbert, D. T., & Ebert, J. E. (2002). Decisions and revisions: The affective forecasting of changeable outcomes. Journal of Personality and Social Psychology, 82, 503–514. Gilbert, D. T., Gill, M. J., & Wilson, T. D. (2002). The future is now: Temporal correction in affective forecasting. Organizational Behavior and Human Decision Processes, 88, 690–700. Gilbert, D. T., Lieberman, M. D., Morewedge, C. K., & Wilson, T. D. (in press). The peculiar longevity of things not so bad. Psychological Science. Gilbert, D. T., Pelham, B. W., & Krull, D. S. (1988). On cognitive busyness: When person perceivers meet persons perceived. Journal of Personality and Social Psychology, 54, 733–740. Gilbert, D. T., Pinel, E. C., Wilson, T. D., Blumberg, S. J., & Wheatley, T. P. (1998). Immune neglect: A source of durability bias in affective forecasting. Journal of Personality and Social Psychology, 75, 617–638. Gilbert, D. T., & Rimsky, S. E. (1999). [Adroitness at reducing dissonance as a predictor of belief in God.] Unpublished raw data, Harvard University. Gilbert, D. T., & Wilson, T. D. (2000). Miswanting: Some problems in the forecasting of future affective states. In J. Forgas (Ed.), Feeling and thinking: The role of affect in social cognition (pp. 178–197). Cambridge: Cambridge University Press. Gilovich, T. (1991). How we know what isn’t so: The fallibility of human reason in everyday life. New York: The Free Press. Gilovich, T., Griffin, D. W., & Kahneman, D. (Eds.) (2002). The psychology of judgment: Heuristics and biases. New York: Cambridge University Press. Gilovich, T., & Medvec, V. H. (1995). The experience of regret: What, when, and why. Psychological Review, 102, 379–395. Goethals, G. R., & Reckman, R. F. (1973). The perception of consistency in attitudes. Journal of Experimental Social Psychology, 9, 491–501. Greenwald, A. G. (1980). The totalitarian ego: Fabrication and revision of personal history. American Psychologist, 35, 603–618. Griffin, D. W., & Ross, L. (1991). Subjective construal, social inference, and human misunderstanding. In M. Zanna. Advances in experimental social psychology (Vol. 24, pp. 319–356). New York: Academic Press. Hamann, S. B., Ely, T. D., Hoffman, J. M., & Kilts, C. D. (2002). Activation of the human amygdala in positive and negative emotion. Psychological Science, 13, 135–141. Hawkins, S. A., & Hastie, R. (1990). Hindsight: Biased judgments of past events after the events are known. Psychological Bulletin, 107, 311–327. Hawthorne, N. (1937). N. H. Pearson, et al. (Ed.). The complete novels and selected tales of Nathaniel Hawthorne. New York: Random House (originally published 1846). Heider, F. (1958). The psychology of interpersonal relations. New York: Wiley. Helson, H. (1964). Adaptation-level theory. New York: Harper & Row. Higgins, E. T. (1996). Knowledge activation: Accessibility, applicability, and salience. In E. T. Higgins & A. Kruglanski (Eds.) Social psychology: Handbook of basic principles, New York: Guilford.

406

WILSON AND GILBERT

Hilgetag, C. C., Lomber, S. G., & Payne, B. R. (2001). Neural mechanisms of spatial attention in the cat. Neurocomputing, 38–40, 1281–1287. Hochschild, A. R. (1979). Emotion work, feeling rules, and social structure. American Journal of Sociology, 85, 551–575. Hodges, S. D. (1997). When matching up features messes up decisions: The role of feature matching in successive choices. Journal of Personality and Social Psychology, 72, 1310–1321. Holmberg, D., & Holmes, J. G. (1994). Reconstruction of relationship memories: A mental models approach. In N. Schwarz & S. Sudman (Eds.), Autobiographical memory and the validity of retrospective reports (pp. 267–288). New York: Springer-Verlag. Holyoak, K. J., & Simon, D. (1999). Bidirectional reasoning in decision making by constraint satisfaction. Journal of Experimental Psychology: General, 128, 3–31. Houston, D. A., Sherman, S. J., & Baker, S. M. (1991). Feature matching, unique features, and the dynamics of the choice process: Predecision conflict and postdecision satisfaction. Journal of Experimental Social Psychology, 27, 411–430. Hsee, C. K., & Abelson, R. P. (1991). Velocity relation: Satisfaction as a function of the first derivative of outcome over time. Journal of Personality and Social Psychology, 60, 341–347. Isen, A. M. (1993). Positive affect and decision making. In M. Lewis & J. M. Haviland (Eds.), Handbook of emotions (pp. 261–277). New York: Guilford. Johnston, W. A., & Schwarting, I. S. (1997). Novel popout: An enigma for conventional theories of attention. Journal of Experimental Psychology: Human Perception and Performance, 23, 622–631. Jones, E. E., & Davis, K. E. (1965). From acts to dispositions: The attribution process in social psychology. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 2, pp. 219–266). New York: Academic Press. Kahneman, D. (1994). New challenges to the rationality assumption. Journal of Institutional and Theoretical Economics, 150, 18–36. Kahneman, D., Fredrickson, B. L., Schreiber, C. A., & Redelmeier, D. A. (1993). When more pain is preferred to less: Adding a better ending. Psychological Science, 4, 401–405. Kahneman, D., & Snell, J. (1990). Predicting utility. In R. Hogarth (Ed.), Insights in decision making, Chicago: University of Chicago Press. Kahneman, D., & Tversky, A. (1979). Prospect theory: An analysis of decision under risk. Econometrica, 47, 263–291. Kahneman, D., & Tversky, A. (1984). Choices, values, and frames. American Psychologist, 39, 341–350. Kahneman, D. & Tversky, A. (Eds.) (2000). Choices, values, and frames. Cambridge: Cambridge University Press. Kammann, R., & Flett, R. (1983). Affectometer 2: A scale to measure current level of general happiness. Australian Journal of Psychology, 35, 257–265. Kelley, H. H. (1967). Attribution theory in social psychology. In D. Levine (Ed.), Nebraska symposium on motivation (Vol. 15, pp. 192–238). Lincoln, NE: University of Nebraska Press. Kimmel, H. D., Van Olst, E. H., & Orlebeke, J. F. (1979). The orienting reflex in humans. Hillsdale, NJ: Erlbaum. Koehler, J. J. (1996). The base rate fallacy reconsidered: Normative, descriptive, and methodological challenges. Behavioral and Brain Sciences, 19, 1–53. Klaaren, K. J., Hodges, S. D., & Wilson, T. D. (1994). The role of affective expectations in subjective experience and decision making. Social Cognition, 12, 77–101. Kunda, Z. (1990). The case for motivated reasoning. Psychological Bulletin, 108, 480–498.

AFFECTIVE FORECASTING

407

Langer, E. J., & Rodin, J. (1976). The effects of choice and enhanced personal responsibility for the aged: A field experiment. Journal of Personality and Social Psychology, 34, 191–198. Larsen, J. T., McGraw, P., & Cacioppo, J. T. (2001). Can people feel happy and sad at the same time? Journal of Personality and Social Psychology, 81, 684–698. Larsen, R. J., & Fredrickson, B. L. (1999). Measurement issues in emotion research. In D. Kahneman, E. Diener, & N. Schwarz (Eds.), Well-being: The foundations of hedonic psychology (pp. 40–60). New York: Russell Sage Foundation. Latane´ , B., & Darley, J. M. (1970). The unresponsive bystander: Why doesn’t he help? Englewood Cliffs, NJ: Prentice Hall. LeDoux, J. (1996). The emotional brain: The mysterious underpinnings of emotional life. New York: Simon & Schuster. Liberman, N., Sagristano, M., & Trope, Y. (2002). The effect of temporal distance on level of mental construal. Journal of Experimental Social Psychology, 38, 523–534. Linville, P. W., & Fischer, G. W. (1991). Preferences for separating or combining events. Journal of Personality and Social Psychology, 60, 5–23. Loewenstein, G. (1987). Anticipation and the valuation of delayed consumption. The Economic Journal, 97, 666–684. Loewenstein, G. (1994). The psychology of curiosity: A review and reinterpretation. Psychological Bulletin, 116, 75–98. Loewenstein, G. (1996). Out of control: Visceral influences on behavior. Organizational Behavior and Human Decision Processes, 65, 272–292. Loewenstein, G. (2001). A visceral account of addiction. In P. Slovic (Ed.), Smoking: Risk, perception, & policy (pp. 188–215). Thousand Oaks, CA: Sage. Loewenstein, G., & Adler, D. (1995). A bias in the prediction of tastes. The Economic Journal, 105, 929–937. Loewenstein, G., & Frederick, S. (1997). Predicting reactions to environmental change. In M. H. Bazerman, D. M. Messick, A. E. Tenbrusel, & K. A. Wade-Benzoni (Eds.), Environment, ethics, and behavior (pp. 52–72). San Francisco, CA: New Lexington Press. Loewenstein, G., Nagin, D., & Paternoster, R. (1997). The effect of sexual arousal on sexual forcefulness. Journal of Research in Crime and Delinquency, 34, 443–473. Loewenstein, G., O’Donoghue, T., & Rabin, M. (1999). Projection bias in predicting future utility. Unpublished manuscript, Carnegie Mellon University, Pittsburgh, PA. Loewenstein, G., & Prelec, D. (1993). Preferences for sequences of outcomes. Psychological Review, 100, 91–108. Loewenstein, G., Prelec, D., & Shatto, C. (1998). Hot/cold intrapersonal empathy gaps and the under-prediction of curiosity. Unpublished manuscript, Carnegie Mellon University, Pittsburgh, PA. Loewenstein, G. F., & Schkade, D. (1999). Wouldn’t it be nice?: Predicting future feelings. In D. Kahneman, E. Diener, & N. Schwartz (Eds.), Well-being: The foundations of hedonic psychology (pp. 85–105). New York: Russell Sage Foundation. Loewenstein, G., Weber, E. U., Hsee, C. K., & Welch, N. (2001). Risk as feelings. Psychological Bulletin, 127, 267–286. Lovallo, D., & Kahneman, D. (2000). Living with uncertainty: Attractiveness and resolution timing. Journal of Behavioral Decision Making, 13, 179–190. Lowe, R. E., & Steiner, I. D. (1968). Some effects of the reversibility and consequences of decisions on postdecision information preferences. Journal of Personality and Social Psychology, 8, 172–179. MacDonald, T. K., MacDonald, G., Zanna, M. P., & Fong, G. T. (2000). Alcohol, sexual arousal, and intentions to use condoms in young men: Applying alcohol myopia to risky sexual behavior. Health Psychology, 19, 290–298.

408

WILSON AND GILBERT

Mandler, G. (1975). Mind and emotion. New York: Wiley. Martin, L. L., & Stapel, D. A. (1998). Correction and metacognition: Are people naive dogmatists or naive empiricist during social judgments? In V. Yzerbyt, G. Lories, & B. Dardenne. Metacognition: Cognitive and social dimensions (pp. 228–247). New York: Sage. McFarland, C., Ross, M., & DeCourville, N. (1989). Women’s theories of menstruation and biases in recall of menstrual symptoms. Journal of Personality and Social Psychology, 57, 522–531. Mellers, B. A. (2000). Choice and the relative pleasure of consequences. Psychological Bulletin, 126, 910–924. Mellers, B. A., & McGraw, A. P. (2001). Anticipated emotions as guides to choice. Current Directions in Psychological Science, 10, 210–214. Mellers, B., Schwartz, A., & Ritov, I. (1999). Emotion-based choice. Journal of Experimental Psychology: General, 128, 332–345. Milgram, S. (1974). Obedience to authority: An experimental view. New York: Harper & Row. Mitchell, T. R., Thompson, L., Peterson, E., & Cronk, R. (1997). Temporal adjustments in the evaluation of events: The ‘‘rosy view.’’ Journal of Experimental Social Psychology, 33, 421–448. Mischel, W., Cantor, N., & Feldman, S. (1996). Principles of self-regulation: The nature of willpower and self-control. In E. T. Higgins & A. W. Kruglanski (Eds.), Social psychology: Handbook of basic principles (pp. 329–360). New York: Guilford. Myers, D. G. (1999). Close relationships and quality of life. In D. Kahneman, E. Diener, & N. Schwarz (Eds.), Well-being: The foundations of hedonic psychology (pp. 374–391). New York: Russell Sage Foundation. Newtson, D. (1973). Attribution and the unit of perception of ongoing behavior. Journal of Personality and Social Psychology, 28, 28–30. Nisbett, R. E., & Kanouse, D. E. (1969). Obesity, food deprivation, and supermarket shopping behavior. Journal of Personality and Social Psychology, 12, 389–294. Nisbett, R. E., & Ross, L. D. (1980). Human inference: Strategies and shortcomings of social judgment. Englewood Cliffs, NJ: Prentice Hall. Nisbett, R. E., & Wilson, T. D. (1977). Telling more than we can know: Verbal reports on mental processes. Psychological Review, 84, 231–259. Norem, J. K. (2001). The positive power of negative thinking: Using defensive pessimism to manage anxiety and perform at your peak. New York: Basic Books. Norem, J. K., & Cantor, N. (1986). Defensive pessimism: Harnessing anxiety as motivation. Journal of Personality and Social Psychology, 51, 1208–1217. Ortony, A., Clore, G. L., & Collins, A. (1988). The cognitive structure of emotions. New York: Cambridge University Press. Osberg, T. M., & Shrauger, J. S. (1986). Self-prediction: Exploring the parameters of accuracy. Journal of Personality and Social Psychology, 51, 1044–1057. Parducci, A. (1995). Happiness, Pleasure, and judgment: The contextual theory and its applications. Mahwah, NJ: Erlbaum. Piaget, J. (1952). The origins of intelligence in children. New York: International Universities Press. Piaget, J., & Inhelder, B. (1969). The psychology of the child. New York: Basic Books. Pittman, T. S. (1998). Motivation. In D. Gilbert, S. Fiske, & G. Lindzey (Eds.), The handbook of social psychology (4th ed., Vol. 1, pp. 549–590). New York: Random House. Pohl, R. F., & Hell, W. (1996). No reduction in hindsight bias after complete information and repeated testing. Organizational Behavior and Human Decision Processes, 67, 49–58. Pyszczynski, T. (1982). Cognitive strategies for coping with uncertain outcomes. Journal of Research in Personality, 16, 386–399.

AFFECTIVE FORECASTING

409

Pyszczynski, T., Greenberg, J., & Solomon, S. (1997). Why do we need what we need? A terror management perspective on the roots of human social motivation. Psychological Inquiry, 8, 1–20. Quinn, P. C., Eimas, P. D., & Tarr, M. J. (2001). Perceptual categorization of cat and dog silhouettes by 3- and 4-month old infants. Journal of experimental child psychology, 79, 78–94. Rachman, S. (1994). The overprediction of fear: A review. Behavior Research and Therapy, 32, 683–690. Rachman, S., & Arntz, A. (1991). The overprediction and underprediction of pain. Clinical Psychology Review, 11, 339–355. Radloff, L. S. (1977). The CES-D Scale: A self-report depression scale for research in the general population. Applied Psychological Measurement, 1, 385–401. Read, D., Loewenstein, G., & Kalyanaraman, S. (1999). Mixing virtue and vice: Combining the immediacy effect and the diversification heuristic. Journal of Behavioral Decision Making, 12, 257–273. Read, D., & van Leeuwen, B. (1998). Predicting hunger: The effects of appetite and delay on choice. Organizational Behavior and Human Decision Processes, 76, 189–205. Robinson, M. D., & Clore, G. L. (2001). Simulation, scenarios, and emotional appraisal: Testing the convergence of real and imagined reactions to emotional stimuli. Personality & Social Psychology Bulletin, 27, 1520–1532. Robinson, M. D., & Clore, G. L. (2002). Belief and feeling: Evidence for an accessibility model of emotional self-report. Psychological Bulletin, 128, 934–960. Roese, N. J., & Olson, J. M. (1996). Counterfactuals, causal attributions, and the hindsight bias: A conceptual integration. Journal of Experimental Social Psychology, 32, 197–227. Ross, L., & Nisbett, R. E. (1991). The person and the situation: Perspectives of social psychology. New York: McGraw-Hill. Ross, M. (1989). Relation of implicit theories to the construction of personal histories. Psychological Review, 96, 341–357. Ross, M., & Newby-Clark, I. R. (1998). Construing the past and future. Social Cognition, 16, 133–150. Sanna, L. J. (1996). Defensive pessimism, optimism, and stimulating alternatives: Some ups and downs of prefactual and counterfactual thinking. Journal of Personality and Social Psychology, 71, 1020–1036. Schkade, D. A., & Kahneman, D. (1998). Does living in California make people happy? A focusing illusion in judgments of life satisfaction. Psychological Science, 9, 340–346. Schmidt, N. B., Jacquin, K., & Telch, M. J. (1994). The overprediction of fear and panic in panic disorder. Behavior Research and Therapy, 32, 701–707. Schulz, R. (1976). Effects of control and predictability on the physical and psychological wellbeing of the institutionalized aged. Journal of Personality and Social Psychology, 33, 563–573. Seligman, M. E. P. (1975). Helplessness: On depression, development, and death. San Francisco: Freeman. Sherman, S. J. (1980). On the self-erasing nature of errors of prediction. Journal of Personality and Social Psychology, 39, 211–221. Sieff, E. M., Dawes, R. M., & Loewenstein, G. (1999). Anticipated versus actual reaction to HIV test results. American Journal of Psychology, 112, 297–311. Simon, D., Pham, L. B., Le, Q. A., & Holyoak, K. J. (2001). The emergence of coherence over the course of decision making. Journal of Experimental Psychology: Learning, Memory, and Cognition, 27, 1250–1260.

410

WILSON AND GILBERT

Simonson, I. (1990). The effect of purchase quantity and timing on variety-seeking behavior. Journal of Marketing Research, 27, 150–162. Smith, A. (1853). The theory of moral sentiments. London: H. G. Bohn (originally published 1759). Snell, J., Gibbs, B. J., & Varey, C. (1995). Intuitive hedonics: Consumer beliefs about the dynamics of liking. Journal of Consumer Psychology, 4, 33–60. Solomon, R. L. (1980). The opponent-process theory of acquired motivation. American Psychologist, 35, 691–712. Spencer, K. M., Dien, J., & Donchin, E. (2001). Spatiotemporal analysis of the late ERP responses to deviant stimuli. Psychophysiology, 38, 343–358. Steele, C. M. (1988). The psychology of self affirmation: Sustaining the integrity of self. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 21, pp. 261–302). New York: Academic Press. Tatarkiewicz, W. (1976). Analysis of happiness. E. Rothert & D. Zielin´ ska, Trans. Warszawa: PWN/Polish Scientific Publishers (original work published 1962). Taylor, S. E. (1989). Positive illusions. New York: Basic Books. Taylor, S. E., & Brown, J. D. (1988). Illusion and well-being: A social psychological perspective on mental health. Psychological Bulletin, 103, 193–210. Taylor, S. E., & Pham, L. B. (1996). Mental simulation, motivation, and action. In P. M. Gollwitzer & J. A. Bargh (Eds.), The psychology of action: Linking cognition and motivation to behavior (pp. 219–235). New York: Guilford. Tesser, A. (2000). On the confluence of self-esteem maintenance mechanisms. Personality and Social Psychology Review, 4, 290–299. Thompson, S. C., Armstrong, W., & Thomas, C. (1998). Illusions of control, underestimations, and accuracy: A control heuristic explanation. Psychological Bulletin, 123, 143–161. Trope, Y., & Liberman, N. (in press). Temporal construal. Psychological Review. Tversky, A. (1972). Elimination by aspects: A theory of choice. Psychological Review, 79, 281–299. Tversky, A., & Kahneman, D. (1974). Judgment under uncertainty: Heuristics and biases. Science, 185, 1124–1131. Vaillant, G. (2000). Adaptive mental mechanisms: Their role in positive psychology. American Psychologist, 55, 89–98. van Hout, W. J. P. J., & Emmelkamp, P. M. G. (1994). Overprediction of fear in panic disorder patients with agoraphobia: Does the mismatch model generalize to exposure in vivo therapy?. Behavioral Research and Therapy, 32, 723–734. Varey, C., & Kahneman, D. (1992). Experiences extended across time: Evaluation of moments and episodes. Journal of Behavioral Decision Making, 5, 169–185. Vinogradova, O. S. (2001). Hippocampus as comparator: Role of the two input and two output systems of the hippocampus in selection and registration of information. Hippocampus, 11, 578–598. Wasserman, D., Lempert, R. O., & Hastie, R. (1991). Hindsight and causality. Personality and Social Psychology Bulletin, 17, 30–35. Watson, D., Clark, L. A., & Tellegen, A. (1988). Development and validation of brief measures of positive and negative affect: The PANAS scales. Journal of Personality and Social Psychology, 54, 1063–1070. Wechsler, B. (1992). Stereotypes and attentiveness to novel stimuli: A test in polar bears. Applied Animal Behaviour Science, 33, 381–388. Wegener, D. T., & Petty, R. E. (1997). The flexible correction model: The role of naive theories of bias in bias correction. In M. P. Zanna (Ed.), Advances in experimental social psychology (Vol. 29, pp. 141–207). San Diego, CA: Academic Press.

AFFECTIVE FORECASTING

411

Weinstein, N. D. (1980). Unrealistic optimism about future life events. Journal of Personality and Social Psychology, 39, 806–820. Wilson, T. D. (2002). Strangers to ourselves: Discovering the adaptive unconscious. Cambridge, MA: Harvard University Press. Wilson, T. D., & Brekke, N. C. (1994). Mental contamination and mental correction: Unwanted influences on judgments and evaluations. Psychological Bulletin, 116, 117–142. Wilson, T. D., Centerbar, D. B., & Gilbert, D. T. (2002). [Positive social feedback and the pleasures of uncertainty]. Unpublished raw data, University of Virginia. Wilson, T. D., Gilbert, D. T., & Centerbar, D. B. (in press). Making sense: The causes of emotional evanescence. In J. Carrillo & I. Brocas (Eds.), Economics and psychology. New York: Oxford University Press. Wilson, T. D., Gilbert, D. T., & Salthouse, T. (2001). [Predicted emotional reactions across the adult life span.] Unpublished raw data, University of Virginia. Wilson, T. D., Houston, C. E., & Meyers, J. M. (1998). Choose your poison: Effects of lay beliefs about mental processes on attitude change. Social Cognition, 16, 114–132. Wilson, T. D., Kermer, D., & Gilbert, D. T. (2002). [Random acts of kindness and the pleasures of uncertainty]. Unpublished raw data, University of Virginia. Wilson, T. D., & Klaaren, K. (1992). The role of affective expectations in affective experience. In M. S. Clark (Ed.), Review of personality and social psychology: Emotion and social behavior (Vol. 14, pp. 1–31). Newbury Park, CA: Sage. Wilson, T. D., Kurtz, J., & Gilbert, D. T. (2002). [The pleasure of uncertainty over positive events]. Unpublished raw data, University of Virginia.. Wilson, T. D., & LaFleur, S. J. (1995). Knowing what you’ll do: Effects of analyzing reasons on self-prediction. Journal of Personality and Social Psychology, 68, 21–35. Wilson, T. D., Laser, P. S., & Stone, J. I. (1982). Judging the predictors of one’s own mood: Accuracy and the use of shared theories. Journal of Experimental Social Psychology, 18, 537–556. Wilson, T. D., Lisle, D. J., Kraft, D., & Wetzel, C. G. (1989). Preferences as expectation-driven inferences: Effects of affective expectations on affective experience. Journal of Personality and Social Psychology, 56, 519–530. Wilson, T. D., Meyers, J., & Gilbert, D. T. (2002). Lessons from the past: Do people learn from experience that emotional reactions are short lived? Personality and Social Psychology Bulletin, 27, 1648–1661. Wilson, T. D., & Stone, J. I. (1985). Limitations on self-knowledge: More on telling more than we can know. In P. Shaver (Ed.), Review of personality and social psychology (Vol. 6, pp. 167–183). Beverly Hills, CA: Sage. Wilson, T. D., Wheatley, T. P., Kurtz, J., Dunn, E., & Gilbert, D. T. (2002). Ready to fire: Preemptive rationalization versus rapid reconstrual after positive and negative outcomes Unpublished manuscript, University of Virginia. Wilson, T. D., Wheatley, T. P., Meyers, J. M., Gilbert, D. T., & Axsom, D. (2000). Focalism: A Source of durability bias in affective forecasting. Journal of Personality and Social Psychology, 78, 821–836. Woodzicka, J. A., & LaFrance, M. (2001). Real versus imagined gender harassment. Journal of Social Issues, 57, 15–30. Zeelenberg, M., van Dijk, W. W., Manstead, A. S. R., & van der Pligt, J. (2000). On bad decisions and disconfirmed expectancies: The psychology of regret and disappointment. Cognition and Emotion, 14, 521–541. Zuckerman, M., & Lubin, B. (1965). Multiple Affect Adjective Check List. San Diego, CA: Educational and Industrial Testing Service.

INDEX

A Abandonment avoiding through sexual behavior, 124 suppression of thoughts related to, 85 Academic Motivation Scale, 310 Accuracy mindset, 266 Achievement, requiring both high locomotion and high assessment orientation, 313, 332–334 Achievement orientation, 317–318 Action/adventure films, preferred by high locomotors, 301 Action-Control Scale, 308, 314 Action-Decision subscale, 308, 314 Action set, of perceiver in task-oriented interactions, 166 Action vs. state orientations, 313–314 Activating triggers, in attachment system, 58 Active influence, based on prototypicality of leader, 10–11 Activity affordances, 334–336 Activity engagement and performance, 329–337 Activity flow, and locomotion/assessment orientation, 208–313 Activity involvement, correlation with high locomotion orientation, 311 Adolescent sexuality studies, 123–124 Adult attachment bonds, 137–138 Adult attachment styles, 88–126 Advice to negotiators, 238 Advisory strategic influence, and assessment orientation, 323 Affective forecasting correction for unique influences in, 360–362 defined, 346 errors in, 346–353 413

expectation effects in, 362–364 andfailuretolearnfromexperience,393–396 focalism in, 366–369 framing effects in, 356–358 hot/cold intrapersonal empathy gaps in, 364–366 illusion of external agency in, 384–386 impact bias and, 393–396 and keeping options open, 386–388 ordinization neglect in, 374–380 and pleasures of uncertainty, 391–392 process of, 354–384 and the psychological immune system, 380–383 recall and affective theories in, 358–360 sources of error in, 354–384 types of, 346–353 undermined by sense making processes, 369–380 valence predictions, 401 Affective responses, effect of attachment styles on, 114–115 Affectometer, 352 Affiliation, need for, 251 Affiliation goals, 124 Affiliation system, 124–125 Age, and length of emotional impact, 397–399 Agreeableness, 251 Agreement, quality of, 242 Ainsworth, Mary, 54 Alertness, in stigmatized group members, 174 Allocative behavior, of leaders, 26–27 Ambiguity, discomfort with, in high assessors, 306 Anchor information, irrelevant, 244 Anchor values, in negotiation, 265 Anchoring-and-adjustment, 244–245, 264, 266–267

414

INDEX

Anger experience and management of, 106–108 of mothers of avoidant infants, 91 Antisocial behavior, associated with Machiavellianism, 251 Anxiety and attachment styles, 115 cognitive consequences of, in behavioral confirmation, 173 in social interactions, and assessment orientation, 302–303 Anxious attachment behaviors, 55, 66 cognitive accessibility of attachment themes/figures, 90 and death anxiety, 102 description of parents by persons with, 92 experience and management of anger in, 107 hyperactivating strategies, 77 in infants, 67 and loss of social identity after death, 103 in mothers of infants with congenital heart defects, 101 and rumination, 102 self-disclosure and, 98 Appointed leaders, 19–20 Archetypes, of disputants, 237, 269–272, 279 Aspiration, level of, 263–264 Assessment orientation, 296–298 and advisory strategic influence, 323 and anxiety in social interactions, 303 correlation with achievement, 313 correlation with conscientiousness, 309 correlation with emotional instability, 306 correlation with first- and only-borns, 305 correlation with number of decision alternatives, 326 correlation with performance goal orientation, 307 correlation with sense of personal failure, 305 correlation with stress load, 304 and discomfort with ambiguity, 306 discriminating from locomotion, 299–313 and fear of failure, 317–318 and full evaluation strategy, 320 and heterogeneity of judgment, 327 and Ideal/Ought self-guides, 305 as independent mode from locomotion, 294–298 in Japan and South Korea, 338–339

negative correlation with decisiveness, 309 negative correlation with learning goal orientation, 307–308 negative correlation with optimism, 304 negative correlation with self-esteem, 304 and self-consciousness, 302 semantic differential scales for, 303 and shyness, 302 situational affordances for, 335 Assessment set, of perceiver in gettingacquainted interactions, 165 Asymmetry between confirmation and disconfirmation, 201–203 between perceiver and target levels of social categorization, 179–180 Attachment anxiety, 68 and attitudes toward task-oriented groups, 115 and attitudes toward work, 119 cognitive recall of attachment themes/figures, 90 and compulsive caregiving, 122 and family cohesion/flexibility, 118 and hostile reactions to outgroups, 113 and need for social acceptance/approval at work, 119 and pursuit of attachment vs. affiliation goals, 125 and scores on Rosenberg self-esteem scale, 109 and self-devaluation, 110 sexual behavior and, 124 Attachment avoidance, 68. See also Attachment styles; Avoidant behaviors Attachment behavioral system, 56–61 activation of, 73–76 and affiliation system, 124–125 analogies to flight-or-fight reactions, 63 attachment strategies in, 62–63 attachment styles in, 66–71 biological function of, 57–58 brain regions activated in, 128 cognitive and neural substrates of, 126–128 cognitive substrate of, 59–60 empirical assessments of, 88–126 and exploration system, 119–121 individual differences in, 61–71 mental representations in, 63–66

415

INDEX and other behavioral systems, 60–61, 119–125 set-goal of, 59 and sexual system, 123–125 Attachment categories, 66 Attachment dynamics, systemic model of, 136–137 Attachment-figure availability, 55, 63, 128–129 blocking monitoring of, 84 exaggerated monitoring of, by anxious individuals, 96 monitoring of, 73 and security-based strategies, 77–80 and self-regulation, 131 studies in, 92–95 Attachment figures, 57 heightened access to thoughts of, in anxious individuals, 90 hierarchy of, 138 interactions with, 61–62 lack of concern with, in avoidant individuals, 91 measuring accessibility of representations of, 89 psychotherapists functioning as, 140 unavailability of, 83 Attachment-related psychodynamics, 55 Attachment-related threats, management of, 104–106 Attachment security, 69 broaden and build cycle of, 77 Attachment strategies, 64 interpersonal and intrapersonal manifestations of, 95–109 and relationship quality, 117–119 secondary, 62–63 Attachment styles, 66–71, 88 Attachment-system activation, 88–92 Attachment-system dynamics, in adulthood, 71–88 Attachment theory, 55 and issues of development, stability, and change, 138–141 normative components of, 56–61 Attention, correlation with high locomotors, 309 Attentional control, 309 Attribution error, 12 Attribution processing, 2, 10, 11–14 Autocratic leadership style, 322–324

Autonomic nervous system, activation in attachment system, 127 Autonomy, 129–132 Availability, trust in, 78 Avoidant behaviors, 55, 66 in behavioral confirmation, 160–161, 215 compulsive closure in persons with, 97 deactivating strategies of, 77 descriptions of parents by persons with, 92 and detachment from caregiving, 122 dismissing attachment and affiliation goals in friendships, 125 and distancing coping responses to stress, 116 due to maternal anger, 91 experience and management of anger in, 107 and family cohesion/flexibility, 118 and fear of the unknown after death, 103 fearful, 134–136 frontal lobe association with withdrawal tendencies in, 128 and group-specific attachment avoidance, 95 in Hindu-Muslim interactions in India, 169 and hostile reactions to outgroups, 113 in infants, 66, 91, 134 lack of attachment-related worries in, 90–91 and mate poaching, 124 and one-night stands, 123 and reactions to task-oriented groups, 115 reduced access to emotional memories, 109 and relationship exclusivity, 124 and self-inflation, 110 short-circuiting expression of anger, 91 unmoderated by partner availability, 94 B Behavior generation, 57 Behavior over time task, 320 Behavioral confirmation, 154 affect on stereotypes and intergroup relations, 197–210 avoidance behaviors in, 160–161 confirmatory strategy in, 159–160 defined, 153 dominance in, 161–162 and intergroup anxiety, 172–175 and nondeferential behaviors, 191 and perceiver prejudice, 170–172

416

INDEX

Behavioral confirmation (continued) processes involved in, 158–160 reciprocation strategy in, 158–159 as a result of power differences, 194 self- and other-categorization in, 163–170 and stereotype content, 175–176 and stigmatization, 157 Behavioral decision theory, 276 Behavioral disconfirmation, 157 Behavioral expectations, 154 Behavioral inhibition, 86 Behavioral intentions, 303 Behavioral leeway, for leaders, 28 Behavioral system, defined, 56–57 Benevolence, in parents of persons with secure attachment styles, 93 Biases and heuristics based on social value orientation, 257 due to cognitive limitations, 244–245, 264–265 fixed-pie thinking, 243 in judgment and decision making, 243 undermining constructive negotiation processes, 243 Bluffs, as negotiation tactic, 254 Bond, James, as high locomotor, 298 Boundaries, permeability, and stereotype maintenance/change, 211, 217 Boundary-heightening, 207 Bowlby, John, 4, 53 Boys camp studies, 169 Broaden and build cycle, of attachment security, 77 C Caregivers as attachment figures, 57 of avoidant infants, 133–134 during infancy, 58 mental representations of, in attachment style studies, 92–95 Caregiving system, 121–123 Categorization, of ingroup vs. outgroup members, 205 Category salience, 202, 204, 215, 216 Change leading, as function of group leadership, 39–40

openness to, and locomotion/assessment orientation, 311–313 orientation for, 295–296 positive expectancy about, 312 resistance to, 40 of state, and locomotion, 296 Charismatic leadership, 3–4, 12, 29–31 Child development, and negotiation, 235 Chronic back pain, coping with stress of, 100 Churchill, Sir Winston, 260, 280 Clinging behavior, 83 Close relationships and attachment styles, 96–99 avoidance through work, by avoidant individuals, 119 search for, by anxious individuals, 96 Coercion, 28, 321 in evolution of leader power, 39 as negotiation tactic, 254 Cognition in conflict and negotiation, 243–249 need for, 263 Cognitive biases, in attachment strategies, 78 Cognitive closure, and time pressure, 264 Cognitive dissonance, in negotiation, 236 Cognitive limitations anchoring-and-adjustment, 243 bias due to, 244–245, 264–265 causing erroneous reasoning in negotiations, 243 Cognitive neuroscience, 54 Cognitive processes, and negotiation outcomes, 236, 255–256 Collective self, 4 Collective strategies, of stereotype change, 187–192, 192–193, 216 Collectivist culture, vs. individualistic culture, 256 Combat training exercise, coping with stress of, 100 Comedy films, preferred by high locomotors, 301 Commitment, of leader to group, 22–24 Community development, and negotiation, 235 Comparing process, and assessment orientation, 297 Compensation strategy in behavioral confirmation, 158–159 by targets in stigmatized groups, 177

417

INDEX Competence, expectations of, in task-oriented interactions, 177 Competition and cooperation, in negotiation, 236, 253 and naive realism in negotiation, 248–249 and selfish motivation, 250 Competitive incentive, in negotiation, 238 Compliance, with group leader’s suggestions, 11 Compliance-based influence tactics, 28 Compulsive closure, in persons with avoidant attachment style, 97 Compulsive self-reliance, 63, 69 Concession making, resistance to, 245, 254, 277 Conduct disorders, and attachment styles, 15 Confirmatory construal, 243, 246–247, 259 Confirmatory strategy, 159–160, 215 Conflict cognition and motivation in, 243–249 expectations of, reducing problem-solving abilities, 258 issues, outcomes, and tasks in, 237–242 Conflict escalation, 83 Conflict management and attachment styles, 96–99 in ingroup-outgroup relations, 191 Conformity, 6, 9 Conformity-based influence tactics, 28 Congenital heart defects (CHD), coping styles by mothers of infants with, 100–101 Connectedness, intense search for, by anxious individuals, 112 Conscientiousness, correlation with high assessment, 309 Consensual liking, 11 Consensual prototypicality, 38 Consensus bias, false, 112 Constituents, accountability to, 254 Construal errors, in affective forecasting, 354–356 Contact comfort, 59 Contextual activating triggers, 57 Contingency theories, of leadership, 3 Control behaviors, 85 and avoidantpersons’ responsesto threats of death, 104 as chief goal of avoidant individuals in relationships, 98 dominance in behavioral confirmation, 161–162

Control orientation, 251 Control theory, 293 Cooperation and competition, in negotiation, 236, 253–254 with elected leaders, 20 influence of egoistic and prosocial motives on, 250 and prosocial motives, 250 Cooperative incentive, in negotiation, 239 Cooperative misers, 270 Coping strategies attachment styles and stressful events, 99–102 constructive, 79 emotion-focused, 79 in securely attached individuals, 116 Counterphobic activities, 87 Counterstereotypic behavior, 202 Critical appraisal, and assessment orientation, 297 Criticism, sensitivity to, 302 D Dating game experiment, 389–390 Deactivating strategies, 69, 84–87, 86, 87, 132–134 of avoidant persons, 77 and defensive, self-enhancing attributions, 111 and detachment from caregiving, 122 in infants of angry mothers, 91 in romantic relationships, 96–99 self-disclosure and, 97 used postemptively to repress and forget, 105 used preemptively for separation-related thoughts, 105 Deactivation, 64 of attachment system, 62–63 Death anxiety coping strategies in anxious individuals, 102 and increased judgments/punishment strategies of anxious or avoidant persons, 103–104 suppressed in avoidant individuals, 103 Deceleration rate, of emotional intensity, 350 Decision alternatives, correlation between number and assessment orientation, 326

418

INDEX

Decision making and judgment correlation with regulatory modes, 319–328 expectancy vs. value emphasis, 324–326 Decision-Related Action vs. State Orientation subscale, 308 Decisiveness, correlated with high locomotors, 308–309 Decomposed game methodology, 251, 269 Defensive projection, 86 by avoidant individuals, 112 Deferential behavior, by target in behavioral confirmation, 159–160 Deliberation vs. implementation orientation, 314–316 Democratic leadership style, 322–324 Dependence, 83 Depersonalization, in stereotype maintenance and change, 203 Depersonalized social attraction, 10–11 in high-salience groups, 31–32 Depression, and attachment styles, 115 Detachment, emotional, 85 Dimensions of value in negotiation, 238 in single-issue tasks, 239 Directive leadership style, 322–324 Disabled individuals, avoidance behaviors in interactions with, 161 Disclosure anxious persons unresponsive to partner’s, 98 and attachment styles, 96–98 Discomfort With Ambiguity subscale, 306 Disconfirming behavior, 181 failure by perceivers to acknowledge, 199–201 requirements of, 187–188 and stereotypes, 189, 210 vs. stereotype change, 213 Discrimination, 156 sought by stigmatized group members, 211–212 Dismissing avoidants, 70 Disorganized/disoriented attachment behaviors, 66 in infants, 67 Dispositional attributions, 12 Disputant archetypes, 237, 269–272, 279 Dispute resolution. See Negotiation Dissociated anger, 107

Distance, pursuit of interpersonal, 85 after trust violation, in avoidant individuals, 99 in behavioral confirmation, 160–163 as coping strategy in avoidant individuals, 116 in secure mothers of infants with congenital heart defects, 101 and stress coping behaviors in avoidant individuals, 100 Distancing coping, 85 Distinctiveness bias, false, 112 Distress management negative beliefs about, 81 optimism about, 78 Distributive fairness as motivator in negotiation, 242 of outcomes, 249 Distributive issues, 241 Distributively fair leaders, 24 Distrust, 83 Divorce, coping with stress of, 100 Dominance in behavioral confirmation, 161–162, 215 as indication of high task competence, 196 by men, in task-oriented interactions with women, 168 Dramatic films, preferred by high assessors, 301 Dual concern theory, 250, 254, 255, 276 Durability bias, 349–350 Duration of emotions forecasting, 346 predicting, 349–351 Dyadic Adjustment Scale, 118 Dyadic relationships, leaders and individual followers, 5 Dyadic self-fulfilling prophecies, 198 E Eating disorders, and attachment styles, 115 Economic models, of bargaining, 250, 253 Effort investment, 334–336 Ego defensiveness, 243, 248–249, 256–260 as barrier to dispute resolution, 236–237, 247–248 moderated by social motivation, 256–257, 259 in motivated information processing model, 268

419

INDEX Egocentric self-assessments, 247 Egoistic motivation, 249 in negotiation, 242 Elected leaders, 19–20 Emotion-focused coping, 79, 81–82 in mothers of infants with congenital heart defects, 101 as stress response in anxious individuals, 116 Emotional distance, 69 in caregiving situations, 122 Emotional evanescence ability to learn from, 397–399 undermining affective forecasting, 369–370 Emotional instability, correlation with assessment orientation, 306 Emotional memories, cognitive activation and architecture of, 108–109 Empathy, as prosocial motivation, 250 Empowerment, through group support, 191 Endowment effect, 259 Environment development, and negotiation, 235 Epistemic motivation as aid to reducing fixed-pie perceptions, 265–266 defined, 262 due to individual differences or situational cues, 266 and need for cognitive closure, 264 in negotiation, 236, 237 origins of, 262–264 reduced by time pressure, 264 Errors, in affective forecasting, 346–353 Escapism, dealing with anger via, 107 Essentialism, 12 Ethnocentrism, 6 of leader, 11 less prevalent in securely attached individuals, 120 Evaluation and assessment orientation, 297 need for, 316–317 Evoked reaction potential (ERP), 127–128 Excitatory links, 64 Expansion of the self, 131 Expectation-congruent behaviors, 200 Expectation effects, in affective forecasting, 362–364 Expectation-incongruent information, 120

Expectations, of target in getting-acquainted interactions, 176 in task-oriented interactions, 177 Expert power, 321–324 Exploitative skills, 252 Exploration system, 119–121 External agency controlling negative emotional reactions via belief in, 402 illusion of, in affective forecasting, 384–386 Extrinsic vs. intrinsic flow, and locomotion/ assessment orientation, 311–312 Extroversion, correlation with locomotion orientation, 318 Eye contact reduced by target in dominant interactions, 162 reduced in avoidance interaction, 160–161 F Failure correlation with assessment orientation, 305, 308 fear of, and assessment orientation, 317–318 Family cohesion measurements, 118 Family flexibility, 118 Fatigue, 265 influencing epistemic motivation in negotiation, 263 Fearful avoidance, 70, 134–136 and childhood physical or sexual abuse, 139 Felt security, 59 Fight or flight mechanism, analogies to attachment system, 63 Financial success, as extrinsic goal in motivation, 251 Firstborns, and assessment orientation, 305 Fixed-pie assumptions, 243, 246, 253, 256, 265–266 Flow, and locomotion/assessment orientation, 309, 324–326 Focalism, 401 in affective forecasting, 366–369 and retrospective impact bias, 394 Framing effects and errors in affective forecasting, 356–357 types of forecasting errors caused by, 358 Freud, Sigmund, 53 Frustration, chronic, 83

420

INDEX

Full evaluation strategy, and assessment orientation, 320 Functional Impulsivity Scale, 308 Functional magnetic resonance imaging (MRI), 127–128 Functionally equivalent behaviors, 57 Future events, misconstruing emotional reactions to, 354–356

Groups high-salience, 31 low-salience, 41 Guiding hand, and affective forecasting, 384 Gulf War, coping with stressful events in, 99–100 H

G Game of agreement, 241 Game of coordination, 241 Gender stereotypicality, 35 Getting-acquainted interactions, 165–170 Getting-along behavior, 177, 211–213 Getting started, correlation with locomotion vs. assessment orientation, 329–331 Glass ceiling, 34–36, 42–43 Goal Orientation Scale, 307 Goal-oriented behavior, action phases in, 314–315 Goal pursuit correlation with high locomotion orientation, 310–311 and locomotion, 296 preferred orientation to, 299 Goal setting, vs. goal striving, 315 Good will negative beliefs about, 81 optimistic beliefs about, 81 Group cohesiveness, 39 and anxious persons’ reaction to mortality salience, 104 empowering individuals, 191 and instrumental functioning, 94–95 Group identification, 23 and group member identification, 26–27 and leader allocation behavior, 26–27 Group membership salience, 14–18 Group perceptions and attitudes, maintenance and change in, 203–209 Group processes, 4 Group prototypicality, 17–18 Group representativeness, 20 Group salience, and leader prototypicality, 30–31 Group-specific attachment anxiety, 94–95 Group stereotypes, 179

Hard influence tactics, 28 Hawthorne, Nathaniel, 374 Helplessness, in adulthood, 55 Heuristics and biases anchoring-and-adjustment, 244–245 availability heuristic, 244 in judgment and decision making, 243 in motivated information processing model, 268 representativeness heuristic, 244 High assessors, 297. See also Assessment orientation discomfort with ambiguity in, 306 and high locomotors, 298 High locomotors, 295 and high assessment orientation, 298 negative correlation with performance goal orientation, 306–307 preference for action/adventure films, 301 High-salience groups, 31–32 Hindu-Muslim interactions, 169 Hobbes, Thomas, 249 Holmes, Sherlock, as high assessor, 298 Hopelessness-depressive pattern of attributions, 111 Hostility, and attachment styles, 115 Hot/cold intrapersonal empathy gaps, in affective forecasting, 364–366 House Committee on Un-American Activities, 38 Hyperactivating strategies, 80–84, 82, 132–134 and hopelessness-depressive pattern of attributions, 111 of anxious persons, 77 and compulsive caregiving, 122 self-disclosure and, 98 Hyperactivation, 64 of attachment system, 62–63 Hypervigilance, 83

INDEX I Ideal self-guides, and assessment orientation, 304 Immune neglect, 381 Impact bias, 351, 353 functionality of, 399–401 and learning from experience, in affective forecasting, 393–396 over human lifespan, 397–399 Impression motivation, 274–276 Independence, parental fostering of, and attachment style, 93 Individual differences, as factor in negotiation outcomes, 272 Individual mobility, 193 Individual ownership effect, 259 Individuating information, vs. stereotyping, 173 Infants, attachment styles of, 66–67 Influence, vs. power, 38–39 Influence tactics, hard vs. soft, 28 Information exchange, enhanced by prosocial motivation, 269 Information-gathering strategy, defined, 153 Information processing, 10, 11–14 influenced by social motives, 258–259 Ingroup favoritism, 6, 24, 27 Ingroup leaders, 18–19 Ingroup loyalty, 11 by leaders, 21 Ingroup-outgroup relations, 191 categorization and, 205 in workplace settings, 206–207 Inhibitory links, 64 Instrumental functioning, and group cohesion, 94–95 Insults, emotional effects of minor vs. major, 388–389 Integrative agreements, 240 higher for prosocial than selfish negotiators, 253 native realism interfering with, 246 Integrative bargaining, 241 Integrative model, of attachment system dynamics, 71–72 Intensity of emotions forecasting, 346 predicting, 349–351 Interaction bids, 58–59

421

Interaction distance, large in avoidance behaviors, 160–161 Interaction strategies, individual vs. collective, 192–193 Interaction styles avoidance, 160–161 dominance, 161–162 Interactions, structured vs. unstructured, 215–216 Interactive strategies, consequences of prejudice for, 171 Interdependence, sense of, 263 Intergroup anxiety, 169, 215 in behavioral confirmation, 172–175 Intergroup discrimination, 6 Intergroup relations, 4 behavioral confirmation and disconfirmation effects on, 197–210 discrimination in, 6 large-scale, 2 motivation in, 7 positive self-esteem in, 7 prototypes in, 5 Intergroup stereotypes, persistence of, 156 International relations, and negotiation, 236 Interpersonal adjustment, and stereotype confirmation, 193, 211–213 Interpersonal conflicts catastrophic appraisal of, 83 management of, according to attachment style, 96–99 Interpersonal relationships, between leader and follower, 13–14 Interpretation, and assessment orientation, 297 Interruption in avoidance behaviors, 160–161 by men, in task-oriented interactions with women, 168 Intersubjectivity, 53 Intimacy related to trust, 98 seeking, in response to fear of death, 103 and self-disclosure styles, 97 Intrinsic vs. extrinsic flow, and locomotion/ assessment orientation, 310–311 Irrevocable choices, 387 Isolation effect, 356–358

422

INDEX J

Jealousy, in persons with anxious attachment styles, 96 Job Involvement Scale, 311 Joint gain, as motive in prosocial negotiators, 249 Judgment and decision making correlation with regulatory modes, 319–328 heterogeneity vs. homogeneity, 326–328 L Laboratory tasks, in study of negotiation, 241 Leader behavior allocative, 26–27 commitment to group, 20–21, 22–24 fairness, 24 favoring ingroup members, 24 group-oriented, 20–25 group-serving, 11 influence tactics, 27–29 interpersonal aspects of, 14 leading change, 39–40 normative, 11 outgroup conflicts among insecure leaders, 37–38 as predictor of leadership effectiveness, 2 procedural fairness, 22–24 prototypicality as component of, 12, 14–18 punishment-oriented, 22–23 relationship oriented, 25 self-sacrificial, 22–24 Leader categorization theory, 3 Leader effectiveness, and self-sacrifice for group, 21 Leader-follower relations, 31–32 Leader influence tactics, 27–29 Leader-member exchange (LMX), 4, 29, 42 and prototypicality, 31–32 Leader schemas, 29 congruence manipulations, 22 congruence of, 14–18, 29 general and task-specific, 3 Leader-subordinate relations, 28 Leadership appointed vs. elected, 19–20 charismatic and transformational, 3–4 contingency theories of, 3

glass ceiling obstacles to advancement of, 34–36 historical review of research in, 2–6 ingroup vs. outgroup, 18–19 as relational term, 1 social identity analysis of, 14–18, 30–31 Leadership effectiveness determinedby leader-follower relationships, 13–14 perceived, 16 and prototypicality, 25–29 Leadership endorsement, and demographic mismatches, 34–36 Leadership research, future directions for, 33–40 Leadership styles, 322–324 Learning from experience, in affective forecasting, 393–396 Learning goal orientation, and high locomotors, 307–308 Legitimate power, 195, 321–324 Leviathan, The, 249 Lexical decision task evaluating attachment style, 89 evaluating trust associations, 98–99 Life Orientation Test Revised, 304 Likert scales, 327 Linguistic divergence, 191 Locomotion and assessment scales, 300 Locomotion and coalition games, 242 Locomotion orientation correlation with achievement, 313, 317–318 correlation with activity involvement, 311 correlation with extroversion, 318 correlation with goal pursuit, 310–311 correlation with high attention, 309 correlation with high decisiveness, 308–309 correlation with high persistence, 309 correlation with optimism, 324–326 correlation with time commitment, 311 correlation with Type A behavior, 317 defined, 293, 295–296 discriminating from assessment orientation, 299–313 and feelings of personal success, 305 in Germany and U.S., 338–339 and homogeneity of decision making, 327 as independent from assessment orientation, 294–298

423

INDEX inversely related to shyness and self-consciousness, 302 negative correlation with performance goal orientation, 307–308 and progressive elimination strategy, 319–321 semantic differential scales in measuring, 303 situational affordances of, 335 in Spain and Italy, 338 Loneliness in adolescents and adults, 55 and attachment styles, 115 Loss, suppression of thoughts related to, 85 Loss aversion, in negotiation, 245 Low assessors, 297 Low intelligence, perceptions of, in behavioral confirmation studies, 162 Low locomotors, 295 M Machiavellians, 251 Marginal leaders, 26 Marital satisfaction, negotiation and, 235 Mate poaching behavior, 124 McCarthy, Joe, 38 Measuring and affective forecasting, 352–354 and assessment orientation, 297 Memory retrieval time, 108–109 poor for past emotional reactions, 393 Mental contamination, 360–361 Mental health and adjustment, studies on attachment styles and, 115–117 Mental representations, 63–66 Met stereotype, 182 clear activation of, as prerequisite for compensation, 187 Minorities, glass ceiling obstacles to advancement of, 34–36 Misconstrual problem, 354–356 types of errors caused by, 356 Mixed-motive interaction, 236 Monitoring, as feedback in assessment, 297 Moods, effect of attachment styles on, 14–115 Motivated information processing model, 268–274 of negotiation, 237

Motivational goals, 243–249 in negotiation, 242 shifts in, 273–274 Multiple Affect Adjective Check List, 352 N Naive realism, 243, 248–249, 255–256, 265–266 as barrier to dispute resolution, 236–237, 246–247 in motivated information processing model, 268 Need for Cognitive Closure Scale, 306–307, 308–309 Need to Evaluate Scale, 316–317 Negative cognitions, 82, 83 Negative emotions cognitive consequences of, 107–108 in persons with anxious attachment styles, 96 Negative events defanging, 402 making sense of, and the psychological immune system, 380–383, 395 books providing advice on, 238 cognition and motivation in, 243–249 Negotiation issues, outcomes, and tasks in, 237–242 reward structure in, 237–238 in social conflict, 235 Negotiation outcomes cognitive processes and, 236 higher for prosocial than selfish negotiators, 253 NEO-Five Factor Inventory, 306, 308 Neuroendocrine responses, and attachment-system activation, 127 Noise, high level of, 264 Nondeferential behavior, 186 due to collective identity consciousness, 191 linguistic divergence as, 191 Nonprototypical leaders and charismatic attributes, 31 endorsed with group-oriented behavior, 27 Nonpunitiveness, in parents of persons with secure attachment style, 93 Nonstigmatized individuals, 160–163 Normative behavior, 6, 9

424

INDEX O

Object relations, 53 Occupational segregation, 36 One-night stands, 123 Only children, and assessment orientation, 305 Optimism about distress management, 78 correlation with locomotion orientation, 324–326 negative correlation with assessment orientation, 304 and overconfidence, in negotiation, 246, 256 Options, keeping open, and affective forecasting, 386–388 Ordinization neglect, 401 causing errors in affective forecasting, 374–380 Organizational psychology, 2–4 and negotiation, 235 and social identity research, 5 Other-categorization, in behavioral confirmation, 163–170 Other-concern, 254 Ought self-guides, and assessment orientation, 304 Outgroup leaders, 18–19 Outgroup stereotypes, 164 Outgroups demonizing, 37–38 hostile responses to, and attachment styles, 113 perceived as undifferentiated and homogeneous, 202 Overdependence, on a partner, 82 P Parental behavior, and development of attachment styles, 92–93 Participative leadership style, 322–324 Partisan mindset, 266 Partner behavior contextual availability and supportiveness, 94 perceptions by anxious and avoidant individuals, 111–113 perceptions by securely attached individuals, 111

relation between perception and attachment styles, 93 Path-Goal Leadership Questionnaire, 323 Perceived legitimacy, 192–193, 217 Perceiver-target dyad, 158 Perceiver’s perspective avoidant style, in getting-acquainted interactions, 167 in behavioral confirmation studies, 163–177 dominant style, in task-oriented interactions, 167 level of intergroup anxiety, 172–175 limitations of disconfirming behaviors on, 199–201 maintenance and change of stereotypes, 201–203 persistence of stereotypes in, 203 prejudice, in behavioral confirmation, 170–172 resistance to stereotype change, 207–209 Perceptual confirmation, 154, 213–214 necessity of dissociation from stigmatized group and, 200 Perceptual gulf, between leader and followers, 17 Perfectionism, in avoidant individuals, 110 Performancegoalorientation,inhighassessors, 307–308 Persistence, correlation with high locomotors, 309 Person-environment transactions, 60 Person perception studies, 111–113 Personal attraction, vs. social attraction, in groups, 13–14 Personal expectation confirmation, 155 Personal Fear of Invalidity Scale, 308 Personality disorders, and attachment styles, 115 Personality traits, in charismatic leadership, 30 Persuasion, 28 Persuasive arguments, 254 Physical abuse, and fearful avoidance behaviors, 139 Physical attraction, emphasized in anxious attachment behaviors, 96 Physical threats, 74 Poirot, Hercule, as high assessor, 298 Polarization, within groups, 9 Position-based power, 39 Positional commitments, 254

425

INDEX Positive and Negative Affect Schedules, 352 Positive self-image, as barrier in negotiation, 243, 247 Postdecisional dissonance reduction, 387 Posttraumatic stress disorder, 55 avoidance-related symptoms of, 86 Posture backward lean in avoidance interactions, 160–161 relaxed in dominant interactional styles, 162 Power, 216 attempts to exercise, in behavioral confirmation studies, 162 barriers to, and stereotype change, 193 and enactment of collective strategies of stereotype change, 208 as influence on epistemic motivation, 266 relative, 263 vs. influence, 38–39 Power claims, and behavioral confirmation, 160 Power differences, 194–197 impact on negotiation outcomes, 237 Power imbalance, in groups, 13 Power motivation, 274–276 Prediction, of future feelings, 345–346 Preemptive rationalization, 389–390 Pregnancy and birth, coping with stress of, 100 Prejudice, 36, 215 expectation of, by stigmatized individuals, 181 by perceiver, in behavioral confirmation, 170–172 Primacy effect, 120 Primary attachment strategy, 58–59 deactivation of, 85 Prisoner’s dilemma-type game, 242, 256–257 Proactive coping, 400 Problem-solving ability reduced by expectations of conflict, 258 enhanced by prosocial motivation, 269 and information exchange in negotiation, 254 and rationality, 243 withdrawal of efforts toward, 85 Procedural fairness as leader attribute, 21, 24 studies of, 22–24 Process accountability, 263

impact on negotiation outcomes, 270, 275 Productivity, enhanced by ingroup-favoring leadership, 24 Progressive elimination strategy, and locomotion orientation, 319 Projection, of negative self-views, 83 Projection bias, 360 Proofreading study, 331–332 Prosocial behavior, vs. Machiavellianism, 251 Prosocial misers, 237, 270 Prosocial motivation, 250 in negotiation, 242 Prosocial motives, 249 Prosocial thinkers, 237, 270 Prototypes enhancing attributions of leadership, 12 and group leader identity, 10 and increasing group salience, 42 in intergroup relations, 6 and leader ability to influence, 10–11 management of, 37–39 and social identity salience/social identification, 13–14 and social influence, 7–10 strategic management of, 39 Prototypicality and charismatic leadership, 30–31 as determinant of effective leadership, 29–33 and group-oriented behavior, 25–29 and leader-member exchange (LMX), 31–32 paradox of, 37–39 proxies for, 18–20 Proximity-seeking behavior, 59, 127 with death anxiety, 103–104 monitoring viability of, 73 and secondary attachment strategies, 80–87 and trust, in secure individuals, 98 Proximity themes, 89 Psychic numbing, 86 Psychological immune system, 380–383, 395 Psychological threats, 75 perception of, in behavioral confirmation, 173 R Rapid speech, 162 Rappaccini’s Daughter, 374 Rationality, and problem solving, 243

426

INDEX

Rationalization and recovery from negative events, 381–382 timing of, 389–391 Reactive devaluation, 248 in negotiations, 243 Recall and affective theories, in affective forecasting, 358–360 Reciprocation strategy in behavioral confirmation, 158–159, 215 in negotiation behaviors, 273 vs. defection, in negotiation, 252 Reference points, in negotiation, 245 Referent power, 321–324 Regulatory mode theory consequences of locomotion and assessment modes in, 318–336 influence of situational variables on mode, 299 vs. other orientations, 313–318 Regulation, definition, 294 Regulatory Mode Questionnaire, 299 Rejection reaction of secure individuals to issues of, 90 suppression of thoughts related to, 85 Relationship exclusivity, 124 Relationship Rating Form, 117–118 Relationship satisfaction measurements, 117–119 Relative power, 263 Relative values, in negotiation, 244 Representativeness heuristic, 244 Reverse discrimination, 36 Reward motivation, in stereotype enactment, 184 Reward structure, in negotiation, 237–238 Romantic attachment, 55, 68 studies of attachment styles in, 96–99 Rosenberg self-esteem scale, 109, 304 Rule-based inferences, in information processing, 261–262 Rumination in anxious individuals, following trust violations, 99 correlation with assessment orientation, 309 and hyperactivating strategies, 102 S Safe haven, 59 Satisfaction with Life Scale, 352

Schelling, Thomas, 236 Scholastic Aptitude Test, 332 Secondary attachment strategies, 80–87, 132–134 Secure attachment behavior, 66 and access to negative/positive emotional memories, 109 broaden and build cycle of, 77 core beliefs of, 78 description of parents of persons with, 92 and high family cohesion, 118 in infants, 66 and lack of projective mechanisms, 113 not occupied with attachment themes/ figures, 90 and proneness to caregiving, 121–122 and search for symbolic immortality, 103 Secure base, 59 Security attainment, 64 as chief goal of anxious attachment style, 98 Security-based strategies, 77 and attachment-figure availability, 77–80 and self-regulation and autonomy, 129–132 Segregated systems, 86 Selective attention, and affective forecasting, 371 Self-appraisal and attachment styles, 109–111 egoistic, in negotiation, 247 high self-esteem in avoidant individuals, 110 negative by anxious individuals, 110 vulnerability to, and assessment orientation, 304 Self-categorization theory (SCT), 13, 164, 184 and behavioral confirmation, 163–170 as ingroup member, 164 Self-conception, within groups, 6 Self-concern, 254 Self-consciousness in behavioral confirmation, 173 and social comparison, 302 Self-criticism, 82 in anxious individuals, 109–110 in avoidant individuals, 110 Self-disclosure and attachment styles, 97 low in avoidance interactions, 160–161 low in avoidant interactions, 160–161 Self-efficacy in dealing with threats, 78, 81, 84

INDEX heightened perceptions of, 85 Self-enhancement strategy, 186 Self-esteem high, in avoidant individuals, 86, 110 negative correlation with assessment orientation, 304 Self-evaluative concerns, 302–308 Self-fulfilling prophecy, in negotiation, 247 Self-guides, strong in persons with both high locomotion/high assessment, 305 Self-image, and ego defensiveness, as impediment to negotiation, 236 Self-preoccupation, 82 Self-projection, by anxious individuals, 112 Self-regulation, 129–132 defined, 293 Self-regulatory competence, correlation with high locomotors, 309 Self-reliance, 85 and avoidant persons’ reactions to thoughts of death, 104 Self-sacrifice as element of charismatic leadership, 30 by leaders for groups, 21 Self-serving evaluation, 247–248 Self-stereotyping, in negotiation, 245 Selfish misers, 237, 270 Selfish motivation dominant over prosocial motivation, in negotiation, 274 in negotiation, 250 Selfish thinkers, 237, 270 Semantic differential scales, in measuring locomotion and assessment orientation, 303 Sense making processes dynamics of, 384–392 timing of rationalization, 389–391 undermining affective forecasting, 369–380 Separation reaction of anxious individuals to issues of, 90 reaction of secure individuals to issues of, 90 suppression of thoughts related to, 85 Separation-related thoughts, ability to suppress in avoidant persons, 105 Set-goals, of the attachment system, 59

427

Sexual abuse, and fearful avoidance behaviors, 139 Sexual harassment, and errors in affective forecasting, 355–356 Sexual system, and attachment behavioral system, 123–125 Shyness, and assessment orientation, 302 Silence, in avoidance interactions, 160–161 Single-issue negotiation tasks, 239 Sixteen Personality Factor (PF) Questionnaire, 306, 308 Skill development, through group empowerment, 191 Skin conductance tests, 105 Sleep problems, in posttraumatic stress disorders, 87 Small group dynamics, 2 Smith, Adam, 369 Sociability, of target in getting-acquainted interactions, 176 Social anxiety, and assessment orientation, 304 Social appraisal, and social norms, 302–304 Social attraction, vs. personal attraction, in groups, 14 Social attractiveness, of leader, 10–11 Social categorization, 163 and depersonalization, 6–7 at group vs. individual level, 179 via behavioral confirmation, 154 Social change, role of leadership in, 43 Social cognition theory, 4–6, 53–54 Social cognitive neuroscience, 54 Social comparison, and self-consciousness, 302 Social conflict laboratory tasks in study of, 241 and negotiation, 235 Social identity and change through leadership, 40 and leadership, 10–14 and leadership analysis, 14–18 and prototypicality, 13–14 salience, and stereotype maintenance/ change, 202 and social influence, 6–10 and social stereotype confirmation, 155 theory of leadership, 17, 30–31 Social identity analysis, and contemporary leadership theories, 29–33 Social identity research, and organizational psychology, 5

428

INDEX

Social identity salience, 13–14, 29 Social influence and social identity, 6–10 and strategic preference, 321–324 Social motivation, 249–250 affecting negotiator cognition, 271 moderating ego defensiveness and naive realism, 249 moderating impact of ego defensiveness, 256–257 in negotiation, 236, 237 origins of, 250–253 Social Network Inventory (SNI), 327 Social norms, social appraisal and, 302–304 Social psychology, vs. organizational sciences, 2–4 Social relations, 53 Social stereotypes and behavioral confirmation, 154, 155 modification of, 157 Social value orientation, 250–251 Socioeconomic status (SES), and stereotype confirmation, 197 Sociological self-fulfilling prophecies, 198 Soft influence tactics, 28 Solidarity, 6 Specific emotions forecasting, 346 predicting, 347–349 skill in predicting, 401 Speech rate, as indicator of high task competence, 196 Status and prestige order, 195, 216 Status-based structural differentiation, in groups, 12–13 Status characteristics, 195–196 Status differences, 194–197 Stereotype change, 193, 207–209 collective strategies for, 188–192, 203, 208 on perceiver’s part, 201–203 and stereotype disconfirmation, 207 Stereotype compensation, 182, 185–188, 213 Stereotype confirmation, 162, 216 cognitive consequences of anxiety and, 173 and social change, 209–210 and socioeconomic status (SES), 197 and stereotype enactment, 182 in stigmatized groups, 179–181 Stereotype content, 175–176 ambivalent, 184

disconfirming negative, 186 positive aspects of, 213–214 unfavorable, 184 Stereotype enactment, 182–185, 213–214 Stereotype maintenance, 204–207 Stereotype threat, 183 Stereotyping, 6, 264 affected by stigmatization, 157 as cognitive limitation in negotiations, 243, 245 less likely among secure individuals, 120 Stereotype disconfirmation, 189, 214 Stigma consciousness, 177, 216 consequences of acknowledging, 189 exacerbation of stereotype confirmation with lack of, 180–181 and getting-along behavior, 211–213 target awareness of, 181–192 Stigmatized individuals/groups, 160–163 alertness in, 174 collective strategy advantages for, 192–193, 208 and resistance by nonstigmatized group members, 197 self-perceptions by targets in, 179 view of position as illegitimate or discriminatory, 190 Strange situation assessment procedure, 55 Strategic choice, 253–255, 255–256 Strategic interaction. See also Negotiation laboratory tasks in study of, 241 Strategicpreference,as targetof socialinfluence of others, 321–324 Stress avoidant strategies for coping with, 55 correlation with assessment orientation, 304 reactions to persistent, 101 Stressful events, attachment styles of coping with, 99–102 Stroop color-naming task, 89 Subjective appraisal, of threats, 74 Subliminal testing reactions of anxious individuals to, 90 reactions of avoidant individuals to, 90–91 reactions of secure individuals to, 89 security-related words and pictures, 93 threat-related words, 89 Submissive style, in dominant social interactions, 162 Suboptimal strategies, in negotiation, 244–245

429

INDEX Substance abuse, and attachment styles, 115 Success, feelings of, in high locomotors, 305 Support-seeking strategies in mothers of infants with congenital heart defects, 101 as stress response in secure individuals, 116 Suspicion, 83

related tocontrol,inavoidantindividuals, 98 related to intimacy, 98 in secure attachment behavior, 78 Trust-violation memories, correlations with attachment styles, 98–99 Type A behavior, correlation with locomotion orientation, 317

T

U

Target behavior, perceiving and explaining, 199–291 Target perspective, 177–194 Task competence and dominant behavior patterns, 196 low, perceived in behavioral confirmation studies, 162 Task-oriented groups, as perceived by attachment-anxious individuals, 115 Task-oriented interactions, 166–167 Threat-related thoughts, activating attachment system, 75, 90 Threatening events activating attachment system, 74 construed by persons with anxious attachment styles, 100 downplaying by avoidant behaviors, 84 inner sources of, 74 monitoring and appraisal of, 73 Threats constructive transformation of, 103–104 dismissal of, 85 exaggerated appraisal of, in anxious individuals, 116 use in negotiation, 254 Time commitment, correlation with locomotion orientation, 311 Time pressure impact on negotiation outcomes, 237, 265 and resistance to concession making, 254 Time to completion, correlation with locomotion vs. assessment orientation, 331–332 Toughness, in negotiation, 254 Transactional leadership, 30 Transformational leadership, 3–4, 29–31 Transmuting internalization, 131 Trust in love relationships, and attachment styles, 96, 98–99

Ultimatum games, 242 Uncertainty, pleasures of, in affective forecasting, 391–392 Unexpected events, more intense emotional reactions to, 371 Union-management negotiations, 239 Unique influences correcting for, in affective forecasting, 360–362 focalism, in affective forecasting, 366–369 hot-cold intrapersonal empathy gaps, in affective forecasting, 364–366 V Valence of emotions, 346, 392 predicting, 347 skill in predicting, 401 Validity, of affective forecasting, 352–354 Verbal behaviors in avoidant interactions, 160–161 competitive linguistic style, 191 in dominant interactions, 161–162 loud and firm voice in dominant interactional style, 162 Vertical individualism, 251 Vietnam war, 235 Visual dominance, 162 by men, in task-oriented interactions with women, 168 Voice tone, as indicator of high task competence, 196 Volition inhibition, correlation with high assessors, 309 W Warmth, effect of perceiver stereotype on interpersonal, 183 Win-win agreements, 240

430

INDEX

Withdrawal, from stressors, by avoidant individuals, 100 Women and glass ceiling obstacles to leadership, 34–36 in task-oriented interactions with men, 168 Work, as exploration in adulthood, 119 Work Preference Inventory (WPI), 310 Work team performance, and negotiation, 235

Working alliance, 140 Working models, in attachment behavioral system, 60 Worry, in anxious and avoidant individuals, following trust violations, 99 Y Yom Kippur War, 101–102