ADVANCES IN CHILD DEVELOPMENT AND BEHAVIOR
VOLUME 11
Contributors to This Volume Paul B. Baltes
Jacob L. Gewirtz
G...
18 downloads
1530 Views
15MB Size
Report
This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form
ADVANCES IN CHILD DEVELOPMENT AND BEHAVIOR
VOLUME 11
Contributors to This Volume Paul B. Baltes
Jacob L. Gewirtz
Gladys B. Baxley
Aletha Huston-Stein
Harry Beilin
Judith M. LeBlanc
Elizabeth F. Boyd
W. Stuart Millar
Orville G. Brim, Jr.
Leo Montada
Michael J. Chandler
Stephen W. Porges
Nancy W. Denney
Hayne W. Reese
Sigrun-Heide Filipp
John C. Wright
ADVANCES IN CHILD DEVELOPMEN-T AND BEHAVIOR edited by Hayne W. Reese Department of’Psycho1og.v West Virginia Universit,v Morgantown, West Virginia
VOLUME 11
ACADEMIC PRESS
*
New York
- San Francisco
A Subsidiary of Harcourt Brace Jovanovich, Publishers
London
C O P Y R I G H T 0 1976, BY A C A D L M I C PRESS, INC. A1 1. RlGHl S RESERVED. h O PkR7 OF 1111s PUBLICATION MAY B E REPRODUCED OR TRANSMITTED I N ANY F O R M O R BY ANY MEANS. EI FCTRONIC OR MECHANICAL, INCLUDING PHOTOCOPY, RECORDING, OR ANY INFORMATION STORAGE AWD R r T R l E V A L SYSTEM, W I T H O U T PLRbllSSlON IN WRITING F R O M T H F PUB1 ISHER.
ACADEMIC PRESS, INC.
1 1 1 Fifth Avenue, New York, New York 10003
United Kingdom Edition published b y ACADEMIC PRESS, INC. (LONDON) LTD. 2 4 / 2 8 Oval Road. London N W l
LIBRARY OF CONGRESS CAIALOG CARD N U M B E R63-23237 : ISBN 0-12-009711-7 PRINTED IN T H E UNITED STATES OF AMERICA
Contents
..................................................
ix
............................................................
xi
List of Contributors Preface
The Hyperactive Child: Characteristics, Treatment, and Evaluation of Research Design I. 11. 111. IV. V. VI.
GLADYS B. BAXLEY AND JUDITH M. LEBLANC Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Characteristics and Diagnosis of Hyperactivity ......................... Etiology of Hyperactivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Evaluation and Assessment of Pharniacotherapeutic Effects . . . . . . . . . . . . . . . Analysis of Drug Research with Ilvperactive Children . . . . . . . . . . . . . . . . . . . Summary and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2 3 11 14 20 28 29
Peripheral and Neurochemical Parallels of Psychopathology: A Psychophysiological Model Relating Autonomic Imbalance to Hyperactivity, Psychopathy, and Autism STEPHEN W. PORGES 1. Overview ..................................... 11. Hyperactivity . . . . . . . . . . . . . . . . . .
111. IV. V. VI. VII. VIII.
Psychopathy . . . . . . . . . . . .... Relationship between Central and Peripheral Autonomic Activity . . . . . . . . . . Statistical Model of Autonomic Balance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Autonomic Responses and Attention .......................... Autism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Summary and Conclusions ................................... .... ... References . . . . . . . . . . . . . . . . . .
40 43
60 62
Constructing Cognitive Operations Linguistically HARRY BEILlN I. Language as Representation ....................................... 11. The Mechanisms of Conservation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
68 69 V
vi
Contents
.
111 IV . V. VI .
Methodological Issues in Piagetian Training Research .................... The General Training Model: Methodological Issues ..................... Linguistic Training . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Summary of Results and Theoretical Considerations .................... References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
71 75 82 96 102
Operant Acquisition of Social Behaviors in Infancy: Basic Problems and Constraints I. I1. I11. IV .
W . STUART MILLAR Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Social Responses Reinforced by Social Feedback ....................... Mixed Social and Nonsocial Contingencies: the Case for Constraints . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
108 109 125 135 136
Mother-Infant Interaction and Its Study I. I1 . I11. IV . V. VI .
VII .
JACOB L. GEWIRTZ AND ELIZABETH F. BOYD Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Direction of Influence in Child Socialization .......................... Mechanistic and Organismic Models of Development and Interaction . . . . . . . . Study of Interactions between Mother and Infant in Natural Settings . . . . . . . Learning Analysis of “Simultaneous” Behaviors ........................ An Experimental Analysis of Mother-Infant Interaction . . . . . . . . . . . . . . . . . Summary and Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
142 142 146 149 152 153 159 160
Symposium on Implications of L ife-Span Developmental Psychology for Child Development
Introductory Remarks PAUL B. BALTES. CHAIRMAN Text . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
167
Theory and Method in L i fe-Span Developrnental Psychology: Implications for Child Development ALETHA HUSTON-STEIN AND PAUL B. BALTES 169 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I1. Prototypical Issues in Life-Span Developmental Theory . . . . . . . . . . . . . . . . . . 171 I11. Prototypical Issues in Life-Span Developmental Methodology . . . . . . . . . . . . . 179
1V. Summary References
..................................................... ....................................................
183 183
The Development of Memory: Life-Span Perspectives I. I1 . Ill IV V. VI .
. .
HAYNE W . REESE Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Memory Development in Childhood ................................. Memory Development in Addthood and Old Age ...................... Comparison between Childhood and Old Age .......................... Implications for Childhood Memory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Summary and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
190 190 195 200 202 207 209
Cognitive Changes during the Adult Years: Implications for Developmental Theory and Research NANCY W. DENNEY AND JOllN C. WRIGHT Research on Adult Cognition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11. Implications for Developmental rheory .............................. 111. Implications for Research Procedures with Children ..................... IV Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
I
.
.
213 215 220 222 222
Social Cognition and Li fe-Span Approaches to the Study of Child Development
.
MICHAEL J CHANDLER I . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I1. Pluralistic versus Monistic Views of Adulthood . . . . . . . . . . . . . . . . . . . . . . . . I11. Social Cognition and Multilincar Development . . . . . . . . . . . . . . . . . . . . . . . . . IV . Social versus Impersonal Cognition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
225 226 229 233 237 238
Life-Span Development of the Theory of Oneself: Implications for Child Development ORVILLE G . BRIM. JR . I . The Sense of Self Viewed as Theory about Oneself . . . . . . . . . . . . . . . . . I1. Culture and Socialization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I11. The Child's Developing Theory of Self ............................. IV . Two Illustrations of Possible Systematic Changes in Theories of Onesclf through the Life-Span . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
242 244 245 247
viii
Cot2rents
V . Setting the Research Agenda . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Refcrenccs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
250 251
Implications of LifeSpan Developmental Psychology for Childhood Education LEO MONTADA AND SIGRUN-HEIDE FILIPP 1. Introduction
...................................................
I1 . Contributions of Developmental Psychology to the Establishment of Educational Goals ............................................... 111. Design of Intervention Programs .................................... 11’. Problems in Transforming Empirical Knowledge into Intervention Programs . . V . Toward Research in Childhood Intervention from Life-Span Perspectives . . . . VI . Concluding Comments ........................................... References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
254 255 257 258 262 263 264
Author Index
.......................................................
267
Sub~ectIndex
.......................................................
277
...........................................
280
Contents of Previous Volumes
List of Contributors Numbers in parentheses indicate the pages on which the authors’ contributions begin.
PAUL B. BALTES College of Human Development, The Pennsylvania State University, University Park, Pennsylvania (167, 169) GLADYS B. BAXLEY Department o f Psychology, University o f Illinois, Champaign, Illinois ( 1 )
HARRY BEILIN City University of New York, Graduate School and University Center, New York, New York (67) ELIZABETH F. BOYD Laboratory of Developmental Psychology, National Institute of Mental Health, Bethesda, Maryland (141) ORVILLE G. BRIM, JR. Foundation for Child Development, New York, New York (241) MICHAEL J. CHANDLER Department of Psychology, University of Rochester, Rochester, New York (225) NANCY W. DENNEY Department of Psychology, Universit-vo f Kansas, Lawrence, Kansas (213) SIGRUN-HEIDE FILIPP Department of Psychology, University o f Trier, Trier, West Germany (253) JACOB L. GEWIRTZ Laboratory of Developmental Ps-vchology,National Institute of Mental Health, Bethesda, Maryland (141 ) ALETHA HUSTON-STEIN College of Human Development, The Pennsylvania State University, University Park, Pennsylvania (169) JUDITH M. LEBLANC Department of Human Development, University of Kansas, Lawrence, Kansas ( I ) ix
X
List of Contributors
W. STUART MILLAR' Department of Psychology, University of Strathclyde, Glasgow, Scotland (107)
LEO MONTADA Department of Psychology, University of Trier, Trier, West Germany (253) STEPHEN W. PORGES Department of Psychology, University of Illinois, Champaign,Illinois (35) HAYNE W. REESE Department of Psychology, West Virginia University, Morgantown, West Virginia (189) JOHN C. WRIGHT Department of Human Development, University of Kansas, Lawrence, Kansas (213)
' Present address: Department of Psychology, University College London, Cower Street, London, W.C.l., United Kingdom.
Preface The amount of research and theoretical discussion in the field of child development and behavior is so vast that researchers, instructors, and students are confronted with a formidable task in keeping abreast of new developments within their areas of specialization through the use of primary sources, as well as being knowledgeable in areas peripheral to their primary focus of interest. Moreover, there is often simply not enough journal space to permit publication of more speculative kinds of analyses which might spark expanded interest in a problem area or stimulate new modes of attack on the problem, The serial publication Advances in Child Development and Behavior is intended to ease the burden by providing scholarly technical articles serving as reference material and by providing a place for publication of scholarly speculation. In these documented critical reviews, recent advances in the field are summarized and integrated, complexities are exposed, and fresh viewpoints are offered. They should be useful not only to the expert in the area but also to the general reader. No attempt is made to organize each volume around a particular theme or topic, nor is the series intended to reflect the development of new fads. Manuscripts are solicited from investigators conducting programmatic work on problems of current and significant interest. The editor often encourages the preparation of critical syntheses dealing intensively with topics of relatively narrow scope but of considerable potential interest to the scientific community. Contributors are encouraged to criticize, integrate, and stimulate, but always within a framework of high scholarship. Although appearance in the volumes is ordinarily by invitation, unsolicited manuscripts will be accepted for review if submitted first in outline form to the editor. All papers-whether invited or submitted-receive careful editorial scrutiny. Invited papers are automatically accepted for publication in principle, but may require revision before fmal acceptance. Submitted papers receive the same treatment except that they are not automatically accepted for publication even in principle, and may be rejected. I wish to acknowledge with gratitude the aid of my home institution, West Virginia University, which generously provided time and facilities for the preparation of this volume. I also wish to thank Drs. Norman Cavior, Hiram E. Fitzgerald, John H. Flavell, Norah M. Gutrecht, David Klahr, Lewis P. Lipsitt, Willis F. Overton, and Marcia S. Scott for their editorial assistance. Hayne W. Reese xi
This Page Intentionally Left Blank
THE HYPERACTIVE CHILD: CHARACTERISTICS, TREATMENT, AND EVALUATION OF RESEARCH DESIGN'
Gludys B. Baxley U N l V E K S I T Y OF I L L I N O I S Ulld
drrditli M . LeBlanc UNIV1:KSITY O F KANSAS
I. INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11. CHARACTERISTICS AND DIAGNOSIS OF HYPERACTIVITY . . A. DESCRIPTIVE CHARACTERISTICS . . . . . . . . . . . . . . . . . . . . B. METHODS O F DIAGNOSIS . . . . . . . . . . . . . . . . . . . . . . . . . . . 111. ETIOLOGY O F HYPERACT!VITY . . . . . . . . . . . . . . . . . . . . . . . . A. ORGANIC FACTORS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. NONORGANIC FACTORS . . . . . . . . . . . . . . . . . . . . . . . . . . . . C. SUMMARY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
IV. EVALUATION AND ASSESSMENT OF PHARMACOTHERAPEUTIC EFFE,C'TS . . . . . . . . . . . . . . . . . . . . . . . . . . . A. SYMPTOM RATINGS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. PSYCHOLOGICAL M1:ASURES . . . . . . . . . . . . . . . . . . . . . . . . C. OBJECTIVE LABORATORY TASK PERFORMANCE . . . . . .
2 3 3 5 11
11 13 13
14
14 15 16
' T h e preparation of this manuscript was supported in part by Public Health Service training grant HD 00183, by CORIl Grant HD 02528 from the National Institute of Child Health and Human Development t o thc Kansas Center for Research in Mental Retardation and Human Development, by Grant 40-32-66-301 from the University of Illinois Research Board to the first author, and by Grant HD 05951 from the National Institute of Chld Health and Development to the University of Illinois Institute for Child Behavior and Development. We wish especially lo thank Drs. Stephen W . Porges, Donald M. Baer, Nancy K . Squires, and William H. Redd for their critical evaluation of previous drafts of this paper. 1
2
Gladys B. Baxley and Judith M.LeBlunc
D. BEHAVIORAL MEASURES . . . . . . . . . . . . . . . . . . . . . . . . . . . E. PHYSIOLOGICAL MEASURES . . . . . . . . . . . . . . . . . . . . . . . . F. ACADEMIC PERFORMANCE . . . . . . . . . . . . . . . . . . . . . . . . .
17 18 19
V. ANALYSIS OF DRUG RESEARCH WITH HYPERACTIVE CHILDREN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A. INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. LIMITATIONS OF GROUP-STATISTICAL DESIGNS . . . . . . . C. ADVANTAGES OF INDIVIDUAL ANALYSES . . . . . . . . . . . .
20 20 21 25
VI. SUMMARY AND CONCLUSIONS . . . . . . . . . . . . . . . . . . . . . . . . .
28
REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
29
I. Introduction A hyperactive, or hyperkinetic, c h l d is characterized by short attention span, restlessness, distractability, and overactivity (American Psychiatric Association, 1968). It is apparent that the syndrome is defined essentially in terms of its behavioral manifestations. In addition, hyperactive children often display learning deficits presumably related to their lack of concentration and focus of attention. Bradley (1937) is generally credited as being the first to study the effects of psychoactive drugs with hyperactive children. He demonstrated that Benzedrine@, a central nervous system stimulant, produced spectacular calming effects on the behavior of hyperactive children. Although Molitch and Eccles (1937) published a similar report in the same year, it was Bradley’s research that was instrumental in the establishment of pediatric psychopharmacology as a formal discipline emphasizing the investigation of drug effects in children. Since Bradley’s demonstration, psychoactive drugs have been prescribed increasingly to control the behavior of school children, especially those diagnosed as hyperactive (Bazell, 1971; Krager & Safer, 1974). This increased use of drugs to control behavior has had a profound effect on pharmacology, in that it has generated a great deal of research with children, so much so that there presently exists an extensive body of research in pediatric psychopharmacology. However, a major problem is encountered in much of this research because of the manner in which the results are analyzed and interpreted. Possibly because of a tacit assumption that group data accurately reflect the performance of individuals who compose a group, pediatric drug researchers
The Hyperactive Child
3
consistently adopt group experimental designs and statistical analyses of the group data for interpretation of drug effects. The increased use of drugs to treat hyperactive children stems in part from the experimentally demonstrated beneficial effects of drugs for such children. Thus, inferences about the potential effects of drugs on individuals are derived from effects interpreted from group averages, with little attention paid to the effects on the individuals within the group. The sole use of group designs, however, may not necessarily be appropriate for psychopharmacology research. The purposes of this paper are to examine existing pharmacological research in terms of its implications for treatment and diagnosis and t o propose individual analysis procedures as a functional research addition for assessing the effects of drugs on hyperactive children.
11. Characteristics and Diagnosis of Hyperactivity A. DESCRIPTIVE CHARACTERISTICS
1. Diversity of Labels Confusion surrounding the hyperactive disorder is reflected in the variety of labels given to it. For example, the disorder has been referred to as the “hyperkinetic syndrome,” “hyperkinetic impulse disorder,” and “hyperactive child syndrome” (American Psychiatric Association, 1968; Burks, 1960; Comly, 1962; Knobel, 1962; Laufer & Denhoff, 1957; Laufer, Denhoff, & Solomons, 1957; Millichap, 1968; Sainz, 1966; Stewart, 1967; Warren, Karduck, Bussaratid, Stewart, & Sly, 1971; Werry, 1968a; Werry, Weiss, & Douglas, 1964; Werry, Weiss, Douglas, & Martin, 1966). Other labels that have been used are based on theories of underlying neuropathology : “minimal brain dysfunction syndrome,” “organic deviation,” and “organic learning and behavior disorder” (Bradley, 1957; Clements, 1966; Clements & Peters, 1962; Conners, 1967; Millichap & Fowler, 1967; Solomons, 1965; Wender, 1971; Weiss, Werry, Minde, Douglas, & Sykes, 1968; Werry, 1 9 6 8 ~ ) In . the final analysis, however, the label of “hyperactivity” may be more appropriate, because it is the children’s overt behavior that distinguishes them from other children (even emotionally disturbed or learning-disabled children).
2. Behavioral Descriptions In spite of the diversity of labels, behavioral descriptions of the symptoms of the hyperactive disorder are remarkably constant among investigators. One of the most outstanding features of the disorder, and one for which children are typically referred for treatment, is excessive motor activity (Burks, 1960; Chess, 1960; Laufer & Denhoff, 1957; Millichap, 1968; Patterson, Jones, Whittier, & Wright, 1965; Sainz, 1966; Werry, 1968a; Werry & Sprague, 1970). Chess (1960), for example, defined the hyperactive child simply as one who carries out
4
Gladys B. Baxley and Judith M. LeBlanc
activities at a higher rate than the average child, as one who is constantly in motion, or both. Werry (1968c), in contrast, defined hyperactivity as a total daily motor activity (movement of the body or any portion of it) that is significantly greater than the norm. Laufer and Denhoff (1957) defined the disorder in terms of a standard level of activity: hyperactivity is involuntary and constant overactivity that greatly surpasses the normal. In addition, most investigators include in the hyperactivity syndrome other more loosely specified accompanying characteristics such as short span of attention and concentration, impulsivity, distractability, low frustration tolerance, poor emotional control and emotional lability, hyperexcitability, aggressiveness, and specific cognitive defects (Burks, 1960; Comly, 1962; Eveloff, 1970; Millichap, 1968; Sainz, 1966; U.S. Department of Health, Education, and Welfare, 1971; Werry, 1968a; Werry et al., 1966). When these characterizations are combined, the hyperactive child is defined as one who chronically exhibits sustained above-normal levels of gross motor activity that is generally accompanied by secondary behaviors (e.g., short attention span) as well as specific cognitive deficiencies (such as memory impairments). Hutt, Hutt, and Ounsted (1963) suggested a technique to distinguish braininjured from other hyperactive children. Whereas brain-injured children are said to exhibit abrupt changes of activity accompanied by abrupt synchronous changes of visual fixation, other hyperactive children do not show this same synchronous change of fixation. Instead, the latter continue one activity while visually attending to another activity, often for extended periods of time. In addition to these general characteristics, there are other features somewhat correlated with hyperactivity. For example, although some hyperactive children are mentally retarded, most are generally of average or above-average intelligence (Millichap, 1968; U.S. Department of Health, Education and Welfare, 1971). Hyperactive children are more frequently male than female, and, as Stewart (1967) reported, they represent approximately 4% of the grade-school population in the United States. Unfortunately, there is no clear evidence of the relationship between the incidence of hyperactivity and the socioeconomic status of the children so labeled. Although hyperactivity has been shown to result from lead poisoning (Silbergeld & Goldberg, 1974), which is frequently a problem among children from low-income groups, at the present time there are no exact data available to support an argument of a higher incidence of hyperactivity in this group as compared to other socioeconomic groups. Stevens, Sachdev, and Milstein (1968), however, have estimated that roughly 40% of all school children referred to mental health clinics because of behavioral disturbances are diagnosed as hyperactive.
3. Social Concomitants Parents of hyperactive children complain that their children are seldom quiet, are always on the go, have boundless energy, and require much less sleep than
The ffyperactive Child
5
their peers or siblings (Laufer & Denhoff, 1957; Werry, 1968a). Such children present a picture of perpetual m(jtlon. Comly (1962) reported that hyperactive children, as toddlers, are difficult to handle, tend to wake up from naps early, and frequently climb out of cribs, playpens, yards, etc. Thus, these children are not only a constant nuisance but are frequently in danger of physical harm (Werry, 196%). Parents and teachers seldom complain about the general activity level of hyperactive children but rathei about their overactive behavior in specific situational contexts. A teacher, for example, does not complain about overactive behavior on the playground during recess but has difficulty tolerating it during the spelling period. Therefore, hypcractive children are judged deviant or maladjusted at least as much by the social inappropriateness of their behavior as by any generalized quantitative exccss (Werry, 1968a). Thus, according to Zentall (1975), “the best definition o f hypeiactivity remains a social judgment of the frequency and intensity of specified behaviors displayed at ‘inappropriate’ times” (p. 549).
4. Academic Concomitants Hyperactive children are viewed by school personnel as distinctly different from their classmates. They are described as highly distractable, restless, and disruptive, and they are typically below average in academic performance. The poor academic performance is considered to result from poor concentration, impairment of motor, memory, and speech functions, and other cognitive deficits such as reading problems, reversals in reading and writing, and difficulties in arithmetic (Laufer & Denhoff, 1957; Millichap, 1968; Werry, 1968a; Werry et al., 1966). Hyperactive children can also be an academic paradox by their erratic academic behavior. Sometimes they compl-te work assignments correctly and neatly; other times, the same children either do not complete assignments at all or the work is messy and disorganized. On standardized intelligence tests these children frequently score higher than would be predicted from their academic performance and typically score higher on verbal than on performance measures of intelligence (Paine, 1968). B. METHODS OF DIAGNOSIS
The diagnosis of hyperactivity is shrouded in confusion because different authorities have used different criteria for diagnosis. A hyperactive child is quite likely t o have more than one diagnosis, particularly if the child has been seen in more than one clinic. The diagnosis is often derived from rather arbitrary criteria. Knobel (1962), for example, differentiated hyperactivity stemming from an organic etiology from that with a psychogenic cause. He labeled as “organic” those children who, in addition to exhibiting the behavioral character-
6
Gladys B. Baxley arid Judith M. LeBlanc
istics of hyperactivity, also yielded positive findings on at least two o f three clinical measures: the neurological examination; electroencephalographic recordings; and psychological tests which are purported to show signs of “organicity.” Hyperactive behavior with an organic basis is said t o be erratic, without direction, situationally nonspecific, and accompanied by impulsivity and aggression that show neither intentionality nor direction. The child whose hyperactivity is more psychologically based, however, is said t o show impulsive and aggressive behaviors that are intentional in nature and have apparent direction. Laufer and Denhoff (1957) proposed that the behavioral manifestations of hyperactivity are characteristic enough t o be diagnostic in and of themselves. However, other investigators have reasoned that the overt behaviors associated with hyperactivity could also be indicative of emotional disturbance. From this latter viewpoint some investigators believe that a diagnosis of hyperactivity must be supported by evidence of brain dysfunction (Menkes, Rose, & Menkes, 1967). Werry ( 1 9 6 8 ~ )however, pointed out that the function of the diagnosis is not simply t o label a child but is also t o identify the appropriate population, t o facilitate treatment, and, if possible, to provide a basis for estimating prognosis. In other words, the diagnostic procedure should distinguish hyperactive children from children with other disturbances (e.g., brain damage) of which hyperactivity may be only a small part. With such a selection procedure, a treatment strategy could be based on characteristics common to the populatid5as a whole, and the prognosis thus could be determined from treatment results with other already identified hyperactive children. Several procedures are currently used to diagnose hyperactivity; these may be grouped into three major categories: (1) diagnostic procedures based on physiological and/or neurological factors: (2) procedures based on psychological factors; and (3) procedures based on behavioral factors. I . Diagnostic Procedures Based on Physiological andlor Neurological Factors The electroencephalogram (EEG) has contributed much t o the understanding of central nervous system function (Bradley, 1957). Although its initial value was primarily in the study of convulsive states, it is now commonly used in diagnosing hyperactivity. One reason is that EEG tracings are frequently abnormal in children who exhibit hyperactivity (Paine, 1968). Burks (1960), for example, noted that hyperactivity associated with abnormal EEG is one of the most widely accepted criteria for a diagnosis of organic brain pathology. He proposed that organic pathology of the cortex (or surface) of the brain often distorts or changes the nature of brain activity in typical ways that can be recorded by the EEG. Thus, the EEG is thought to provide positive evidence of hyperactivity through characteristically distorted brain-wave patterns. However, EEG recordings in children are often unstable, and anxiety or tension can disrupt their
The Hyperactive Child
I
pattern (Werry et al., 1964). Therefore, reliance only on the EEG could lead to an erroneous diagnosis of hyperactivity. In a study of 97 children referred for evaluation of severe behavior disorders, Stevens et al. (1968) found that a generally abnormal EEG did not correlate with specific disorders. However, there was a correlation between specific regional and wave-form abnormalities and behavior disorders. Solomons (1 965) stated that EEG does not correlate directly with specific behavior patterns, intelligence, or the presence or absence of known cerebral trauma. Paine (1968) further acknowledged that abnormalities evidenced in the EEG are usually minor and probably do not contribute substantially to the diagnosis of hyperactivity. Thus, it is not surprising that Werry et al. (1964) viewed the EEG as a rather “crude” instrument. Because hyperactivity is sometimes equated with abnormal brain structure or abnormal function (Clements, 1966), neurological tests have been used for diagnostic purposes. However, conventional neurological examinations usually reveal no abnormalities of major or “hard” signs, such as in cranial nerves and reflexes (Paine, 1968), but only of minor or “soft” neurological signs (e.g., impaired muscle coordination, poor visual motor function, or difficulties in performing skilled acts) in a large proportion of hyperactive children (Werry, 1 9 6 8 ~ ) .For example, of the children studied by Hertzig, Bortner, and Birch (1969), 95% showed some abnormality on clinical examination, with the overwhelming majority having two or more “soft” signs of neurological disturbance. Thus, minor neurological deviations are sometimes taken as positive evidence for the existence of cerebral impairment i n the diagnosis of hyperactivity. Millichap (1968) also included in this list other findings such as motor impersistence or the inability to maintain gaze or posture, motor incoordination, impaired alternating movements of the forearms, gait ataxia, inability to hop, Babinski responses, and many others. By considering hyperactivity in terms of its associated neurological deviations, experts have been able to achieve consensus as to its diagnosis. However, as Werry ( 1 9 6 8 ~ )has pointed out, in the neurological sense there is as yet no evidence that any of these central nervous system deviations are in fact valid or reliable indices of cerebral dysfunction. Previous investigations have shown that specific kinds of deviations appeared on only those measures which were specifically designed to measure them (Paine, Werry, & Quay, 1968; Rodin, Lucas, & Simson, 1963; Schulman, Kaspar, & Throne, 1965). In other words, neurological, medical, and psychological measures reflected different and independent dysfunctions rather than a single dysfunction. Thus, neurological evaluations may have little relevance for a diagnosis of hyperactivity unless this information is combined with other kinds of information (Wender, 1971). The medical and developmental histories of the child do play major roles in the diagnosis of hyperactivity (Bradley, 1957; Burks, 1960; Comly, 1962). This
8
Gladys B. Baxley and Judith M. LeBlanc
style of diagnosis of hyperactivity requires a detailed behavioral history (Wender, 1971). Information gained from specific inquiry into the child’s level of functioning at various stages in development may have more predictive value for an accurate diagnosis than any other kind of information. Bradley (1957) maintained that history is especially helpful when there are indications of illness or injury. However, he conceded that such positive indications are not always available, and specific inquiry into the health of the mother during pregnancy, conditions during the birth itself, and the child’s health during infancy may also yield “significant” information. Denhoff and Novack (1967) also suggested that information concerning family history may contribute substantially t o a diagnosis of hyperactivity. This is especially true in families that exhibit a high incidence of neurological disability, presumably because such facts may indicate hereditary influences (given the presupposition that such ilifluences are relevant). Burks (1 960) constructed a questionnaire to obtain information regarding the developmental histories o f a group of fourth-grade students. From the questionnaires completed and returned by 204 parents, one particularly interesting finding was that unusual circumstances surrounding the birth process were more frequently reported in the histories of hyperactive children than in those of a group of normal children or a group of retarded readers showing n o evidence of the hyperactive disorder. Circumstances noted in births of hyperactive children were unusually long or short labor, anoxia, breech or unusual presentation, cord about neck, misshapen head, etc. Additionally, more than five times as many premature births were reported for hyperactive children as for normals. Severe infectious disease and blows to the head were also more common among the hyperactive group.
2. Diagnostic Procedures Based on Psychological Factors A psychological examination is typically included in the diagnosis of hyperactivity. These examinations are considered t o be most valuable if they are conducted by experienced clinical psychologists (Bradley, 1957), presumably because in addition t o test performance the child’s general behavior in the testing situation provides valuable diagnostic clues (Comly, 1962). Experienced clinical psychologists supposedly can identify such behaviors as short span of attention and concentration, as well as low frustration tolerance in the testing situation. Measures of general intelligence are sometimes employed for diagnostic purposes. Tests such as the Wechsler Intelligence Scale for Chi!dren (WISC) and the Stanford-Binet often reveal high verbal ability but low motor performance in hyperactive children (Comly, 1962; Denhoff & Novack, 1967). C o d y (1962) indicated that performance quality varies in hyperactive children, as denionstrated in frequent shifts from “insightful” approaches t o test items t o nonattentive trial-and-error performance, particularly on nonverbal tasks.
The Hyperactive Child
9
The Bender Visual-Motor-Gestalt Test and the Goodenough Harris Draw-aPerson Test often show discrepancies between the responses of hyperactive children who have some degree of neurological impairment and hyperactive children who d o not (Comly, 1962; Denhoff & Novack, 1967). Those in the former category often do not appreciate the relationship of parts of figures to each other and have difficulty with tasks involving spatial relationships (Comly, 1962). Immediate visual recall (from the Bender test) is also frequently shown to be impaired in such children, as is short-term auditory memory (from the Digit Span subtest of the WISC). The Rorschach Inkblot Test, the Children’s Apperception Test, and the Goodenough Harris Draw-a-Person Test are sometimes used to identify children suspected of being borderline psychotic (Wender, 1971). These tests are also considered to be useful for identifying hyperactive children, who often exhibit symptoms similar t o borderline psychotic children. Despite widespread use, the value of psychological tests for diagnosing hyperactivity is questionable because no satisfactory correlation has been established between performance on sucli measrires and the behavioral or neurological attributes of hyperactivity (Wenc!er, 1971). In other words, performance on psychological tests, be it normal or abnormal, may tell very little about the behavioral or neurological status of the child. Thus any predictive and prescriptive value of psychological tests, especially for therapy, has yet t o be established.
3. Diagnostic Procedures Based o t i Behavioral Factors The major identifying characteristic of hyperactive children is the behavioral manifestation of their disorder (e.g., excessive motor activity, restlessness, distractability, short attention span). Consequently, diagnosis is frequently based on observations of those overt behaviors. Diagnostic procedures that emphasize the behavioral components 01‘ hyperactivity are generally used t o assess the seriousness of the behavioral disturbance i n terms of the number of settings in which the hyperactive behavior causes problems. These diagnostic procedures sometimes involve questionnaires or rating scales, direct observation, and direct measurement of physical movcnient. The questionnaire or rating scale approach is very common (Peterson, 1961 ; Werry, 196%; Werry & Sprague, 1970; Werry et al., 1966). It requires that adults (e.g., parents, teachers, ward attendants) or peers rate the target child on items that pertain to the child’s motor behaviors (Werry & Sprague, 1970). The rater is generally someone who is with the child for a long period of time, on a regular basis. This diagnostic technique, however, has one major disadvantage for which it has received considerable criticism. The raters are individuals with whom the hyperactive child frequently interacts, and it is these same individuals with whom the hyperactive child typically comes into conflict. Therefore, these individuals are likely t o have biased opinions of the child’s behavior. Werry and
10
Gladys B. Baxley and Judith M. LeBIanc
Sprague (1970), however, proposed some advantages offered by the rating method as a diagnostic technique. They noted that the method is simple and inexpensive to use, and they also pointed out that questions about the chld’s behavior are at least “operationalized” and presented on a quantitative basis. Therefore, it is possible to assess their characteristic interrater reliability (Werry & Sprague, 1970; Werry et al., 1966). Yet, despite the fact that a number of published rating scales exist (Burks, 1960; Werry et al., 1966), no adequate tests have been made of their validity, their reliability, or their practicality. Direct behavioral observation, another diagnostic tool, requires that an observer watch and record the occurrence of specific motor movements at particular times and places. Such observations have been made in the child’s natural environment, such as the classroom (Becker, Madsen, Arnold, &Thomas, 1967; Patterson et al., 1965; Werry & Quay, 1969), or in a laboratory setting (Doubros & Daniels, 1966; Hutt et al., 1963). The structure of this technique is “freefield,” in that it can allow the children being observed to behave as they ordinarily would in usual situations. The major strength of this diagnostic technique lies in its objectivity and in its capacity to yield “real-life’’ data (Werry & Sprague, 1970). The observers do not interact with the child, thus minimizing problems of observer bias. In addition, relatively unskilled persons can quickly be trained to record these observations. The major weaknesses of this technique, however, are related to problems of inadequate sampling of the behaviors in question and the high costs involved in employing (and training) observers. The implementation of this technique also presents problems. It is difficult to have observers record the behavior of a child in many of the settings where the child is likely to be. However, as indicated by Werry and Sprague (1970), this technique is advantageous if at least a portion of clinical diagnostic time is devoted to it. Directly measuring the physical movement of children is now used increasingly for diagnosis. Such techniques rely on mechanical devices to measure activity as it naturally occurs. However, these devices measure only physical movement and no other behavioral manifestations of the problem. Some of these mechanical devices are self-winding wristwatches (Bell, 1968; Schulman & Reisman, 1959), photoelectric counters (Ellis & Pryer, 1959), ballistographic or stabilimetric chairs (Sprague & Toppe, 1966; Werry & Sprague, 1970), ultrasonic devices (McFarland, Peacock, & Watson, 1966), movies (Lee & Hutt, 1964), and telemetric devices (Werry & Sprague, 1970). With the possible exception of the self-winding watches, the majority of these devices are cumbersome, expensive, and have little relevance for routine clinical work (Werry, 1 9 6 8 ~ ) The . wristwatches, however, although relatively accurate, are difficult to calibrate, do not produce a permanent record, and record only movement of the limb to which they are attached. Nevertheless, with further refinement, this instrument may prove to have considerable diagnostic utility.
The Hvperactive Child
11
4. Summary There appears to be little agreement concerning appropriate and relevant means for diagnosing hyperactivity in children, and the most commonly used diagnostic procedures (i.e., those based on pliysiological/neurological, psychological, and behavioral factors) have not provided an adequate solution to this dilemma. A major difficulty is that the diagnostic procedures currently in use do not provide consistent diagnoses when different diagnostic criteria are used. Some authorities rely primarily on physiological and/or neurological test findings, and others base the diagnosis on performance on various psychological tasks; still others diagnose hyperactivity from the outward (behavioral) manifestations of the disorder, and often use expensive and cumbersonle mechanical devices to measure the behavior. Thus, it is not surprising that at present there is little agreement and much uncertainty about a diagnosis of hyperactivity.
111. Etiology of Hyperactivity The exact determination of the etiology of hyperactivity is still unknown, and of course there are almost as many theories concerning its etiology as there are individuals who propose them. Theories range from organic t o psychogenic causes and from disordered metabolism to inborn temperament (Satterfield, Cantwell, Lesser, & Podosin, 1972). A. ORGANIC kACTORS
In a significant number of cases, tlie etiology of hyperactivity is thought to be direct insult t o or dysfunction within the brain or within the central nervous system as a whole (Boelsche, 1965; Clements, 1966; Eveloff, 1970; Menkes e l al., 1967; Millichap, 1968; Minde & Webb, 1968; Nichamin & Barahal, 1968). Pasamanick, Rogers, and Lilienfeld ( 1 956) and Millichap (1968) have proposed that hyperactivity results from brain damage sustained prenatally or at birth, or from encephalitis in infancy or early childhood. Support for this view is found mainly in prenatal and perinatal histories provided by parents of hyperactive children. Other investigators have proposed that hyperactivity develops through an imbalance between cortical and subcortical areas of the brain (Eveloff, 1970; Knobel & Lytton, 1961), with specific dysfunction in the diencephalon (Denhoff, 1961; Laufer & Denhoff, 1957; Solonions, 1965). The component of the reticular activating system that passes through the diencephalon acts as a sorting and routing mechanism for incoming sensory impulses prior to their transmission t o higher levels of the brain (cT. Grinspoon Sr. Singer, 1973). When functioning properly the primary function of the diencephalon is to inhibit flooding of tlie
12
Gladys B. Baxley and Judith M. LeBIanc
cortex with irrelevant stimuli. Thus, dysfunction within the diencephalon may cause the cortex t o be bombarded with stimuli it cannot deal with efficiently. Such dysfunction would cause incoming stimulus inipulses t o spread out of the usual pathways into large, not yet mature cortical areas, causing a sensory “overload” and typical hyperactive behavior. However, Lytton and Knobel (1959) and Knobel and Lytton (1961) suggested that the diencephalon theory is untenable because, in their view, lesions in this region should produce a more complicated symptomatology . They proposed instead that the structural development of the cortex, slower than that of the diencephalon, makes the cortex more susceptible to damage. Thus, they suggested that consideration should be given to disorders of the cortex or of the subcortical pathways in the etiology of hyperactivity. Another theory that has considerable empirical support, one that is closely related to a theory of brain dysfunction, is that which attributes the etiology of hyperactivity to underarousal of the central nervous system (Satterfield, Cantwell, & Satterfield, 1974; Zentall, 1975). The abnormally high levels of motor activity, short attention span, and distractability of hyperactive children and the effectiveness of stimulant drug treatment have been interpreted as evidence that such children have an underaroused central nervous system and hence lack the capacity for selective inhibition of motor responses to environmental stimuli. Satterfield et al. (1974), for example, have suggested that hyperactive children who show a favorable clinical response to stimulants are those who have abnormally low levels of autonomic arousal as evidenced by lower skin conductance levels, higher mean EEG amplitudes, and larger evoked cortical-response amplitudes when compared to those who show a poor response to stimulants. These and other physiological measures, such as heart rate, are typically equated with autonomic activation and responsivity. Thus, the authors concluded that changes in these physiological measures following the administration of stimulants were consistent with an underarousal theory. The apparently heavy emphasis on brain damage or dysfunction as a possible cause of hyperactivity may be related to the belief that children with various behavior disorders share important common fcatures with the brain-damaged child (Conners, Eisenberg, & Sharpe, 1964). Yet Birch (1964) suggested that the term “brain damage,” as it is applied to hyperactive children, is an unfortunate label because it implies the existence of etiologic knowledge where none exists. In an effort to resolve this complex issue, he proposed that brain damage not be regarded as a neurological designation, but rather as a behavior pattern. In this view, any cerebral damage might cause similar patterns of deviant behavior, but the presence of hyperactivity is not a reliable variable for inferring brain damage (Becker, Englemann, & Thomas, 1971). Werry and Sprague (1970) theorized that brain damage is not necessarily a cause of hyperactivity. They, with others, usually attribute the problem to factors such as the following: genetic variation, “constitutional” determinants,
The Hyperactive Child
13
or chromosome abnormalities (Bakwin & Bakwin, 1966; Chess, Thomas, Rutter, & Birch, 1963; Stewart, Pitts, Craig, & Dieruf, 1966; Warren et al., 1971); metabolic disturbance (Burks, 1960; Comly, 1962; Lindsley & Henry, 1942; Martin, 1967); emotional disturbance (Eisenberg, 1957; Schulman et al., 1965; Solomorfs, 1965; Werry et al., 1966); or central nervous system dysfunction (Satterfield et al., 1972, 1974). Hyperactivity is especially likely to be attributed to these factors when there is no evidence of existing neurological defect (Solomons, 1965). €3.
NONORGANlC FACTORS
In recent years there has been a renewed interest in a theory relating the incidence of hyperactivity in childrcn to the ingestion of foods containing food coloring and other additives (Feingold, 1974). The first report linking hyperactivity to food additives appeared more than 50 years ago: Shannon (1922) reported an improvement in the behavior of seven hyperactive children when one or more foods containing ccrtain additives were eliminated from their diet. Furthermore, when the foods were reintroduced, the hyperactive symptoms returned. More recently, Crook ( 1 974) reported the case of a 4-year-old child whose hyperactivity was treated by an elimination diet. Over a period of several weeks specific foods were eliminated fro111 the child’s diet until the hyperactive symptoms disappeared. However, when the foods were reintroduced, one at a time, the hyperactivity sometimes reappeared, but only after the child ate those foods containing sugar. The author concluded that this child’s hyperactivity was caused by an allergic reaction to cane sugar; in order to control the symptoms, no foods containing cane sugar should be eaten by the child more than once or twice a week. According to Crook (1974), the foods that have a special potential to produce an adverse reaction are foods containing milk, chocolate, cola, cane and beet sugar, cereal grains, eggs, and citrus (foods containing food coloring and other additives should also be avoided). Along with interest in the food-additives theory of hyperactivity, there has also been a great deal of controversy about it. For example, Brown (1974) criticized work in this area as being either incomplete or “so poorly designed as to be worthless, even harmful.” (p. 6 2 ) . However, because so much remains unknown about the causes of hyperactivity, especially any nonorganic causes, virtually no theory can be rejected until disproved. However, every hypothesis should also be developed into a rigorous program of objective research. <’. SUMMARY Based upon what is presently known about hyperactivity in children, there appear to be many questions that remain to be answered about the etiology of the problem. There are several extant theories about the cause of hyperactivity,
14
Gladys B. Buxley and Judith M. LeBlunc
but none o f them is completely satisfactory. For example, it is believed by some that hyperactivity is the result of damage to the brain or central nervous system. More specifically, some authorities propose that damage (or dysfunction) within the reticular activating system of the central nervous system causes hyperactivity. Others, however, do not believe that brain damage would necessarily cause hyperactivity, and propose instead that it can be attributed to genetic and/or metabolic disturbances. or to emotional disturbance. A more recent (and highly controversial) theory attributes hyperactivity t o allergic reactions t o certain foods, especially foods containing artificial coloring and food additives. At the prcsent time, however, the etiology of hyperactivity remains open t o question. Yet the etiology, especially remote ctiology, although of interest, need not play a significant role in the diagnosis and management of a hyperactive child (Werry, 1 9 6 8 ~ ) Currently, . etiology, like the label “hyperactive,” tells little about the most appropriate treatment strategies t o be adopted. Although drug therapy is the primary medical treatment of hyperactivity, investigators should emphasize the importance of an accurate evaluation of the hyperactive symptoms and also the outcomes of all possible treatments.
1V. Evaluation and Assessment of Phariiiacotherapeutic Effects In the vast literature pertaining t o the pharmacological tieatment of hyperactive children, several methods have been used to evaluate treatment effectiveness. The methods can be classified into six categories: symptom ratings; psychological measures; objective laboratory task performance ; behavioral measures; physiological measures; and academic performance. A.
SYMPTOM RATINGS
Many clinical reports have contained analyses of positive drug effects in terms of reported symptomatic changes in the hyperactive child. Often, changes reported by physicians and clinicians are based on subjective and sometimes arbitrary ratings, such as “better,” “no change,” or “worse.” Because rating scales are so frequently employed in diagnosing hyperactivity (Werry er al., 1966), they are also felt t o be useful in the measurement of treatment effectiveness. For example, Conners, Eisenberg, and Barcai (1967), in a double-blind study, found that teachers rated the subjects’ behavior as significantly more improved following administration of dextroamphetamine than following placebo. Likewise, Alexandris and Lundell (1 968), using a five-point rating scale of worse, no change, mild improvement, marked improvement, and complete improvement, found that the effects of thioridazine and amphetamine on hyperactive children were rated superior t o placebo. These ratings were made by
The Hyperactive Child
15
nurses, teachers, physicians, and the principal investigators once per week for 6 months. In another study, by Conners, Taylor, Meo, Kurtz, and Fournier (1972a), 96% of children treated with dextroamphetamine were rated by parents and teachers as either “improved” or “much improved,” while only 30% of those treated with placebo were rated as “improved.” A number of other studies have all contained evidence of improvement in symptom ratings (Conners, Rothschild, Eisenberg, Schwartz. & Robinson, 1969; Finnerty, Soltys, & Cole, 1971; Knights & Hinton, 1969; Sattcrfield et al., 1972; Sprague, Barnes, 8t Werry, 1970; Weiss et al., 1068; Werry, 1968b; Winsberg, Bisler, Kupietz, & Tobias, 1972). In spite of the fact that the rating scales used in these studies yielded data that support the effectiveness of drug treatment with hyperactive children, one cannot be certain how much of that effectiveness was actually due to the drug and how much was due to the perceptions and expectations of the raters. Even though the raters may have been “blind” to the specific experimental conditions in effect at any given time, their perceptions of and expectations for the behavior of the subjects could have influenced their ratings. In other words, the blind may not actually have been a blind because it may not have precluded the raters’ expecting changes in the subjects’ behavior and, hence, identifying the treated group. Moreover, a frequent criticism of the blind procedure as a control technique is that it may be broken when the active drug is administered either because of the development of unwanted (side) effects in some subjects or because the raters are extremely sensitive t o drug changes (cf. Rickels, Lipman, Fisher, Park, & Uhlenhuth, 1970). Therefore, merely providing a blind as an experimental control may not guarantee rater objectivity. One means for determining the degree of rater objectivity might be to have a second person rate the subjects along with the primary rater. However, the second rater would not know that the ratings were being made to determine drug-treatment effects. Therefore, with the primary rater blind to the specific experimental conditions in effect and the second rater blind to the purpose of the ratings, the product of the two ratings could then be compared for rater agreement. B. PSYCIIOLOGICAL MEASURES
A variety of different psychological measures are used t o determine the effects of psychoactive drugs. The Wechsler Intelligence Scale for Children (WISC), Porteus Maze Test, the Goodenough Harris Draw-a-Person Test, and many others have been used in drug rescarch I n separate investigations of the effects of dextroamphetamine, Weiss et al. (1968), Conners et al. (1969), and Finnerty el al. (1971) all observed no significant differences between active medication and placebo in the subjects’ performance on the WISC. Epstein, Lasagna, Conners, and Rodriguez (1 968), however, found that dextroamphetamine
16
Glndys B. Baxley and Judith M. LeBInnc
decreased scores on the verbal subtests of the WISC but increased scores on the performance subtests. An analysis of the effects of methylphenidate on WISC performance (Knights & Hinton, 1969) showed that this amphetamine-like drug produced a significant improvement over placebo in the WISC performance scores but not in the verbal scores. In the latter comparison, methylphenidate was found t o be no more effective than placebo. Other measures, such as the Goodenough Harris Draw-a-Person Test, have also yielded conflicting data on various drug effects. For example, Conners and Rothschild (1968) found no significant differences between amphetamine and placebo on the Draw-a-Person Test, but Conners (I971 b) found that both niethylphenidate and dextroamphetamine produced highly significant improvements in performance over placebo on this test. Finally, Alexandris and Lundell (1968) found no significant differences between thioridazine, amphetamine, and placebo on any of the psychological measures they employed. These measures included the Bender Visual-Motor-Gestalt Test, the Goodenough Harris Draw-aPerson Test, and several subtests from the WISC. Thus, given the ambiguity of these findings, and the apparent lack of relevance of psychological measures for diagnostic purposes, it is not possible to make any unequivocal statements regarding the effects of psychoactive drugs on performance of psychological tasks by hyperactive children. C. OBJECTIVE LABORATORY TASK PERFORMANCE
Laboratory tasks have frequently been devised to assess the effects of various drugs in hyperactive children (Baxley & LeBlanc, 1974; Baxley & Marks, 1974). These tasks sometimes yield information that is difficult to obtain in the less structured natural environment but may have much relevance for the education of the children. For exatnple, Freibergs, Douglas, and Weiss (1 968) demonstrated that long-term administration of moderate dosage levels of chlorpromazine for hyperactivity had no apparent adverse effects on the capacity of the children to acquire object and number discriminations. The authors found no significant differences between subjects’ performance under chlorpromazine and placebo conditions. Sprague, Werry, and Scott (1967) found that the response latencies of subjects in a two-choice discrimination task decreased following treatment with amphetamine (Dexedrine@). In a later study Sprague et al. (1970) studied the effects of niethylphenidate and thioridazine on recognition-task performance in a group of hyperactive emotionally disturbed boys. In addition to performance accuracy, the reaction times of the subjects were also measured, as well as their activity levels during the experimental sessions. Two dosage levels of both methylphenidate and thioridazine were used, together with a placebo control condition. The results of this study indicated that methylphenidate was superior t o either
The Hyperactive Child
17
thioridazine or placebo in significantly increasing the subjects' correct responses and decreasing their reaction tinies. Cohen, Douglas, and Morgenstern (1 97 1 ) found a significant difference between the reaction times of hyperactive children when they received methylphenidate and when they received placebo. Reaction times measured under drug conditions were significantly faster. The authors concluded that methylphenidate improves attention. In another investigation of reaction time, Sykes, Douglas, and Morgenstern (1972) found that hyperactive chldren were able to sustain their attention longer, detect more significant task stimuli, and make fewer incorrect responses following treatment with methylphenidate. In light of these examples, it seems clear that stimulants such as methylphenidate and amphetamine tend to enhance performance of laboratory tasks. However, the effects of other classes of psychoactive drugs, such as chlorpromazine, are not as clear. No doubt fuithcr research could clarify this issue. D. BEIIAVIORAL MEASURES
Behavioral measures provide a means for reliably determining the efficiency of pharmacologic therapeutic techniques. For example, in a study of the delayed speech and language of children of normal intelligence but diagnosed as having cerebral dysfunction, Creager and Van Riper (1967) compared the effects of methylphenidate with those of a placebo. In this double-blind investigation, three samples of speech were obtained for each subject: one at the beginning of the study, and one each after the administration of methylphenidate and placebo. During each experimental session, subjects were required to respond to a series of pictures and, in an effort to elicit less structured verbalizations, they were asked t o respond to two general questions. The authors found that verbal productivity (i.e., the total number of words spoken, the total number of responses, and the total numbcr of incomplete utterances) was significantly increased with the use of methylphenidate. Anton and Greer (1969) reported that in a group of hyperactive mentally retarded boys treated with dextroamphetamine, several of the children exhibited an increased awareness of environmental stimuli, engaged in less random activity, and increased the frequency of their vocalizations during drug therapy. This behavioral response was observed during a 3-week period when the subjects were orally administered 10 nig of dextroamphetamine per day. The authors noted, however, that this favorable response abated when the drug was continued for an extended period of time. Finally, data gathered by Sykes, Douglas, Weiss, and Minde (1971) indicated that the ability of hyperactive children to attend, as measured by their responses on a Continuous Performance Test, was significantly improved when they were treated with methylphenidate. In this study, 40 hyperactive children were
18
Gladys B. Baxley and Judith M. LeBlanc
administered either methylphenidate or placebo. The drug dosage levels ranged between 30 and 40 mg per day and were individually titrated for each subject until an optimal clinical effect was obtained. The experimental task consisted of the individual presentation of series of letters, the subjects being required to select a specific letter whenever it appeared. The authors found that the hyperactive children who received the active drug had better performance in that they were able to detect more of the relevant stimuli with fewer impulsive and erroneous responses than were similar children who received placebo. E. PHYSIOLOGICAL MEASURES
Various studies have been conducted in an attempt to determine the treatment effects of psychotropic compounds as measured by physiological factors (e.g., Porges, Walter, Korb, & Sprague, 1975; Satterfield & Dawson, 1971). Physiological measures are frequently analyzed from the standpoint of the selection of “good” and “poor” responders to stimulant medications. In other words, hyperactive children who obtain a good clinical response t o stimulants are found to differ from poor responders on several physiological measures such as skin conductance level, EEG response, and auditory evoked cortical responses (Satterfield, Cantwell, Saul, & Lesser, 1973; Satterfield et al., 1972, 1974). Epstein et al. (1968), for example, attempted to determine the response of hyperactive children to 10 t o 20 mg of dextroamphetamine per day. Based on a variety of clinical data, the subject population was divided into an “organic” (probable brain damage) and a “nonorganic” group. Although no consistent or significant differences in physiological measures were obtained for either group following drug treatment, the “organic” group demonstrated a greater tolerance for higher dosage levels of dextroamphetamine with fewer side effects. However, there was an increased rate of excretion of the drug in the urine of the “organic” children, perhaps accounting for their greater tolerance level. In another study, Knights and Hinton (1969) found significant changes in weight, diastolic blood pressure, and heart rate among 40 children treated with methylphenidate. However, when the children were grouped according to whether they had a history of probable brain damage or not, no clear relationship was found between etiology and response to the drug. Similarly, Satterfield et al. (1974) demonstrated a promising technique for classifying hyperactive children. Using EEG recordings and measures of skin conductance level, the authors found significant differences between hyperactive subjects who responded well to methylphenidate treatment and those who responded poorly. The best drug responders had higher mean resting EEG amplitudes, greater resting EEG amplitude ranges, and a higher number of EEG movement artifacts prior to drug treatment than did the worst drug responders. Best and worst response to treatment was based on improvement in scores, or
The Hyperactive Child
19
the lack of it, o n a teacher-rating scale. Furthermore, the skin conductance levels of the best-response group were lower than those of the worst-response group. The findings in this study and others (Anton & Greer, 1969; Conners et al., 1967) are suggestive of methods for separating “good” from “poor” stimulant drug responders before such treatment is attempted. The photo-Metrazol technique is an ingenious procedure that has strong implications for determining psychoactive drug treatment effects (Laufer ct al., 1957). It is used as a means of empirically establishing the relationship between diencephalic dysfunction and hyperactivity, where it is assumed that dysfunction in the diencephalon region 01‘ die brain is related t o the hyperactivity. This technique consists of determining the tlireshold dose, in milligrams of Metrazol (pentylenetetrazol) per kilogram of body weight, that is sufficient to elicit EEG spike-wave bursts and myoclonic: jerking of the forearms when an individual is visually stimulated by a stroboscopic light. Among 50 children studied with this technique, Laufer et al. found ;I clear separation between hyperactive and nonhyperactive children. Based on these data, Solomons (1965) concluded that hyperactive children have significantly lower photo-Metrazol thresholds than d o other nonhyperactive children o f comparable ages. Laufer et al. (1957) later l‘ouncl that when hyperactive children were treated with dextroamphetamine t o control their behavior, their photo-Metrazol threshold increased to normal levels. C‘onversely. when dextroamphetamine was discontinued, there was a decline in photo-Metrazol levels. Thus, the authors suggested that stimulant drugs (in this case, amphetarnine) restore the “normal” functioning of the diencephalon, thereby relieving hyperactivity. F. ACADtMIC PERFORMANCE
Because poor academic perfo~-manceof‘ hyperactive children is frequently the basis for their referral for treainient, it seems only fitting that an analysis of academic behavior be included in the assessment of psychotropic drug treatment effect. However, few rcsearctiers have systematically investigated the various parameters o f academic performaiicc. One exception is a study by Sprague et ul. (1970) in which tlie authors I‘otind a greater effect of methylphenidate on certain classroom behaviors of hyperactive boys than of either thioridazine or placebo. Specifically, ine~liylpliciiidatewas associated with significantly fewer classroom isolation or punishment periods, as well as higher percentages of on-task or task-related behaviois. In contrast, Alcxandris and Lundell (1968) could find n o significant tliflkrenccs between the effects of amphetamine, thioridazine, o r placebo on tlie rcading. spelling, and arithmetic performance of hospitalized hyperactive chiltlr-cn In view of the evidence presented, i t appears that psychoactive drugs generally have clear, positive effects on a t least some measures used t o assess treatment
20
Gladys B. Baxley and Judith M. LeBlanc
effect (e.g., symptom ratings, laboratory task performance, and behavioral measures). However, drug effects on psychological test performance and academic performance are not as clear-cut. Therefore, it appears that before any statements can be made concerning the effect of drugs on measures such as school performance, more definitive research, especially in this area, is needed.
V. Analysis of Drug Research with Hyperactive Children A. INTRODUCTION
The foregoing discussion of hyperactivity and its treatment through pharmacologic means has been presented in light of some of the major issues related to psychotropic drug research and usage with children. The primary objective of this discussion has been to present a coherent picture of the characteristics that define hyperactivity, the means by which it is diagnosed, and the theories concerning its etiology that form the basis for drug treatment. This discussion will serve to undergird the subsequent argument for the individual analysis of psychotropic drug data in the treatment of hyperactivity. Without a clear understanding of the history of hyperactivity as a significant clinical and educational problem, this latter discussion would be difficult to place in perspective. In spite of the abundance of experimental evidence attesting to the apparent effectiveness of psychotropic drugs on the behavior of hyperactive children, criticisms have been raised which seriously challenge the adequacy of the research base. In many instances, the enthusiasm over drug therapy for hyperactive children has been based on research that in total yields mixed results and tenuous conclusions. Because several of the most frequently cited criticisms of pediatric drug research have been the subjects of a number of previous reviews (e.g., Conners, 1972b; Freeman, 1966; Grant, 1962; Sprague & Werry, 1971; Sprague, Werry, & Davis, 1969), it is perhaps unnecessary t o reiterate those criticisms here. Therefore, in the next few pages, consideration is given to an examination of the fundamental issue of design in drug research. More specifically, attention is given to an analysis of and justification for the inclusion of individualized (or single-subject) approaches in the design of drug research. A primary objective in the design of any psychopharmacological research is to provide a description of the functional relationship between the administration of a chemical agent and specific behavior change. Traditional attempts to establish such a relationship have involved the comparison of means obtained for groups of subjects exposed to different values of the independent variable. In fact, nearly all drug research with hyperactive children has included group designs with nontreatment groups employed to control for effects obtained in
The Hyperactive Child
21
the treatment groups. Depending on the nature of the research question, a common practice of drug researchers is t o employ several groups of subjects to whom different levels of the particular drug are administered. The etfects on the dependent variables are then measured, and the results are analyzed statistically (the group-statistical approach). Alternatively, a repeated-measures procedure may be used in wliicli each gloiip of subjects participates in every experimental condition. The order in which the conditions are introduced may be counterbalanced across groups to control for (but not eliminate) effects of the sequence of administration; but any differences obtained between drug conditions are still analyzed statistically. The intent of this analysis is t o determine whether there are differences in ct'fect between the groups based on the administration of the drug. B . LIMITATIONS 01 GROUP-STATISTICAL DESIGNS
Although the group-statistical research design is the one most commonly employed in pediatric drug research, it has several limitations which iniflit militate against its sole use in research on hyperactivity. These limitations might be conceptualized in the following manner. Because data for all subjects treated similarly are lumped together in a group design, such a design precludes the specification of differential drug responsiveness in individual subjects. Owing to this fact, then, the group design does not allow thc identification of homogeneous subgroups (i.e., hyperactive children who are more similar on certain dimensions than other hyperactive children). Therefore, because of its limitations for identifying similar subjects, a group design also does not facilitate the determination o f idiosyncratic response styles and response patterns. But let LIS consider each of these points in turn. 1. Specification of Differential Drug Responsiveness Pediatric drug researchers make several assumptions about the nature of hyperactivity and psychotropic drug effects when they statistically analyze and interpret data from groups of hyperactive children. One assumption is that the response to drug treatment of individual hyperactive children corresponds t o the average drug response of the entire sample from which the individuals were drawn. This, however, may not be the case, because some subjects in the sainple might show very divergent responses. For example, in a saniple of 50 hyperactive children, 35 may improve following drug treatment, 10 may show no change a t all, while 5 may actually show varying degrees of deterioration. Where most of' the subjects in the group show a strong positive response to treatment, a few show a relatively weak response o r no response at all, and still others deteriorate. By using a group-statistical analysis procedure, the differential responsiveness of a few subjects is thereby masked by the strong response of the majority. The
22
Glndys B. Baxley and Judith M. LeBlanc
data, however, are interpreted in terms of the statistical significance of the overall group effect, and inferences are made about the functional relationship between psychotropic drugs and specific behavioral changes based on this interpretation. But by drawing inferences based on the average group effect, the researcher obviates the possibility of identifying “nested” groups or subgroups of hyperactive children who may react in divergent fashion from the majority of subjects in the sample. In essence, psychoactive drugs may cause different behavioral consequences in different subjects. Dose-response data for individual subjects may also show differential effects from one subject to another at different dosage levels, yet the average group response may not adequately reflect this fact. When the data are interpreted from the average response of the group, the interpretation does not represent the range of effect across different dosage levels of the drug. Thus, when only the data of the group are considered in a statistical analysis, it is impossible t o determine which subjects in the group actually benefit from treatment, as well as the magnitude and direction of their response at each dosage level.
2. Identification o f Homogeneous Subgroiups Another assumption of drug researchers concerns the makeup of the experimental sample, namely, that the individuals who compose the sample are homogeneous with respect t o their defining characteristics. In other words, hyperactive children are assumed t o represent a specific subgroup identifiable by some generally agreed upon behavioral characteristics. However, from the myriad of labels applied t o the hyperactive disorder (Clements, 1966), and from the fact that there is presently very little consensus on either its definition or diagnosis (Conners, 1972b), it seems abundantly clear that hyperactive children are not a homogeneous group, and conclusions about hyperactivity based on group experimental data are, of necessity, tenuous. Although the diagnostic label is the same, the sequelae that contribute t o the label may vary widely from subject t o subject and thereby seriously challenge the validity of the label. Although hyperactive children do present a similar clinical picture, they constitute a very heterogeneous group. In fact, considering the variety of theories about the etiology of the disorder, hyperactive children probably should differ greatly with respect t o a multitude of factors, such as physical and neurological status, psychological functioning, neurophysiological history, and family interaction patterns. To average the effects of a drug across all the members of a group of such children would not only dampen the differential effects of the drug but would also obscure the importance of the diversity of factors which distinguish individual hyperactive children and define homogeneous subgroups. When the relevant dependent variable of the experimental sample is a particular behavioral characteristic (defined in this case as “hyperactivity”), some of the factors which comprise that characteristic may be unevenly distrib-
The Hyperactive Child
23
uted among all the members of that sample. For example, in a comparison of. two groups of hyperactive children, factors indicative of brain damage which are associated with hyperactivity (e.g., neurological abnormalities, learning deficiencies, and a neurophysiological history suggestive of brain damage) may be i-ound at a much higher frequency among the subjects in one group. However, other factors associated with hyperactivity (e.g., excessive motor activity, sliort attention span) may be more evenly distributed among all of the subjects of both hyperactive groups. If, following drug treatment, statistically significant differences arise between the two groups (e.g., the “brain-damage’’ group improves with treatment, while the other group does not), what then can one conclude about that difference based on the statistical analysis? It is only possible to conclude that either: (1) the drug was more effective with hyperactive children who are brain damaged; or ( 2 ) the drug was least effective with hyperactive children who are not. Finer delineations about drug effects on specific kinds o f brain damage are also not possible from a group-statistical analysis. A problem that is closely related to that of homogeneous subgroups, and one which drug researchers tend t o overlook, is that of interpreting drug-effect data from experimental comparisons of hyperactive and “normal” children. A statistical comparison of groups of hyperactive and nonhyperactive children would assume homogeneity of variance ( i t . , the variance is the same for both groups o f children). This assumption may not be valid, however, in that comparisons of hyperactive and normal children may be statistically inappropriate. One would expect, as a function of the multiplicity of defining characteristics of hyperactivity, that the variability o n a specific dimension (e.g., attention span) would be greater among the subjects in the hyperactive group than among the nommal children. This fact, then, violates an assumption of homogeneity of variance. However, an analysis comparing only groups of hyperactive subjects (assuming random selection of those assigned to the various treatment conditions) is not affected by this assumption. This type of analysis assumes that the variance of the groups is stable over repeated testing situations. However, the assumption of stable variance over repeated measures may also be faulty. [For a more elaborate discussion of this problem, see Greenhouse and Geisser (1959)l.
3. Determination of Idiosyncratic Response Styles and Response Patterns By their sole use of group-statistical experimental designs, drug researchers assume that their findings will generaliLe from one group of hyperactive children to other hyperactive groups. This assumption implies the validity of interpretations of quantitative as well as qualitative similarities of drug response betwecn independent hyperactive groups. But group-statistical designs fail t o account for idiosyncratic response styles and response patterns of individual subjects. Because hyperactive children differ in their response t o psychotropic drugs (cf.
24
Gladys B. Baxley and Judith M. LeBlanc
Conners & Eisenberg, 1963; Millichap, 1973; Millichap, Aymat, Sturgis, Larsen, & Egan, 1968; Satterfield, 1973; Satterfield et al., 1973), a delineation of the specific parameters along which they vary may be important for efficient treatment. When a drug has a statistically significant influence, that influence is partially determined by: (1) the magnitude of the subjects’ responses, and (2) the quality of their responses. However, the one aspect of a drug’s effect that has important implications for treatment outcome and is not adequately addressed in the literature is that of qualitative differences between groups, or more specifically, the direction of the effect. The distinction between quantity and quality of drug response is merely the distinction between the amount or magnitude of change as a function of the drug and the direction of that change, either positive or negative. Because the same drug at the same dosage level may, on a physiological level, affect individuals differently, there are likely to be extremes of response on a behavioral level as well for which average data are not representative. To carry this argument a step farther, suppose a sample of hyperactive children contains small homogeneous subsamples in which the subjects respond in a fashion consistent with one another but respond in a divergent fashion from the mean of the overall sample. If all the subjects are averaged together, the group curve may not mimic any actual individual, and the idiosyncratic response styles and patterns either of individuals or of the subsamples may be lost. Sidinan (1952) and others have demonstrated that, given a mean group-response pattern, the forms of individual response patterns are not uniquely specified. So, in essence, no inductive inference from a group mean t o individual response patterns is possible because, as suggested by Estes (1956), the shape of the group response pattern could have arisen from any of an infinite variety of subsamples of individual patterns. Thus, the sole use of a group-statistical approach t o the design and analysis of drug data is attended by considerable risk in any subsequent interpretation for individuals. Group-statistical drug researchers also assume that valid statements about the generality of experimental findings are possible when a large enough sample is used t o allow interpretations of the representativeness of those findings. In other words, as a sample of hyperactive children approaches the size of the population, the characteristics of the sample approach the characteristics of the population. This may be true theoretically, but as Sidrnan (1960) has argued, representativeness is an actuarial problem for which the currently prevalent group-statistical design is not appropriate. In the absence of some specification of the number of subjects who actually yield each value of the dependent measure, the use of groups adds little or nothing to the “representativeness” of the experimental results; instead, it merely allows the identification of small mean differences as “significant” differences.
The Hyperactive Child
25
C. ADVANTAGES 01: INDIVIDUAL ANALYSES
1. Specification ofDiffereiztia1Drug Responsiveness In light of the limitations of group-statistical research designs, the inclusion of an individual analysis approach t o the analysis and interpretation of drug research seems a reasonable addition to a statistical design. There are a number of strong arguments which support this point of view. By employing a few carefully selected subjects and individually analyzing their responses, it becomes possible t o address some of the problems inherent in group-statistical designs. For example, an individual analysis of drug response allows the identification of subjects who show a differential response. A study by Millichap and Boldrey ( 1 967) is an example of this point. With the use of an actoineter t o measure gross motor activity of hyperactive children, they compared the effects of methylphenidate and phenobarbital to placebo on the hyperactivity and involuntary movements of 14 children referred for treatment. The actometer was a self-winding calendar wristwatch that was modified so that the pendulum was attached directly t o the hands of the watch. The apparatus was then worn by the subjects o n the wrist of their dominant side. The authors found that acute doses of methylphenidate (.25 t o .5 mg/kg) produced an increased rather than a decreased level of activity in four o f the subjects. However, because 10 of the subjects in this study continued t o receive other regularly prescribed medications, specific data on the effects of methylphenidate and phenobarbital were confounded, thus making it impossible to determine why methylphenidate should be associated with an increased activity level. This particular finding serves t o emphasize the importance of individually tailored drug assessment procedures. In a study by Baxley and LcBlanc (1974) differential effects were found on the short-term memory of five mentally retarded hyperactive children. The study was designed to determine whether or not five different dosage levels of methylphenidate were substantially different from placebo in their effects on the subjects’ performance of an instruction-following and digit-recall task. The authors found that the active drug had very different effects from placebo on each subject’s performance, at each of the several different dosage levels. In other words each subject showed a differential responsiveness to the drug, and the individual analysis design afforded the opportunity t o determine not only specific individual effects, but also the dosage levels at which those effects occurred. It allowed for a determination of which subjects actually responded t o the drug treatment. Sleator (Personal communication, 1975) found that in a group of 27 hyperactive children the mean increase in heart rate following treatment with methylphenidate was roughly 8.4 beats per minute. However, she reported that very
26
Gludys B. Baxley and Judith M. LeBlanc
divergent responses were obtained by a few of the subjects following drug administration. For example, one child showed a 40 beats per minute increase in heart rate when the active drug was administered, while another child actually showed a 17 beats per minute rate decrease. These data also support the utility of an individual analysis of drug effects.
2. Identification of Homogeneous Subgroups When data of individual responsiveness t o drugs can be derived from a statistical distribution defined according to specific factors related t o hyperactivity, the subsample of which the individual is a part would be a nearly homogeneous one. The power of an individual drug effect could then serve a predictive function by showing the utility of the drug for the treatment of that subsample. Satterfield ct ul. (1974) demonstrated this point when they studied the effects of methylphenidate on a group of hyperactive children, testing the hypothesis that children who respond to stimulant medication are those who, prior t o treatment, arc found t o have low central nervous system arousal levels. In their study, a group of 14 hyperactive children was treated for 3 weeks with individually titrated doses of methylphenidate. Measures of resting EEG brain-wave patterns, auditory evoked cortical responses, and skin conductance levels were obtained as indices o f autonomic arousal prior t o drug treatment. Teacher ratings of classroom behavior were obtained before and after drug treatment, and changes in these ratings were used as measures of improvement. The children who showed 70% or better improvement on the teacher ratings (best responders) were compared to those with 30% or less improvement (worst responders). Using a nested (or hierarchical) design (cf. Winer, 1962), Satterfield et al. (1974) found that the subgroup of children who showed the best response t o treatment were those who were also found initially t o have low central nervous system arousal levels. They concluded from this finding that there was apparently an identifiable subgroup of hyperactive children with low levels of autonomic arousal whom one would predict t o be good responders to stimulant medication. This conclusion, incidentally, supports a neurophysiological model of hyperactivity; specifically, the analysis of idiosyncratic behavioral response styles and physiological response patterns may facilitate treatment of individual hyperactive children. It becomes possible t o increase the precision of the estimate of treatment effect because such an estimate is based on within-subject variability and, under certain conditions (e.g., where drugs are known to have differential effects), the within-subject variance is much less than the variance between subjects. Thus, repeated administrations of a drug which produce a consistent response within one individual or subsample could produce a very different but consistent response in a second individual or subsample. Therefore, thc diversity of symptoms associated with hyperactivity would no longer be a problem with which researchers must contend, but rather a valid foundation on
The Hyperactive Child
21
which predictions about drug effects could be based. Important information is gained when a few subjects are studied whose hyperactivity (the characteristic by which the sample is defined) can be attributed to one of a number of different causes, but who nevertheless yield a similar response to the drug treatment. If a lawful relationship holds within a subsample of individuals who are similar in terms of the specific parameters of their diagnostic label, drug researchers can generate relevant data for the efficient treatment of that subsample, rather than bemoan the “heterogeneity” of the hyperactive population as a whole.
3. Determination of Idiosyncratic Response Styles and Response Patterns An individual analysis is not based on the assumption that generality is necessarily attributed to the data of a single subject; but, on the contrary, information that a given drug produces a reliable and replicable effect in individual members of a subsample has considerable generality. Porges et al. (1975), for example, hypothesized and found reliable differential physiological responsiveness to methylphenidate in groups of hyperactive children selected as a function of attentional measures. Using a nested design in a similar fashion to that of Satterfield et al. (1974), Porges and associates divided their subjects into two groups based on their reaction time performance: subjects who had attentional deficits (i.e., poor reaction time performance), and those who did not (good reaction time performance). The children in the two groups were then compared on teacher ratings for changes in classroom social behavior as a function of the drug treatment. Subjects who had attentional deficits prior to treatment showed improvement on the reaction time task following methylphenidate, but this subgroup did not show a corresponding improvement in classroom social behavior. In addition, the children initially found to be “good attenders” showed no improvement in reaction time performance following drug treatment, but their social behavior improved. Thus, the particular variables affected by drug treatment may have been a function of the initial deficit of each individual child. In the experiment of Porges et al. (1975) [as in that of Satterfield e t al. (1974)l , it was possible to categorize an a priori idiosyncratic behavior into two subgroups by an analysis of individual subject performance on a task. In designs which preclude such analyses (group-statistical designs), the identification of these subgroups would be impossible. The nested design allowed the decomposition of the entire sample of subjects according t o an identifiable characteristic. On a continuum from single-subject to group designs, the nested design may be viewed as containing aspects of both. By identifying more homogeneous subjects and categorizing them into subgroups, the nested design allows a more accurate estimate of the individual while preserving the utility of group statistics. Despite the fact that there may be quantitative as well as qualitative response differences
28
Gladys B. Baxley and Judith M. LeBlanc
be tween group members, a primary concern of pediatric drug researchers is the generality of lawful cause-and-effect relationships and not the representativeness of specific pieces of data that are taken out of context (Sidman, 1960). The single-subject analysis, then, recognizes idiosyncratic differences with respect t o influences of independent variables and serves t o strengthen the validity of those individual effects. An argument for the individual analysis of drug effects, however, does not necessarily imply a small number of subjects for its own sake. The same logic that refutes the claim that the larger the group the more generalizable the findings argues also that an incomplete analysis of data for a single subject provides just as little basis for predicting drug effects for even that individual. However, when the responses of a few well-chosen subjects show an orderly and consistent relationship, adding more subjects t o the sample provides relatively little new information (Boren, 1966). A small number of subjects is often quite sufficient to permit a reasonable estimate of the adequacy of the experimental procedures.
VI. Summary and Conclusions In a n effort t o present a balanced picture of the hyperactive behavior disorder, we have attempted t o describe the hyperactive child on the basis of information gatliered from a variety of sources. The primary focus of this review has been the analysis of experimental literature related t o hyperactivity. More specifically, we have presented a discussion of: (1 j the behaviors which characterize hyperactivity; (2) the methods by which the disorder is diagnosed; ( 3 ) the theories of the etiology of the disorder; and (4) the methods used t o assess pharmacological treatment. Finally, we have presented an argument for the marriage of an individual-analysis approach with that of the traditional group design as a means of interpreting drug effect data for individual hyperactive children who may be categorized into a specific subgroup. Because children are labeled hyperactive for a variety of very complex reasons, attempts t o delineate and understand those reasons have generated much research activity. However, out of that activity has come much confusion about the trcatment of hyperactivity through pharmacological means, especially the research base which underlies that treatment. One indication of this confusion is evidenced by the variety of labels used t o describe the hyperactive child. Clements (1966) has identified no fewer than 38 different terms that refer t o the hyperactive disorder, but Cruickshank (1971) has placed the number at more than 40. Some of these terms are suggestive of neuropathology, while others refcr to the behavioral manifestations of the disorder, and still others focus on the intellectual concomitants of' hyperactivity. Despite the diversity of these labels, all appear t o be based on theories of the presumed etiology of the hyperactive disorder.
The &peractive Child
29
Of the theories presently available t o account for the etiology of hyperactivity, those related to dysfunction within the brain and central nervous system have received the most research attention. Dysfunction within the diencephelon region of the reticular activating system as well as underarousal of the central nervous system are two theories which show some promise of elucidating the cause of the hyperactive disorder. Researchers are particularly interested in exploring these areas because of the evidence that stimulant drugs may exert their primary influence within the reticular activating system. There is also considerable evidence from stimulant drug therapy t o support the view that hyperactivity may be the result of attempts to gain sufficient stimulation by a child who is experiencing insufficient sensory stimulation due t o an abnormally low level of autonomic arousal. These theories have yet t o be substantiated, but future research endeavors should serve L O clarify this issue. This discussion of hyperactivity is by no means exhaustive but is intended to highlight some of the most pertinent questions surrounding the use of psychoactive drugs in the treatment and management of the hyperactive disorderespecially the research basis for drug trcatment. An argument has been made for increased, but not exclusive, use of an individual or single-subject approach in pediatric drug research, because the sole use of group-statistical research designs is less appropriate in certain expcrimcntal situations. However, a unification of the two approaches might well be a means of enhancing our knowledge of hyperactivity as well as of hci1it:iting its treatment in individual children. This dual analytic approach offers a viable alternative path for pediatric drug researchers t o take. REFERENCES Alexandris, A., & Lundell, F. W. Fffect of tldoridazine, amphetamine and placebo on the hyperkinetic syndrome and cognitive area in mentally deficient children. Canadian Medical Association Journal, 1968, 98, 92-96. American Psychiatric Association. Diagnostic and statistical manual of mental disorders. (2nd ed.) Washington, D.C.: American Psychiatric Association, 1968. Anton, A . H., & Creer, M. Dextroaniphetarnine, catecholamines, and behavior: The effect of dextroamphetamine in retarded children. Archives of Neurology, 1969, 21, 248-252. Bakwin, H., & Bakwin, R. M . Clinical rrianagernerzt of behavior disorders in children. Philadelphia: Saunders, 1966. Baxley, G . B., & LeBlanc, J. M. 1-ffccts o f psychotropic drugs on the short-term memory of mentally retarded children. Paper presented at the 82nd annual convention of the American Psychological Association, New Orleans, September 1974. Baxley, C . B., & Marks, L. A . Tlic ctl'ccts of rnethylphenidate on the acquisition of complex visual discrimination tasks in hyperactive children. Unpublished manuscript, University of Illinois, 1974. Bazell, R. J . Panel sanctions amphetamines for hyperkinetic children. Science, 1971, 171, 12-23. Becker, W. C., Englemann, S., & Thomas, D. R. Teaching: A course in applied psychology. Chicago: Science Research Associates, 1971.
30
Gladys B. Baxley and Judith M. LeBlanc
Becker, W. C., Madsen, C. H., Jr., Arnold, C. R., & Thomas, D. R. The contingent use of teacher attention and praise in reducing classroom behavior problems. Journal of Special Education, 1967, 1,287-307. Bell, R. Q. Adaptation of small wrist watches for mechanical recording of activity in infants and children. Journal of Experimental Child Psychology, 1968, 6, 302-305. Birch, H. The problem of “brain damage” in children. In H. Birch (Ed.), Brain damage in children: Biological and social aspects. Baltimore: Williams & Wikins, 1964. Boelsche, A. Early recognition of the child with minimal brain dysfunction. Post Graduate Medicine, 1965, 38, 321-326. Boren, J. J. The study of drugs with operant techniques. In W. K. Honig (Ed.), Operant behavior: Areas of research and application. New York: Appleton-Century-Crofts, 1966. Bradley, C. The behavior of children receiving Benzedrine. American Journal of Psychiatry, 1937,94,577-585. Bradley, C. Characteristics and management of children with behavior problems associated with organic brain disease. Pediatric Clinics of North America, 1957, 4, 1049-1060. Brown, G. W. Food additives and hyperactivity. Journal of Learning Disab 62-63. Burks, H. F. The hyperkinetic child. Exceptional Children, 1960, 27, 18-26. Chess, S. Diagnosis and treatment of the hyperactive child. New York Journal o f Medicine, 1960,60, 2379-2385. Chess, S., Thomas, A., Rutter, M., & Birch, H. G. Interaction of temperament and environment in the production of behavioral disturbances in children. American Journal ofPsychiafry, 1963,120,142-148. Clements, S. D. Minimum brain dysfunction in children. NINDB Monograph No. 3. Washington, D.C. U.S. Government Printing Office, 1966. Clements, S. D., & Peters, J. E. Minimal brain dysfunctions in the school-age child. Archives of General Psychiatry, 1962,6, 185-197. Cohen, N. J., Douglas, V. I., & Morgenstern, G. The effect of methylphenidate on attentive behavior and autonomic activity in hyperactive children. Psychopharmacologia, 1971, 22, 282-294. Comly, H. H. Clinical management of the hyperkinetic impulse disorder; etiology, diagnostic criteria: Comparison of dextroamphetamine and methylphenidate effects. Reprint of paper presented at a meeting of the American Orthopsychiatric Association, Los Angeles, March 1962. Conners, C. K. The syndrome of minimal brain dysfunction: Psychological aspects. Pediatric Clinics of North America, 1967, 14, 749-766. Conners, C. K . Drugs in the management of children with learning disabilities. In L. Tarnopol (Ed.), Learning disorders in children: Diagnosis, medication and education. Boston: Little, 1971. (a) Conners, C. K . The effect of stimulant drugs on human figure drawing in children with minimal brain dysfunction. Psychopharmacologia, 1971, 19, 329-333. (b) Conners, C. K. Behavior modification by drugs: 11. Psychological effects of stimulant drugs in children with minimal brain dysfunction. Pediatrics, 1972, 49, 702-708. (a) Conners, C. K . Pharmacotherapy of psychopathology in children. In H. C. Quay and J. S. Werry (Eds.), Psychopathological disorders o f children. New York: Wiley, 1972. (b) Conners, C. K., & Eisenberg, L. The effects of methylphenidate on symptoniatology and learning in disturbed children. American Journal of Psychiatry, 1963, 120, 458-464. Conners, C. K., Eisenberg, L., & Barcai, A. Effects of dextroamphetamine on children: Studies on subjects with learning disabilities and school behavior problems. Archives of General Psychiatry, 1967, 17,478-485.
The Hyperactive Child
31
Conners, C. K., Eisenberg, L., & Shape, L. Effects of methylphenidate (Ritalin) on paired-associate learning and Porteus maze performance in emotionally disturbed children. Journal of Consulting Psychiatry, 1964,28, 14-22. Conners, C, K., & Rothschild, G. H . Drugs and learning in children. In J. Hellmuth (Ed.), Learning disorders. Vol. 3. Seattle: Special Child Publications, 1968. Conners, C. K., Rothschild, G., Eisenberg, L., Schwartz, L. S., & Robinson, E. Dextroamphetamine sulfate in children with learning disorders. Archives of General Psychiatry, 1969,21,182-190. Conners, C. K., Taylor, E., Meo, G., Kurtz, M . A., & Fournier, M. Magnesium pemoline and dextroamphetamine: A controlled study in children with minimal brain dysfunction. Psychopharmacologia, 1972,26, 32 1-336. Creager, R. D., & Van Riper, C. The effects of methylphenidate on the verbal productivity of children with cerebral dysfunction. Journal of Speech and Hearing Research, 1967, 10,623-628. Crook, W. G. An alternative method of managing the hyperactive child. Letters to the Editor. Pediatrics, 1974, 53, 656. Cruickshank, W. M. (Ed.) Psychology of exceptional children and youth. Englewood Cliffs, N.J.: Prentice-Hall, 1971. Denhoff, E. Emotional and psychological background of the neurologically handicapped child. Exceptional Children, 1961,21, 347-349. Denhoff, E., & Novack, H. S. Syndromes of cerebral dysfunction: Medical aspects that contribute t o special education methods. In N. G. Haring and R. L. Schiefelbusch (Eds.), Methods in special education. New York: McGraw-Hill, 1967. Doubros, S. G., & Daniels, G. J . An experimental approach to the reduction of overactive behavior. Behaviour Research and Therapy, 1966, 4, 251-258. Eisenberg, L. Psychiatric implications of brain damage in children. Psychiatric Quarterly, 1957, 31,72-92. Ellis, N. R., & Pryer, R. S. Quantification of gross bodily activity in children with severe neuropathology. American Journal of Mental Deficiency, 1959,63, 1034-1037. Epstein, L. C., Lasagna, L., Conners, C. K., & Rodriguez, A. Correlation of dextroamphetamine excretion and drug response in hyperkinetic children. Journal of Nervous and Mental Disorders, 1968, 146, 136-146. Estes, W. K. The problem of inference from curves based on group data. Psychological Bulletin, 1956, 53, 134-140. Eveloff, H. H. Pediatric psychopharmacology. In W. G. Clark and J. Del Giudice (Eds.), Principles of psychopharmacology. New York: Academic Press, 1970. Feingold, B. F. Why your child is hyperactive. New York: Random House, 1974. Finnerty, R. J., Soltys, J. J . , & Cole, J. 0. The use of d-amphetamine with hyperkinetic children. Psychopharrnacologia, 1971, 21, 302-308. Freeman, R. D. Drug effects on learning in children: A selective review of the past thirty years. Journal of Special Education, 1966, 1, 17-43. Freibergs, V., Douglas, V. I., & Weiss, G. The effect of chlorpromazine on concept learning in hyperactive children under two conditions of reinforcement. Psychopharrnacologia, 1968,13,299-310. Grant, Q. R. Psychopharmacology in childhood emotional and mental disorders. Journal of Pediatrics, 1962, 6 1 , 6 2 6 4 3 7 . Greenhouse, S . W., & Geisser, S. On methods in the analysis of profile data. Psychometrika, 1959,24,95-112. Grinspoon, L., & Singer, S. B. Amphetamines in the treatment of hyperkinetic children. Harvard Educational Review, 1973, 43, 5 15-555.
32
Gladys B. Baxley and Judith M. LeBlanc
Hertzig, M. E., Bortner, M., & Birch, H. G. Neurologic findings in children educationally designated as “brain-damaged.” Ainerican Journal of Orthopsychiatry, 1969, 39, 437446. Hutt, C., Hutt, S. J., & Ounsted, C. A method for the study of children’s behavior. Developmental Medicine and Child Neurology, 1963, 5, 233-245. Knights, R. M., & Hinton, G. G. The effects of methylphenidate (Ritalin) on the motor skills and behavior of children with learning problems. Journal of Nervous and Mental Disorders, 1969, 128, 6 4 3 4 5 3 . Knobel, M. Psychopharmacology for the hyperkinetic child. Archives of General Psychiatry, 1962,6,198-202. Knobel, M., & Lytton. G. J. Stimulants in psychiatric disorders of children. Abstract, Kansas City Mental Health Foundation, 2200 McCoy Street, Kansas City, Missouri, 1961. Kmger, J. M., & Safer, D. J. Type and prevalence of medication used in the treatment of hyperactive children. New England Journal of Medicine, 1974, 291, 11 18-1 120. Laufer, M. W., & Denhoff, E. Hyperkinetic behavior syndrome in children. Journal of Pediatrics, 1957, 5 0 , 4 6 3 4 7 4 . Laufer, M. W., Denhoff, E., & Solomons, G. Hyperkinetic impulse disorder in children’s behavior problems. Psychosomatic Medicine, 1957, 19, 38-49. Lee, D., & Hutt, C. A play-room designed for filming children: A note. Journal of Child Psychology and Psychiatry, 1964, 5, 263-265. Lindsley, D. B., & Henry, C. E. The effect of drugs on behavior and the clectroencephalograms of children with behavior disorder. Ps.vchosomatic Medicine, 1942, 4, 140-149. Lytton, G. J . , & Knobel, M. Diagnosis and treatment of bchavior disorders in children. Diseases of the Nervous System, 1959, 20, 334-340. Martin, D. M. Hyperkinetic bchavior disorders in children: Clinical results with nietliylphenidate hydrochloride (Ritalin). Western Medicine, 1967, 8 , 23-27. McFarland, J. N., Peacock, L. J., &Watson, J. A. Mental retardation and activity level in rats and children. American Journal of Mental Deficiency, 1966, 71, 381-386. Menkes, M. M., Rose, J. S., & Menkes, J. H. A twenty-five year follow-up study on the hyperkinetic child with minimal brain dysfunction. Pediatrics, 1967, 39(3), 393-399. Millichap, J. G. Drugs in management of hyperkinetic and perceptually handicapped children. Journal of the American Medical Association, 1968, 206, 1527-1530. .Millichap, J. G. Drugs in management of minimal brain dysfunction. Annals of the New York Academy of Science, 1973, 205, 321-334. Mithhdp, J. G., Aymdt, F., Sturgis, L. H., Larsen, K. W., & Egan, R. A. Hyperkinetic behavior and learning disorders. American Journal of Diseases of Childhood. 1968, 116, 2 35-244. Millichap, J . G., & Boldrey, E. E. Studies in hyperkinetic behavior: 11. Laboratory and clinical evaluations of drug treatments. Neurology, 1967, 17, 4 6 7 4 7 1 . Millichap, J . G., & Fowler, G. W. Treatment of “minimal brain dysfunction” syndromes. Pediatric Clinics o f N o r t h America, 1967, 14, 161-771. Minde, K., & Webb, G. Studies o n the hyperactive child: Vl. Prenatal and paranatal factors associated with hyperactivity. Developmental Medicine and Child Neurology, 1968, 10, 355-363. Molitch, M., & Eccles, A. The effect of Benzedrine Sulfate o n the intelligence scores of children. American Journal of Psychiatry, 1937, 94, 587-590. Nichamin, S. J., & Barahal, G. D. Faulty neurologic integration with perceptual disorders in children. Michigan Medicine, 1968, 67, 1071-1075. Paine, R. S. Syndromes of “minimal cerebral damage.” Pediatric Clinics of North America, 1968,15, 779-801.
The Hyperactive Child
33
Paine, R. S., Werry, J. S., & Quay, H . C . A study of “minimal cerebral dysfunction.” Developmental Medicine and Child Neurology, 1968, 10, 505-520. Pasamanick, B., Rogers, M., & Lilienteld, A. M. Pregnancy experience and the development Journal of Psychiatry, 1956, 112, 613-61 8. of behavior disorder in children. ,~ltneric~arz Patterson, G. R., Jones, R., Whittier, J., & Wright, M. A. A behavior modification technique for the hyperactive child. Behavioitr Research and Therapy, 1965, 2, 21 7-226. Peterson, D. R. Behavior problems o f middle childhood. Journal of Consulting Psychology, 1961,25, 205-209. Porges, S. W., Walter, G. F., Korb, R. J., & Spraguc, R. L. The influences of methylphenidate on heart rate and behavioral measures of attention in hyperactive children. Child Development, 1975, 46, 727-7 33. Rickels, K., Lipman, R . S., I’islit.r, S . , Park, L. C., & Uhlenhutli, E. H . Is a double-blind clinical trial really double-blind’! A report of doctor’s medication guesses. Psychopharmacologia, 1970,16, 329-336. Rodin, E., Lucas, A , , & Simson, C. A study of behavior disorders in children by means of general purpose computers. Proceedings of the conference on data acquisition and processing in biological medicitie. Oxford: Pergamon, 1963. Sainz, A . Hyperkinetic disease of children: Diagnosis and therapy. Diseases of the Nervous System, 1966, 21,48-50. Satterfield, J. H. EEG issues i n L,liildren with minimal brain dysfunction. Seminar in Psychiatry, 1973, 5, 35-46. Satterfield, J. H., Cantwell, D. P., Lesser, L. I . , & Podosin, R. D. Physiological studies of the hyperkinetic child: I . American Journal ofPsvchiatry, 1972, 128, 102-108. Satterfield, J. H., Cantwell, D. P., & Satterfield, B . T. Pathophysiology of the hyperactive child syndrome. Archives of (;enera/ Psychiatry, 1974, 31, 839-844. Satterfield, J. H., Cantwell, D. I)., Saul, R . E., & Lesser, L. I. Response to stimulant drug treatment in hyperactive children: Prediction from EEG and neurological findings. Journal of Autism and Childhood Schizophrenia, 1973, 3, 36-48. Satterfield, J. H., & Dawson, M. I . 13ectrodermal correlates of hyperactivity in children. Psychophysiology, 197 1 , 8, I 9 1-1 97. Schulman, J. L., Kaspar, J . C., & Throne, J. M. Brain damage and behavior: A clinical-experimental study. Springfield, 111.: Thomas, 1965. Schulman, J. L., & Reisman, J. M . A n objective measure of hyperactivity. American Journal of Mental Deficiency, 1959, 6 4 , 4 5 5 4 5 6 . Shannon, W. R. Neuropathic manifestations in infants and children as a result of anaphylactic reactions t o foods contained in their dietary. American Journal of Diseases of Childhood, 1922, 24, 89. Sidman, M . A note on functional relations obtained from group data. Psychological Bulletin, 1952,49, 263-269. Sidman, M . Tactics of scientific research. New York: Basic Books, 1960. Silbergeld, E. K., & Goldbcrg, A . M. Lead-induced behavioral dysfunction: An animal model of hyperactivity. Experimental Neurology, 1974, 42, 146-157. Solomons, G. The hyperactive child. .Journal o f Iowa Medical Society, 1965, 55, 464-469. Sprague, R. L., Barnes, K. R., & Werry, J. S. Methylphenidate and thioridazine: Learning, reaction time, activity, and classroom behavior in disturbed children. American Journal of Orthopsychiatry, 1970,40,615-628. Sprague, R. L., & Toppe, L. K . Relationship between activity level and delay of reinforcement. Journal of Experimental Child l’.qdiology, 1966, 3, 390-397. Sprague, R. L., & Werry, J . Methodology of psychopharmacological studies with the retarded. In N. R. Ellis ( .), International review of research in mental retardation. Vol. 5 . New York: Academic Press, 1971.
34
Gladys B. Baxley and Judith M. LeBlanc
Sprague, K. L., Werry, J. S., & Davis, K. Psychotropic drug effects o n learning and activity level of children. Paper presented at the Gatlinburg Conference o n Research and Theory in Mental Retardation, March 1969. Sprague, R. L., Werry, J. S., & Scott, K. G. Effects of dextroamphetamine on activity level and learning in retarded children. Paper presented at the meeting of the Midwestern Psychological Association, Chicago, May 1967. Stevens, J. R., Sachdev, K., & Milstein, V. Behavior disorders of childhood and the EEG. Archives o f Neurology, 1968, 18, 160-177. Stewart, M. A. Hyperactive child syndrome recognized 100 years ago. Journal of the American Medical Association, 1967, 202, 28-29. Stewart, M. A., Pitts, F., Craig, A., & Dieruf, W. The hyperactive child syndrome. American Journal of Orthopsychiatry, 1966, 36,861-867. Sykes, D. H., Douglas, V. I., & Morgenstern, G. The effect of methylphenidate (Ritalin) o n sustained attention in hyperactive children. Psychopharmucologia, 1972, 25, 262-274. Sykes, D. H., Douglas, V. I., Weiss, C., & Minde, K. K. Attention in hyperactive children and the effect of methylphenidate (Ritalin). Journal of Child Psychology and Psychiatry, 1971, 12, 129-139. US. Department of Health, Education, and Welfare. Report of the conference on the use of stimulant drugs in the treatment of behaviorally disturbed young school children. HEW News, 1971, March 10. Warren, R. J., Karduck, W. A., Bussaratid, S., Stewart, M. A,, & Sly, W. S. The hyperactive child syndrome. Archives of General Psychiatry, 1971,24, 161-162. Weiss, G., Werry, J. S., Minde, K., Douglas, V. I., & Sykes, D. Studies on the hyperactive child: V. The effects of dextroamphetamine and chlorpromazine o n behavior and intellectual functioning. Journal of Child Psychology and Psychiatry, 1968, 9, 145-156. Wender, P. H. Minimal brain dysfunction in children. New York: Wiley (Interscience), 1971. Werry, J . S. Developmental hyperactivity. Pediatric Clinics of North America, 1968, 15, 581-599. (a) Werry, J . S. Studies of the hyperactive child: IV. An empirical analysis of the minimal brain dysfunction syndrome. Archives of General Psychiatry, 1968, 19,9-16. (b) Werry, J. S. The diagnosis, etiology, and treatment of hyperactivity in children. In J. Hellmuth (Ed.), Learning disorders. Vol. 3. Seattle: Special Child Publications, 1968. (c) Werry, J. S., & Quay, H. C. Observing the classroom behavior of elementary school children. Exceptional Children, 1969, 35,461-472. Werry, J . S., & Sprague, R. L. Hyperactivity. In C. G. Costello (Ed.), Symptoms of psychopathology. New York: Wiley, 1970. Werry, J. S., Weiss, G., & Douglas, V. I. Studies o n the hyperactive child: I. Some preliminary findings. Conadian Psychiatric Association Journal, 1964,9, 120-1 30. Werry, J. S., Weiss, G., Douglas, V. I., & Martin, J. Studies on the hyperactive child: 111. The effect of chlorpromazine upon behavior and learning. American Academy of Child Psychiatry Journal, 1966,5,292-312. Winer, B. J. Statistical principles in experimental design. New York: McGraw-Hill, 1962. Winsberg, B. G., Bisler, I., Kupietz, S., & Tobias, J. Effects of imipramine and dextroamphetamine on behavior of neuropsychiatrically impaired children. American Journal of Psychiatry, 1972, 128, 1425-1431. Zentall. S. Optimal stimulation as theoretical basis of hyperactivity. American Journal of Orthopsychiatry, 1975,45, 549-563.
PERIPHERAL AND NEUROCHEMICAL PARALLELS OF PSYCHOPATHOLOGY: A PSYCHOPHYSIOLOGICAL MODEL RELATING AUTONOMIC IMBALANCE TO HYPERACTIVITY. PSYCHOPATHY. AND AUTISM
Stephen W . Porges U N I V E R S I T Y OF ILLINOIS
I . OVERVIEW
..........................................
I1. HYPERACTIVITY
36
.....................................
37
I l l . PSYCHOPATHY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
38
IV . RELATIONSHIP BETWEEN CENTRAL AND PERIPHERAL AUTONOMIC ACTIVITY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
40
V . STATISTICAL MODEL OF AUTONOMIC BALANCE . . . . . . . . . A . PHYSIOLOGICAL ASSUMPTIONS . . . . . . . . . . . . . . . . . . . . . B. STATISTICAL TECIINIQUES . . . . . . . . . . . . . . . . . . . . . . . . . C . TEST O F MODEL W I T H HYPERACTIVE CHILDREN . . . . . .
43 43 44 45
VI . AUTONOMIC RESPONSES AND ATTENTION . . . . . . . . . . . . . . A . TWO-COMPONENT MODEL O F ATTENTION . . . . . . . . . . . . B. RELATION BETWEEN TIlE TWO-COMPONENT MODEL AND AUTONOMIC AND NEUROTRANSMITTER ACTIVITY
47 50
VII . AUTISM
.............................................
55
VIII . SUMMARY AND CONCLUSIONS . . . . . . . . . . . . . . . . . . . . . . . . .
60
REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
51
62 35
36
Stephen W.Porges
1. Overview The balance between the antagonistic components of the autonomic nervous system has been of interest for many years. Many theories have evolved which implicate autonomic balance with behavioral and psychological pathologies. Eppinger and Hess (1 9 1 5) theorized that specific nervous disorders may be associated with pathological autonomic balance and that autonomic balance may b e controlled by the central nervous system. Thus, speculating that the pathologies of autonomic balance were merely reflections of pathologies in the central nervous system, Eppinger and Hess used primitive pharmacological tests with vagotropic and sympathicotropic substances as a basis for their diagnostic procedures and as support for their theory. Subsequent tests of the theory produced inconsistent results. Wilder (1931), attempting t o implicate autonomic balance as an indicator of autonomic responses, formulated the Law of Initial Values, which drew attention to the importance of the state of the autonomic nervous system prior t o stimulus manipulations. The ‘‘law’’ has never been adequately applied t o psychophysiological research, since it requires the assumption of a linear regression between base-level activity and response direction and magnitude. Autonomic response systems are seldom determined solely by either sympathetic or parasympathetic influences but often by a relationship between these two antagonistic systems, and therefore a linear additive model may not be appropriate. Wenger (1941) utilized factor-analytic procedures t o assess autonomic balance. This approach is important in psychophysiological research, where multiple measures of the autonomic nervous system are taken and where correlations among variables are often low. Wenger modified and extended the Eppinger and Hess hypothesis of autonomic balance. The restated hypothesis consisted of two parts: (1) either phasic or chronic predominance of the adrenergic (sympathetic) or cholinergic (parasympathetic) branches of the autonomic nervous system may create a state of autonomic imbalance; (2) the relationship between adrenergic and cholinergc branches of the autonomic nervous system is distributed continuously about a central tendency which operationally defines autonomic balance. Research on autonomic balance has not been extremely fruitful, partly because the techniques developed to assess autonomic balance have been inadequate. However, the concept of autonomic balance is still theoretically associated with behavioral disorders characterized by defective attention. The balance in the autonomic nervous system has been theorized t o parallel the balance between the catecholaminergic and cholinergic neurotransmitter systems in the brain (Anisman, 1975; Broverman, Klaiber, Vogel, & Kobayashi, 1974: Gellhorn, 1957). Thus, the excitatory activity of the catecholamines may be “reflected” in the sympathetic nervous system and the inhibitory activity of central acetylcholine may be “reflected” in the parasympathetic nervous system.
Peripheral Parallels of Psychopathology
37
There appear t o be behavioral parallels of autonomic imbalance. For example, many psychopathologies that have as a defining characteristic an attentional or response deficit are associated with atypical autonomic nervous system activity. This chapter covers three of these psychopathologies: hyperactivity, psychopathy, and early infantile autism. Hyperactivity and psychopathy are discussed as examples of behavioral pathologies that may be theoretically explained in terms of autonomic imbalance. Early infantile autism is discussed after prcsentation of a statistical model of autonomic balance and a model of attention which links autonomic response to neurotransmitter activity. The models are used to predict an underlying physiological substrate of autism, and the predictions are evaluated against the available psychophysiological and biochemical research.
11. Hyperactivity Hyperactivity is a generalixd symptom which has been used to categorize a population of individuals who exhibit a lack of control of spontaneous activity. A diagnosis of hyperactivity is often associated with abnormally high levels of motor activity, short attention span, low frustration tolerance, hyperexcitability, and an inability to control itnpulses (Clements & Peters, 1962; Douglas, 1972; Laufer & Denhoff, 1957; Millichap & Fowler, 1967). Several physiological models have been proposed to explain hyperactivity. The high activity level has been interpreted as a parallel of an overaroused or highly aroused central nervous system (Freibergs & Douglas, 1969; Laufer, Denhoff, & Solomons, 1957), as a compensatory behavior to raise the arousal of a suboptimally aroused individual via an increase in proprioceptive sensory input (Satterfield & Dawson, 197 I : Stewart. 1970; Werry, Sprague, Weiss, & Minde, 1970), or as a correlate of defective cortical inhibitory mechanisms (Dykman, Ackerman, Clements, & Peters, 197 1 ). In a recent study, Porges, Walter, Korb, and Sprague (1975) demonstrated that hyperactive children who exhibited attentional deficits, defined b y long response latencies, also had heart-rate responses that were theoretically incompatible with sustained attention. Heart-rate responses theoretically associated with sustained attention are characterized by parasympathetic influences on autonomic activity (reduced heart rate and reduced heart -rate variability). When methylphenidate was administered, reaction-time performance improved, heart rate changed from acceleration t o deceleration, and heart-rate variability changed from increases t o decreases. The foregoing research summary might suggest that the hyperactive child is deficient in levels of parasympathetic activity and is dominated by activity of the sympathetic system. Centrally. the neurochemical balance might reflect high catecholaminergic levels and low cholinergic levels. Anisnian (1 975) suggested this pattern and discussed the S U C C ~ S So f sympathomimetic drugs in treating
38
Stephen bf, Porges
behavioral hyperactivity. He suggested that these catecholnminergic agonists elicit a cholinergic rebound which may facilitate mediated inhibitory behaviors. Unpublished research in our laboratory has produced strong evidence that the vagally mediated relationship between heart rate and respiration (a measure of parasympathetic influence) increases in the hyperactive child when low doses of methylphenidate (a sympathomimetic drug) are administered. Moreover, the component of lieart-rate activity that is not vagally mediated (i.e., is sympathetically controlled) decreases with methylphenidate. This pattern of results suggests that the hypotheses relating sympathetic dominance to hyperactivity may be accurate and that, when the autonomic balance shifts from sympathetic t o parasympathetic, behavior becomes more nearly normal.
111. Psychopathy Although the term “psychopath” is often equated with criminal or asocial behavior, a number of behavioral characteristics define this psychopathology (Clcckley, 1955). It is generally believed that the psychopath does not exhibit signs of psychosis, does not hallucinate, and does not have delusions. The psychopath is often charming and intelligent, yet he violates laws even though he has been previously punished for these same behaviors. He often denies blame for misdeeds and frequently lies and cheats. Of importance in defining the psychopath is his resistance t o social conditioning; social reinforcers seem t o be ineffective. Moreover, he is relatively free from anxiety. Quay (1965) proposed a “stimulus-seeking” theory of psychopathic behavior. I n this theory, the psychopath seeks stimulation or increased stimulus input because he finds himself in a state similar t o stimulus deprivation. Quay (1976) speculated that the psychopath exhibits an autonomic dysfunction which may be characterized by low reactivity. This dysfunction would suggest that the psychopath is parasympathetically dominant and exhibits low levels of Sympathetic activity. In support of this hypothesis, Wheeler (1974) hypothesized that the adrenergic-cholinergic balance of the psychopath’s autonomic nervous system is in the direction of cholinergic dominance. Wheeler reported behavioral data supporting this hypothesis. He demonstrated that psychopaths perform poorly on automatization tasks and well on restructuring tasks. Automatization tasks are simple repetitive tasks; restructuring tasks require restructuring of obvious percepts in order to identify some hidden, less obvious stimulus relationship. These types of tasks, according t o Broverman, Klaiber, Kobayashi, and Vogel (1968), are not only associated with individual differences in cognitive styles, but are also indicators of underlying physiological and biochemical differences. Strong automatization is related t o a dominance of central autonomic nervous system adrenergic functioning, and weak automatiza-
Peripheral Parallels of Ps.vchopatholog,v
39
tion coupled with strong restructuring performance is related t o a dominance of central autonomic nervous system cholinergic functioning. Thus, Wheeler’s behavioral data may be used, according t o Broverman et al. (1968), as support for the hypothesis that psychopaths exhibit cholinergic dominance in the peripheral and central autonomic nervous system. Other investigators have reported reduced sympathetic activity in the psychopathy. Lykken (1957) reported that “primary” sociopaths exhibit both reduced amplitude and conditionability of the galvanic skin reflex (CSR). Lippert and Senter (1966) reported that psychopaths exhibit a lower frequency of spontaneous GSRs during rest, smaller increases over trials, decreases below resting level following the experimental manipulation, and more rapid habituation to repeated stimulation. Two behavioral characteristics have been associated with the psychopath: (a) limited reactivity t o environmental stimuli, and (b) an inability t o learn social contingencies, especially avoidance learning. It is possible that these two characteristics have a common mechanism which is related to autonomic imbalance. The limited reactivity of the psychopath may be postulated t o parallel attenuated sympathetic nervous system and central catecholaminergic activities, since these systems have been implicated in excitatory behavior. To explain the poor performance in avoidance learning, a general model relating the neurotransmitters of the central autonomic nervous system t o behavior has been postulated by Anisman (1975). According t o Anisman, when an aversive stimulus is presented, there is, in normal individuals, a catecholamine reaction which should facilitate avoidance behavior. This biochemical shift elicits a cholinergic rebound resulting in the behavioral inhibition associated with poor avoidance performance. Applying this model t o the psychopath, it is possible to speculate that individuals who are deficient in catecholaminergic activity or who have dominant cholinergic systems will perform inadequately on avoidance learning tasks. Thus, the psychopath who is hypothesized to have a less reactive catecholaminergic system and a dominant cholinergic system would be expected to exhibit attenuated responses t o external stimulation and poor avoidance behavior. Another characteristic of the psychopath is a limited ability t o express affect. Is it possible that this affective “flatness” is related to a dysfunction in autonomic or central (adrenergic-cholinergic) balance? Again, it is first necessary t o describe a relationship between adrenergic-cholinergic balance and emotion in a normal individual and then t o project this model to the psychopath. Solomon and Corbit (1974) have generated a model of emotional behavior which assumes that specific systems in the brain “suppress or reduce all excursions from hedonic neutrality” (p. 143). These systems attempt to maintain an emotional homeostasis when the organism is confronted with situations that are appetitive or aversive. The systems operate t o decrease the intensity of the subjective
40
Stephen W.Porges
“liedonic quality” (i.e., emotion, affect, arousal) of the objective reinforcing properties of stimuli. The systems are automatic and are activated when there is a significant departure from “affective equilibrium” as a consequence of stimulation. Solomon and Corbit postulated an “opponent-process” which is precipitated by and is opposed to the stimulus-aroused affective state. They presented heart-rate conditioning data which suggest that the “opponent-process” is associated with vagal activity (parasympathetic/cholinergic) while tlie stimulus-aroused state is associated with vagal inhibition or generalized increased sympathetic activity. There is a similarity between the models proposed by Anisinan (1975) and b y Solomon and Corbit (1974). According to Anisman, the cholinergic system acts in response to the stimulus-aroused catecholaminergic state. Moreover, the cholinergic rebound is more sluggish and longer lasting. I n concordance with the above statements, Solomon and Corbit’s “Opponeiit-process” may parallel parasympathetic/cholinergic influences, whereas the “stimulus-aroused affective state” may parallel sympatlietic/catecholaniinergic influences. If the psychopath actually exhibits an autonomic imbalance characterized b y parasympathetic dominance and attenuated sympathetic activity, one would expect the psychopath to exhibit a resident or tonic “opponent-process” that would negate or reduce the affective response to ariy stimulus situation. Thus, the psychopath would behave as if lie were not affected by emotional events, whether pleasant or unpleasant.
IV. Relationship between Central and Peripheral Autonomic Activity Gellhorn (1957) demonstrated that not only does a balance exist between the sympathetic and parasympathetic systems, but also excitation of one system might result in the activation of the complementary system. Specifically, the parasympathetic reaction could be directly related to tlie iiiagnitude (intensity, frequency, or duration) of the preceding sympathetic stimulation. The function of this “principle of reciprocity” is t o maintain neurochemical homeostasis between the antagonistic portions of the autonomic nervous system. Moreover, the rebound presumably is mediated by the baroreceptors of the carotid sinus and aortic arch, since transection of nerves connecting the baroreceptors with the medulla prevents the occurrence of rebound effects. Recently, it has been observed that sympathetic excitation may result in a central cholinergic rebound similar to that stipulated by Gellhorn regarding parasympathetic activity. In addition, the balance resulting from the central innervation may modulate behaviors such as general activity and avoidance
Peripheral Parallels v f Psychopathology
41
behavior. Manto (1967) observed that 1 hour after injection of epinephrine (a peripheral sympathominietic drug), there was a decline in general activity. The decline in activity was reversed with scopolamine (an anticholiriergic drug with central sites of action) but not by the peripheral analog methylscopolamine (a parasympathetic blocker). This finding suggests that peripheral adrenergic stiniulation niay result in an inhibitory central cholinergic rebound and that the parasympathetic dominance that s e e m to exist following intense sympathetic activity can be attributed to a central rather than a peripheral process. Another neurotransmitter, serotonin, has been implicated in the mediation of behavior. Williams, Hamilton, and Carlton (1974) presented data which suggest that two different categories of behavior, reflexive and voluntary, may be pharmacologically extracted from a coninion habituation task. Specifically, they found that the norrnally observed decrenient of tlie reflexive response to the change in stimulation was relatively unaffected by an anticholinergic drug, scopolamine, while the decrement in voluntary exploratory behavior was profoundly attenuated. In contrast, when para-cliloroplienylalanine (PCPA), an antiserotonergic drug, was administercd, the decrement of the reflexive response was attentuated and the decrement in voluntary exploratory behavior was unaffected. These data suggest that the habituation of reflexive exploratory behavior was unaffected; they also suggest that tlie habituation of reflexive behaviors is mediated by serotoninergic systems and that the habituation of voluntary behaviors is mediated by cholinergic svstenis. Moreover, lesions to the medial septum, which contains areas sensitive t o both serotonin and acetylcholine, attenuated both reflexive and exploratory decrements (Williams et aL, 1974). These lesions may have disrupted the activities of both serotonin and acetylcholine. In developing models o f the relative influences of serotonergic and cholinergic systems on behavior, it is clear that both systems are inhibitoryserotonergic systems inhibiting reflexive behaviors, and cholinergic systems inhibiting voluntary or operant bcliaviors. However, it is very difficult to assess their independent effects on behavior, since most behaviors are complex and contain both reflexive and voluntary components (Swonger & Rech, 1972). Phylogenetically and ontogenetically, one niay observe the sequential development of three neurotransmitter systems: catecholaminergic, serotonergic, and cholinergic. Anatornically, the distribution of these neurotransmitters may reflect the evolution of these self-regulatory systems from the brainstem (exhibiting a high density of catecliolaniiiiergic areas) to the midbrain (exhibiting a high density of serotonergic areas), and then to the forebrain (exhibiting a high density of cholinergic areas). The distribution of the neurotransmitters in the brain is very complicated, and there is evidence of abundant interactions and overlapping distributions among tlicse systems. A simplistic view of their influences is presented in Table I.
Stephen W.Porges
42
TABLE 1 Autonomic and Behavioral Parallels of Neurotransmitter Activity
Neurotransmitters
Influences o n behavior
Peripheral autonomic activity
Increases in general activity
Sympathetic responses
Serotonin
Decreases amplitude of reflexive behavior
Attentuation of sympathetic and parasympathetic reflexive responses
Acetylcholine
Inhibits spontaneous activity and is associated with voluntary behavior
Parasympathetic responses
Catecholamines
(epinephrine; norepinephrine; dopamine)
Earlier in this chapter we discussed the relationship between two behavioral disorders, hyperactivity and psychopathy, and autonomic imbalance. The autonomic imbalance hypothetically parallels an imbalance in the neurotransmitter systems. Thus, it might be predicted that the hyperactive individual would exhibit high levels of catecholaminergic activity relative to cholinergic activity and that the psychopath would exhibit the converse. Davis, Janowsky, ElYousef, and Sekerke (1973) have also speculated that mental health may be a function of a balance between these neurotransmitter systems. They believe that “it is reasonable t o consider that there may be a dimension in men which goes from an excited-activated-euphoric state t o an inhibited-retarded-depressed state, and that these behavioral states may have as their basis, a balance between adrenergic and cholinergic factors” (p. 680). Behavioral conditions, often observed in children suffering from defective attention, are states of hyperreactivity and hypoactivity. Hyperreactivity might be associated with deficient serotonergic systems which function t o inhibit reflexive responses to changes in stimulation. Hypoactivity might be associated with weak catecholaminergic systems, as in the psychopath. The behavioral consequence of weak catecholaminergic activity may be partially a function of high cholinergic levels which tend to inhibit catecholaminergic activity. Low levels of activity in both the catecholaminergic and cholinergic systems might result in lethargic behavior and a lack of structured cognitive activity. Retarded children exhibited low levels of autonomic activity in both sympathetic and
Peripheral Parallels oJPs.vchopatholog),
43
parasympathetic systems in a preliminary study conducted recently by the writer, suggesting low levels of b o t h catecholaniinergic and cholinergic activity. Low levels of catecholaniinergic activity might result in a compensatory behavior of stimulus seeking (proposed by Quay for the psychopath), whereas normal or high levels of cholinergic activity should be associated with proficient performance on complicated mediated cognitive tasks (Brovernian et al., 1968). The physiological relationship between central cholinergic-catecholaminergic balance and autonomic parasynipathetic-sympathetic balance is quite strong, as shown by the following points: ( I ) the transniitters in the two antagonistic peripheral systems biochemically parallel the central neurotransmitters; and (2) changes in the levels and actions of the central neurotransmitters are reflected in parallel changes in activity in the peripheral autonomic nervous system. Pharmacological interventions which directly affect central neurotransmitter activity also influence the peripheral autonomic nervous system. Pharmacological interventions which affect only peripheral autonomic activity have only slight influences on central neurotransmitter activity. This relationship suggests that although the autonomic nervous systeni may afford a noninvasive opportunity to monitor the balance of central catecliolaminergic and cholinergic systems, its direct influence on central nervous system activity-and b y inference, behavioris not great. Thus, the clear demonstration that sympathetic and parasynipathetic responses are spontaneously mediated centrally b y the central neurotransmitter systems enables psycliophysiologists t o utilize measures of autonomic balance as viable elements for understanding behavior.
V. Statistical Model of Autonomic Balance Recently, the author devised a simple noninvasive technique to assess autonomic balance b y measuring respiration and heart rate. 'The technique is extremely simple, although the statistical and physiological assumptions are extremely complicated. A . PHYSIOLOGICAL ASSUMPTIONS
Autonomic balance may be assessed by investigating the rhythm of the heart, since that rhythm ( R H )is deternnned by three causes: influence of the sympathetic nervous system (S); influence of the parasympathetic nervous system (P); and effects intrinsic t o the heart ( I ) . This relation may be expressed as:
44
Stephen W.Porges
If we assume that the intrinsic component is determined solely by individual differences in heart tissue independent of autonomic influences, the parameters of the intrinsic influence should be randomly distributed relative to the autonomic influences. Thus, the mean of any subpopulation selected by sympathetic nervous system or parasympathetic nervous system levels should result in the following function (K is a constant):
since
Z(Q=K
(3)
when I is random relative t o the fluctuations in the sympathetic nervous system and the parasympathetic nervous system. Given the relationship in Eq. 2, the problem is to partition the relative contributions of the parasympathetic nervous system and the sympathetic nervous system. Since the respiratory influences on heart-rate activity are mediated by the parasympathetic nervous system via the vagus (Chess, Tam, & Calaresu, 1975; Clynes, 1960; Eckoldt & Schubert, 1975; Katona & Jih, 1975; Levy, DeCeest, & Zieske, 1966), it might be possible t o assess the relative parasympathetic nervous system and sympathetic nervous system influences on heart-rate activity by quantifying the respiratory influence on heart rate. Chess et al. have demonstrated through pharmacological and surgical procedures with an animal preparation that the respiratory influences on heart-rate activity are vagally mediated and that the remaining heart-rate activity is primarily sympathetically mediated. Thus, the assessment of respiratory influences on the spontaneous pattern of heart rate may be used as a noninvasive measure of parasympathetic influences. B. STATISTICAL TECHNIQUES
Spectral analysis (a time-series statistical procedure; see, e.g., Box & Jenkins, 1970) may be used to detect rhythmic variations in a record of physiological events by decomposing the record into a series of sinusoidal functions of different frequencies. Frequency in time-series statistics refers to the duration of each sinusoidal function transformed in terms of the proportion of the function occurring within 1 second. Thus, an event (pattern of increases and decreases in heart rate or respiration amplitude) which could be described by a sine wave with a 5-second duration has a frequency of .2 Hz, since one-fifth of the function would occur within 1 second. The spectral-density function describes (with regard to frequency) t!ie distribution of physiological events.
Peripheral Parallels of Psychopathology
45
When two variables such as heart rate and respiration are simultaneously recorded, cross-spectral analysis may be used t o investigate whether frequency components in the two variables are related. The cross-spectral estimate displays peaks corresponding to the frequencies of coincident peaks in the individual spectral estimates, and their amplitudes are related t o the product of the amplitudes of the individual peaks. The measure emphasized in the present paper is called coherence and is related t o the influence of any observed rhythm in one variable on an observed rhythm in the second. Coherence is a normalized function with values between 0 and 1 . The coherence function is analogous t o a correlation coefficient defined at each frequency in the spectrum. The writer has developed a modification of the coherence measure which is weighted as a function of the actual heart-rate activity occurring at the specific frequencies associated with respiration. This weighted coherence measure identifies the percentage of heart-rate activity shared with respiration. The percentage is then used as an estimate of autonomic balance. The following equation is used t o define C,,, , the weighted coherence measure:
where coherence squared (C’) is the shared variance of heart rate and respiration and FH is the power density of heart rate (a measure of variance) at each frequency (A) at which respiration normally occurs (.2 to .5 Hz or 12 to 30 times a minute). The product of these values is summed over the dominant frequencies of respiration. The natural logarithm of the numerator, which represents the variance of heart rate shared with respiration, is operationally defined as the parasympathetic component. The denominator minus the numerator represents the variance of heart rate not accounted for by respiration, and the natural logarithm of this difference is operationally defined as the sympathetic component. C. TEST OF MODFL WIT11 t!YPERACTIVE CHILDREN
To test the proposed method of assessing autonomic balance and the model relating autonomic balance to behavioral psychopathology, a sample of children diagnosed as hyperactive was studied. To assess autonomic balance, spontaneous heart rate and respiration were recorded before and after a task demanding attention. Earlier, in the discussion of hyperactivity (Section In), it was postulated that if behavioral pathology is related t o autonomic balance, the hyperactive child might exhibit a balance reflecting sympathetic dominance. When heart-rate responses during attention were assessed in an earlier study with hyperactive children (Porges rt ul., 19751, the assumed parasympathetic response
46
Stephen Ct’. Porges
components associated with decreased heart rate and reduced heart-rate variability were observable only with the pharmacological intervcntion of niethylphenidate administration. The next study was addressed to the question of whether autonomic balance assessed by an estimate of weighted coherence (Eq. 4) would reflect the hyperactive child’s defective behavior and whether this measure of autonomic balance is sensitive to drug treatments that are known to have clear behavioral consequences. Weighted coherence was calculated for scven hyperactive subjects, each testcd in the following five conditions: (a) placebo; (b) .5 nig/kg methylphenidate; (c) . l nig/kg dextroamphetamine; (d) .?S nig/kg dextroamphetamine; and ( c ) off drug after these treatments. The sequence of conditions. excepting the Posttreatment condition, was randomized. Seven normal cluldren. matched in age with the hyperactive children, were tested without drugs. As illustrated in Fig. 1 , weighted coherence varied as a function of the specific drug treatment. Although both methylphenidate and dextroamphetamine are classified as stimulant drugs, the weighted coherence of the hyperactive children was higher during the methylphenidate condition than during the other drug treatments. However, the weighted coherence of nornial age-matched control children was higher than all the drug treatment levels for the hyperactive children. T i e data from this study may be interpreted to reflect two points: (1) The hyperactive child’s weighted coherence reflects a sympathetic dominance, whereas the coherence of the normal age-matched child reflects a greater parasympathetic influence. (2) Administration of methylphenidate, the preferred treatment for hyperactivity, selectively influences autonomic balance, but treatment with the other stimulant, dextroamphetamine, does not. A further study was undertaken t o answer two questions: First, do dosage levels of methylphenidate differentially influence the measure of autonomic balance? Second, is there an indication of autonomic imbalance in the hyperactive child prior to drug intervention? T o answer these questions, 13 hyperactive chldren were tested in the following conditions: (a) pretreatment; (b) .5 mg/kg methylphenidate; (c) 1.O mg/kg methylphenidate; (d) placebo; (e) . l mg/kg dextroamphetamine; (0.25 mg/kg dextroamphetamine; and (g) off drug after these treatments. All conditions except the pre- and posttreatment conditions were randomized. As illustrated in Fig. 2, the weighted coherence was significantly higher with the low dosage of methylphenidate than in the other conditions, which did not differ from each other. Of interest is the effect of the high dosage of methylphenidate, which, although not affecting the autonomic balance, actually suppressed autonomic activity by lowering both the sympathetic and parasympathetic components. Behaviorally, the preferred dosage has been low (between .3 and S mg/kg), since it has a positive influence on cognitive and social behavior (Sprague & Sleator, 1975); on the other hand, the high dosage often results in lethargic behavior as well as suppressed autonomic activity.
Peripheral Parallels of Psychopathology
4 ‘1 I
LDDPX
tl,llrr
f’Li
Rlt
Post
NOll71
Driig
Fig. 1. Estimate of weighted cohcrence for seven hyperactive children on .I mglkg dextroamphetamine (LoDex), .25 tnglkg dextroamphetamine (HiDex), placebo (Plac), .5 mglkg methylphenidate (Rit), mid off drug after the preceding drug treatments (Post), compared with seven norinal age-rnatched children (Norm).
It also appeared in the aforementioned pilot studies that, as autonomic balance shifted toward reduced sympathetic dominance, the hyperactive child’s behavior improved socially and cognitively . Based upon teacher ratings, the hyperactive child who is being treated with a low dosage of methylphenidate exhibits improved control of impulsive behavior and greater attentiveness within the classroom situation. Moreover. the proposed, technique of assessing autonomic balance seems t o be sensitive to the pharmacological intervention, and the model relating autonomic balance to behavioral psychopathology seems t o be reflected in the hyperactive child’s behavior.
VI. Autonomic Responses and Attention The term attention appears to most psychologists to be a vague construct; it is generally not defined and is used to explain behavior that can not be explained
48
Stephen W. Porges
501
Y
1
P~P
1 loDri
1 UtUex
I PIX
I LoRil
HiRit
Post
Drug
Fig. 2. Weighted coherencp for 13 hyperactive children in the following conditions: pretreatment (Pre), .I mgfkg dextroamphetamine (LoDex), .25 mglkg dextroamphetamine (HiDex). placebo (Plac), .5 mglkg methylph en idate (LoRit), 1.0 mglkg methylph en idate (HiRit), and off drug after the preceding drug treatments (Post).
in terms of a simple input-output or stimulus-response relationship. The term often implies some type of internal or cognitive process which has been used to explain overt behavior. Attention is commonly used in psychology to describe the active selection of information from the environment or the processing of information from internal sources. Within psychology, there are many instances of such varied usage: e.g., the child responded when he attended t o the appropriate stimulus contingency; or the subject attended intensely to the instructions until he understood the task; or he attended t o the thoughts of yesterday’s meeting; or the mother attended t o the baby’s cry. Obviously, attention as a construct has been used to label diverse processes. These processes range from the reflexive-reactive responses to massive changes in stimulation, at one end of the scale, t o the active instrumental behavior associated with cognitive or psychological processes, at the other. The theoretical orientation of the psychologist greatly influences the processes he associates with the construct of attention. In modern psychology, there are two common views of the human organism. These two outlooks have resulted in two families of psychological theory. The first is a derivative of the reductionism propagated by the English Empiricists, and the second is a derivative of the continental philosophy that emphasized the autonomy of the organism. Reese and Overton (1970) have described two general models which fit these theories: the first, formulated within the context of the mechanistic world hypothesis, portrays the
Peripherul Parallels o f Psychopathology
49
human as a reactive organism; the second, formulated within the context of the organismic world hypothesis, views the human as an inherently and spontaneously active organism. Within the framework of these two models, the construct of attention may be broadly defined. The mechanistic model would define attention t o stimulus change in terms of a reactive organism. In this model, attention is not a voluntary behavior but is a function of stimulus variables. Berlyne (1960) was concerned with the identification of these variables, which included novelty, complexity, conflict, surprise. and uncertainty. These collative variables were, according to Berlyne, the stimulus characteristics that determined attention. In contrast, the organismic model would define attention in terms of a spontaneously active organism which selects information from the environment. The critical difference betwecii these models is whether attention is seen as a voluntary or involuntary process. Perhaps psychologists today aIe so theoretically bound to one model that they are unable t o understand the other. According to Overton and Reese (1973), a model causes the experimenter to see the data with a specific focus. Therefore, the model may influence both the selection and the interpretation of research problems. Thus, many psychologists may be unable t o deal with the concept of attention from both the mechanistic and organismic models and therefore find it difficult t o accept the possibility that the construct of attention may deal with both voluntary and involuntary behavior. The notion that there may be different psychological processes associated with the construct of attention is not new. William James (1 890) distinguished between t w o broad categories of attention: passive-involuntary and activevoluntary. Passive-involuntary attention was defined as always immediate or reflexive and related only to objects that directly affect the sensory systems. Voluntary attention was associated with the concept of interest or selection and was assumed to be directed toward objects of the senses or toward ideational or symbolic objects. Psychophysiology as a research discipline is primarily concerned with physiological parallels of psychological processes. At different levels of investigation, psychophysiologists may concern themselves with physiological activity. Psychophysiologists often investigate such areas as physiological correlates of attention, physiological predictors of attentioiial performance, or physiological responses associated with the orienting reflex. Since the concept of attention may be associated with different psychological processes, there is a great need to be more precise in order to identify the appropriate physiological parallels. The possibility exists that there may be different physiological responses associated with the various kinds of attention. It has been a common practice, however, to use the same physiological measures to define and infer all attention-related behaviors. In lay terms, the psychophysiologist has correlated the same physio-
50
Stephen W.Porges
logical measures with attention, regardless of whether the attention process was reactive or voluntary. A. TWOCOMPONENT MODEL 01: ATTENTION
In a series of psychophysiological studies (Porges, 1972, 1974; Porges & Raskin, 1969; Porges et al., 1975; Walter & Porges, 1976) different heart-rate response components have been associated with different types of attention. Based upon observations from these studies arid the use of factor analysis techniques (Cheung, 1973), a two-component model of attention has been postulated that parallels the two categories of attention identified by William James. Within the context of a two-component model of attention, physiological responses may be partitioned into reactive and sustained components. Two reactive responses have been identified: the first is an immediate response (within 2 seconds of the stimulus change) that may parallel Sokolov’s (1963) limited definition of the orienting reflex as an intensity sensor; the second is a short-latency response which may last for 4 or 5 seconds and may be a function of the qualitative nature of the stimulus. The sustained attention response may be characterized by a reduction of heart-rate variability (the interbeat intervals become more constant) and a generalized inhibition of motor and respiratory activity (either reduced respiratory amplitude or a temporary cessation of breathing). This response persists as long as the subject attends. There is a physiological basis for partitioning the heart-rate response pattern. According t o Levy er al. (1966), the latency of the effect of vagal stimulation upon heart rate is considerably shorter than for the effect induced by sympathetic activation. Furthermore, after cessation of stimulation, the effects of sympathetic stimulation decay much more gradually than d o those of vagal activation. The latency for cardiac deceleration after a single shock t o the cervical vagus is 240 to 300 msec in the dog and 100 t o 160 msec in the cat. With sympathetic stimulation, cardiac acceleration first appears after at least 2 seconds. These data suggest that, if there is a similarity between the autonomic nervous systems of the dog or cat and man, the influences of the autonomic nervous system on the heart-rate response pattern may be temporally partitioned into the following components: ( l a ) a short-latency parasympathetic response occurring within 1 second after stimulation; ( l b ) a longer-latency response with sympathetic excitation or vagal inhibition occurring between 2 and 6 seconds after stimulation (the sympathetic component which occurs after 2 seconds would take longer t o decay relative t o the first short-latency parasympathetic response and may parallel the sympathetically mediated electrodermal responses); and (2) a subsequent, longer-duration parasympathetic response. Table I1 summarizes the description, the hypothesized psychological function, and the hypothesized physiological mechanism of these specific response components.
51
B. RELATION BE1 WI I N 1 HI TWO-COMPONENT MOD1 L A N D AUTONOMIC ANI) N I l ~ K O T R A N S M l T TI
It is hypothesized that geiicral levels of catecliolaiiiinergic a:id cliolinergc activity will be reflected in b a d e v e l Lrutoiiomic activity, wliicli in turn will influence the behavioral and autonomic I-esponses during attention. For example, when individuals are atteiiipting to sustain their attention, there is a tendency to suppress heart-ratc variability m d respiratory amplitude. However, the linkage between respiration and lieart rate (a function of parasympathetic activity) is a statistical determiriant o f the reduction in heart-rate variability observed during sustained attention. Moreover, it appears that individuals who d o not have a strong linkage. t h a t is. individuals in whom there is a weak parasympathetic influence, exliibit poorei performance relative to individuals who exhibit strong linkage. I n Section V , the hyperactive cliild-who has a weak linkage, reflected by a low weighted coherence--was described as exhibiting an inability to suppress heart-rate vai-iability. Given a low dosage of methylphenidate, the linkage became strongel and the behavior improved. It appears that the sustained attention response rcllcctccl in suppression of heart-rate variability is vagally mediated and parallels tlic cholinergic activity in the brain. Moreover the short-latency reactive response is ;~lsovagally mediated and therefore should be larger in subjects with high spoiitaiieous levels of parasympathetic activity. Pilot data with adult subjects support this contention. General levels of catecholaiiiiiici-gic activity should be mapped into the sympathetic nervous system. It is speculatccl t h a t high levels of sympathetic activity should result in conditions sucli ;is 114 [pel-activity. The heart-rate responses during attention should be influeiiced I,y high levels of sympathetic activity. The longer-latency reactive resporisc slioultl be enhanced and result in larger heartrate accelerations compatiblc with the sympathetic innervation of the heart, whereas the sustained response slrould be attenuated since the sympathetic activity is antagonistic to the piirasympathetic influences which define the response. The influence of serotonin on the Iicart-rate responses associated with attention is much more speculative. since serotonin does not have the clear autonomic parallels associated with the catechol,~iiiinesand acetylcholine. However, it is possible that since serotonin is related to the inhibition of reflexive behavior, it might be related t o the habituation of the reactive coinponents of attention. Thus, the short-latency deceleration associated with the parasympathetic nervous system and the longer-latency acceleration associated with both vagal inhibition and sympathetic excitatioii may habituate as a function o f level of serotonin, high levels resultiiig in rapid habituation and low levels resulting in hyperreactivity to changes in stimulation. It is possible that the three iieIirotransiiiitters discussed in Section 1V result in orthogonal influences on behavior (sec Fig. 3): catecholaminergic system along a
TABLE I1 Two-Component Model of Attention
Description
Behavioral function
Physiological mechanism
f 1. Reactive component a. Short-latency response (generally decelerative); latency -within approximately 1 second of change in stimulation b. Longer-latency response (directional changes in heart rate); lafency-within approximately 2 and 6 seconds following change in stimulation
Intensity sensor (orienting reflex in Sokolov’s limited definition) Magnitude and direction of response is a function of stimulus signal value: accelerative in response t o important or intense stimuli; decelerative in response t o novel or mild stimuli
VagalljJmediated
Primarily vagally mediated: when acceleration is observcd, the vagal inhibition is probably coupled with generalized sympathetic activation; when deceleration is observed, the vagal discharge masks the sympathetic contribution
2. Sustairzed component Tonic sustained heart-rate responses-these responses are more active (instrumental) than reactive (passive or reflexive): response is a reduction in heart-rate variability and respiratory-amplitude variability (respiratory inhibition, either more frequent shallow breathing or temporary cessation of breathing) during sustained attentional demands; heart rate is often reduced, although it is possible t o observe a reduction of heart-rate variability independent of directional heart-rate changes; latency-response is generally observable approximately 6 seconds following onset of stimulus and continues for duration of sustained at tention
Facilitates information and stimulus processing: this response reflects the organism’s ability to inhibit ongoing activity and is associated with what is generally called sustained attention
Vagally mediated: increased parasympathetic influences on heart rate, which result in a greater coupling between heart rate and respiration
Stephen W. Porges
54
C 0
HIPO
Hyper
*
Activity
Fig. 3. Simplistic model of the behavioral dimensions influenced bv the three neurotransmitter systems: catecholarninergic (activity): serotonergic (reactivity); cholinergic (inhihition).
dimension from hypoactivity to hyperactivity; cholinergic system along a dimension from hypoinhibition to hyperinhibition; serotonergic system along a dimension from hyporeactivity t o hyperreactivity. These orthogonal influences, in various combinations, might result in the same behavior. Since deviant behavior is the primary criterion for classification into a specific psychopathology, individuals with different etiologies might be diagnosed as suffering from the same psychopathology. This might result in conflicting reports regarding the success or failure of specific treatments with psychopathological populations. Assumptions made in group designs (see Baxley & LeBlanc, this volume) regarding the homogeneity of subjects are violated in research with psychopathological individuals. Group designs utilized to study the physiological substrate of various psychopathologies assume that all members of the group share a similar physiological substrate. By viewing the deviant behavior as being caused by the combined dysfunction of more than one physiological system, it night be possible to obtain a more accurate diagnosis: one might investigate the physiological indices of the three neurotransmitter systems and use this information t o reclassify the subjects into more homogeneous subgroups. The research described earlier (Section V, C) that involved autonomic balance measures of hyperactive children might serve as a method of identifying more liomogeneous subjects for a study on the influences of various drug treatments. To illustrate this point, the data from the second hyperactive study may be described. Although the weighted coherence for the group increased with a low dosage of methylphenidate, the three subjects who had the lowest weighted
Peripheral Parallels ojPsychopatho1og.y
55
coherence off drug not only increased with a low dosage of methylphenidate but, unlike the other subjects, their weighted coherence was even higher with the high dosage. In contrast, the two subjects with the highest weighted coherence exhibited little change with the drug treatments. This may indicate that there were three subgroups of hyperactive children in this study: (a) those that would benefit from a high dosage of methylphenidate; (b) those who would not benefit from any type of pharmacological intervention; and (c) those (the majority of the group) who would benefit from only a low dosage of methylphenidate. It is obvious that virtually any psycliopathological population is characterized by individuals with varied etiologies. However, there are indications in the literature that specific behavioral and pliysiological characteristics are commonly associated with individuals in specific diagnostic categories. The problem i s not that group designs give us n o information regarding the substrate of psychopathology but that they limit us by precluding investigations o f individuals who consistently respond in patterns not characterized by the group average. Thus, a technique for assessing more Iiomogeneous subgroups within a given psychopathology would benefit the treatment of individuals who exhibit idiosyncratic response patterns relative t o their diagnostic category.
VII. Autism Autism, a psychopathology associated with abnormal attention, may also be discussed within the proposed model of parallel behavioral (attentional) and autonomic (peripheral and central) dysfunction. The behavior of the autistic child will be discussed, and then the central and peripheral autonomic activity will be predicted as reflections of the observed behavior. Available psychuphysiological and psychopharmacological evidence will be used to test t!iese predictions. Kanner (1943) was the first t o label as infantile autism the syndrome characterized by an enduring failure to recognize and t o respond with affection t o others. The symptoms of autism have been grouped into five categories of disturbances: perception; motility; developmental rate; relationships to persons and objects; and language (Ritvo, Yuwiler, Geller, Ornitz, Saeger, & Plotkin, 1970). Autistic children are often diagnosed as suffering from childhood psychosis or schizophrenia. The distinction between autism and other psychoses is very difficult to make, and many practitioners rely on the Rimland E-2 scale (Rimland, 1964). The scale is based on a detailed questionnaire filled out by the patient’s parents. The cluster of scores above +20 on the E-2 scale is believed t o represent the t o p 10% of psychotic children who manifest autistic symptoms and who correspond t o Kanner’s definition of early infantile autism (Kimland,
56
Stephen W.Porges
1964, 1971). Often in research, however, all psychotic children have been clustered together. Central t o the theme of this chapter is the commonly observed characteristic of autistic children as having a response or attentional deficit. Rutter (1966, 1971) has described an absence of response both t o sound, which has often resulted in the autistic child being diagnosed as deaf, and to pain. The autistic child’s response deficit is generally manifested in a lack of responsiveness, but at times the child may exhibit excessive or erratic responses (Ornitz & Ritvo, 1968; Rindand, 1964; Rutter, 1966, 1971). Thus, situations exist in which an autistic child who may appear deaf t o loud sounds may suddenly overrespond, behaviorally and emotionally, t o a soft distant sound with the appearance of extreme distress. Autistic children may also manifest abnormal stimulus selectivity; one modality of a stimulus complex might dominate to the extent that others will be completely ignored (Lovaas & Schreibman, 197 1 ; Lovaas, Schreibman, Koegel, & Rehm, 1971). It is clear from this brief review of the literature that autistic children suffer a deficit or dysfunction on the dimension earlier described as reactivity. This dysfunction is evident from the observed hypo- and hyperreactiveness t o the environment. Given these observations, it might be predicted that autonomic parallels of reactivity would also be abnormal. In the peripheral autonomic nervous system the abnormality might be manifested in the magnitude of the orienting reflex and/or trials to habituation. In the central autonomic nervous system, the abnormality nught be manifested in abnormal levels of serotonin. There are very few investigations of the possible parallels of autism in the peripheral autonomic nervous system, although a relationship between psychosis and autonomic imbalance has been hypothesized. Cellhorn (1957) postulated that psychosis is a behavioral parallel of a primary disturbance of the central mechanism that regulates autonomic activity. Rubin (1 962) studied the pupillary response, which is mediated by parasympathetic constrictors and sympathetic dialators, and found it t o be heterogeneous in schizophrenics. Rubin suggested that chronic schizophrenics are less reactive than normal individuals and that they exhibit greater variance in the autonomic nervous system than do normal individuals. Rubin proposed a model based upon the balance of adrenaline and acetylcholine, the peripheral autonomic transmitters. He assumed that functional psychoses are associated with a disorder o f homeostasis. paralleled b y patterns of neurohumoral imbalance in the central and peripheral representations of the autonomic nervous system. He suggested that normal children have a significantly greater capacity for increased adrenergic outflow (sympathetic activity) than autistic children. The result would be greater rcactivit-v of normal children t o the environment. When normal and autistic children were compared with respect t o the rate of pupillary constriction (parasympathetic activity), it was found that the rate was significantly faster for the normal children. Rubin
Peripheral Parallels of Psychopathology
51
interpreted this finding as indicating that the rate of cholinergic activity was also more rapid in normal children. These results suggest that normal children, relative t o autistic children, have a greater reactivity to change in stimulation and a greater capacity to inhibit this reactivity and return t o a normal homeostatic balance. Bernal and Miller (1970) have suggested that the galvanic skin response (CSR) might be a useful index of stimulus detection with a schizophrenic or autistic child who may not be capable o f responding verbally. By comparing the CSR response of schizophrenic children with normal children, these investigators found that the schizophrenic children failed t o maintain habituation over the stimulus series and could be differentiated from normal children on the basis of a lower mean GSR response. Bernal and Miller interpreted these findings as evidence that the schizophrenic child has diminished sympathetic responsivity, which is detectable through the GSR. These findings and interpretations are consistent with those of Rubin (1962), suggesting that the observed decrease in sympathetic responsiveness is an autonomic correlate of the behavioral unresponsiveness characteristic of schi/ophrenic and autistic children. Experimental reports bearing upon the reactivity of schizophrenic or autistic children t o sensory stimuli have generally emphasized a lack of behavioral or autonomic responsiveness. Other dimensions related t o reactivity, such as central nervous system activity measured by an electroencephalogram (EEG), have been investigated in autistic and scliizophrcnic children. Reports of neurophysiological dysfunction have described a lack of responsiveness in the EEC (Walter, Aldridge, Cooper, O'Gorman, McCall~iii,& Winter, 1971) and a high incidence of abnormal EECs (Small, 1975). These studies showed that the pathological children were less neurophysiologically responsive to both internal regulatory mechanisms (normal cyclic rhythniic) and external stimuli (evoked cortical responses). Consistent with the abundance of reports depicting autistic and schizophrenic children as having limited or defective patterns of reactivity, there has been a trend toward the study o f ;I possible relationship between defects in the serotonergic neurotransmitter system and autism. Serotonin is the neurotransmitter which has been theoretically implicated with the proposed behavioral dimension of reactivity. Earlier (in Section V1, B), it was suggested that levels of serotonin might be reflected on the behavioral dimension of reactivity, high levels associated with hyporeactivity and low levels with hyperreactivity. Since autistic children characteristically exhibit a hyporeactivity t o environmental events, it might be predicted that autistic children have higher levels of serotonin than normal children. A major problem in investigating levels of neurotransmitter activity is the difficulty in assessing central nervous system levels, since levels in blood sampled from the periphery may not be isomorphic with brain levels. Given this possibil-
58
Stephen W.Porges
ity, caution must be exhibited in interpreting research on the biochemistry of autism. The most common method of estimating central levels in humans is t o investigate peripheral blood platelets. The platelet model, presented by Pletscher (1968) and Paasonen (1968), is based upon the similarities between the properties of blood platelets and monoaminergic storage vesicles in the central nervous system (the catecholamines and serotonin are monoamines). Several investigations have compared monoamine uptake (absorption) and efflux (outflow) from the in vitro platelets of autistic and nonautistic children t o identify whether the pathological reactivity observed in autistic children is a function of a defect in serotonin binding. Unfortunately, this line of research, which will be reviewed in the following paragraphs, has produced conflicting results. A second method of assessing serotonin levels is merely t o measure levels of serotonin in the blood. A third method is to assess the acid metabolites of serotonin in cerebrospinal fluid as an index of the turnover (and, by inference, production levels) of serotonin in the central nervous system. A fourth method is to assay concentrations of metabolic by-products in the urine. The first study relating serotonin t o autism was reported by Schain and Freedman (1961), who noted that 6 out of 2 3 autistic children diagnosed in accordance with Kanner’s (1943) criteria had consistently elevated blood serotonin concentrations and higher platelet counts than age-matched control subjects. Subsequently, a study involving 24 patients and 35 control subjects demonstrated that, as a group, young autistic children have significantly elevated blood serotonin (Ritvo et al., 1970). This study also demonstrated that young autistic children had significantly higher platelet counts and a greater variability of serotonin concentrations per platelet. The finding that autistic children may have higher blood serotonin concentrations raises the question of what clinical changes might occur if the level of serotonin is reduced. Ritvo and his associates (Ritvo, Yuwiler, Geller, Kales, Rashkis, Schicor, Plotkin, Axelrod, & Howard, 1971) attempted t o answer this question. Through the use of biochemical techniques that involved the infusing of precursors of specific neurotransmitters into patients, these investigators effectively reduced blood serotonin concentrations in a few patients. However, the patients’ clinical status did not change; the amount of motility disturbances (handflapping), percentage o f REM (rapid-eye-movement) sleep time, and levels of endocrine activity were unaffected by the experimental manipulation. In two studies, Boullin and his associates (Boullin, Coleman, & O’Brien, 1970; Boullin, Coleman, O’Brien, & Rimland, 197 1 ) studied possible abnormalities in platelet serotonin in patients with infantile autism. The first study (Boullin et al., 1970) included six children diagnosed as autistic (scores above +21 on the Rimland E-2 scale). Five of the six children demonstrated an abnormally enhanced efflux of serotonin from blood platelets. Overall, there was slightly greater serotonin uptake but markedly greater efflux in the platelets from the six
Peripheral Parallels of'Psychopathology
59
autistic children, as compared t o the platelets from normal children. I n the second study (Boullin ef al., 1971) the investigators attempted to use the serotonin efflux and uptake chatacteristics of platelets t o provide a biochemical diagnosis of autistic children froin a population of psychotic children and then t o predict the E-2 score. Predictions were made on the basis of enhanced serotonin loss (efflux) from platelets compared t o normal controls. Using the criterion that efflux had t o be at least three times greater than the control value, correct predictions (based on the E-2) were made on nine of the ten psychotic children. The biochemical results correlated with the E-2 score and predicted autism correctly in six out of scven cases and the absence of autism in three out of three cases. Other than the Boullin ef al. (1970) study, which revealed a slight increase in serotonin uptake of platelets from autistic children relative t o normal children, most studies have shown no differences (Boullin ef al., 1971; Lucas, Krause, & Domino, 1971 ; Yuwiler, Ritvo, Geller, Glousani, Schneiderman, & Matsuno, 1975). The more coniiiion finding is that there is an enhanced serotonin efflux from the platelets of infantile autistic children, which is often interpreted t o be the result of defective binding of serotonin in the central nervous system (Boullin et al., 197 1 ). In two recently published studies, data have been reported which d o not support the view that measurement of platelet serotonin activity can be used as a biochemical index t o aid in the clinical diagnosis of infantile autism. Yuwiler et al. (1975) studied the platelets from young normal, autistic, and hospitalized control children, and found that these gi-oups did not differ in serotonin uptake or efflux. Boullin, Bhagavan, Coleman, O'Brien, and Youdim (1975) found no significant difference between normal and autistic children in the rates of metabolism of serotonin and dopaminc. Another approach used t o study central nervous system biogenic aniine activity (the activity of serotonin and the catecholamines) is t o assay cerebrospinal fluid for metabolites of serotonin and dopamine as a reflection of rate of monoamine formation in the brain. I n a study of 35 children suffering from various pathologies (autism, atypical development, epilepsy, and movement disorders), metabolites of serotonin and dopamine were investigated in samples of cerebrospinal fluid (Cohen, Shaywitz, Johnson, & Bowers, 1974). There were statistically significant differences in the metabolites of both dopamine and serotonin among the groups, and the psychotic children (autistic and atypical combined) had higher levels than the children with epilepsy and movement disorders. It has been demonstrated previously that chronic schizophrenic adults excrete a metabolic by-product of serotonin, the psychogenic substance N,N-dimethyltryptamine (DMT), in their urine while normals do not (see Narasimhachari & Himwich, 1975). In the hope of establishing whether a similar defect is associated with the etiology of early infantile autism, Narasimhachari and Hiniwich
60
Stephen W.Porges
studied concentrations of DMT in the urine of autistic children. As with the adult schizophrenics, children diagnosed as autistic excreted DMT, while none of the normal children did. However, seven of the ten parents of the five autistic children also were positive for DMT in their urine. The behavior of the parents, tested with typical assessment instruments such as the Minnesota Multiphasic Personality Inventory, the Psychiatric Assessment Interview, and the Rorschach Inkblot Test, indicated no evidence of psychopathology. Although the literature just reviewed reveals some conflicting findings, there appears t o be a trend relating defective serotonergic activity with autism. The most consistent findings are that blood levels and platelet efflux are enhanced in the autistic child. As stated earlier, the abnormal levels and activity of serotonin may be interpreted to reflect defective binding or metabolism of serotonin in the brain. Two physiological characteristics appear t o parallel the response and attentional deficit of the autistic child: (1) the peripheral autonomic nervous system reflects a diminished sympathetic responsivity. (2) The serotonergic neurotransmitter system seems t o be defective and characterized by abnormally high levels of serotonin. Both findings might have been predicted from the proposed model as autonomic and biochemical parallels of hyporeactivity. If serotonin were also related t o perceptual thresholds, then the autistic child’s capacity t o ignore painful stimuli might be explained. Although there appear to be no studies relevant t o this hypothesis with human subjects, there is a recent study with rats as subjects. Harvey, Schlosberg, and Yunger (1975) demonstrated that increased pain sensitivity and enhanced reflexive responses to painful stimuli are behavioral consequences of serotonin depletion. The authors suggested that serotonin may not play any specific role in specific categories of behavior, including responses to pain, but may act merely as an inhibitory transmitter that reduces the effects of arousing stimuli. This interpretation is consistent with the role of serotonin proposed earlier in this chapter. If serotonin levels are abnormally high, one can predict the limited reactivity t o stimuli, including painful stimuli, observed in the autistic child. The incidents in which the autistic child is observed t o be hyperreactive may be related to abnormal cyclic depletion of serotonin (normal individuals show diurnal changes in serotonin levels).
VIII. Summary and Conclusions For inany years there have been hypotheses relating both autonomic imbalance t o behavioral disorders (Eppinger & Hess, 1915; Quay, 1976; Wenger, 1941; Wilder, 1931) and neurotransmitter balance t o mental health (Davis et al., 1973). The philosophical roots of these theories may be traced back t o the
Peripheral Parallels of Psychopathology
61
concepts of an “internal milieu” and “homeostasis” postulated by Claude Bernard (1 878) and Walter Cannon (1929), respectively. The link among psychopathology, defective attention, and atypical autonomic function proposed in this chapter appears t o be a reasonable speculation. This link beconies intuitively clear if attention, as einphasized here, is viewed as a state which may facilitate or impede information intake. This view of attention partitions the concept of an attcntiona2 state which facilitates or impedes stimulus intake from the in tellectual components manifested in information processing-the intellectual components hypothetically being associated with cortical activity and the state components being associated with tlie central and peripheral autonomic nervous system. Consistent with this view, individuals suffering from specific psychopathologies described earlier are often physiologically defective in the systems that mediate attentional states. An attempt has been made here to generate a testable model interrelating the biochemical theories of psychopathology and the psychophysiological theories of attention. The model, partially supported by experimental data, might lead to a better understanding of the mechanisnis mediating individual differences in attention and might, through future research, explain why and how specific pharmacological interventions are successful. The success of this approach is based upon the parallelism between behavior and physiological activity. This multivariate approach is the basis f o r pxycliophysiological research. The pathologies in the autonomic nervous system observed in the three psychopathologies discussed earlier may have a varied genesis. It is possible that there might be genetic components interacting with environmental factors. Physiologically, certain cells in the central nervous system have the potential t o respond t o a variety of neurotransmitters; that they actually respond t o one transmitter or another is, in part. a function of environmental influences. Although three psychopatliologies were selected t o demonstrate the relationship between behavior and physiological activity, the use of developmental populations for this purpose would have been as effective. There is a series of developmental trends in the au tvnomic nervous system that parallels the trends observed in behavior. For example, the behavior of a child as he matures, regardless of model of developiiient , shifts from a predominance of stiniulusdetermined reflexive activities (reactive behavior) and unlearned homeostatic mechanisms t o behaviors which are more organized and cognitively mediated (see Reese & Porges, 1976). Developmeiitally, this is characterized by subordination and integration of primitive reflexive behavior t o more complex voluntary behavior. Paralleling this shift is a n increased influence of the parasympathetic nervous system and tlie central cliolinergic systems. Thus, if one investigated various developmental populations from birth tlirougli old age, one would expect t o find that the capacity to attend (reactive and sustained attention) would be linked t o the changing balance in the autonomic nervous system. It is
62
Stephen W. Porges
possible that autonomic imbalance, relative to age norms, might be a precursor of specific psychopathologies characterized by defective attention and that early detection through measurement of autonomic balance might facilitate positive and successful intervention. ACKNOWLEDGMENTS The preparation of this chapter was supported, in part, by Research Scientist Development Award K02-MH-0054 from the National Institute of Mental Health and by grants NEC-00-3-0013 from the National Institute of Education, 524 from the State of Illinois Department of Mental Health. The hyperactive subjects were tested through collaborative research with R. L. Sprague; the research was funded in part by MH 18909 from the National Institute of Mental Health awarded to R. L. Sprague. The research described was aided by Michael N. Cheung, George J. Franks, Elwood Higley, Leighton E. Stamps, and Gary F. Walter. The weighted coherence formula and time-series analyses described in the manuscript are a product of interactions with my colleagues Robert E. Bohrer and Gideon Keren. I wish especially t o thank Brenda Brown for typing the manuscript, Michael N. Cheung for drawing the figures, and Carol Sue Carter-Porges for critically evaluating previous drafts and for stimulating the expression of my ideas in this chapter.
REFERENCES Anisman, H. Time-dependent variations in aversively motivated behaviors: Nonassociative effects of cholinergic and catecholaminergic activity. Psychological Review, 1975, 82, 359-385. Bernal, M. E., & Miller, W. H. Electrodermal and cardiac responses of schizophrenic children t o sensory stimuli. Psychophysiology, 1970, 7, 155-168. Berlyne, D. E. Conflict, arousal and curiosity. New York: McGraw-Hill, 1960. Bernard, C. Lessons on the phenomena of life common to animals and vegetables. Paris: Baillike e t Fils 1878. Boullin, D. J., Bhagavan, H. N., Coleman, M., O’Brien, R. A,, & Youdim, M. G. H. Platelet monoamine oxidase in children with infantile autism. Medical Biology, 1975, 53, 21S213. Boullin, D. J., Coleman, M., & O’Brien, R. A. Abnormalities in platelet 5-hydroxytryp tamine efflux in patients with infantile autism. Nature (London) 1970, 226, 371-372. Boullin, D. J., Coleman, M., O’Brien, R. A., & Rimland, B. Laboratory predictions of infantile autism based on 5-hydroxytryptamine efflux from blood platelets and their correlation with the Rimland E-2 score. Journal of Autism and Childhood Schizcphrenia, 197 1, 1, 63-7 1. Box, G. E. P., & Jenkins, G. M. Time series analysis: Forecasting and control. San Francisco: Holden-Day, 1970. Broverman, D. M., Klaiber, E. L., Kobayashi, Y.,& Vogel, W. Roles of activation and inhibition in sex differences in cognitive styles. Psychological Review, 1968, 7 5 , 23-50. Broverman, D. M., Klaiber, E. L., Vogel, W., & Kobayashi, Y. Short-term versus long-term effects of adrenal hormones on behaviors. Psychological Bullerin, 1974,81,672-694. Cannon, W. B. Organization for physiological homeostasis. Physiological Review, 1929, 9, 399-431.
Peripheral Parallels of Psychopathology
63
Chess, G. F., Tam, M. K., & Calaresu, F. R. Influence of cardiac neural inputs on rhythmic variations of heart period in the cat. American Journal of Physiology, 1975, 228, 775-780. Cheung, M. N. Cardiac responses as measures of the orienting reflex: A factor analytic approach. Unpublished master’s thesis, West Virginia University, 1973. Cleckley, H. The mask of sanity. St. Louis: Mosby, 1955. Clements, S. D., & Peters, J. E. Minimal brain dysfunction in the school-aged child. Archives of General Psychiatry, 1962,6, 185-197. Clynes, M. Computer analysis of reflex control and organization: Respiratory sinus arrhythmia. Science, 1960, 13, 300-302. Cohen, D. J., Shaywitz, B. A., Johnson, W. T., & Bowers, M. Biogenic amines in autistic and atypical children. Archives of General Psychiatry, 1974, 31,845-85 3. Davis, J. M., Janowsky, D. S., El-Yousef, M. K., & Sekerke, H. J. The psychopharmacology of methylphenidate in man. In T. A. Ban, J. R. Boissier, G. J. Gessa, H. Heimann, L. Hollister, H. E. Lehman, I. Munkvad, H. Steinberg, F. Sulser, A. Sundwall, & 0. Vinar (Eds.), Psychopharmacology, sexual disorders and drug abuse. Prague: Avicenum, Czechoslovak Medical Press, 1973. Pp. 6 7 5 4 8 4 . Douglas, V. I. Stop, look and listen: The problem of sustained attention and impulse control in hyperactive and normal children. Canadian Journal of Behaviour Science, 1972, 4, 259-282. Dykman, R. A., Ackerman, P. R., Clements, S. D., & Peters, .I. E. Specific learning disabilities: An attentional deficit syndrome. In H. R. Mylebust (Ed.), Progress in learning disabilities. New York: Grune 8~Stratton, 1971. Pp. 56-93. Eckoldt, K., & Schubert, E. The influence of the tidal volume on the sinus arrhythmia of the heart. Acta Biologica et Medica Germanica, 1975, 34, 167-771. Eppinger, H., & Hess, L. Vagotonia. Journal of Nervous and Mental Disease, Monograph Series, 19 15, 20. Freibergs, V., & Douglas, V. I. Concept learning in hyperactive and normal children. Journal of Abnormal Psychology, 1969,74, 388-395. Gellhorn, E. Autonomic imbalance and the hypothalamus. Minneapolis: University of Minnesota Press, 1957. Harvey, J. A., Schlosberg, A. J., & Yunger, L. M. Behavioral correlates of serotonin depletion. Federation Proceedings, 1975, 34, 1796-1801. James, W. Principles of psychology. New York: Holt, 1890. Kanner, L. Autistic disturbances of affective contact. Nervous Child, 1943,2,217-250. Katona, P. G., & Jih,F. Respiratory sinus arrhythmia: Noninvasive measure of parasympathetic cardiac control. Journal of Applied Physiology, 1975, 39,801-805. Laufer, M. W., & Denhoff, E. Hyperkinetic behavior syndrome in children. Journal of Pediatrics, 1957,50,463-474. Laufer, M. W., Denhoff, E., & Solomons, G. Hyperkinetic impulse disorder in children’s behaviour problems. Psychosomatic Medicine, 1957, 19, 3 8 4 9 . Levy, M. N., DeGeest, H., & Zieske, H. Effects of respiratory center activity on the heart. Circulation Research, 1966, 18,67-78. Lippert, W. W., & Senter, R. .I. Electrodermal responses in the sociopath. Psychonomic Science, 1966,4,25-26. Lovaas, 0. I., & Schreibman, L. Stimulus over-selectivity of autistic children in a twostimulus situation. Behaviour Research and Therapy, 1971, 9, 305-310. Lovaas, 0. I., Schreibman, L., Koegel, R., & Rehm, R. Selective responding by autistic children to multiple sensory input. Journal of Abnormal Psychology, 1971, 71, 211222. Lucas, A. R., Krause, R. R., & Domino, E. F. Biological studies in childhood schizophrenia:
64
Stephen W. Porges
Plasma and RBC cholinesterase activity. Journal of Autism arid Childhood Schizophrenia, 1911, 1, 72-81. Lykken, D. T. A study of anxiety in the sociopathic personality. Journal ofAbnormaIand Social Psychology, 1957, 55, 6-10. Manto, P. G. Blockade of epinephrine-induced decrement in activity by scopolamine. Psychonomic Science, 1967, 7, 203-204. Millichap, J . G., & Fowler, G. W. Treatment of “minimal brain dysfunction” syndrome. Pediatric Clinics of North America, 1967, 14, 767-117. Narasimhachari, N., & Himwich, H. E. Biochemical studies in early infantile autism. Biological Psychiatry, 1975, 1 0 , 4 2 5 4 3 2 . Ornitz, E. M., & Ritvo, E. R, Perceptual inconstancy in early infantile autism. Archiiws of General Psychiatry, 1968, 18, 76-98. Overton, W. F., & Reese, H. W. Models of development: Methodological implications. In J. R. Nesselroade & H. W. Reese (Eds.), Life-span developmental psychology: Methodological issues. Ncw York: Academic Press, 1973. Pp. 65-86. Paasonen, M. K. Platelet 5-hydroxytryptamine as a model in pharmacology. Anrzales Medicinae Experinzentalis et Biologiae Fenniae, 1 9 6 8 , 4 6 , 4 1 6 4 2 2 . Pletscher, A. Metabolism, transfer and storage of 5-hydroxytryptamine in blood platelets. British Journal of Pharmacology and Chemotherapy, 1968, 32, 1-1 6. Porges, S. W. Heart rate variability and deceleration as indexes of reaction time. Journal of Experimental Psychology, 1972,92, 103-1 10. Porges, S. W. Heart rate indices of newborn attentional rcsponsivity. Merrill-Palmer Quarterly, 1914, 20, 231-254. Porges, S. W., & Raskin, D. C. Respiratory and heart rate components of attention. Jouriial ofExperimenta1 Psychology, 1969, 81,497-503. Porges, S. W., Walter, G. P., Korb, R. J., & Sprague, R. L. The influences of methylphenidate on heart rate and behavioral measures of attention in hyperactive children. Child Development, 1915, 46, 127-133. Quay, H. C. Psychopathic pcrsonality as pathological stimulation-seeking. American Journal ofpsychiatry, 1965, 122, 18C-183. Quay, H. C. Psychopathic behavior: Reflections on its nature, origins, and treatment. In F. Weizman & I. Uzgiris (Eds.), The structuring of experience. New York: Plenum, 1916, in press. Rcese, H. W., & Overton, W. 1:. Models of development and theories of development. In L. R. Goulet & P. B. Baltes (Eds.), Life-span developmeiital psychology. Theory and research. New York: Academic Press, 1970. Pp. 115-145. Reese, H. W., & Porges, S. W. Development of learning processes. In V. Hamilton & M. Vernon (Eds.), The development of cognitive processes. London: Academic Press, 1976, in press. Rimland, B. Infantile autism. New York: Meredith, 1964. Rimland, B . The differentiation of childhood psychoses: An analysis of checklists for 2,218 psychotic children. Journol of Autism and Chitdhood Schizophrenia, 197 1. 1, 161-174. Ritvo, E. R., Yuwiler, A., Geller, E., Kales, A., Rashkis, S., Schicor, A,, Plotkin, S., Axelrod, R., & Howard, C. Effects of 1-dopa in autism. Journal of’ Autism and Childhood Schizophrenia, 197 1, 1, 190-205. Ritvo, E. R., Yuwiler, A., Geller, E., Ornitz, E. M., Saeger, K., & Plotkin, S. Increased blood serotonin and platelets in early infantile autism. Archives o.f General Psychiatry, 1970, 23,566-572. Rubin, L. S. Patterns of adrenergic-cholinergic imbalance in the functional psychoses. Psychological Review, 1962, 69,501-5 19.
Peripheral Parallels of Psychopathology
65
Rutter, M. Behavioural and cognitive cliaracteristics of a series of psychotic children. In J . K. Wing (Ed.), Early childhood autisni: ('lirrical educational arid social aspects. Oxford: Pcrgamon, 1966. Pp. 5 1-75. Rutter, M. Thc description and classification o f infantile autism. In D. W. Churchill, G. D. Alpern, & H. K. DeMeycr (Eds.), ltrfaritilc autism: Proceedings of Indiana University colloquiurn. Springfield, 11.: Thomas, 197 1. Satterfield, J . R., & Dawson, M. F,. Electrotlernial correlates of hyperactivity in children. Psychophysiology, 1971, 9, 191-197. Schain, R. J., & Freedman, D. Studies of' 5-hydroxyindole metabolism in autistic and other mentally retarded children. .lourrial of Pcdiafrics, 1961, 68, 315-320. Small, J . G. EEG and neuropliysiologica1 studies of early infantile autism. Biological Psvchiatry, 1975, 10, 385-397. Sokolov, E. N. Perception and tlrc contfifioneclre/lex. New York: Macmillan, 1963. Solomon, R. L., & Corbit, J. D. An opponent-process theory of motivation: I. Temporal dynamics of affect. Psychological licvicw, 1974, 81, 119-156. Spraguc, R. L., & Sleator, E. K. What is thc proper dose of stimulant drugs for children. International Journal ofMental Ilealrlr. 1975. 4, 75-104. Stewart, M. A. Hyperactive children. Scictrrific, 4nwricaii, 1970, 222, 95-98. Swonger, A. K., & Rech, R. 13. Serotoncrgic and cholinerpic involvement in habituation of activity and spontaneous alternation of i a i s in a Y maze. Journal of Comparative and Pli.vsiologica1 Ps~vchology,1972. 8 I , 509-5 22. Walter, G. F., & Porges, S. W. Heart rate and respiratory responses as a function o f task difficulty: The use of discriminant aniilysis in the selection of psychologically sensitive physiological responses. P ~ ? ~ c h o p l r ~ ' s / o l 1976, o ~ ? , , in press. Walter, G. W., Aldridge, V. J., Cooper, R., O'Gorman, G., McCallum, C . , & Winter, A. L. Neurophysiological correlates of apparent defects of sensory-motor integration in autistic children. In D. W. Churchill, G. D. Alpern, & M. K. DeMyer (Bds.), Infantile aicfisnz: Proceedings o f Indiana U n i i ~ w i fcolloquiunz. j~ Sprinpficld, 11.: Thomas, 197 1. Wenger, M. A. The measurement of individual differences in autonomic balance. Psychosomatic Medicine, 1941, 3, 427-434. Werry, J. S., Sprague, R. L., Weiss, G., & Mindc, K. Some clinical and laboratory studies of psychotropic drugs in children: An overview. In W. L. Smith (Ed.), Dru,Fs and cortical function. Springfield, 11.: Thomas, 1970. Wheeler, C. A. The relationship betwecn psychopathy and the weak automatization cognitive style. K I R e s e a r c h Reports, 1914, 6, Monogr. No. 2. Wilder, J. Das "Ausgangswert-Gesctz." cin unbcachtctes biologisches Gcsetz und seine Bedeutung fur Forschung und Praxis. Lcitsclirift fiir die Kesartirnte Neurologic wid Psychiatric, 1931, 137, 317-339. Williams, J . M., Hamilton, L. W., & ('artton, P. L. Pharmacological and anatomical dissociation of two types of habituation. .lorcrrial of Comparative and Ph.ysiologica1 Ps~~cliology, 1914,87,724-132. Yuwiler, A., Ritvo, E., Geller, E., Gloucman, R., Schneiderman, G., & Matsuno, D. Uptake and efflux of serotonin from platclcts o f autistic and nonautistic children. Jourrral of Autism and Childhood Schizoplrrcnm, 1975, 5, 83-98.
This Page Intentionally Left Blank
CONSTRUCTING COGNITIVE OPERATIONS LINGUISTICALLY
1lurr.v Beilin CITY UNIVERSITY O F NEW Y O R K
I . LANGUAGE AS REPRESENTATION . . . . . . . . . . . . . . . . . . . . . . I1 . T H E MECHANISMS OF CONSERVATION . . . . . . . . . . . . . . . . . .
68 69
Ill . METHODOLOGICAL ISSUES IN PIAGETIAN TRAINING RESEARCH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A . EARLY HISTORY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. METHODOLOGICAL ISSUES . . . . . . . . . . . . . . . . . . . . . . . . .
71 71 74
IV . THE GENERAL TRAINING MODEL: METHODOLOGICAL ISSUES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A . PRETEST . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B . TRAINING . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C . POSTTEST . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . D . TRANSFER AND DECALAGE . . . . . . . . . . . . . . . . . . . . . . . . . E . STRONG VERSUS WEAK CRITERIA . . . . . . . . . . . . . . . . . . .
75 75 76 77 77 78
V . LINGUISTIC TRAlNlNG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A . VERBAL RULE INSTRUCTION . . . . . . . . . . . . . . . . . . . . . . . B . LEXICAL TRAINING AND LEXICAL KNOWLEDGE . . . . . . . C . SUMMARY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
82 82 91 95
VI . SUMMARY OF RESULTS AND THEORETICAL CONSIDERATIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A . SOCIAL LEARNING THEORY . . . . . . . . . . . . . . . . . . . . . . . . B . REPRESENTATIVE FUNCTIONALISM . . . . . . . . . . . . . . . . .
96 97 98 61
68
Harry Beilin
C. LINGUISTIC S T R U C T U R A L I S M . . . . . . . . . . . . . . . . . . . . . . . D. V E R B A L R U L E INSTRUCTION: T H E O R Y . . . . . . . . . . . . . . .
100
REFERENCFS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
102
99
I. Language as Representation Cognitive structure in Piaget’s theory derives from the child’s action on objects and from internalized activity that is the analog of overt action. Language is also said to have its origin in action. The child’s first words-verbal signs such as “mommy” and “daddy,” and “semiverbal signs” such as “tch-tch”-are said t o “really represent complex schemes of actions, either related t o the subject or partly objective” (Piaget, 1962, p. 219). The process of naming an object is interpreted too as a statement of possible action (p. 222). The advance of true language over intuitive action, which is represented in images, and over the very earliest language, which is characterized as preconceptual, is that it constitutes a form of representation of concepts (pp. 279-280). This representational function of language remains for Piaget its most important characteristic. Language thus becomes the principal vehicle for representing concepts and operations, although it is not itself the source of the concepts and operations basic to the conception of space, time, causality, and other aspects of the real world. The by now well-known conservation phenomenon, for example, has its origin in the elaboration of elementary sensorimotor schemes. The rudimentary conservations are first realized (at 12 t o 18 months) in the concept of a permanent object and are fully attained (at about 7 years) in the construction of concrete operational schemes. Language enters into this process only for representing structures and functional relations already developed. Although the child does not attain the object concept until it is capable of representative activity (i.e., can symbolically represent the object that is not present), the form of representation which is needed for its realization is not linguistic but the image (Piaget, 1954a). This is evident in the child’s ability t o conceive of the object’s displacement through a series of detours behind or under screens. When knowledge of conservation is acquired later, it is again not language which ensures its construction but the composition of operations that entails internal symbolic action whose properties (such as reversibility) are basic to all logical structures and systems (Piaget & Inhelder, 1974). Much effort has been directed to testing and exploring the implications of Piaget’s theory, but no aspect of the theory has received more attention than Piaget’s characterization of conservation. And one of the principal vehicles for
Constructing Cognitive Operations Linguistically
69
exploring the mechanisms of conservation has been the training study. The first efforts t o train conservation interestingly enough were initiated by the Genevans themselves, in an effort t o show that Piaget’s equilibration theory better accounts for cognitive change than other explanations, particularly enipiricistbehaviorist theories. Non-Genevan attempts t o account for conservation acquisition are not confined t o behaviorist or neobehaviorist theories, however. Among the alternatives [which are dctailcd by Beilin (1971a, in press)] are explanations based on a significant place given to the role of language. The purpose of this review is t o describe research, including that of the present author, designed t o show how conservation can be acquired by linguistic means, despite assertions of Piaget and H. Sinclair (the principal Genevan investigator of language development) that conservation is not acquired in this fashion. This paper goes one step further, however, in showing that in addition to its empirical demonstration, there is a theoretical basis for expecting such achievements t o occur, based in fact on Piaget’s more recent views on the relation between language and thought. Before describing the language-oriented training studies, we will set forth Piaget’s evolving views on conservation as background for the later discussion. We will also consider some of the more significant methodological issues that have arisen in Piagetian training research.
11. The Mechanisms of Conservation One of the marked characteristics of Piaget’s developmental theory is its emphasis on the construction of logical, semilogical and infralogical structures. These constructions extend over many years, with their most advanced level ordinarily achieved in technologically advanced societies in late adolescence or early adulthood, although the end point in the process is not really known. The natural process of construction, accoiding t o the theory, results from a selfregulating (i.e., equilibrating) niechanism that integrates the products of the subject’s own actions (experience) with internal change that occurs through another kind of biological regulation (maturation). The structural product of this interaction, according t o Piaget, is defined exclusively by neither nativist nor empiricist formulations. Piaget’s theory details the nature of the cognitive structures and the functional principles by which they are constructed. But it provides only a nonspecific characteriLation of the nature of experience and, similarly, a nonspecific characterization of natively given mechanisms, except for the reflexes observed at birth (Beilin, 1971b). Conservation is both the process and product of one such construction. As originally described by Binet (1890), it pointed t o the child’s difficulty in differentiating the number of elements in an array of objects from the length of the array. As Piaget’s studies disclose, this phenomenon is not confined t o the
I0
Harry Beilin
conservation of number but applies to all concepts that embody quantitative invariants (Piaget & Inhelder, 1974). In his account of conservation that refined and expanded upon the original formulation, Piaget (1 968) asserted that conservation was possible only when a composition of quantitative variations occurred. These compositions are of two kinds. One takes the form of a composition of transformations (higher X thinner = the same amount) and is manifest in the conservation of liquid (or continuous) quantity. The other takes the form of additive composition (nothing added, nothing taken away = the same amount). This form is evident when a wire in the form of an arc is extended to a straight line (or the reverse). The composition here is based on the intervals between the end points which form a system of additive inclusions (Piaget, 1968, p. 27). The additive and compensatory compositions plus the notion of operational identity, which is a necessary component of the compositions, are evident too in the child’s verbal justifications of his conservation judgments. These justifications involve two reversibility notions plus the identity argument. One is reversibility by inversion (a return to the original state, as when the child suggests that a row of counters, after being expanded to a wider length, can be contracted to return them to their original state); the other, reversibility by compensation, in which a change in one dimension is compensated for by a change in the other dimension (as in the pouring of a liquid from one jar to another of different dimensions with concomitant change in water level). Both reversibilities embody the logical negation of an operation carried forward in one “direction.” Conservation in essence requires the ability to concurrently or successively carry out an operation and then undo (reverse) it. Operational identity is inherent in the total scheme of conservation. In contrast to preoperational identity, which is concerned with maintaining the identity of the object (e.g., it is the same water), it is concerned with the identity of quantities. Entailed is an “identity operation” of an operational grouping ( t o - O = O ) that is only associated with other operations as part of a total system (the additive composition, etc.). Conservation thus requires a conjunction of operations with reversibility and (operational) identity coordinated in a system of additive and compensatory structures. The 1968 formulation emphasized the significant role of transformations in conservation, as did the theory prior to that. In fact, Piaget (1968) asserted, “where there is no transformation we cannot speak of conservation” (p. 18). Piaget has now altered his view of the necessity of transformation in conservation and other cognitive processes (in an address to the Jean Piaget Society, Phdadelphia, June 1975). He now proposes a significant role for correspondences that are sought by the subject in the matching and comparing of states before being able to respond adequately to transformations themselves. Transformations thus emerge as no longer sufficient for an understanding of cognition. Piaget presents three hypothesized relations between correspondences and trans-
Constructing Cognitive Operations Linguistically
71
formations: (1) Correspondences pave the way for transformations (i.e., fixed states precede transformation). (2) There is an interaction between correspondences and transformations. (3) In operational structures there is an entirely new kind of necessary correspondence between a transformation and its inverse. In correspondences the relation between states is that of reciprocity. Conservation is offered as an example in which correspondences play a role. A comparison between states, e.g., a piece of clay observed in one shape compared with another in which a part has been transposed to the other end, while the body of the clay remains untransformed, leads to conservation judgments. In contrast to transformations, correspondences entail no inverse operation (and thus reversibility) but are related to one another when two states are compared in a relation of reciprocity. Beilin (1969) previously distinguished between two types of conservation-transformational invariance and fixed-state (or static) invariance, depending upon whether invariance in a quantitative concept is maintained in spite of a transformation or whether invariance is achieved by an implicit process of object transposition. Fixed-state invariance was demonstrated in a series of studies on area conservation, in which segmented squares of different configuration were compared, with the subject required to judge the equality of identical areas when they “appeared” to be different. This type of conservation ostensibly achieved by correspondence was referred to as quasiconservation, although it would appear now that Piaget accords this the status of conservation. Piaget reports that these correspondence processes lead to “precocious” conservation, although in Beilin’s experiments subjects experienced considerable difficulty in the task and were not successful in it until long after transformational conservation. This may have been due in part, however, to the dkcalage imposed by the task’s concern with area in contrast with the conservation of mass (solid quantity) o r number, or to the nature of the subjects, who were predominantly lower class. In sum, Piaget’s developmental theory, as well as the specific theory of conservation, has undergone a further significant development. This newer recognition of the significance of correspondences embodies the basis of a possible rapprochement with information-processing theories based on matching and comparison processes.
111. Methodological Issues in Piagetian Training Research A . EARLY HISTORY
Piaget’s theory of a cognitive system constructed by virtue of equilibrating mechanisms is not easy to test. Some of its assertions, such as those dealing with
72
Harry Beilin
assimilation and accommodation, are quite abstract and general. These difficulties with the general theory probably account for conservation’s having been so popular with researchers. It is an easily observablc, intuitively interesting phenomenon about which Piaget makes very specific claims. At the same time it is informed by the most abstract elements of Piaget’s theory. The principal means of testing Piaget’s ideas of conservation has been through attempts to train conservation. As already indicated, the Genevans themselves embarked on such studies in the mid-1950s to show the superiority of the equilibration model over alternative empiricist explanations. These studies did not become widely known to American audiences until the appearance of Flavell’s (1963) book on Piaget’s theory, the reports of Smedslund’s (1961a, 1961b, 1961c, 1961d, 1961e, 1 9 6 1 0 studies in the Scandinavian Journal of Psychology, and Wohlwill’s (Wohlwill & Lowe, 1962) studies in American journals. Many other studies carried out in the United States and in the Soviet Union seemed to the Genevans to be focused merely on accelerating the acquisition of conservation and other cognitive functions, which Piaget derided. Though some of these studies did in fact have that emphasis from the start, others were addressed to the nature of conservation mechanisms (Beilin, I97 la. in press). Many of the training studies reflect dissatisfaction with Genevan methods of inquiry. What used to be called the “clinical method” and is now more aptly characterized as the “method of critical exploration” (Inhelder, Sinclair, & Bovet, 1974)-which we shall call here the “critical method”-has often been criticized as unstandardized and overly verbal. Consequently, much non-Genevan training research attempts t o provide a standardized means of eliciting and gauging the child’s cognitive performance. These more standardized procedures have their origin, however, in the earlier Genevan experiments. The earliest training studies (Apostel, 1959: Morf, 1959; Piaget, 1958, 1959a, 1959b, 1959c; Smedslund, 1959, 1961b; Wohlwill, 1959), which were begun in Geneva, and in some cases carried out elsewhere, started simply. In Smedslund’s (1959) case, it was with a procedure that employed a scale to verify that the weight of a plasticine ball was conserved as it was transformed through various deformations of the material. Feedback provided by weighing on the scale was said to be comparable to a “reinforcement,” in that it was external information that was used to establish proof of a conservation judgment. The weighing procedure over deformation trials was followed by addition/subtraction trials in which pieces of plasticine were added and subtracted from the original mass. Smedslund observed that learning in this circumstance was very rapid and inferred from this that nonconserving subjects were actually close to spontaneous acquisition (that is, they were transitional conservers). This latter fact was used to denigrate the effects of training (Oleron & Thong, 1968). In Smedslund’s later experiments, through various refinements, the effects of “external rein-
Coristructirig C o p i t i v e Operations Linguistically
13
forcement” were contrasted with a method of conflict-equilibration training based on Piagetian theory. The results of these experiments (Smedslund, 196 l c , 1961d, 1961e, 1961f) suggested that while “reinforcement” could be effective for learning it did not compare with the effectiveness of the conflict-equilibration procedure. Again Smedslund emphasized that when learning did occur it was with subjects who had the rudiments of operational structures already instated. Piaget (19641, taking note ot’ Sinedslund’s studies, observed that while Smedslund was successful in inducing weight conservation he was unsuccessful with the same method in inducing transitivity. Piaget proposed that training is more likely to be successful when physical relations are critical t o the task, as it is in the conservation of weight where weighing itself is required. When logicomathematical relations are involved, however, as in transitivity, and physical experience is not critical to the operation, training should be ineffective. Piaget proposed, additionally, on the basis o f Wohlwill’s (1959) success in inducing number conservation through addition operations, that learning is most likely to occur if more complex structures are approached by way of simpler structures logically related to them. These early training studies led t o the following general conclusions, which set the stage for much subsequent debate: (1) Successful training of relations embodying logicomathematical operations in children who had not acquired them naturally was not likely to occui. (2) If it did occur it was most likely to occur with children who showed some evidence of acquired structure. (3) When training was successful in inducing conservation it would be by methods that were consistent with Piagetian theory of how mathematical cognitive structures are naturally acquired, namely, by a conflict-equilibrium method. Alternative methods associated with behavioristic or neobehavioristic theories of learning would not be effective in creating cognitive structures in children who did not have them. Training of conservation and other cognitive functions by non-Genevans was already underway (e.g., Beilin & Franklin, 19621,’ and alternative interpretations were quickly proposed for the conservation phenomenon (e.g., Zimiles, 1963). The large number- of studies t h a t were t o follow reflect various kinds of dissatisfaction with Piagetian theoi-y. First is the conviction that Piagetian stage theory imposes t o o restrictive a limitation on cognitive learning. The view that each logical structure requires the acquisition of prior structures in strict developmental order impresses some as overemphasizing developmental control and underemphasizing learning.
’This study was motivated by the desire t o test the control of the child’s cognitive level on his ability to profit from training in measurement tasks.
14
Harry Beilin
Secondly, it is argued that training and learning paradigms that are successful in other concept formation and attainment studies should be equally successful with the concepts described by Piaget. The implication is that Piaget’s characterization of learning, principally as “external reinforcement,” is methodologically if not conceptually out of date. Consequently, a number of other experimental approaches have been tested and have turned out to be successful in inducing cognitive change (Beilin, 197 la). A third dissatisfaction is with Piaget’s description of the mechanisms necessary for the development of logical thought, particularly for conservation. Alternatives to the necessary role of reversibility, compensation, and identity operations have been proposed and tested. These reductionist alternatives emphasize one type of Piagetian operation (such as identity or reversibility), instead of the complex of operations proposed by Piaget, or they provide alternative system explanations based on attention, memory, language, etc. These challenges to Piaget’s theory have resulted in periodic clarifications, additions, and changes in the theory, although at the same time the essential validity of the theory is reasserted. The most recent relevant work of the Genevans (Inhelder el al., 1974) manifests a modification of the Piagetian attitude towards training, although their fundamental posture on the relation of learning to development is not altered. B. METHODOLOGICAL ISSUES
Many methodological problems have arisen to complicate the task of the training study in providing insight into cognitive structures and the processes by which they develop. In the prototypic study of the conservation of number, for example, the subject is provided with two rows of objects (abstract objects such as circles, or concrete objects such as dolls). The experiment essentially starts with two rows of objects laid out in one-to-one correspondence, i.e., for each object in one row there is a corresponding object in a corresponding position in the other row, usually below it. The experimenter then determines whether the subject understands that the two rows are the “same” or “different” in number or amount, or whatever. Controls may include unequal numbers of objects so that a response set to “same” is minimized. One of the rows is then either contracted or expanded so it “appears” to be less or more than the intact row. The subject is again asked whether they are the “same” or “different” in number or amount, and once the judgment is made, is asked why. In studies of the conservation of liquid (also referred to in the literature as continuous quantity) two prototypic forms are used. In one, a jar of particular proportions (width X height) is filled with water. A second jar of different proportions is shown to the subject, water is poured from the first jar to the
Constructing Cognitive Operations Linguistically
15
second, and the subject is asked whether the amount of liquid is the “same” now as before. Elkind (1967) refers t o this as the “identity” case and implies that up to the time of his paper it was not used as a standard method of testing conservation. He cites another procedure, the “equivalence” case, which starts with two jars of equal dimensions as the more typical form of testing. With equal amounts of water poured into the two jars, the child is expected t o assert that the amount of liquid is the same in both. The water in one of the jars is then poured into a third jar of different dimensions while the first jar is kept within the child’s sight, and the standard question is asked. Elkind holds that the latter equivalence procedure is more difficult to conserve than the identity procedure and involves in fact a different form of conservation. Some controversy has arisen over whether different types of conservation are involved in the two procedures (e.g., Brainerd & Hooper, 1975; Moynahan & Click, 1972), but this issue will not be pursued here except to note later Piaget’s criticism of Jerome Bruner’s approach to the problem (Section IV, E).
IV. The General Training Model: Methodological Issues The procedure for each test o f conservation (i.e., length, area, volume. etc.) differs depending upon the attributes involved, but they generally follow the prototype procedures just outlined. The procedures for training conservation are much more varied, however. These depend mainly on the theoretical conception of the mechanisms involved in conservation and the conservation type. A . PRETEST
Training studies for the most part now converge on a common design. They usually start with a pretest to establish the child’s pretraining cognitive level. This is not always so, however, for pretesting is said to entail procedures that provide training prior to the testing itself (e.g., Beilin, 1965), or it is argued on the contrary that the pretest has negative transfer value in that it may increase the salience of “misleading” cues, particularly to transformation itself (Greitzer &Jeffrey, 1973). To assess for potential pretesting effects some studies include a no-pretest control group. The more usual practice, however, is to include a pretest to provide an assessment of the subjects’ pretraining cognitive status. This pretest is used as a device for selecting nonconservers and/or partial or transitional conservers for training. In some cases, however, subject selection is made o n the basis of the subject’s age. This procedure may be justified when gross difference in age (e.g., 5-year-old versus 8-year-old subjects) is taken as a gross indication of cognitive level difference. Brainerd (1972) makes the point
16
Harry Beilin
that age and pretraining cognitive level are often confounded in training research. Unfortunately, in the two cases he cites t o support this point, the authors in question have made an issue of showing that performance is more critically related t o pretest cognitive level than t o age. Age status is introduced as a rule only when it correlates highly with cognitive status (e.g., 4-year-old subjects are rarely anything but nonconservers on pretest). Much has been made by experimentalists of the essentially verbal nature of Piaget’s assessment methods. Nonetheless, even when experimental training is ostensibly nonverbal or minimally verbal, it usually involves response to questions with comparative and related lexical constituents such as “same,” “more,” “less,” “amount ,” and “number.” This quantitative and comparative vocabulary may not in fact be known t o the subjects of these experiments (Beilin, 1964, 1965, 1975; Griffiths, Shantz, & Sigel, 1967). In this case, lack of conservation may be falsely attributed t o a child when the deficiency may be linguistic. This can occur in spite of the fact that progression in knowledge of the lexicon (such as the number lexicon) dcvelops in parallel with cognitive development (Beilin, 1975). (This issue will be treated more extensively in Section V.) It is evident too from recent studies that nonquantitative meanings of comparative terms such as “more” and “less” are known before quantitative meanings (Beilin, 1968; Bloom, 1970). Consequently, training studies increasingly utilize pretest procedures for assessing the child’s linguistic knowledge of relevant terms; nonknowledgeable subjects are dropped from the study or else trained in the use of the appropriate terms. B. TRAINING
Training may be given to one experimental group or it may be contrasted with other types of training, in which case there would be as many experimental groups as there are pure or mixed training procedures. Training procedures ordinarily utilize materials somewhat different from the pretests, though the same pretest materials are often used in training with a change made in the posttest or transfer tests. The more desirable procedure is one with different materials t o ensure that posttest change is a consequence of the concepts trained and not of familiarity with the materials used or of response learning attuned t o the specific set of objects. Pretests, including both assessment of cognitive status (or knowledge of the task) plus linguistic knowledge, are followed by training with as many experimental groups as is appropriate t o the nature of the study and at least one no-training control group. A number of studies fail to include controls, sometimes for legitimate reasons, sometimes not. The function of control groups, as traditionally interpreted, is t o ensure that whatever change occurs in posttest is a function of the experimental procedures and not a function of pretest experi-
Constructing Cognitive Operations Linguistically
I1
ence or learning that occurs from sources other than the manipulation of the independent variable. A separate control group may be necessary t o partial out the effect of pretesting from other effects, however. C. POSTTESTS
Training is usually followed by at least one posttest given immediately after training or following a short interval. Posttests may also follow after a longer delay, usually from 1 week to several months. Two types of task appear in posttests as a rule: (1) those testing specific (or “near”) transfer, which repeat the pretest if training is with different materials or entail a test of the same concept with different materials from those used in pretest and training; and (2) those testing nonspecific (“far”) transfer, tasks different in kind from the conservation training task, such as area, weight, or length tasks when training is with number conservation. D. TKANSI’EK A N D DkCALAGE
Posttest transfer tasks are used to provide a “strong” test of the effectiveness of training. If cognitive change were said to occur by virtue of learning, the change would have t o b e demonstrated to be nontrivial and stable. Transfer tests are considered t o provide a measure of such stability. Some special problems have arisen with transfer in Piagetian experiments, however. Nonspecific transfer is predicated by Piagetian theory on the basis of common structure. Inasmuch as conservations share a common structure or common mechanism (those of additive and compensatory composition, reversibility, and operational identity and possibly reciprocity in correspondences), conservations should be achieved normally in the same stage period. Thus when one conservation is learned, transfer should readily occur t o the others. This expectation has created much debate because the so-called horizontal ddcalage phenomenon, also posited by Piaget, appears to contradict this expected transfer by virtue of common structure. Empirically, dCcalage is manifest in the regularly ordered acquisition of different types of conservation, starting with conservation of number and ending with conservation of volume as much as 4 or 5 years later. The de‘calage is accounted for theoretically by the presumed common logical structure in each case of Conservation conjoined with a secondary or specific logical or physical knowledge structure particular to the quantitative concept involved. The structures involved in the conservation of length, for example, are more complex and more difficult to attain than those of number (as is spelled out in detail in Inhelder, et al., 1974). Each conservation, then, has two components-a common structure coniponent shared with other conservations, and a specific component that is not shared with other conservations but is
I8
Harry Beilin
shared instead with other cognitive structures (number, for example, has component properties that are shared with number classifications and number relations, etc.). This distinction is reminiscent of g and s factors in factorial theories of intelligence. One difficulty which the common factor versus de‘calage conflict engenders is that it complicates the definition of “stage,” in that stage in Piagetian theory is defined by both common structure and vertical ddcalage (i.e., the difference in age of attainment between common structures, for example, concrete operations vs. formal operations). Difficulty arises empirically when horizontal and the so-called oblique dCcalages2 merge into the vertical de‘calages, making clear-cut patterns of achievement difficult t o discern. In training studies, horizontal and oblique decalage becomes an issue in the seeming paradox that if the basic conservation structures are attained they should transfer-yet at the same time should not, because of the complications of horizontal and oblique ddcalage. If nonspecific transfer to another type of conservation does not occur, Piagetians are inclined to attribute it t o the de‘calage, although it might be argued contrariwise that training in respect to the common structure plus specific context knowledge should generalize to all appropriate contexts with sufficient training. The results of such attempts have not been consistent, although, in the main, training does lead to both specific and nonspecific transfer, for most types of training (see Beilin, 1971a, in press). E. STRONG VERSUS WEAK CRITERIA
Delayed posttests also provide a strong test of training. The need for such strong tests arises in part from Piaget’s criticism of training studies that are claimed to have induced cognitive attainments from relatively weak experimental evidence. Piaget puts it this way: “But, when I am faced with these facts, I always have three questions which I want to have answered before I am convinced. The first question is, ‘Is this learning lasting?’ What remains two weeks or a month later? . . . The second question is, ‘How much generalization is possible?’ . . . Then there is the third question, ‘. . . what was the operational level of the subject before the experience and what more complex structures has this learning succeeded in advancing?’ ” (Piaget, 1964). The experimental methods already discussed deal with three of these four questions: the pretest establishes the subjects’ operational level; specific and nonspecific transfer posttests establish whether there is generalization; and the delayed posttest establishes whether it is lasting. Another posttest procedure is often used t o answer the question as to whether what is learned is lasting and also t o explore
* Oblique ditcalage is indicated in differential achievement between conservation, for example, and classification and ordering tasks, which are informed by common concrete operational structures but different component structures.
Constructing Cognitive Operations Linguistically
19
what structures have been learned. Smedslund ( 1 9 6 1 ~ )used a method designed to challenge the security of the child’s judgment by presenting him with a situation that is in conflict with his conservation judgment. The impetus for this was Piaget’s claim that when conservation becomes operational the child feels that conservation is necessary (i.e., “logically” necessary) and in fact will often ridicule the experimenter for presenting him with questions relating to the “obvious” equality of the presented quantities. Thus, if conservation were truly attained in a training experiment, the argument goes, trained conservers would resist extinction as d o natural conservers. After establishing that two clay balls weighed the same before and after deformation, Smedslund ( 1 9 6 1 ~ )surreptitiously removed some clay from one of the balls. The child was again tested t o see if he maintained conservation after weighing the clay or reverted to a nonconservation judgment and explanation. Extinction tests or trials have become part of many studies, even though Smedslund found that all l l subjects trained in his study extinguished in the countersuggestion test, and 6 of the 13 natural conservers resisted extinction. [See Miller (197 1) and Miller and Lipps (1 973) on this issue.] Possibly the most controversial issue in conservation training research concerns the knowledge acquired in a training experiment and the criteria by which one judges the attainment of such knowledge. Ostensibly, if training is successful, the general conservation structures or mechanisms already referred t o are acquired as well as the particular context knowledge associated with the particular type of conservation trained. Two behavioral criteria are employed by the Genevans in assessing such knowledge. One is a correct judgment or series of judgments for the particular quantitative concept trained, with the critical response usually that the “amount” of material is the “same” in respect to number, area, substance, etc. The other is the subject’s justification or explanation of his judgment. Both judgtnent and appropriate justification are required to establish the subject’s cognitive level. Controversy arises as t o whether both criteria are needed for minimal evidence of conservation. Because of the potential inhibition and/or confounding of performance when verbal means of testing are used, the effort is sometimes made to judge the presence of operations by nonverbal means (e.g., Beilin, 1965; Braine, 1959). kxplanations or justifications can only be verbal but “judgments” may be verbal or nonverbal (i.e., the subject can make a motor response that would be equivalent t o “same” and another that would signal “different” or “more.” As already indicated, however, some verbal instructions are necessary to establish the experimental conditions for the subject. Since a child could be a conserver and at the same time lack insight into why he is performing as he does, or because of limited language facility is incapable of expressing what he understands, an argument is made for using the judgmental criterion alone. This, however, is not the sole source of difficulty concerning the criterion problem. There is a more general question regarding the
80
Harry Beilin
use of strong versus weak criteria in accepting evidence of conservation. Weak as against strong can be defined by the criteria suggested by Piaget: the strong are those satisfying all four of Piaget’s criteria; the weak are those satisfying fewer than four. Weak and strong criteria can be defined as well, however, by the nature of the unit of description and analysis. An individual subject’s responses can be considered in two ways. One is to take the subject’s response(s) as a unit (a within-subject measure). When this is done a categorical decision is taken (conserver versus transitional versus nonconserver) as the basic datum, and the number of subjects in the conservation (or transitional) category before and after training is used as the index of training success. A second criterion is needed, however, for determining when a response is a conserving, transitional, or nonconserving response. This may be a qualitative judgment based on justifications, o r it may be quantitative (based, say, on five of six trials with correct responses), or it may be a combination of these. The other way that subjects’ responses are treated statistically is to add or pool them for the number of subjects in a group (between-subject measure). In this case the number of correct responses in a series of trials is summed and the criterion of successful training is the mean of correct responses in posttest compared with the pretest mean. The number of correct responses both on posttest and pretest could be a small percentage of responses (say, 10% and 20%), however, and still show a significant difference with a large enough sample. In consideration of the fact that a statistically significant change could in this way be recorded, even though not a single subject could be identified as a consemer using a within-subject categorial criterion, one can only identify the between-subject nieasure as providing a weak criterion of conservation training. Nevertheless, many experimenters utilize just such group change data based on conservation responses, without at the same time indicating the number of conservers involved. The fact that the choice of criterion can affect the decision as t o whether conservation is present, o r learning has taken place, is demonstrated by Gruen (1966), among others. There are two aspects to the issuc, one conceptual and the other methodological. If one chooses a conceptual definition for conservation that is different from Piaget’s, can it be legitimately claimed that eonservation is being investigated? And if one uses procedures that d o not expose the ostensible Conservation operations a la Piaget, can one legitimately claim one is studying conservation? An exaniple of the conceptual difficulty is the argument between Piaget and Bruner (and Elkind) over the role of identity in conservation. Bruner (see Bruner, Olver, Greenfield, et al., 1966) holds that knowledge of identity is sufficient for conscrving quantity (as does Elkind, 1967). Piaget (1968) argues, however, that only operational identity (same amount) and not object identity (same water) qualifies as a necessary ingredient of operational conservation. In other words, the way conservation is defined b y Piaget requires operational identity for its comprehension. [Piaget also addresses a methodological limitation in Bruner’s studies that yield false positives by using
Constructing Cogriitive Operations Linguistically
81
a weak procedure (criterion) l.or judging conservation.] The studies of Mehler and Bever (1967) and those of Bryant (1974) appear to share the same conceptual difficulty. What these studies collectively imply is that a “truer” notion of conservation is embodied in a dirfet-ent definition of conservation. Thus, if conservation were identified simply as an invariance problem without the properties identified by Piaget (e.g., reversibility plus operational identity), one might find evidence for it at very young ages. With a conceptual definition different from Piaget’s, different methods 01- standards of measurement become acceptable. Thus, although one could identify a phenomenon as identity conservation, it is not the conservation uf a quantitative concept which Piaget identifies as conservation and justifies in various ways. A number of arguments directed against Piaget’s theory may similarly he reduced to differences in conceptual definition. Another aspect of the issue concerns whether it makes any difference if strong or weak criteria are used to assess the presence of conservation. Brainerd (1973; see also Brainerd & Allen, 1971) argues first that the criterion problem is not significant because strong criteria have been used in both successful and unsuccessful training studies. He argues fiirtlier, however, that only judgments are necessary for assessing subject competence. If the only question addressed in these studies was whether conservation training was successful, then the use of such a criterion might be justified. Furthermore, if one assumed that judgments and explanations probe the s u m structures, then judgments alone would be sufficient t o assess the robustness of such structures, as Brainerd claims (assuming, of course, that other sufficiently strong methodological criteria were adopted). What a judgmental criterion fails to do, however, is fully t o expose the nature of the underlying structure. Brainerd’s and similar proposals neglect to take cognizance of the fact that for Piaget concrete operational thought is more than rule-regulated behavior. Though the judgmental response and equivalent nonverbal response patterns may provide adequate knowledge of the underlying structures that govern choice, since such structures by their nature are nonverbal, it does not reflect the child’s consciousness of the events with which he deals. The verbal justification does this best in demonstrating that a symbolic (and operational) level of underlying structure has been reached that can be represented in a verbal justification. Consciousness of the relations that underlie behavior reflects capacities that go beyond those of rule-governed or structuregoverned behavior alone. I t is n u t an accident that Piaget so often titles his books “The child’s conception o f -.” A judgmental criterion thus ignores an essential ingredient in an adequate assessment of the child’s knowledge (Piaget, 1971, pp. 47-49). Brainerd makes the additional claim that judgments are devoid of built-in errors which are characteristic of explanations and incapable of being eliminated by methodological refinement. [See Recse and Schack (1974) for further comments o n Brainerd’s argument.] The “judgments” Brainerd refers t o do as a rule
82
Harry Beilin
(e.g., in his own studies) embody verbal responses, if not verbal explanations. If one considers the difficulty, as evidenced by a great deal of conflicting research, in determining what lexical terms such as “more” mean to 2- to 5-year-old children (see, e.g., Beilin, 1975, Chapter 5), one would surely hesitate t o say that a judgmental criterion that embodies the use of such terms is readily amenable to methodological refinement, whereas verbal explanations are not (and therefore should be avoided). There are in fact difficulties with both sets of criteria that methodological refinements can improve upon but cannot fully eliminate. Brainerd (1974) again argues that whatever errors (“false positives” and “false negatives”) may appear with either response criterion chosen (judgments or explanations), they are both spuriously supportive of theoretical predictions, and so it does not matter which criterion is chosen. On this basis he holds that the judgmental criterion should be used because it is simpler. Since critical information is missed by omitting justifications, simpler is not necessarily better. Brainerd also argues that the nature of a Piagetian assessment is solely t o determine underlying structure of different concepts, apparently without recognizing that some phenomena such as transitivity and conservation have both common and unconimon structures. He argues further that Piaget theorizes only general synchrony in the achievement of a stage, when in fact the Genevans (see Inhelder et aL, 1974) posit, by virtue of dicalage, asynchronies in development of the very types Brainerd himself proposes. In sum, appropriate research design depends on a number of factors related t o the conceptual criteria by which conservations are defined. These criteria require the following: an assessment of the child’s pretraining operational level; an assessment of pretraining linguistic knowledge; control groups; posttesting for specific and nonspecific transfer; adequate assessment of the child’s posttraining operational level, both behaviorally and verbally; tests delayed a week or 2 months after training; and extinction trials. Not all studies require so strict or so extensive a set of design features. Studies of limited scope that deal with specific issues may require only some of these features. None of the studies that will be reviewed here includes all of the methodological constraints. For the most part they do follow a pretest, training, posttest design and vary in the extent to which they employ strong criteria. Where methodological problems require qualification of the findings, these will be discussed.
V. Linguistic Training A. VERBAL RULE INSTRUCTION
Piaget and his colleagues have made a particular effort to show that language at and before the stage of concrete operations is not critical to the development of
Constructing Cognitive Operations Linguistically
83
cognitive structure; instead, cognition is critical to language development (cf. Beilin, 1975). Thus, for the development of conservation and related concepts, it is argued that language instruction and linguistic experience are unrelated to their construction. Sinclair [as will be discussed more fully in Section V, B,] attempted to verify the Piagetian claim by showing that training nonconservers with linguistic structures typically used by conservers resulted in n o improvement in conservation [see also Sinclair-de Zwart (1967)j. A quite different language training method was used by the present author (Beilin, 1965). Instead of stress on the lexicon, this method was based on the embodiment of conservation operations in rule statements that were coordinated in training with relevant expcrimcntal manipulations. The rules represent reversibility by inversion, reversibility by compensation, and the identity operation, and reflect the typical verbal justifications given by conserving subjects (Piaget, 1968). The verbal-rule instruction procedure was compared in effectiveness with a deformation-eqiiilibration procedure, an attention-orientation method, a method based on creating a nonverbal “learning set,” and a control procedure. Pretests and training werc with length and number conservation, and the posttests included specific and nonspecific transfer (area). Verbal-rule instruction was the most effective procedure in inducing conservation in the specific transfer posttests. There was little transfer, however, to the nonspecific transfer task (static area conservation). (Natural conservers by contrast did transfer t o the static area conservation task.) The lack of nonspecific transfer was attributed to area conservation being (1) a more advanced task in the normal de‘calage; and ( 2 ) , as a static invariance task devoid of transformations, more difficult than the usual transformation task (Beilin, 1969). It was proposed (Beilin, 1965, 197 1a) that learning occurred because verbal-rule instruction provided an instructional algorithm for processing perceptual input. Such an algorithm may or may not lead to true operatory structure, depending on the cognitive level of the child. I f the child had n o available relevant operatory structure t o which the algorithm could be assimilated, it could act as a placeholder until the relevant structures developed naturally. Nevertheless, without cognitive conflict, functional conservation was induced as a consequence of verbal-rule instruction. Operatory conservation was not ruled out, however, as we will consider later. The two linguistic approaches, lexical training in relational terms and verbalrule instruction, were combined in a study (Carlson, 1967a) that found this combination was more effective than “minimal verbal instruction.” Carlson holds (with Sinclair) that verbal instruction acts to orient the child’s attention to relevant features of the stimulus, and also suggests (with Beilin) that verbal training provides the subject an opportunity to use a model rule for processing relevant input. In another study, Carlson (1967b) employed a more stringent delayed posttest in which the combined rules-and-lexical training was contrasted
84
Harry Beilin
with training without (or with a minimum of) these types of training. The provision of both lexical and rule instruction was as effective in this as in the prior study. The verbal-rule procedure was used by Smith (1968) in a conservation of weight study, and the effectiveness of the method was again shown. It was compared with a “reinforced practice” method and an addition/subtraction conflict procedure. Only the verbal-rule group showed significantly better performance than a control. Where the verbal-rule procedure was combined with extension and contraction manipulation of length in a number conservation task (Blum & Adcock, 1969), it resulted in nine of ten nonconserving trained subjects conserving on posttest, with one becoming transitional. In a delayed posttest, six of eight older subjects in the sample maintained their conservation performance, while two regressed. It is not uncommon for verbal-rule instruction to be introduced as a subsidiary part of an experiment where the focus is ostensibly on other procedures. In such cases, as in the Overbeck and Schwartz (1970) study, rule instruction was confounded with the other procedures to such an extent that it is difficult t o isolate its effect even when training was successful. The effectiveness of verbal-rule instruction was further confirmed by Peters (1 970), although a “perceptually-guided cue discrimination” procedure, in which configurations of dots aided the differentiation of number, was also effective. Verbal rules in conservation training usually comprise a number of components, as does conservation itself. Sjoberg, Hoijer, and Olsson (1970) attempt to isolate which of three components would be effective in inducing conservation of weight: “decentering” (change in form, not in weight); reversibility (transformed object returned to original shape and retaining weight); and addition/ subtraction (with nothing added or subtracted, weight must be unchanged). In addition to verbal-rule groups, there was a “reinforcement” control group which saw a demonstration without any experimenter explanation. The posttest included a transfer task (volume) and some extinction items. In training, the subject made a prediction of weight and, irrespective of whether he was correct or incorrect, obtained feedback together with a statement of the specific conservation principle. In training, 90% of each nonconserver training group of 15 was successful. In posttest, the three verbal-rule groups had high rates of success and were significantly better than the control (e.g., reversibility: 13). The demonstration control group had one conserver. Only the addition/subtraction group, however, showed a high degree of transfer ( l l ) , although the demonstration control group improved too (6). Posttested natural conservers dropped from 23 to 11 in extinction trials with a similar drop in the addition/ subtraction group. The other groups regressed much more. Verbal justifications yielded the same principles enunciated in the verbal-rule instructions, even in
Constructing Cogiiitive Operations Linguistically
8i
transfer, although with reduced frequency. Overall, rule instruction proved to be effective, with the addition/suhtraction type of verbal rule, and weight precliction most effective among those used. The authors held that the transfer and extinction test results demonstrated that verbal-rule training led t o true operational change. As with many verbal-rule instruction studies, no suggestion was offered as t o why verbal-rule instruction was effective. Verbal-rule instruction was also tested in a study b y Hamel and De Witt (1971), the results of which were related t o the pretest language level of the subjects, as determined b y three language tests. Subjects predicted water levels in a conservation of liquid experiment. Liquid was poured by either the ?xperimenter or subject, followed by visual verification of the subject’s prediction. Incorrect response was followed by verbal-rule instruction (as in Beilin, 1965). Posttests showed considerable improvement, from 4 subjects correct on the first trial t o 25 on the fourth trial, with about 30% overall conserving. Pretest language level was significantly related t o conservation performance. The inore verbal subjects started at higher conservation levels and acquired more conservation as a consequence of rule instruction (53% versus 38% of the less verbal group). In a later study (Hamel & Riksen, 1973), two groups of nonconservers were each given a different kind of verbal-rule instruction. One focused on identity, the other on reversibility for discontinuous and continuous quantity conservation. A common base of rule instruction was given to both groups (“same amount of water”), with identity instruction (“it’s the same ~vatcr”) given t o one group and reversibility (“when you pour it back it reaches the same level”) given t o the other. Both training groups did better than a control group, but there was n o difference between the trained groups on posttest. The identity instruction group showed greater gains in nonspecific transfer (to space, number, substance, and weight), whereas the reversibility group showed greater gains from pretest t o a delayed posttest. Equality and inequality conditions were included in the posttests to negate the possibility that the results were due to simple response sets, a type of control not always included in conservation experiments. In this study, then, rule instruction led to effective learning with nonspecific transfer that carried over to delayed posttests. Hainel is iriclined to an interpretation of his results that supports Bruner’s identity reduction thesis, but since training included rule instruction for invariance of quantity (“same amount of water”) in addition to object identity, it is difficult t o accept the reductionist thesis. (Bruner, it should be noted, used the same instructions.) Inasmuch as invariance of quantity is the critical aspect of the Piagetian thesis, its very addition t o the object identity rule precludes rejecting the Piagetian thesis even if the effect of the object identity rule exceeds the effects of the reversibility rule. A study that included a verbal-rule procedure (Figurelli & Keller. 1972) was designed primarily t o determine whether training differentially affected middle-
86
Harry Beilin
versus lower-class children. Posttest effects of training in number, continuous and discontinuous quantity, size, and substance were significant, independent of socioeconomic status, although on transfer middle-class subjects were better than lower-class subjects, as they were on pretest. On the whole, nonspecific transfer to area and length was not effective. In addition, training was effective with pretest partial conservers, but not with nonconservers, as is often claimed by the Genevans (e.g., Inhelder e f al., 1974). A study of English mentally retarded children who ranged in age from 9 to 16 years (IQs 47 t o 81) was concerned principally with the dCcalage among conservation tasks (Lister, 1972). Additionally, however, two groups of subjects were trained by verbal-rule instruction-one group in understanding quantity in several attribute contexts; the second, in area conservation. A third group was a control. The experimental groups were given four types of rule statement: (1) quantitative identify, “nothing added or taken away”; (2) appearance-reality, “looks different but is really the same”; (3) compensation, “one is more this way, the other is more that way”; and (4) reversibizity, “you can get it back as it was before.” Rule giving was in conjunction with observation of transformations, manipulation of reversible transformations, and verification. In the quantity training group and in the area training group 15 of 17 subjects conserved on every posttest item. The two subjects in each group who did not profit from training did not conserve at all. This was even true for number, although 11 other subjects who had n o conservation on pretest did profit from training. The four nonlearners were given additional training in various conservation problems and then caught up with the others. All trained conservers gave “clear” justifications. The most frequent explanation was “nothing added or taken away”; compensation and material identity were rarely given as justifications. None of the 17 control children showed conservation on any test. Thus, educable mentally retarded children as well as normal children were capable of learning to conserve by means of verbal-rule instruction. It is possible t o make rule instruction contingent solely on incorrect response, or it can be given after both incorrect and correct response. Independently, informational feedback as t o the correctness of the response can be offered or omitted in an experiment. Beilin (1965) provided rule instruction only after incorrect responses, and corrective feedback after all responses. In other studies rule instruction and feedback have been given after both correct and incorrect responses. Two studies were specifically directed to the relation of verbal-rule instruction to feedback. In one, Siegler and Leibert (1972) followed the Beilin rule procedure with four training groups (feedback versus no feedback; rules versus n o rules) in training conservation of liquid quantity. Pretests and posttests additionally tested for solid quantity (clay) and length. Nonconservers ( N = 40) were selected for training from those who failed in liquid Conservation. Verbal rules given depended on the nature of the trials: “same amount”; quantitative
Constructing Cognitive Operations Linguistically
XI
identity (nothing added or taken away); compensation (thinner and u.idci-); object identity; reversibility-inverse. Feedback was either positive o r neutral (ix.. the child was not told he was “wrong”). The results showed that the rule plus feedback treatment was best in all test trials: 70% of the subjects consei-vctl 111 conservation trials; 50% were correct in nonconservation, perceptually discrepant trials; and 50% of nonconservation, perceptually consistent tlials wci-e responded t o correctly. Of rule-trained subjects, 35% gave both correct judgments and correct explanations in liquid and solid quantity conservation (Xi7 for clay). Only 10% of the feedback-only group did as well. At the saiiit t i i i w . only one subject generalized t o length conservation. Thus, while feedback ~ i i d rule instruction were each effective in improving performance, the conjuiictioii of the t w o was even more effective on b o t h judgments and explanations. l l i e results, the authors point out, could not be attributed t o “mere niimiciv” 0 1 response bias t o the word “same,” as performance improved in rclevaiit I I O I I C I servation trials where a “same” response was not called for. In addition. 6 t i t :‘O subjects generalized in specific transfer trials (solid quantity) although i i o t i i ! ‘I nonspecific transfer (length) task. In a second study, Siegler (1973) used simple one-sentence rules rathci tlian rules that include “descriptions and justification.” Subjects were pretestcd on solid and liquid quantity and trained on liquid quantity. There was a rulc plus feedback group and a control group. The rule used stated that “when w c p o i i t all the water from one glass t o an empty glass there is the same amount ofwaier to drink as before,” with variations for the amount of liquid poured. On s o - d l e t l “all plus some” trials, the rule was that “when we pour all the watcr fro111 oiic glass plus some water from another glass into an empty jar we havc moIc w a i ~ ‘ t in the new glass than there is in any of the other glasses.” Similarly, i n the “ o i r l ! some” trials (i.e., with the rule “. . . pour only some water . . . there is less. . .”) subjects received feedback whether they were right or wrong. Negative fecdl)acl\ was employed in this study: “That is not the right answer.” Again v e r M riiic plus feedback training resulted in significant learning. Of nine trained S L t h j ~ C i S . eight succeeded on all posttest problems or on none at all. I n noiisp~,cil’i~: transfer (to clay), however, there was a drop in conservation. Since there wiis I I O direct reference in the rules t o reversibility in this experiment, Siegler believes it is not a necessary component in conservation and thus the experiment is s:ii(l i o disconfirm Piaget’s view of conservation. Training is said instead to providi’ :\ strategy of attending to manipulations and connecting them with thcir c o i ~ v quences, which is close to Sinclair’s notion (see Inhelder ef al., 1074) 1ll:iI linguistic training merely leads t o chsnges in attention, although Siegler tags on ;I more traditional learning explanation. The rejection of reversibility as s co1llp(jnent in conservation is questionable, however, since in the experiment there IS x i i implied reference t o reversibility in that part of the rule statement refcrriiig to “the same amount of water to drink as before.” In addition, since Pi:iyt jii
88
Harry Beilin
presently holds that conservation can be attained in situations where correspondences are noted between initial and end states with no focus on the transformation itself, it may be that Siegler’s rule statement embodies a comparison that establishes such a correspondence. Siegler’s interpretation also differs from the I’iagetian opinion in holding to the more traditional view that generalization is due t o similar perceptual transformations rather than t o generalization by coninion structure. Sullivan (1969) has been interested not only in rule instruction but also in the effectiveness of models in filmed presentations of conservation experiments. In one study, a filmed model demonstrated conservation together with verbal statement of the conservation principle (rule), and in another case simple conservation judgments were stated without rule specification. The verbal rule spelled out compensation, identity, and the appearance-reality distinction. In the no-principle condition the filmed model would merely respond “same” (correctly) after a transformation but give no rule information in response t o a question put by the film’s interrogator. The posttests, which included repetition of the pretest (substance: water or orange juice; wire), a nonspecific transfer test (clay), and an extinction test (the Smedslund method) showed rule instruction t o be better in all posttests than no-principle modeling (7 versus 3 subjects conserving; 7 versus 2; 7 versus 2-011 the respective tests). The results of this study differed f r o m a prior study in which the no-principle group also learned. In another experiment, rule instruction had no differential effect, but t h s result was discounted by Sullivan as due t o artifacts in the experiment. The inconsistent nature of these “modeling” plus rule training results is evident in a later study on training conservation of substance (Waghorn & Sullivan, 1970). Again the materials were wire, water, and clay in a pretest-posttest generalization design. The critical groups were shown conservation modeling films that had two actors, an interrogator and an adult model. In the verbal-principle group the model conserved liquid quantity and gave reversibility, identity, and compensation arguments in his justifications. The no-principle model conserved without explanation. There was substantial improvement in both principle and no-principle groups, not differing from each other, but differing from the control. In contrast t o Sullivan’s earlier study, rule instruction was interpreted as of little consequence relative t o the effect of simply modeling correct answers. Another series of studies on the effects of modeling with verbal-rule procedures has been interpreted within the context of social learning theory. In one experiment, Rosenthal and Zimmerman (1972), like Siegler, varied both feedback and rule instruction to yield four experimental groups plus a control. The subjects were divided on the basis of pretests into a nonconserver group (providing four or fewer conservation responses) and a conserver group (with five o r six conservatiun responses). The latter group was reserved for extinction training. The verbal rule used was, “They were the same in the first place,” which the
model said in response t o the cxpcriinenter’s request for an expianation of‘ a conservation manipulation. All ti-ealincnts led t o equal gains in posrtests arid also in transfer, using a judgments-only criterion. With an explanation criteiion the verbal-rule group showed a signiiicant training effect, whether o r not feedback was given during training. (Feedback was detined as verbal praise from the model, not as corrective feedback.) The fact that the feedback n o d e suhlccts were better than the controls was attributed t o the effect of simply observing the model. In a second part of tlic study, partially conserving pretest su1)jccts who observed tlie model fail t o give conscrvation judgments were tested t o see whether they extinguished their conservation responses. For both judgments and judgments plus explanation criteria there was a drop in conservation pcrtorniance and a drop in the level of conservation explanations, indicating partial extinction. In a third experiincnt with economically disadvantaged Mexican-American children there was a noiiniodeling verbal instruction variation i n which the experimenter rather than the model stated the rule plus the correct judgment. The modeling group was supcrior t o the verbal instruction group in both posttest and transfer. In a fourth expeiiinent cariied out with very young children (age: 4 years, 2 months to 4 years, 9 months; median: 4 years, 6 months), verbal-rule instruction and modeling led to significant increases in correct responding but no increase in verhal justifications. In Rosenthal and Zimmerman’s (1972) experiments the m success was the mean of correct 1-esponses and not a subject criterion. One does not know then how many subjects responded with sufficient consistency to be considered conservers. Nevertheless, the authors conclude that observing the model conserve without giving explmations was “indicative of inferential thinking elicited by modeling” in accoi-d with the assumptions of social lear-ning theory. In a further study, Zimniernian and Rosenthal (1974a) again assessed the effects of modeling, this time with corrective feedback, in length, number, and “2-dimensional space.” The subjects were those who failed on pretest (mean age: 5 years, 7 months). Both equality and inequality conservation were tested. Inequality items were transformed so that they still appeared unequal? but in reverse of the starting configuration. Inequality and equality trials were pooled in the analyses. Training entailed the use of a model plus rule instruction. The rule statement was the same as in the prior experiment, except for the substitution in the inequality condition o f “had more or larger.” The training gi-oups were as follows: (1) positive feedback with rule; (2) model plus correction and verbal rule; (3) model, no correction; and (4) control. All training gi-oups improved with both judgments ;ind judgments plus explanation criteria. The most effective procedure was tlie model plus correction (and verbal ride), followed by correctioiz-no modcl, next model-no correction, and last the control. The subjects were also tested on the frequency with which they correctly returned transformed materials t o their initial conditions. Again the
‘)(I
iiiotirl
Harry Beilin
phis correction (and verbal rule) group was best, and the poorest was
r u ~ ~ ~ d -corrcctiorz. no The study thus shows that verbal rule instruction was the
inost effective method, with corrective feedback more effective than modeling :ilone. Thus, contrary t o their earlier study with Chicano youngsters, the additJon of the model was not the critical element in training. A later study (Zimmernian & Lanaro, 1974) was concerned with relating verhnl rule instruction to the (nonverbal) demonstration of a reversibility response. with 4-year-old subjects, in response t o Beilin’s (1971 a ) observation that no i:hiltlren in that age range or younger had been shown t o that time t o profit i r o i n training to conserve. There were three experimental groups: modeling ~ I ~ i n modeling e; plus reversibility; and a control. All modeling included rule instruction. The same materials were used as in the prior experiments. In the ii,vt.tsihility condition, the subject observed the transformation in materials; in ttic others, not. Treatment and control conditions were repeated t o enhance “t-c.c.all” with a different set of materials 2 days after the first training session. Tlic p a t e s t gains from pretest t o posttest, in judgment alone, was for the model plits reversibility group relative to the modeling alone and control groups. In the i!cl;iyetl posttest the modeling alone group was superior. With the judgments plus cxpl:ination criterion, the training effect was significant, with slightly greater s c ‘ o i - c ~on the delayed posttest. There was, however, no difference between the ttioJclin~yplus reversibility and modeling alone groups. Transfer occurred and w s best in the modeling alone group. The authors concluded that 4-year-old :,uh,jects can be trained with rule instruction plus models and that the acquisition is stable. The authors also concluded that learning was not dependent upon the children’s cognitive level. One difficulty with the Zimmerman and Lanaro experiment is that no evidence is provided that the 4-year-old subjects had the requisite knowledge of such terms as “same length” and “as before,” although they were more likely t o understand “longer,” which the data d o show. With verbal instruction the child riiay simply have learned an algorithm-answer t o “before” and then applied it to tlic perceived transformations, for there is evidence that children at this age have olily ;I very limited conception of “before-after’’ time relations (Beilin, 1975). Another, and more serious, difficulty with the experiment is that although the chxnge data reported are statistically significant in respect t o responses, they are ver.y ineager and not enough t o suggest that individual children were conserving. None, in fact, appeared to have reached an above-chance performance level. With ;in expected chance level of 3.5 responses correct, and obtained mean response It.vels u t 2.5 correct, subjects on the whole were not responding even at chance level. Training was clearly “effective,” t o some extent, but whether it made conservers out of nonconservers is questionable. ‘I’he conclusions to be drawn from studies utilizing verbal-rule instruction will be presented later. First we wish t o consider those studies concerned with lexical kiiowledge and conservation.
Constructing Cognitive Operations Linguistically
91
B. LEXICAL TRAINING AND LEXICAL KNOWLEDGE
Piaget’s position on the development of language vis-A-vis cognition i s that knowledge of the lexicon appropriate to an experimental task such as conservation is not expected t o be attained and understood prior t o the development of related cognitive operations. In other words, linguistic competence cannot be in advance of cognitive operations; in fact it should be a reflection of them (e.g., Piaget, 1954b). The role that the lexicon plays in knowledge of conservation is deeply rooted, then, in the general relation between language and cognition. Nevertheless, the argument is made by some authors (e.g., Griffiths et al., 1967) that success in conservation tasks may not become evident or can be inhibited because the subject lacks knowledge of the testing vocalary that includes such terms as “same amount,” “more,” and “less.” Sinclair (see Sinclair-de Zwart, 1967; Inhelder et al., 1974), in an attempt t o confirm the Piagetian claim, examined the words typically used b y subjects in conservation experiments. These fell into two classes, descriptive (“one,” “quantity,” etc.) and comparative (‘‘rnoi-e,” “less,” etc.). They were also characterized by Sinclair as “scalar” and “vector” terms, respectively. A study was then made of how these terms were understood and used in a conservation of quantities experiment (with clay). Though most terms used by the experimenter were understood by even the youngest subjects (age: 4 years, 6 months) in the experiment, their use differed depending upon whether the subject was a conserver or a nonconserver. Most nonconservers’ statements employed two-part (coordinated) structures embodying the use of scalars (e.g., “a lot,” “not a lot”) and much less use of vectors (“more,” “less”). The nonconservers also typically used undifferentiated terms (length terms in one instance; thickness terms in the other), whereas conservers used pairs of opposites (long/short ; thick/thin). In a conservation of length task (with pencils) conservers used four-part constructions: “This is long but thin, the other short but fat”; whereas nonconservers said, characteristically, “This pencil is long, that one is short, this one i s thin, that one is fat.” In a second experiment, subjects were trained in the use of the terms and constructions Sinclair found in the speech of conservers. Children were “encouraged” t o use examples proposed by the experimenter, e.g., “Give more t o one and less to the other,” in respect to dolls possessing different amounts of clay and marbles. Of 31 pretest subjects, 28 were nonconservers on conservation of liquids; in the first and second posttests after training, 18 remained nonconservers, 7 were intermediate, and 3 became conservers. Sinclair concluded that linguistic training did not lead to progress in conservation. If it did anything, it aided some children who had already acquired conservation t o give clearer verbal justifications. That is, language t nining facilitated rapid verbal coding and the efficient storage and retention of information (Inhelder er al., 1974, p. 215). Piaget has often commented that although cognitive structures develop in a
92
Harry Beiiin
fail-1y fixed order. their rate of development can be affected by social and
cultural experience. The same is true for conservation, and cross-cultural eviilerice is cited that points t o such differences in rate of acquisition. In one such instance, I’iaget, noting the way the comparative form is used in French and English (Sinclair found they were the same), observed that Turkish, which lacks :i w n i p a r ~ h l ecomparative form, is apparently associated with later acquisition of coiiscrvation t h a n is observed in Geneva. Sevinc and Turner (in press) purS I I ~ I I tliis observation with two groups of bilingual migrant Cypriote children residing in London-one predominantly Creek-speaking, the other predciiiiinantly Turkish-speaking. The Greek comparative is comparable to the English: the Turkish is different. A third group of native English speakers was ;rlso studied. (Subjects ranged from 4 t o 1 I years of age.) The Creek and Turkish subjects thus shared the same migrant Cypriote bilingual cultural background hilt dit‘fered in comparative type, whereas the English and Greek children t1it‘ti.i-edi n cultural experience but were similar in their linguistic use of comparatives. The children all spoke English, however, which confounds the results t o sonic extent in that they all had common knowledge of the comparative in English. The subjects were administered a linguistic pretest in their native language, based on Sinclair’s assessment of conservation language (see Sinclair-de %wart, 1‘367) delineating scalar (“much”) and vector (comparative-“more”) knowledge in the three groups, as well as three types of cognitive task: conservat i o n (six tasks), seriation. and multiple classification (nine matrices). (Cognitive tcsis were also conducted in the child’s native language.) The results showed that native laIiguage affects cognitive performance in consistent ways. English and Greek, which similarly code object attributes and differences with scalars (de,, and vectors (comparatives), yield organizations of concrete operations .,iiiptivcs) in c1:issification and conservation tasks. For Greek subjects, attainment of coiiiparativcs was predictive of (correlated with) conservation. Turkish has only m e forni for both classification and comparative relations; for the Turkishspeaking subjects, then, there was an overlapping of classification and conservat i m skills, and mastery of the comparative forni was found to be nonprcdictive of conservation performance. The authors suggested that language is neither a necessary nor sufficient condition f o r conservation. They held instead that language has a contributory role in cognitive development, a role nevertheless less influential than “maturation.” An additional finding is that the Turkish children, allhougli they are poorer than the others in conservation, far exceeded them in multiple classification skills. Why this occurred, or how the structure of thc ‘Turkish lexicon can account f o r it. is not discussed. The various ways language is used in conservation experiments lead t o a number of questions concerning its effects on conservation. These have been addressed in a number of studies, sometimes as a subsidiary rather than principal focus. Beilin (1964, I965), for example, showed that the child’s knowledge of ~
Constructing Cognitive Operations Linguistically
93
comparatives (“more,” “less”) and terms of equivalence (“same,” “different”) were not understood equally well by children in conservation experiments. Children who understood ‘‘same number” on pretest were more likely t o conserve on number, length, and area tests both before and after training than those who did not understand the expressions. Children were also found t o understand “more” better than “less,” a finding which has now been extensively investigated in nonconservation contexts (e.g., Donaldson & Balfour, 1968; Palermo, 1973; Weiner, 1974). Beilin (1 968) later proposed that a noncomparative meaning of “more” is understood (as “addition to”) by children before they understand the comparative sense (as “more than”), which in turn is understood before the conserved sense (“more in spite of appearing like less, or equal to”). Bloom (1970) also showed that a noncomparative sense of “more” precedes the comparative, although “more” was said t o be understood first as “repetition.” A study by Pratoomraj and Johnson (1966) on the role of different types of questions employing different lexical ternis-“more,” ‘‘less,’’ “same,” “different”-showed that with three types of conservation task, i.e., prediction, judgment, and explanation, variation in the type of question (lexical term) had little effect on performance, whereas the type of task did. Griffiths et al. (1967), in contrast, hold that lack of knowledge of comparatives can interfere with conservation performance. Going in another direction, Shantz and Sigel(l967) tested the effects on conservation of labeling of object attributes in addition t o learning simple classification skills. Such labeling plus classification proved successful, but it was not clear whether this was due t o attribute labeling or to classification skill learning. Lumsden and Kling (1969) found that training for “bigger” led t o improvement in size conservation, but with older rather than younger subjects (7 years, 6 months versus 6 years, 6 months). They propose that teaching concepts should not be pursued prior to the child’s being able t o handle their appropriate verbal representations. A study by Russell (1975) was concerned with whether verbal competence is a reflection of the level of “operational” or “cognitive equilibrium,” although it was conducted not with conservation as such but with a task involving comprehension of the phrase “sanie amount of room.” The subject was presented first with a standard figure and had to choose one with the same amount of room from a set of comparison figures. In an alternative task he was required t o adjust an apparatus of fixed width and variable height t o a standard. Following the pretest, subjects were trained in one of two procedures (manipulative or conceptual) and posttested. Successful and unsuccessful subjects were given manipulative training plus corrective feedback in which they transferred squares or cubes t o selected figures “with a little help from the experimenter.” Conceptual training was verbal and was intended to “redraw” the chld’s concept of area and volume and t o “discourage” the interpretation of “same amount” as same shape or length. There was also variation in two-dimensional (area) and three-dimen-
94
flarry Beilin
sional figures (volumes). Five-year-old subjects, in the pretest, tended t o interpret “sanie amount” as height when they varied tlie test figure t o conform t o the standard; they interpreted it as “same shape” when they selected from comparison shapes. The latter intcrpretation is questionable, however, since the data o n which it is based were not from tlie “coi-rwt” choices of the same shape (which were chosen at just about chance level) but from those close to it (which are statistically “dominant” when tlie correct choices are omitted from the statistical analysis). Russell also reported that conceptual training was effective in one condition, whereas manipulation training was not effective at all. However, when pretest differences arc cquated in an analysis of covariance, conceptual training was also ineffective. The most stable finding from the study was that 5-year-old subjects misinterpret the conservation expression ‘‘same amount of room” when tested in a nonconservatiori context. Russell, in light of other results, attributes the child’s difficulties to other than “sein:uitic” failure. The author holds that tlie notion that conservation failure I-esults from lack of knowlcdge of the meaning of words is too simplistic. Me asserts instead that the failure is inore basic and is related either t o a lack of cognitive operations (Piaget’s explanation) or 3 source related to tlie child’s “consciousness.” Another study combines relatiorial term training (“more,” “less,” “same”) with “language activation” [in fact, rule instruction; the rule given was: “The water is the same height and tlie glasses are the same width, therefore the glasses have the same amount of water” (Hamel, Van Der Veer, & Westerhof, 1972)l. Conservers were bettei- than nonconservei-s on a pretest of relational terms. After training, there were fewer relational term errors on the posttests and concomitantly substantial improvement in performance: 15 of 20 pretest conservers conserved on two posttests, as did 4 nonconservers and 1 control subject. Justifications also changed with conservation change. Hamel supports what he believes to bc Bruner’s position on identity, but offers as justification an argument that sounds more like Piaget’s. Haniel makes no distinction between levels or types of identity (i.e., object versus operational), which taken together with the confounding of rule instruction and lexical training leaves the results ambiguous. One feature of Bruner’s views merits attention here since it concerns language. Bruner cuntends that in the course of development the child increasingly fits his experience t o his language so that the verbal representation of objects and events enables the child t o acquire concepts and operations. The key to his “instrumental functionalism” is the notion of progressive change in the forms of representation that increasingly facilitate and extend the power of thought. The forms of representation~eiiactive,ikonic, and symbolic-involve different instruments of both representation and expression. Symbolic representation, the most advanced form of representation, is primarily linguistic, and speech is its princi-
Constructing Cognitive Operations Linguistically
95
pal expression. Bruner’s experiments on conservation (Bruner et al., 1966) are interpreted as demonstrating that the linguistic representation of particular perceptual relationships results in resistance t o the effects of misleading perceptual cues (as in the screening experiments in conservation). The Genevans (Inhelder et al., 1974) argue, however, that Bruner’s verbal training combined with other procedures directs attention t o relevant features of the experiment, the result of which is correct local response (pseudoconservation) but not true operational change. The resistance t o misleading perceptual cues that linguistic representation provides is seen by Beilin (1969) as a form of algorithm learning in which the acquired linguistic algorithm provides the child with a processing procedure for dealing with perceptual and conceptual inputs. The notion of representation-bound cognitive structure is also held by Aebli (1970) and Steiner (1974). Their view is that the form (media) in which objects are represented affects their comprehension. Aebli suggests for example that the (linguistic) form in which the Piagetian “clinical” interview is conducted prestructures the problem posed by the experimenter (Steiner, 1974, p. W S ) , and affects the ease with which cognitive structures are achieved in Genevan experiments. C. SUMMARY
The research on lexical training and lexical effects on conservation indicates first that lexical training as such does not improve conservation performance,