Explaining Value: and Other Essays in Moral Philosophy

Explaining Value

This page intentionally left blank

Explaining Value and Other Essays in Moral Philosophy

Gilbert Harman

CLARENDON PRESS · OXFORD

Great Clarendon Street, Oxford OX2 6DP Oxford University Press is a department of the University of Oxford It furthers the University's objective of excellence in research, scholarship, and education by publishing worldwide in Oxford New York Auckland Bangkok Buenos Aires Cape Town Chennai Dar es Salaam Delhi Hong Kong Istanbul Karachi Kolkata Kuala Lumpur Madrid Melbourne Mexico City Mumbai Nairobi São Paulo Shanghai Taipei Tokyo Toronto Oxford is a registered trade mark of Oxford University Press in the UK and in certain other countries Published in the United States by Oxford University Press Inc., New York © Gilbert Harman 2000 The moral rights of the authors have been asserted Database right Oxford University Press (maker) First published 2000 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, without the prior permission in writing of Oxford University Press, or as expressly permitted by law, or under terms agreed with the appropriate reprographcs rights organization. Enquiries concerning reproduction outside the scope of the above should be sent to the Rights Department, Oxford University Press, at the address above You must not circulate this book in any other binding or cover and you must impose this same condition on any acquirer British Library Cataloguing in Publication Data Data available Library of Congress Cataloging in Publication Data Explaining value and other essays in moral philosophy / Gilbert Harman. Includes bibliographical references 1. Ethics. I. Title. BJ1012 .H31354 2000 170—dc21 00–035603 ISBN 0–19–823805–3 ISBN 0–19–823804–5 (pbk.)

For My Brothers Bill and Roger

This page intentionally left blank

Acknowledgements Preparation of some of the material in this book was supported in part by a grant to Princeton University from the James S. McDonnell Foundation. I am also grateful to the publishers of the original versions of the essays in this book for permission to reprint them. ‘Moral Relativism Defended’, Philosophical Review 84 (1975), 3–22. ‘What is Moral Relativism?’, in A. I. Goldman and J. Kim (eds.), Values and Morals (Dordrecht: Reidel, 1978), 143–61. ‘Relativistic Ethics: Morality as Politics’, Midwest Studies in Philosophy 3 (1978), 109–21. ‘Justice and Moral Bargaining’, Social Philosophy and Policy 1 (1983), 114–31. ‘Is There a Single True Morality?’, in David Copp and David Zimmerman (eds.), Morality, Reason and Truth (Totowa, N. J.: Rowan & Littlefield, 1985), 27–48. ‘Toward a Theory of Intrinsic Value’, Journal of Philosophy 64 (1967), 792–804. ‘Desired Desires’, in Ray Frey and Chris Morris (eds.), Value, Welfare, and Morality (Cambridge: Cambridge University Press, 1993), 138–57. An earlier version of ‘Intrinsic Value’ appeared as ‘La valeur intrinsèque’, translated by Laurie Calhoun, Revue de Métaphysique et de Morale 99 (1994), 245–55. ‘Human Flourishing, Ethics, and Liberty’, Philosophy and Public Affairs 12 (1983), 307–22. ‘Moral Philosophy Meets Social Psychology: Virtue Ethics and the Fundamental Attribution Error’, Proceedings of the Aristotelian Society 99 (1998–9), 315–31. ‘Moral Agent and Impartial Spectator’, Lindley Lecture (Lawrence Kan.: University of Kansas, 1986). ‘Explaining Value’, Social Philosophy and Policy 11 (1994), 229–48. ‘Moral Philosophy and Linguistics’, Proceedings of the 20th World Congress of Philosophy, vol. I: Ethics, ed. Klaus Brinkmann (Bowling Green, Ohio: Philosophy Documentation Center, 1999), 107–15.

This page intentionally left blank

Contents Introduction Part I. Moral Relativism 1. Moral Relativism Defended 2. What is Moral Relativism? 3. Relativistic Ethics: Morality as Politics 4. Justice and Moral Bargaining 5. Is There a Single True Morality? Part II. Intrinsic Value 6. Toward a Theory of Intrinsic Value 7. Desired Desires 8. Intrinsic Value Part III. Virtue 9. Human Flourishing, Ethics, and Liberty 10. Moral Philosophy Meets Social Psychology: Virtue Ethics and the Fundamental Attribution Error Part IV. Explaining Value 11. Moral Agent and Impartial Spectator 12. Explaining Value 13. Moral Philosophy and Linguistics Bibliography Index of Names Index of Subjects

xi 3 20 39 58 77 103 117 137 151 165 181 196 217 227 235 237

This page intentionally left blank

Introduction These essays have all been previously published. I have modified them to have the same format, making (what seem to me to be) corrections where needed, adding additional references, and tinkering with wording. Throughout these essays I am concerned with explaining aspects of morality. What accounts for the existence of basic moral disagreements? Why do most people think it is worse to injure someone than to fail to save them from injury? Where does the right of self-defence come from? Why do many people think it is morally permissible to treat animals in ways we would not treat people? Why do people disagree about moral relativism to the extent that they do? What is it to value something and what is it to value something intrinsically? How are a person's values (noun) related to what the person values (verb)? How much of morality can or should be explained in terms of human flourishing or the possession of virtuous character traits? For that matter, are there character traits of the sort we normally suppose there are? How do people come to be moral? Is morality something one learns or does it arise in everyone naturally without instruction? I have somewhat arbitrarily divided the essays into four groups, which I have labelled (1) moral relativism, (2) intrinsic value, (3) virtue, and (4) explaining values. Here are summaries of the essays. The first five essays are concerned with moral relativism. In Essay 1, ‘Moral Relativism Defended’, I argue that a version of moral relativism can be formulated as an intelligible thesis, the thesis that morality derives from an implicit agreement and moral judgements are true or false only in relation to such an agreement. I appeal to this thesis to explain otherwise puzzling aspects of some of our own moral views, in particular why we think that it is more important to avoid harm to others than to help others avoid harm. I go on to argue that traditional objections to implicit agreement theories can be met. In Essay 2, ‘What is Moral Relativism?’, I discuss three views that have been considered to be versions of moral relativism—normative moral relativism, moral judgement relativism, and meta-ethical relativism. After discussing what may seem to be conclusive objections to each view, I argue

xii

INTRODUCTION

that the objections can all be met and that all three versions of moral relativism are correct. Essay 3, ‘Relativistic Ethics: Morality as Politics’, formulates the ‘naive view’ that morality is based on certain moral demands that everyone accepts, or at least has reasons to accept, as demands on everyone and on which all moral reasons depend. I argue that the naive view must be rejected because there are no substantive moral demands satisfying those conditions. I suggest that moral conventionalism or ‘morality as politics’ is the best replacement for the naive view. Other conceptions of morality retain less of the content of the naive view and tend toward unrealistic accounts of what morality requires. Extremely agent-centred theories, as in certain forms of existentialism, practically abandon morality as a social enterprise, as do extremely critic-centred theories, like certain forms of emotivism and Hare's theory, which are best seen as rejecting the moral ought in favour of the ought of evaluation. In Essay 4, ‘Justice and Moral Bargaining’, I claim that substantive principles of justice derive from actual agreements that result from actual but usually tacit moral bargaining. I criticize arguments by Gewirth, Nagel, and Kohlberg for thinking that all rational or normal people must recognize certain requirements of justice. I note that mere hypothetical agreements could not give people reasons. I show how moral bargaining might explain moral concern and respect, and show that this explanation does not undermine the moral force of the principles we accept. I go on to advocate that we be more explicit about moral bargaining, suggesting that this will improve moral argument and lead to greater tolerance. In Essay 5, ‘Is There a Single True Morality?’, I ask why people, especially philosophers, disagree in the way they do about moral relativism. Although people appear to be quite divided, with about half being moral relativists and half moral absolutists, there is a tendency to suppose that the view one takes about this is obviously correct and that anyone who takes the opposite view is just denying the plain truth. I trace some of the disagreement to methodology and to the attitude one takes about how ethics fits with a scientific conception of the world. Those who take it to be important how one's moral views cohere with science tend to be moral relativists; those who do not take that issue to be all important tend to be moral absolutists. The next three essays are concerned with value and intrinsic value. In Essay 6, ‘Toward a Theory of Intrinsic Value’, I am concerned with what would or would not be a good or worthwhile or desirable states of

INTRODUCTION

xiii

affairs, as when one says, ‘It would be a good thing if P.’ Some theorists have attempted to provide such a theory by providing a theory of intrinsic value. I examine what was once a standard textbook account of intrinsic value and argue that (a) it is not clear whether a theory of intrinsic value can be developed along the lines of that standard account; (b) if one is to develop such a theory, one will need to introduce a notion of ‘basic intrinsic value’ in addition to the notion of ‘intrinsic value’; and (c) several different theories of intrinsic value may account for the same judgements concerning desirability, and it may be arbitrary to choose one such theory over another. In Essay 7, ‘Desired Desires’, I argue against appeals to second-order desires in order to explain freedom of the will, valuing, or values. Along the way, I defend a number of subsidiary points: (1) Positive intentions are reflexive or selfreferential. (2) Intrinsic desires are not just noninstrumental desires. (3) To value something is in part to want it. (4) There is sometimes a difference between believing that something is good and valuing it. (5) There is a difference between valuing something and having something as one of your values. In Essay 8, ‘Intrinsic Value’, I observe that good news might be desirable to you even though it is not of value to you. I also note that something not valued intrinsically might be valued as contributing to a larger whole that is valued intrinsically, in the sense that the whole with that part is intrinsically preferred to the whole without that part. I apply this point to the value of some good news, to retributive punishment, to Dworkin's account of sacred value, etc. Essays 9 and 10 discuss character traits and virtue ethics. In Essay 9, ‘Human Flourishing, Ethics, and Liberty’, I discuss attempts to derive at least some of morality from a conception of human flourishing and those character traits or virtues that would contribute to human flourishing. I raise serious problems for such an approach to ethics. In Essay 10, ‘Moral Philosophy Meets Social Psychology: Virtue Ethics and the Fundamental Attribution Error’, I discuss results in social psychology that cast doubt on the existence of robust character traits of the sort that people normally take for granted and that virtue ethics must appeal to. I conclude that we should try to stop thinking in terms of character tratis. The final three essays discuss various ways in which we might be able to explain aspects of morality.

xiv

INTRODUCTION

In Essay 11, ‘Moral Agent and Impartial Spectator’, I start with a standard objection to impartial spectator theories of morality, namely, that they offer an aesthetic, critic-centred conception of morality, and so cannot adequately account for moral motivation and the way morality presents itself to an agent. I describe the issue as it arose in the eighteenth Century for Hutcheson, Hume, and Smith, and note that Smith provides an elegant account of why agents might be motivated to act in ways that impartial spectators would approve, an account that is premised on an insightful critique of Hume's theory of sympathy. In Essay 12, ‘Explaining Values’, I consider various ways in which people's moral values might be explained, starting with Hume's account of our different attitudes toward chastity in men and in women. I go on to consider Hume's somewhat utilitarian explanation of moral sentiments in general, various theories of moral development, explanations of aspects of morality in terms of natural selection, Posner's explanation of aspects of morality in terms of wealth maximization, explanations that appeal to principles like double effect, and sociological and historical explanations of aspects of morality. Finally, in Essay 13, ‘Moral Philosophy and Linguistics’, I observe that languages are at least as diverse on the surface as moralities, yet linguistics thrives by studying linguistic universals. I suggest that linguistics provides a model in searching for moral universals.

Part I Moral Relativism

This page intentionally left blank

1 Moral Relativism Defended In the first four essays, I argue that morality arises when people reach an implicit agreement or come to a tacit understanding about their relations with one another. Part of what I mean by this is that moral judgements—or, rather, an important class of them—are true or false only in relation to and with reference to one or another such agreement or understanding. This is vague, but I will try to make it more precise in what follows. In any event, it should be clear that I intend to argue for a version of what has been called ‘moral relativism’. In doing so, I am taking sides in an ancient controversy, discussed further in Essay 5, below. Many people believe that the sort of view which I am going to defend is obviously correct—indeed, that it is the only sort of account that could make sense of the phenomenon of morality. At the same time there are also many who think that moral relativism is confused, incoherent, and even immoral, at the very least obviously wrong. Some arguments against relativism make use of a strategy of dissuasive definition; they simply define moral relativism as an inconsistent thesis. For example, moral relativism might be defined as committed to the following three claims (Williams, 1972: 20–21; M. Singer, 1961: 332): (a) There are no universal principles. (b) One ought to act in accordance with the principles of one's own group. (c) Principle (b) is a universal moral principle. It is easy enough to show that this version of moral relativism will not do, but surely a defender of moral relativism can find a better definition. My moral relativism is a soberly logical thesis—a thesis about logical form, if you like. Just as the judgement that something is large is true or false only in relation to one or another comparison class, so too, I will argue, the judgement that it is wrong of someone to do something is true or false only in relation to an agreement or understanding. A dog may be large in relation to chihuahuas but not large in relation to dogs in general.

4

MORAL RELATIVISM

Similarly, I will argue, an action may be wrong in relation to one agreement but not in relation to another. Just as it is indeterminate whether a dog is large, period, apart from any relation to a comparison class, so too, I will argue, it is indeterminate whether an action is wrong, period, apart from any relation to an agreement. There is an agreement, in the relevant sense, if each of a number of people intends to adhere to some schedule, plan, or set of principles, intending to do this on the understanding that the others similarly intend. The agreement or understanding need not be conscious or explicit. I will not try to say what distinguishes moral agreements from, for example, conventions of the road or conventions of etiquette, since these distinctions will not be important as regards the purely logical thesis that I will be defending. Although I want to say that certain moral judgements are made in relation to an agreement, my present argument does not apply to all moral judgements. It is true that I argue elsewhere that all moral judgements are made in relation to an agreement (Essays 2–4, below; Harman and Thomson, 1996), but that is not what I defend in the present essay. In this essay I argue that there is a way in which certain moral judgements are relative to an agreement but other moral judgements are not. This relativism is a thesis only about what I call ‘inner judgements’, such as the judgement that someone morally ought or ought not to have acted in a certain way or the judgement that it was morally right or wrong of him to have done so. The relativism defended in this first essay is not meant to apply, for example, to the judgement that someone is evil or the judgement that a given institution is unjust. (I do argue that justice is a relative matter in Essay 4.)

1.1 Inner Judgements Many of us believe on reflection that different people have reasons to observe different moral requirements, depending on the moralities those people accept. Those of us who have this ‘relativistic’ view make inner judgements about a person only if we suppose that he or she has reasons to be motivated by the relevant moral considerations. We make other sorts of judgement about those whom we suppose not to have such reasons. Inner judgements include judgements in which we say that someone morally should or ought to have done something or that someone was morally right or wrong to have done something. Inner judgements do not

MORAL RELATIVISM DEFENDED

5

include judgements that someone is a savage, inhuman, evil, a betrayer, a traitor, or an enemy. Consider this imaginary example. Intelligent beings from outer space land on Earth, beings without the slightest concern for human life and happiness. Some of us seem to be able to imagine that the fact that a certain course of action on their part will injure inhabitants of Earth might mean nothing to them and might in particular give them no reason to avoid the action. Those of us who think we can imagine this find it odd to go on to say that the beings ought nevertheless to avoid injuring human beings or that it would be wrong of them to attack human beings. Of course we would want to resist them if they did such things and we would certainly make negative judgements about them; but we would judge that they are dreadful enemies to be repelled and even destroyed, not that they morally ought not to act as they do. Closer to home, suppose that a contented employee of Murder, Incorporated was raised as a child to honour and respect members of the ‘family’ but to have nothing but contempt for the rest of society. His current assignment, let us suppose, is to kill a certain bank manager, Bernard J. Ortcutt. Since Ortcutt is not a member of the ‘family’, the employee in question has no compunction about carrying out his assignment. If we were to try to convince him that he should not kill Ortcutt, our argument would merely amuse him. Thinking of this case, some of us find it plausible that we might not be able to provide the assassin with the slightest reason to desist unless we were to point to practical difficulties, such as the likelihood of his getting caught. Given that assumption about reasons, it strikes us as a misuse of language to say of the assassin that he ought not to kill Ortcutt or that it would be wrong of him to do so. To say such things seems to imply that our own moral considerations carry some weight with him, which they do not. Instead we can only judge that he is a criminal, someone to be hunted down by the police, an enemy of peace-loving citizens, and so forth. It is true that we can make certain judgements about him using the word ‘ought’, even given that he has no reasons not to kill. Investigators who are waiting for the assassin to appear at the bank might use the ‘ought’ of expectation to say, ‘He ought to arrive soon.’ They mean that on the basis of their information one would expect him to arrive soon. The same investigators might use the ‘ought’ of rationality to say that the assassin ought to go in by the rear door. They would mean that it would be more rational for him to do that than to go in by the front door. But in neither of these cases is the moral ‘ought’ in question.

6

MORAL RELATIVISM

There is another use of ‘ought’ that is normative and in a sense moral, but which is distinct from what I am calling the moral ‘ought’, when we say that something ought or ought not to be the case. It ought not to be the case that members of Murder, Incorporated kill their victims; in other words, it is a terrible thing that they do so. The same thought can perhaps be expressed as ‘They ought not to go around killing people’, meaning that it ought not to be the case that they do, not that they are wrong to do what they do. The normative ‘ought to be’ is used to assess a situation; the moral ‘ought to do’ is used to describe a relation between an agent and a type of act that he might perform or has performed (Humberstone, 1971–2). (Thomas Nagel has pointed out to me that, when we use the evaluative ‘ought to be’ to say that something ought to be the case, we often imply that someone ought to do something or ought to have done something about it. To take his example, we would not say that a certain hurricane ought not to have killed fifty people just because it was terrible that the hurricane did so. But we might say this if we had in mind that the deaths from the hurricane would not have occurred except for the absence of safety or evacuation procedures that the authorities ought to have provided.) The sentence, ‘They ought not to go around killing people,’ is therefore multiply ambiguous. It can mean, roughly: (a) (b) (c) (d)

One would expect them not to do so (the ‘ought’ of expectation). It is not in their interest to do so (the ‘ought’ of rationality). It is a bad thing that they do so (the normative ‘ought to be’). It is wrong of them to do so (the moral ‘ought to do’).

For the most part I am here concerned only with the last interpretation, (d). The word ‘should’ behaves very much like ‘ought to’. There is a ‘should’ of expectation (‘They should be here soon’); a ‘should’ of rationality (‘He should go in by the back door’); a normative ‘should be’ (‘They shouldn't go around killing people like that’); and the moral ‘should do’ (‘You should keep that promise’). I am of course concerned mainly with the last sense of ‘should’. ‘Right’ and ‘wrong’ also have multiple uses. I will not try to say what all of them are, but I do want to distinguish using the word ‘wrong’ to say that a particular situation or action is wrong from using the word to say that it is wrong of someone to do something. In the former case, the word ‘wrong’ is used to assess an act or situation. In the latter case, it is used to

MORAL RELATIVISM DEFENDED

7

describe a relation between an agent and an act. Only the latter sort of judgement is an inner judgement. Although those of us I have referred to earlier would not say that it was morally wrong of the assassin to kill Ortcutt, we could say that his action was wrong, and we could say that it is wrong that there is so much killing. To take another example, it sounds odd to many of us to say that Hitler morally ought not to have ordered the extermination of the Jews, or that it was morally wrong of him to have done so. That sounds somehow ‘too weak’ a thing to say. Instead we want to say that Hitler was an evil man. Yet it does not sound odd to us to say, ‘Hitler ought not to have ordered the extermination of the Jews’, if what we mean is that it ought never to have happened; and we can say without oddity that what Hitler did was wrong. Oddity seems to us to affect only the inner judgement that Hitler was wrong to have acted in that way. That is what sounds ‘too weak’ to us. It is worth noting that the inner judgements do not sound too weak simply because of the enormity of what Hitler did. What is relevant is our supposition that in acting as he did he shows that he could not have been susceptible to the moral considerations on the basis of which we make our judgement. He is in the relevant sense beyond the pale and we therefore cannot make inner judgements about him. Consider another mass-murderer, Stalin. Those of us who have trouble making inner moral judgements about Hitler can imagine someone taking a sympathetic view of Stalin. In this view, Stalin realised that the course he was going to pursue would mean the murder of millions of people and he dreaded such a prospect; however, the alternative seemed to offer an even greater disaster—so, reluctantly and with great anguish, he went ahead. In relation to this view of Stalin, inner judgements about Stalin are not as odd to some of us as are similar judgements about Hitler. For we might easily continue the story by saying that, despite what he hoped to gain, Stalin should not have undertaken that course of action. It was wrong of him to have done so. What makes inner judgements about Hitler odd or ‘too weak’ for some of us is not that the acts judged seem too terrible for the words used, but rather that the agent judged seems beyond the pale—in other words, beyond the motivational reach of the relevant moral considerations. It is true that many people find that there is nothing deviant about inner judgements about Hitler. Furthermore, even in my own case I find that it is possible to describe Stalin and Hitler in ways that reverse these points, so that inner judgements about Hitler do not sound odd to me and inner judgements about Stalin do sound odd.

8

MORAL RELATIVISM

It is also true that for various reasons we might pretend that an agent is or is not susceptible to certain moral considerations. For example, a speaker may for rhetorical or political reasons wish to suggest that someone is beyond the pale, that he should not be listened to, that he can be treated as an enemy. On the other hand, a speaker may pretend that someone is susceptible to certain moral considerations in an effort to make that person or others susceptible to those considerations. Inner judgements about one's children sometimes have this function. So do inner judgements made in political speeches that aim at restoring a lapsed sense of morality in government.

1.2 The Logical Form of Inner Judgements Inner judgements have two important characteristics. First, they imply that the agent has reasons to do something. Second, the speaker in some sense endorses these reasons and supposes that the audience also endorses them. Other moral judgements about an agent, on the other hand, do not imply that the agent has reasons for acting that are endorsed by the speaker. If someone S says that A (morally) ought to do D, S implies that A has reasons to do D and S endorses those reasons. If S says that B was evil in what B did, S does not imply that the reasons S would endorse for not doing what B did were reasons for B not to do that thing. In fact, S implies that such reasons were not reasons for B. Let us examine this more closely. If S says that (morally) A ought to do D, S implies that A has reasons to do D that S endorses. I shall be concerned only with any speaker, S, who assumes, controversially, that such reasons would have to have their source in values, goals, desires, or intentions that S takes A to have, and that S approves of A's having those values, goals, desires, or intentions because S shares them. So, if S says that (morally) A ought to do D, there are certain motivational attitudes M which S assumes are shared by S, A, and S's audience. In other words, I am discussing the use of moral language by speakers who assume (contrary to Nagel, 1970) that the possession of rationality is not sufficient to provide a source for relevant reasons, that certain values, desires, goals, or intentions are also necessary. In particular, the speakers in question assume that there might be no reasons at all for a being from outer space to avoid harm to us. They assume that, for Hitler, there might

MORAL RELATIVISM DEFENDED

9

have been no reason at all not to order the extermination of the Jews. They assume that the contented employee of Murder, Incorporated might have no reason at all not to kill Ortcutt. Many speakers who accept such assumptions find that they distinguish inner moral judgements from other moral judgements in the way that I have indicated. Ultimately, I want to argue that the shared motivational attitudes M should be identified with intentions to keep an agreement (supposing that others similarly intend). For I want to argue that inner moral judgements are made relative to such an agreement. That is, I want to argue that, when a relevant speaker S makes the inner judgement that A ought to do D, S assumes that A intends to act in accordance with an agreement that S and S's audience also intend to observe. In other words, I want to argue that the source of the reasons for doing D that S ascribes to A consists in A's sincere intention to observe a certain agreement. I have not yet argued for the stronger thesis, however. I have argued only that S makes his or her judgement relative to some motivational attitudes M that S assumes are shared by S, A, and S's audience. Formulating this as a logical thesis, I want to treat this use of the moral ‘ought’ as involving a four-place relation. ‘Ought (A, D, C, M)’ relates an agent A, a type of act D, considerations C, and motivating attitudes M. The relativity to considerations C can be brought out by considering what are sometimes called statements of prima facie obligation. ‘Considering that you promised, you ought to go to the board meeting. Considering that you are the sole surviving relative, you ought to go to the funeral. All things considered, it is not clear what you ought to do’ (Cf. Davidson, 1969). The claim that there is this relativity to considerations is not, of course, what makes my thesis a version of moral relativism, since any theory must acknowledge relativity to considerations. The relativity to considerations does, however, provide a model for a coherent interpretation of moral relativism as a similar kind of relativity. It is not as easy to exhibit the relativity to motivating attitudes as it is to exhibit the relativity to considerations. Normally, a speaker who makes this sort of moral ‘ought’ judgement intends the relevant motivating attitudes to be ones that the speaker shares with the agent and the audience, and normally it will be obvious what attitudes these are. But sometimes a speaker does invoke different attitudes by invoking a morality the speaker does not share. Someone may say, for example, ‘As a Christian, you ought to turn the other cheek; I, however, propose to strike back.’ A spy who has been found out by a friend might say, ‘As a citizen, you ought to turn me in, but I hope that you will not.’ In these and similar cases a speaker makes

10

MORAL RELATIVISM

a moral ‘ought’ judgement that is explicitly relative to motivating attitudes that the speaker does not share. In order to be somewhat more precise, then, my thesis is this. As used by the relevant sort of speaker, ‘Ought (A, D, C, M)’ means roughly this: given that A has motivating attitudes M and given C, D is the course of action for A that is supported by the best reasons. In judgements using this sense of ‘ought’, C and M are often not explicitly mentioned but are indicated by the context of utterance. Normally, when that happens, C will be ‘all things considered’ and M will be attitudes that are shared by the speaker and audience. I mentioned that inner judgements have two characteristics. First, they imply that the agent has reasons to do something that are capable of motivating the agent. Second, the speaker endorses those reasons and supposes that the audience does too. Now, any ‘Ought (A, D, C, M)’ judgement has the first of these characteristics, but as we have just seen a judgement of this sort will not necessarily have the second characteristic if made with explicit reference to motivating attitudes not shared by the speaker. If reference is made either implicitly or explicitly (for example, through the use of the adverb ‘morally’) to attitudes that are shared by the speaker and audience, the resulting judgement has both characteristics and is an inner judgement. If reference is made to attitudes that are not shared by the speaker, the resulting judgement is not an inner judgement and does not represent a full-fledged moral judgement on the part of the speaker. In such a case we have an example of what Hare (1952: 164–8) calls an inverted-commas use of ‘ought’.

1.3 Moral Bargaining I have argued that the relevant sort of moral ‘ought’ judgements are relational, ‘Ought (A, D, C, M)’, where M represents certain motivating attitudes. I now want to argue that the attitudes M derive from an agreement. That is, they are intentions to adhere to a particular agreement on the understanding that others also intend to do so. I put this forward as a hypothesis to account for an otherwise puzzling aspect of ordinary moral views that, as far as I know, there is no other way to account for. I will use the word ‘intention’ in a somewhat extended sense to cover certain dispositions or habits. Someone may habitually act in accordance with the relevant understanding and therefore may be disposed to act in that way without having any conscious intention. In such a case it may

MORAL RELATIVISM DEFENDED

11

sound odd to say that he or she intends to act in accordance with the moral understanding. Nevertheless, for present purposes I will count that as his or her having the relevant intention in a dispositional sense. I now want to consider the following puzzle about our moral views, a puzzle that has figured in philosophical discussion of issues such as abortion. Many people assign greater weight to the duty not to harm others than to the duty to help others. For example, many believe that a doctor ought not to save five patients who would otherwise die by cutting up a sixth patient in order to distribute healthy organs where needed to the others. People believe this even though they think that the doctor has a duty to try to help as many of his patients as possible. In other words, people think that a doctor has a stronger duty to try not to harm any of his patients (or anyone else) even if by so doing the doctor could help five others (Foot, 1978). This aspect of ordinary moral views can seem very puzzling, especially on the supposition that moral feelings derive from sympathy and concern for others. But the hypothesis that morality derives from an agreement among people of varying powers and resources provides a possible explanation. The rich, the poor, the strong, and the weak would all benefit if all were to try to avoid harming one another. So everyone could agree to that arrangement. But the rich and the strong would not benefit from an arrangement whereby everyone would try to do as much as possible to help those in need. The poor and weak would get all of the benefit of this latter arrangement. Since the rich and the strong could foresee that they would be required to do most of the helping and that they would receive little in return, they would be reluctant to agree to a strong principle of mutual aid. A compromise would be likely and a weaker principle would probably be accepted. In other words, although everyone could agree to a strong principle concerning the avoidance of harm, it would not be true that everyone would favour an equally strong principle of mutual aid. It is likely that only a weaker principle of the latter sort would gain general acceptance. So the hypothesis that morality derives from an understanding among people of different powers and resources can explain why in our morality avoiding harm to others is taken to be more important than helping those who need help. By the way, I am here only trying to explain an aspect of our moral views. I am not therefore endorsing that aspect. And I defer until later a relativistic account of the way in which aspects of our moral view can be criticised ‘from within’. There is no need to suppose that the agreement or understanding in

12

MORAL RELATIVISM

question is explicit. It is enough if various members of society knowingly reach an agreement in intentions—each intending to act in certain ways on the understanding that the others have similar intentions. Such an implicit agreement is reached through a process of mutual adjustment and implicit bargaining. Indeed, it is essential to the proposed explanation of this aspect of our moral views to suppose that the relevant moral understanding is the result of some sort of bargaining, either explicit or implicit. It is necessary to suppose that, in order to further our interests, we form certain conditional intentions, hoping that others will do the same. The others, who have different interests, will form somewhat different conditional intentions. After some sort of negotiation, a compromise is reached. Seeing morality in this way as a compromise based on implicit bargaining could explain why our morality takes it to be worse to harm someone than to refuse to help someone. The explanation requires that we view our morality as an implicit agreement about what to do. This sort of explanation could not be given if we were to suppose, say, that our morality represented an agreement only about the facts (naturalism). Nor is it enough simply to suppose that our morality represents an agreement in attitude, if we forget that such agreement can be reached not only by way of such principles as are mentioned, for example, in Hare's (1952, 1963) ‘logic of imperatives’, but also through bargaining. According to Hare (1952: 18–20, 168–9), to accept a general moral principle is to intend to do something. If we add to his theory that the relevant intentions can be reached through implicit bargaining, the resulting theory begins to look like the one that I am defending. Many aspects of our moral views can be given a utilitarian explanation. We could account for these aspects, using the logical analysis I presented in the previous section of this essay, by supposing that the relevant ‘ought’ judgements presuppose shared attitudes of sympathy and benevolence. We can equally well explain them by supposing that considerations of utility influence implicit agreements, so that the appeal is to a shared intention to adhere to those agreements. Any aspect of morality that is susceptible of a utilitarian explanation can also be explained by an implicit agreement, but not conversely. There are aspects of our moral views that seem to be explicable only in the second way, on the assumption that morality derives from an agreement. One example, already cited, is the distinction we make between harming and not helping. Another is our feeling that each person has an inalienable right of self-defence and self-preservation. Philosophers have not been able to come up with a really

MORAL RELATIVISM DEFENDED

13

satisfactory utilitarian justification of such a right, but it is easily intelligible on our present hypothesis, as Hobbes observed (1957: Pt. I, ch. 14). You cannot, except in very special circumstances, rationally form the intention not to try to preserve your life if it should ever be threatened, say, by society or the state, since you know that you cannot now control what you would do in such a situation. No matter what you now decided to do, when the time came, you would ignore your prior decision and try to save your life. Since you cannot now intend to do something later that you now know that you would not do, you cannot now intend to keep an agreement not to preserve your life if it is threatened by others in your society. This concludes the positive side of my argument. I have argued that inner moral judgements are made in relation to an implicit agreement. I now want to argue that this theory avoids difficulties traditionally associated with implicit agreement theories of morality.

1.4 Objections and Replies One traditional difficulty for implicit agreement theories concerns one's motivation for doing what one has agreed to do. It will, obviously, not be enough to say that one has implicitly agreed to keep agreements, since the issue would then be why one keeps that agreement. And this suggests an objection to implicit agreement theories. But the apparent force of the objection derives entirely from taking an agreement to be a kind of ritual. To agree in the relevant sense is not just to say something; it is to intend to do something—namely, to intend to carry out one's part of the agreement on the condition that others do their parts. If one agrees in this sense to do something, one intends to do it, and intending to do it is already to be motivated to do it. So there is no problem as to why one is motivated to keep one's agreements in this sense. We do believe that in general one ought not to pretend to agree in this sense in order to trick someone else into agreeing. But that suggests no objection to the present view. All that it indicates is that our moral understanding contains or implies an agreement to be open and honest with others. If it is supposed that this leaves a problem about someone who has not accepted our agreement—‘What reason does he have not to pretend to accept our agreement so that he can then trick others into agreeing to various things?’—the answer is that such a person may or may not have such a reason. If someone does not already accept something of our

14

MORAL RELATIVISM

morality, it may or may not be possible to find reasons why he or she should. A second traditional objection to implicit agreement theories is that there is not a perfect correlation between what is generally believed to be morally right and what actually is morally right. Not everything generally agreed on is right, and sometimes courses of action are right that would not be generally agreed to be right. But this is no objection to my thesis. My thesis is not that morality derives from an agreement in moral judgement; the thesis is rather that moral judgements make reference to and are made in relation to an agreement in intentions. Given that people have agreed in this sense, there can still be disputes as to what the agreement implies for various situations. In my view, many moral disputes are of this sort. They presuppose a basic agreement and they concern what implications that agreement has for particular cases. There can also be various things wrong with the agreement that a group of people reach, even from the point of view of that agreement, just as there can be defects in an individual's plan of action even from the point of view of that plan. Given what is known about the situation, a plan or agreement can in various ways be inconsistent, incoherent, or selfdefeating. In my view, certain moral disputes are concerned with internal defects of the basic moral understanding of a group, and what changes should be made from the perspective of that understanding itself. This is another way in which moral disputes make sense with reference to and in relation to an underlying agreement. Another objection to implicit agreement theories is that not all agreements are morally binding—for example, those made under compulsion or from a position of unfair disadvantage—which may seem to indicate that there are moral principles prior to those that derive from an implicit agreement. But, again, the force of the objection derives from an equivocation concerning what an agreement is. The principle that compelled agreements do not obligate, concerns agreement in the sense of a certain sort of ritual indicating that one agrees. My thesis concerns a kind of agreement in intentions. The principle about compelled agreements is part of, or is implied by, our agreement in intentions. According to me, it is only with reference to some such agreement in intentions that a principle of this sort makes sense. Maybe our moral agreement in intentions implies that it is wrong to compel people who are in a greatly inferior position to accept an agreement in intentions that they would not otherwise accept. Maybe some people in an inferior position have been compelled thus to settle for

MORAL RELATIVISM DEFENDED

15

accepting a basic moral understanding that they would not have accepted had they not been in such an inferior position. In that case our basic moral understanding is incoherent, and various suggestions might be made concerning the ways in which this understanding should be modified. But this moral critique of the understanding can proceed from the understanding itself, rather than from ‘prior’ moral principles. In order to fix ideas, consider a society in which there is a well-established and long-standing tradition of hereditary slavery. Let us suppose that everyone accepts this institution, including the slaves. Everyone treats it as in the nature of things that there should be such slavery. Furthermore, let us suppose that there are also aspects of the basic moral agreement which speak against slavery. That is, these aspects together with certain facts about the situation imply that people should not own slaves and that slaves have no obligation to acquiesce in their condition. In such a case, the moral understanding would be defective, although its defectiveness would somehow be hidden from participants, perhaps by means of a myth that slaves are physically and mentally subhuman in a way that makes appropriate the sort of treatment elsewhere reserved for beasts of burden. If this myth were to be exposed, the members of the society would then be faced with an obvious incoherence in their basic moral agreement and might come eventually to modify their agreement so as to eliminate its acceptance of slavery. In such a case, even relative to the old agreement it might be true that slave owners ought to free their slaves, that slaves need not obey their masters, and that people ought to work to eliminate slavery. For the course supported by the best reasons, given that one starts out with the intention of adhering to a particular agreement, may be that one should stop intending to adhere to certain aspects of that agreement and should try to get others to do the same. On the other hand, we might envision a second society with hereditary slavery whose agreement has no aspects that speak against slavery. In this case, even if the facts of the situation were fully appreciated, no incoherence would appear in the basic moral understanding of the society. It would not be true in relation to that understanding that slave owners ought to free their slaves, that slaves need not obey their masters, and so forth. There might nevertheless come a time when there were reasons of a different sort to modify the basic understanding, either because of an external threat from other societies opposed to slavery or because of an internal threat of rebellion by the slaves. Now, it is easier for us to make what I have called inner moral judgements

16

MORAL RELATIVISM

about slave owners in the first society than in the second. For we can, with reference to members of the first society, invoke principles that they share with us and, with reference to those principles, we can say of them that they ought not to have kept slaves and that they were immoral to have done so. This sort of inner judgement becomes increasingly inappropriate, however, the more distant they are from us and the less easy it is for us to think of our moral understanding as continuous with and perhaps a later development of theirs. Furthermore, it may be appropriate to make only non-inner judgements of the slave owners in the second society. We can say that the second society is unfair and unjust, that the slavery that exists is wrong, that it ought not to exist. But it may be inappropriate in this case to say that it was morally wrong of the slave owners to own slaves. The relevant aspects of our moral understanding, which we would invoke in moral judgements about them, are not aspects of the moral understanding that exists in the second society. (However, I come back to the question of slavery below.) Let me turn now to another objection to implicit agreement theories, an objection that challenges the idea that there is an agreement of the relevant sort. For, if we have agreed, when did we do so? Does anyone really remember having agreed? How did we indicate our agreement? What about those who do not want to agree? How do they indicate that they do not agree and what are the consequences of their not agreeing? Reflection on these and similar questions can make the hypothesis of implicit agreement seem too weak a basis on which to found morality. But once again there is equivocation about agreements. The objection treats the thesis as the claim that morality is based on some sort of ritual rather than an agreement in intentions. But, as I have said, there is an agreement in the relevant sense when each of a number of people has an intention on the assumption that others have the same intention. In this sense of ‘agreement’, there is no single moment at which one agrees, since one continues to agree in this sense as long as one continues to have the relevant intentions. Someone refuses to agree to the extent that he or she does not share these intentions. Those who do not agree are outside the agreement; in extreme cases they are outlaws or enemies. It does not follow, however, that there are no constraints on how those who agree may act toward those who do not, since for various reasons the agreement itself may contain provisions for dealing with outlaws and enemies. This brings me to one last objection, which derives from the difficulty people have in trying to give an explicit and systematic account of their moral views. If one actually agrees to something, why is it so hard to say

MORAL RELATIVISM DEFENDED

17

what one has agreed? In response, I can say only that many understandings are tacit in this way. It is often possible to recognize what is in accordance with the understanding and what would violate it, without being able to specify the understanding in any general way. Consider, for example, the understanding that exists among the members of a team of acrobats or a symphony orchestra. Another reason why it is so difficult to give a precise and systematic specification of any actual moral understanding is that such an understanding will not in general be constituted by absolute rules but will take a vaguer form, specifying goals and areas of responsibility. The agreement may indicate that one is to show respect for others by trying where possible to avoid actions that will harm them or interfere with what they are doing. It may indicate the duties and responsibilities of various members of the family, who is to be responsible for bringing up the children, and so forth. Often what will be important will be not so much exactly what actions are done as how willing participants are to do their parts and what attitudes they have—for example, whether they give sufficient weight to the interests of others. The vague nature of moral understandings is to some extent alleviated in practice. One learns what can and cannot be done in various situations. Expectations are adjusted to other expectations. Moral disputes arise, nonetheless. Such disputes may concern what the basic moral agreement implies for particular situations; and, if so, that may be either because of disputes over the facts or because of a difference in basic understanding. Moral disputes may also arise concerning whether or not changes should be made in the basic agreement. Racial and sexual issues seem often to be of this second sort. But there is no clear line between the two kinds of dispute. When the implications of an agreement for a particular situation are considered, one possible outcome is that it becomes clear that the agreement should be modified. Moral reasoning is a form of practical reasoning. One begins with certain beliefs and intentions, including intentions that are part of one's acceptance of the moral understanding in a given group. In reasoning, one modifies one's intentions, often by forming new intentions, sometimes by giving up old ones, so that one's plans become more rational and coherent—or, rather, one seeks to make all of one's attitudes coherent with each other (Harman, 1976). The relevant sort of coherence is not simply consistency. It is something very like the explanatory coherence that is so important in theoretical reasoning (Harman, 1999: 32–5). Coherence involves generality and lack of

18

MORAL RELATIVISM

arbitrariness. Consider our feelings about cruelty to animals. Obviously these do not derive from an agreement that has been reached with animals. Instead it is a matter of coherence. There is a prima facie arbitrariness and lack of generality in a plan that involves avoiding cruelty to people but not to animals. On the other hand, coherence in this sense is not the only relevant factor in practical reasoning. Another is conservatism or inertia. A third is an interest in satisfying basic desires or needs. One tries to make the least change that will best satisfy one's desires while maximizing the overall coherence of one's attitudes. Coherence by itself is not an overwhelming force. That is why our attitudes towards animals are weak and wavering, allowing us to use them in ways we would not use people. Consider again the second hereditary slave society I mentioned above. This society was to be one in which no aspects of the moral understanding shared by the masters spoke against slavery. In fact that is unlikely, since there is some arbitrariness in the idea that people are to be treated in different ways depending on whether they are born slave or free. Coherence of attitude will speak at least a little against the system of slavery. But the factors of conservatism and desire might speak more strongly in favour of the status quo, so that, all things considered, the slave owners might have no reason to change their understanding. One thing that distinguishes slaves from animals is that slaves can organize and threaten revolt, whereas animals cannot. Slaves can see to it that both coherence and desire oppose conservatism, so that it becomes rational for the slave owners to arrive at a new, broader, more coherent understanding, one which includes the slaves. It should be noted that coherence of attitude provides a constant pressure to widen the consensus and eliminate arbitrary distinctions. In this connection it is useful to recall ancient attitudes toward foreigners, and the ways people used to think about ‘savages’, ‘natives’, and ‘Indians’. Also, recall that infanticide used to be considered as acceptable as many consider abortion to be. There has been a change here in moral attitudes, prompted, I suggest, largely by considerations of coherence of attitude. Finally, I would like to say a few brief words about the limiting case of group morality, when the group has only one member—when, as it were, a person comes to an understanding with himself. In my view, a person can make inner judgements in relation to such an individual morality only about himself. A familiar form of pacifism is of this sort. Certain pacifists judge that it would be wrong of them to participate in killing, although they are not willing to make a similar judgement about others.

MORAL RELATIVISM DEFENDED

19

(Such a pacifist is only unwilling to make inner moral judgements about others. Although he is unwilling to judge that those who do participate are wrong to do so, he is perfectly willing to say that it is a bad thing that they participate.) There are, of course, many other examples of individual morality in this sense, when a person imposes standards on himself that he does not apply to others. The existence of such examples is further confirmation of the relativist thesis that I have presented. My conclusion is that relativism can be formulated as an intelligible thesis, the thesis that morality derives from an implicit agreement and moral judgements are true or false only in relation to such an agreement. Such a theory helps to explain otherwise puzzling aspects of some of our own moral views, in particular, why we think that it is more important to avoid harm to others than to help others. The theory is also partially confirmed by what is, as far as I can tell, a previously unnoticed distinction between inner and non-inner moral judgements. Furthermore, traditional objections to implicit agreement theories can be met. Moral relativism is discussed further in Essays 2–5 and 13, below.

2 What Is Moral Relativism? Of the various views that have been called ‘moral relativism’, we can distinguish three plausible versions, which, for the purposes of this essay, I will label ‘normative moral relativism’, ‘moral judgement relativism’, and ‘meta-ethical relativism’. Normative moral relativism asserts that different people, as agents, can be subject to different ultimate moral demands. Moral judgement relativism holds that moral judgements make implicit reference to the speaker or to some other person or to some group or to one or another set of moral standards, etc. Meta-ethical relativism says that conflicting moral judgements about a particular case can both be right.

2.1 Normative Moral Relativism According to my initial statement of normative moral relativism, different people, as agents, can be subject to different ultimate moral demands. Only ultimate differences are supposed to be relevant, since even the nonrelativist will agree that different agents are subject to some different moral demands because of differences in situation. Jack is morally obligated to pay George ten dollars. Mary is not. That is not yet normative moral relativism. Maybe Jack has borrowed ten dollars from George and Mary has not. In that case, Mary and Jack might still be subject to the same ultimate moral demands, including, for example, that one should pay one's debts. Differences in situation can include differences in custom. In England in 1977 Jack ought to drive on the left side of the road. In France, Mary ought to drive on the right. This is again not yet normative moral relativism, since Jack and Mary may still be subject to the same overriding principle—for example, one ought to drive on the side of the road designated by law and custom. Similarly, other differences between what Jack and Mary ought to do may reflect differences in custom concerning

WHAT IS MORAL RELATIVISM?

21

politeness and etiquette, family responsibility, property, and so forth, rather than any difference in ultimate principle. What is an ultimate moral demand? A given demandD is an ultimate moral demand on an agentA if and only if there is no further moral demand D* on A which, given A's situation, accounts forA's being subject toD. However, quite apart from the issue of relativism, it seems to be a possible nonrelativistic view that there are no ultimate moral demands on a person in the following sense: wheneverD applies toA there is always a more fundamentalD* which explains whyD applies. According to my initial formulation of normative moral relativism, anyone who denies that there are ultimate moral principles must be a normative moral relativist. That is wrong. We need a better formulation. Can we take normative moral relativism to be the view that there are no moral demands to which everyone is subject? No, because a nonrelativist can coherently suppose that there are people subject to no moral demands at all, for example, infants and idiots. Can we take moral relativism to say that there are no moral demands that apply to everyone who is subject to at least some moral demands? That would cover only a very strong form of normative moral relativism. We also need to include a weaker version that is compatible with the existence of some universal moral demands. The weaker version claims there can be two people subject to different moral demands and not subject to some more basic demand that accounts for this, given differences in their situation. More formally, it says that there can be two peopleA and B and a moral demand D such that (1) (2) (3) (4)

A is subject to D, B is not subject to D, although B is subject to some moral demands, and there is no demand D* to which A and B are both subject which accounts for (1) and (2) given the differences in situation between A and B.

This formulation allows for some moral universality and is compatible with the claim that there are no ultimate moral demands. It is basically the same view that Frankena (1973) calls ‘normative relativism’. He formulates the view as follows: ‘[W]hat is right or good for one individual or society is not right or good for another, even if the situations involved are similar.’ This is inexact, because any two situations are similar in some respects and different in others. We need to say that the situations are similar in the sense that there are no morally significant differences between the two situations.

22

MORAL RELATIVISM

If we suppose that a difference is morally significant only if it is counted significant by some moral principle or demand, we are led to something like the formulations I have offered for ‘normative moral relativism’. Brandt (1967) uses the term ‘normative relativism’ differently for any view that asserts something is wrong or blameworthy if some person or group thinks it is wrong or blameworthy. He gives two examples of such a view. (1) If someone thinks it is right (or wrong) to do A, then it is right (wrong) for him to do A. (2) If the moral principles recognized in the society of which X is a member imply that it is wrong to do A in circumstances C, then it is wrong for X to do A in C. These are not very plausible views. Nor are they clearly versions of relativism, since they appear to be advocating moral principles that might be taken to have universal applications. In any event, they are certainly not good examples of what I am calling normative moral relativism, which says that two people can be subject to different moral demands and not subject to some more basic demand that accounts for this given their situation. This ends my preliminary discussion of normative moral relativism, which makes a claim about moral agents. I turn now to a relativistic thesis about the meaning or form of moral judgements.

2.2 Moral Judgement Relativism According to moral judgement relativism, moral judgements contain an implicit reference to the speaker or some other person or some group or certain moral standards, etc. One version holds that moral judgements are always implicitly egocentric in the sense that they are always equivalent to judgements containing egocentric terms essentially. Egocentric terms in English include I, me, this, and now. So one example of this sort of moral judgement relativism would be Brandt's (1954), ‘It would be wrong to do X’ means the same as ‘If I were normal, impartial, and fully informed, I should feel obligated not to perform X.’ (Brandt, 1959, abandons this suggestion.)

WHAT IS MORAL RELATIVISM?

23

Notice that this form of moral judgement relativism says that egocentric terms are essential to the equivalent paraphrase. Anyone can agree that ‘Stealing is wrong’ is equivalent to ‘My stealing is wrong and so is everyone else's.’ But that does not guarantee the truth of this form of moral judgement relativism. Notice also that this form of moral judgement relativism holds that all moral judgements are egocentric in this sense. It is obvious that some are, for example, ‘I should not steal’ is, but, again, that is not enough to establish the truth of this form of moral judgement relativism. Another version of moral judgement relativism takes moral judgements to be implicitly relative to one or another morality in something like the way in which a judgement that someone is tall is implicitly relative to one or another comparison class. George can be tall in relation to one such class and not tall in relation to another. For example, George might be tall for a man but not tall for a basketball player. It makes no sense to ask whether George is tall, period, apart from one comparison class or another. Similarly, this form of moral judgement relativism holds that moral judgements make sense (are true or false) only in relation to one or another set of moral demands. Something can be right in relation to one morality and not right in relation to another. ‘As a Christian, you ought to turn the other cheek; I however, propose to strike back.’ The judgement that X is wrong is always incomplete, just as the judgement that P is tall is incomplete. We always understand the latter judgement as the judgement that P is tall for a person, or for a basketball player, etc. Similarly, we must understand the former judgement as saying that X is wrong for a Christian, for a Muslim, for someone who accepts such and such demands, or something. Stevenson (1963b) uses the term ‘relativism’ to indicate something like this second form of moral judgement relativism. On the other hand, Firth (1952) counts an analysis as ‘relativistic’ if it is a version of the first form of moral judgement relativism in its essential use of egocentric terms. An analysis that is relativistic in Firth's sense will not necessarily count as a form of relativism in Stevenson's sense. Nevertheless, both views see a more or less implicit relativity in the logical form or meaning of moral judgements, so both count as forms of moral judgement relativism in my sense. I turn now to a third form of relativism which sees a different sort of relativity in moral judgements, a relativity in the correctness of moral judgements.

24

MORAL RELATIVISM

2.3 Meta-Ethical Relativism According to meta-ethical relativism, there can be conflicting moral judgements about a particular case that are both fully correct. The idea is that two people with different moralities might reach conflicting moral judgements concerning a particular case—for example, one saying the agent was morally right, the other saying the agent was morally wrong—where both opinions are correct. The two judgements must really conflict and therefore must be judgements about the same particular case. It is not enough to contrast a judgement about an act in one society with a judgement about the ‘same act’ in another society. Furthermore, it is not enough for meta-ethical relativism that there should be such conflicting judgements. Both judgements must be correct. In particular, neither judgement can rest on mistakes about the facts of the case, a failure fully to appreciate these facts, incorrect reasoning, or anything else that might distort someone's judgement. Finally, both judgements must be fully correct. Borderline cases are not enough, cases in which one might be equally justified in saying one thing or the opposite. According to meta-ethical relativism, the one judgement is fully correct for the one speaker, the conflicting judgement is fully correct for the other speaker, and it would not be correct for either speaker to make the other judgement. Both Brandt and Frankena use the term ‘meta-ethical relativism’ in roughly this sense. According to Brandt (1967), the meta-ethical relativist ‘denies that there is always one correct moral evaluation’ of a given issue. Elsewhere, Brandt (1959: 272) calls the same view ‘ethical relativism’, which he defines as the thesis that ‘there are conflicting ethical opinions that are equally valid’. However, Brandt (1959: 3278-9; 1967: 75) would allow certain merely apparently conflicting opinions to count as conflicting, if moral judgement relativism should be true. So certain moral judgement relativists will count as meta-ethical relativists as Brandt uses this term but not as I am using it. As Frankena (1973) explains meta-ethical relativism, ‘It holds that, in the case of basic ethical judgements, there is no objectively valid, rational way of justifying one against the other; consequently two conflicting basic judgements may be equally valid.’ Presumably he refers to ‘basic ethical judgements’ to allow for the possibility that less basic judgements might be justified on the basis of more basic judgements, the question then being how the most basic judgements are to be justified. But this part of his definition

WHAT IS MORAL RELATIVISM?

25

is best omitted, since I want to allow both relativists and nonrelativists to be able to deny that there are basic ethical judgements in this sense. This leaves us with the claim that two conflicting ethical judgements may be equally valid, which is essentially the way in which I have defined meta-ethical relativism. (Cf. Frankena, 1976b: 123, where this position is called simply ‘relativism’: ‘[I]f two people hold conflicting normative judgements . . . both judgements may be rational or justified.’) Brandt (1959: 275; 1967: 76) uses the term ‘methodological relativism’ to refer to what he takes to be a particular version of what I am calling meta-ethical relativism, a version which holds that ‘there is no unique rational method in ethics' for assessing moral judgements.’ Stevenson (1963b) uses the term ‘methodological relativism’ slightly differently for the claim that statements about the reasons and justification of moral judgements are themselves implicitly relational, in something like the way in which moral judgements are held to be implicitly relational according to moral judgement relativism. Brandt's but not Stevenson's definition counts as a ‘methodological relativist’ the sceptic who denies that moral judgements are ever correct or justified in any sense. Such a sceptic clearly does not count as a metaethical relativist in my sense, because a meta-ethical relativist holds that conflicting moral judgements can both be correct, whereas such a sceptic denies that any moral judgement is ever correct.

2.4 The Three Versions of Moral Relativism Briey Compared I have distinguished three versions of moral relativism. (1) Normative moral relativism holds that two people can be subject to different moral demands and not subject to some more basic demand that accounts for this given their different situations. (2) Moral judgement relativism says that moral judgements implicitly refer to one or another person, group, or set of moral demands. And (3) meta-ethical relativism holds that conflicting moral judgements about a particular case can both be right. It is clearly possible to accept one of these versions without accepting the others. For example, an emotivist might accept meta-ethical relativism on the ground that people can differ in their moral attitudes without having relevantly different beliefs, without having reasoned incorrectly or failed to appreciate certain things, without being prejudiced or biased, etc. But, as

26

MORAL RELATIVISM

Stevenson (1963b) points out, the emotivist need not (and probably will not) accept moral judgement relativism. And the emotivist can also deny normative moral relativism, since he can (and no doubt will) suppose that the principles he accepts apply to all agents. Moreover, an existentialist might accept normative moral relativism on the grounds that each person is subject only to the principles that person accepts. This does not imply moral judgement relativism or meta-ethical relativism, since the existentialist can suppose that conflicting moral judgements about a particular agent cannot both be correct. Given the moral principles that agent accepts, at most one of the judgements will be right. Finally, a relativistic ideal observer theorist with the sort of view Brandt (1954) defends can accept moral judgement relativism without accepting meta-ethical relativism, since he can suppose that two really (as opposed to merely apparently) conflicting moral judgements cannot both be right. And such a theorist can reject normative moral relativism too, for example, on the grounds that if he were an ideal observer he would apply the same basic principles to everyone. So, it is possible to accept any one of these versions of moral relativism without being committed to the others. It is also possible, consistently, to accept more than one of these versions. Indeed, I am inclined to accept all three. But someone might say that none of these is really a possible view—for each is subject to serious objections that I must now consider.

2.5 An Objection to Normative Moral Relativism According to normative moral relativism there can be two people A and B and a moral demand D such that (1) (2) (3) (4)

A is subject to D, B is not subject to D, B is subject to some moral demands, there is no moral demand D* to which A and B are both subject and which accounts for (1) and (2) given the difference in situation between A and B.

The obvious objection to this is that, if (1), (2), and (3) are true, there must be some reason why A but not B is subject to D, for surely this cannot

WHAT IS MORAL RELATIVISM?

27

be an arbitrary fact. Something must be true of A but not true of B which accounts for why A but not B is subject to D. It follows that there must be some sort of general principle which implies that people with certain characteristics F are subject to D, and others are not. So there must be a general principle P more basic than that expressing the demand D which accounts for (1) and (2) given the difference in situation between A and B, that is, given that A is F and B is not. But then it can seem that (4) must be false. For won't P express just the moral demand D* whose existence (4) denies? Consider, for example, an existentialist who advocates normative moral relativism on the grounds that each person is subject only to moral principles that he or she accepts. Then A is subject to D but B is not, because A accepts a principle expressing the demand D and B does not. But then it may seem that the existentialist is committed to supposing that there is after all a moral demand D* that applies both to A and to B and explains why A but not B is subject to D, namely, the demand of non-hypocrisy, the demand that one should act in accordance with those principles one accepts. And it can seem that the same point will apply to any attempt to defend normative moral relativism, leading to the conclusion that this is not after all a coherent form of moral relativism. I suspect, indeed, that reflections along these lines leads Brandt to use the term ‘normative relativism’ not for what I am calling normative moral relativism but for the view that something is wrong or blameworthy if some person or group thinks it is wrong or blameworthy. For it can seem that the existentialist is more plausibly taken to be advocating normative relativism in Brandt's sense than normative moral relativism in my sense; for it can seem that normative moral relativism in my sense is quite incoherent. I believe that this is wrong, and that normative moral relativism is not only coherent but true. Before saying why, let me mention objections to the two other forms of moral relativism.

2.6 An Objection to Moral Judgement Relativism Moral judgement relativism holds that moral judgements make implicit reference to the speaker, some other person or persons, or some set of moral demands. It follows from this that apparently conflicting moral judgements do not actually conflict if they are made by different speakers

28

MORAL RELATIVISM

or are made in relation to different people or moralities. But, as Stevenson (1963a, 1963b) persuasively argues, this seems to imply that certain genuine moral disagreements are mere pseudo-issues. Recall the suggestion in Brandt (1953) that ‘It would be wrong to do X’ means ‘If I were normal, impartial, and fully informed, I should feel obligated not to perform X.’ Now suppose that A and B are discussing whether abortion is wrong. A maintains that abortion is wrong; B maintains that it is not wrong. A and B imagine that they are disagreeing; they mean to be disagreeing with each other. But Brandt's analysis implies that they are not really disagreeing. A is saying that if she, A, were normal, impartial, and fully informed, she would feel obligated not to have an abortion in certain circumstances. B is saying that if she, B, were normal, impartial, and fully informed, she would not feel obligated not to have an abortion in those circumstances. And both of these claims could be true. It is possible that under these conditions A would feel obligated not to have an abortion and B would not feel obligated. But it is quite implausible that A and B do not disagree. For it seems that if A and B are using words normally they do genuinely disagree. Stevenson (1963a) argues that, even if they fail to disagree in belief, they certainly disagree in their attitude toward abortion. He takes this to be evidence for his view that moral judgements express favourable or unfavourable attitudes rather than beliefs.

2.7 An Objection to Meta-Ethical Relativism Metal-ethical relativism holds that actually conflicting moral judgements about a particular case can both be right in nonborderline cases. An obvious objection is that it is not clear what this could mean. It would seem that to say that a judgement is right is to agree with it. The meta-ethical relativist therefore seems to be committed to agreeing with each of two judgements which, also according to the same theorist, genuinely conflict. This seems to involve an inconsistency. What can the meta-ethical relativist have in mind? I am supposing that the people making the judgements genuinely, and not merely apparently, disagree. So it seems that the meta-ethical relativist, in agreeing with both, must be inconsistent! In saying that both judgements are right, the meta-ethical relativist might mean either that both are true or that both are justified, given the facts and all other relevant considerations. In either case the objection

WHAT IS MORAL RELATIVISM?

29

seems to apply. Reverting to my earlier example, let us suppose again that A maintains that abortion is wrong and that B maintains that abortion is not wrong, and let us agree that this is a genuine disagreement—that these judgements really, and not just apparently, conflict. Then what could it mean to say that both judgements are true? Presumably, abortion is wrong if it is true that abortion is wrong, and abortion is not wrong if it is true that abortion is not wrong. If the meta-ethical relativist says both judgements are true, this seems to imply the contradictory conclusion that abortion is both wrong and not wrong. On the other hand, suppose that the meta-ethical relativist says that, given all the facts and all other relevant considerations, the judgement that abortion is wrong is justified and so is the judgement that abortion is not wrong. What could this mean? It might mean that this is a borderline case, like deciding whether a certain man is bald—perhaps given all the facts, etc., the judgement that he is bald is justified, and so is the judgement that he is not bald. But to say that there are borderline moral issues of this sort is not to say anything interesting—of course there are. I am supposing that the meta-ethical relativist claims something more—that in a case like this, both judgements might be justified, although it is not a borderline case. But what can such a claim mean? Stevenson's (1963b) suggestion might seem to help here. Perhaps the meta-ethical relativist is a methodological relativist in Stevenson's sense, holding that the term ‘justified’ is a relative term here, having application only to one or another person. Neither judgement is justified, period; rather, the one judgement is justified for A but not B, and the other judgement is justified for B but not A. However, as Stevenson observes, this sort of methodological relativism is faced with the same sort of objections raised against moral judgement relativism. For A and B will disagree not only about whether abortion is wrong but also about whether the judgement that abortion is wrong is justified, A maintaining that it is, B denying this. The second disagreement, like the first, seems to be a real disagreement and not the pseudo-issue it is taken to be by this sort of normative relativism. There are, then, serious objections to each of the versions of moral relativism I have distinguished. Against normative moral relativism, it can be said that if A is subject to a moral demand that B is not subject to, then there must be some reason for this, a reason that invokes a demand on both A and B, contrary to what is claimed in normative moral relativism. The objection to moral judgement relativism is that it wrongly treats certain real disagreements as verbal pseudo-issues. Finally, it can be objected

30

MORAL RELATIVISM

that meta-ethical relativism faces a dilemma. On the one hand, it inconsistently agrees with each of two conflicting judgements. Or, on the other hand, in supposing that ‘justified’ is a relative term, it misrepresents real disagreements about justification as merely apparent disagreements. But all these objections can be met. I will now describe what strikes me as a highly plausible view that, in one way or another, involves all three versions of moral relativism and shows how the objections to each version can be avoided.

2.8 An Argument for Normative Moral Relativism Normative moral relativism can be derived from two assumptions. The first is this. ASSUMPTION 1. A moral demand D applies to a person only if that person either accepts D (i.e. intends to act in accordance with D) or fails to accept D only because of ignorance of relevant (nonmoral) facts, a failure to reason something through, or some sort of (nonmoral) mental defect like irrationality, stupidity, confusion, or mental illness. This assumption might be defended by an appeal to the sorts of considerations sometimes offered in support of ‘internalist’ accounts of moral demands. (Frankena, 1976a, usefully discusses such considerations.) It might be said, for example, that ASSUMPTION 1 captures what distinguishes moral demands from demands of other sorts, such as legal demands. Moral demands have to be acceptable to those to whom they apply in a way that legal demands do not. The point might be made in terms of the notion of a reason to do something, because, if a moral demand applies to someone, that person has a compelling reason to act in accordance with that demand. This is not true of legal demands. A legal demand applies to anyone in a given area who satisfies certain conditions, whether or not he or she has a compelling reason to act in accordance with that demand. Now, reasons seem to depend on reasoning. To say that a person has a conclusive reason to do something seems to be to say that there is warranted practical reasoning available to the person that would culminate in a decision to do the act in question. Such reasoning is available in the

WHAT IS MORAL RELATIVISM?

31

sense that the person could reason in that way if he or she was of sufficient intelligence, was rational, knew all the relevant facts, etc. Frankena (1976a: 52) observes that an ‘externalist’ might object to the considerations just mentioned by claiming that they rest on an ambiguity in the term ‘reason’, which might mean either ‘justifying reason’, that is, reason to think one ought to do something, or ‘motivating reason’. The externalist can say that the analysis of ‘conclusive reason to do something’ in terms of practical reasoning culminating in a decision to do it is, perhaps, an analysis of ‘motivating reason to do something’. But this, the externalist can say, is irrelevant to the point that, if a moral demand applies to someone, that person has a compelling reason to act in accordance with that demand. For here the relevant type of reason is a justifying reason: if a moral demand applies to someone, that person has a compelling reason to think he or she ought to do that thing. The internalist will reply that this is an ad hoc manœuvre, since no evidence has been offered for the alleged ambiguity. (Frankena, 1976a, gives examples which show that reasons for doing something might be either moral or nonmoral, e. g. self-interested. But that seems irrelevant to a claim of ambiguity.) Furthermore, the notion of a ‘justifying reason’, as a reason to think one ought, leads to a potential regress. To think one ought to do something is to think one has reasons to do it. These reasons must not in turn be taken to be justifying reasons. Otherwise, thinking one ought to do something would be explained as thinking one has reason to think one ought to do that thing. The internalist point is obscured if talk involving ‘ought’ is not clearly distinguished from talk of obligation. It can happen that, according to the law, one is supposed to do a certain thing, although there is not the slightest reason actually to do it—perhaps there is no penalty for not doing it and one has no moral reasons to observe this particular law. Here we might speak of legal obligation, but we should not say that legally one ought to do the act in question, even though there is a legal ‘justification’ only for doing that. The word ‘ought’ is used to speak of reasons in a way that ‘obligation’ is not. Of course, if, according to morality, one is supposed to do a certain thing, then one ought to do it—one does have a reason in that case. So, even though there is no way to prove the relativist's first assumption, a strong case can be made for it, and it is at the very least a possible view to take. The second assumption can be stronger or weaker, depending on how

32

MORAL RELATIVISM

strong a form of normative moral relativism is to be defended. The strongest assumption is this. ASSUMPTION 2S. For any moral demand D, there is someone subject to some moral demands who does not accept D as a legitimate demand on himself or herself, where this nonacceptance is not the result of any relevant (nonmoral) ignorance on that person's part or any failure to reason something through or any sort of (nonmoral) mental defect such as irrationality, stupidity, confusion, or mental illness. Assumptions 1 and 2S jointly imply that there are no universal moral demands which apply to everyone who is subject to some moral demands. I am interested here in a weaker normative moral relativism that can be based on a correspondingly weaker second assumption. ASSUMPTION 2W. There are two people A and B and a moral demand D such that: (1) A is subject to D. (2) B does not accept D as a legitimate moral demand on him or herself. (3) B's nonacceptance is not the result of any relevant (nonmoral) ignorance on B's part or any failure to reason something through or any sort of (nonmoral) mental defect such as irrationality, stupidity, confusion, or mental illness. (4) B is subject to some moral demands. (5) Neither A nor B accepts a moral demand D* that would account for A but not B being subject to D given the difference in situation between A and B. (6) In neither case is this failure to accept such a D* the result of any relevant (nonmoral) ignorance or any failure to reason something through or any sort of (nonmoral) mental defect such as irrationality, stupidity, confusion, or mental illness. Assumptions 1 and 2W jointly imply normative moral relativism as I have defined it above. What can be said on behalf of assumption 2W? Let us consider a couple of possible examples. Consider first the issue of cruelty to animals as it appears to typical middle-class American citizens and as it appears to Hopi Indians as described by Brandt (1954). It seems plausible to suppose that a moral demand applies to the typical middle-class American citizen

WHAT IS MORAL RELATIVISM?

33

forbidding the infliction of needless suffering on animals. On the other hand, according to Brandt (1954), a typical Hopi Indian did not accept such a strong demand as a legitimate moral demand, where this nonacceptance was not the result of ignorance, irrationality, stupidity, etc. Furthermore, neither the typical middle-class American citizen nor the typical Hopi Indian appears to accept any more basic demand that would account for why this should apply to middleclass American citizens but not Hopi Indians given differences in their situations. And it would seem that in neither case is the failure to accept such a basic principle the result of ignorance, irrationality, stupidity, etc. So this seems to be an example establishing the truth of assumption 2W. (The issue of cruelty to animals is discussed further in Essay 4, section 4.10, and Essay 5, section 5.6.1, below.) I am myself inclined to think there are many examples of this sort even within American society. Consider the moral demand that one not kill other people. Some professional criminals do not seem to accept this demand. These criminals have no qualms about killing other people if there is something to be gained from doing so. In some cases, their lack of acceptance of the relevant principle does not appear to rest on a failure to notice certain facts or incorrect reasoning or any failure to follow out certain reasoning. Nor is it always because of stupidity, irrationality, confusion, or mental illness. Furthermore, clearly this is not because there is some more basic moral demand which accounts, given differences in situation, for why most people but not these criminals are subject to the prohibition against the killing of other people. After thinking over these and similar examples, some people (like me) will suppose that assumption 2W is true. Others will not. Some philosophers have argued—unconvincingly—that a rejection of moral demands of this sort must involve ignorance, irrationality, or some other (nonmoral) mental defect. (See Essay 4, sections 4.6-4.7.) Others might hope that psychology would resolve the issue. ‘Unfortunately,’ as Brandt (1959: 282) remarks about a related issue, ‘psychological theories do not provide a uniform answer to our question.’ Gestalt theory and Piagetian theory point in one way, Hullean learning theory and psychoanalytic theory point in the other way (Brandt, 1959: 114-50; Hoffman, 1970). In any event, I believe it is safe to say that neither philosophy nor psychology has produced a strong case against assumption 2W. Assumption 1 says, roughly, that a moral demand applies to someone only if it is rational for that person to accept that demand. Assumption 2W says, roughly, that it can be rational for different people to accept different

34

MORAL RELATIVISM

moral demands ‘all the way down’. Together these assumptions logically imply, roughly, that different people can be subject to different moral demands ‘all the way down’. How can it be rational for different people to accept different moral demands ‘all the way down’? This might be because what it is rational to accept depends on how one's mind works, and different people's minds work in different ways. A less radical suggestion is that the demands it is rational to accept depend on the demands one already accepts, where different people accept demands sufficiently different to begin with that it is rational for them to end up accepting different demands. This is not to say that if one accepts a demand it is automatically rational for one to do so. Nor is it to say that the only thing relevant to the question whether it is rational to accept a given demand is what demands one already accepts. Presumably one's desires, other intentions, and beliefs are important too. To accept a moral demand, in the relevant sense, is to intend to adhere to it. Whether it is rational to do this is a question of practical reasoning, which in my opinion is to be resolved (roughly) by minimally modifying antecedent intentions, beliefs, desires, etc. in the interest of a certain sort of coherence (Harman, 1976). But the details are not important here. We must now see how my earlier objection to normative moral relativism can be met. The objection was that, if A is subject to D and B is not, then something must be true of A, but not B, which accounts for this. So there must be some sort of general principle which accounts for why people who are F are subject to D, and why those who are not F are not subject to D. This general principle will express a moral demand D* which accounts for why A, but not B, is subject to D. So, there must always be such a demand D*. But normative moral relativism denies that there is always such a demand D*. The answer to the objection is that the principle in question does not always express a moral demand. A, but not B, may be subject to D because it is rational for A to accept D, but not rational for B to do so, and this is sometimes not the result of some more basic moral principle applying to both A and B, but rather due simply to the fact that A and B start out accepting sufficiently different moral demands in the first place. Perhaps even in the key case for the relativist there is a demand which applies both to A and B and which, given the differences in their situations, accounts for why A, but not B, is subject to D. But the demand is not a moral demand—it is simply the demand that one be rational.

WHAT IS MORAL RELATIVISM?

35

2.9 Relativism and Universalizability Frankena (1973: 109) raises a somewhat different objection to normative moral relativism: ‘Such a normative principle seems to violate the requirements of consistency and universalization.’ Elsewhere (p. 25) he refers to the relevant principle as ‘The Principle of Universalizability’: ‘If one judges that X is right or good, then one is committed to judging that anything exactly like X, or like X in relevant respects, is right or good. Otherwise he has no business using these words.’ Here Frankena seems to be suggesting that the use of moral terminology normally presupposes a Principle of Universalizability that is incompatible with normative moral relativism. This implies that normative moral relativism involves a misuse of language. Clearly that is too strong a conclusion. Many people may use moral terminology in a way that presupposes a Principle of Universalizability that is incompatible with normative moral relativism. Perhaps such a principle is even partly constitutive of what these people mean by these terms. Nevertheless, other people—relativists, for example—do not use moral terminology in this way. It must be argued, and not just assumed, that there is something wrong with this second usage. Indeed, suppose that, although one initially uses moral terminology in the first way, one becomes convinced of assumptions 1 and 2W. These assumptions logically imply normative moral relativism. Then one has become convinced of the falsity of a presupposition of one's use of moral terminology. What should one do? There are two options here. One might simply abandon moral terminology altogether in the way that one abandons certain racist or sexist terminology after becoming aware of the false presuppositions of its usage. Or one might modify one's usage so that it no longer involves the presuppositions one takes to be false, in the way that Einstein modified the scientific usage of terms like ‘simultaneous’, ‘acceleration’, and ‘mass’, after arguing that prior usage had involved a false presupposition. This second course, of modifying one's usage of moral terminology so as to eliminate the unwanted presupposition, seems clearly the preferable alternative. How should one's usage be modified? Presumably along the lines suggested by moral judgement relativism. Judgements expressing moral demands can be treated as involving an implicit reference to a person or to a group of people who are presumed to accept certain demands, or reference simply to certain demands themselves. But it is important to avoid

36

MORAL RELATIVISM

or evade the objection raised against moral judgement relativism, namely, that it represents certain real disagreements as mere pseudo-issues. If two people—even two relativists—disagree about whether it is morally wrong to cause pain and injury to animals, they really do disagree and are not just talking past each other. It is not just that one says that causing pain and injury to animals is wrong in relation to our morality, while the other says that it is not wrong in relation to Hopi morality. Within a relativistic theory, the objection can be partially met as follows. When a relativist makes a simple moral judgement with no explicit reference to one or another morality, the speaker makes this judgement in relation to a morality he or she accepts. The speaker also presupposes that this morality is accepted by anyone to whom the judgement is addressed. The speaker may also presuppose that it is accepted by any agents referred to in the judgement (if what is at issue is whether it is wrong of them to do a certain thing). A relativist can make moral judgements in relation to moralities he or she (or the audience) does not accept, but it is normally a misuse of moral language—even when it has been relativized—to do this without making it clear that one is doing so. It is, furthermore, always a misuse of language to make certain moral judgements about an agent in relation to a morality not sufficiently accepted by the agent. (The relevant judgements are the ‘inner judgements’ discussed in Essay 1.) If two relativists disagree, one saying simply that it is morally wrong to cause pain and injury to animals, the other saying simply that this is not wrong, they do mean to be disagreeing with each other. They presuppose that they are making these judgements in relation to the same relevant moral demands. Of course, they may be mistaken about that, in which case they really are talking past each other, despite their intentions. If they come to see that they are speaking in relation to relevantly different moralities, they will have to stop saying what they are saying or indeed be guilty of misuse of language. A relativist can intelligibly disagree with an absolutist over whether something is wrong if the relativist makes his judgement in relation to a morality the absolutist accepts, in the same way that a relativity-theory physicist can disagree with a Newtonian over what the mass of a certain object is, if the relativity theorist makes his judgement in relation to an appropriate inertial framework. A relativistic judgement made in relation to a given morality concerning the moral demands applicable to a particular agent presupposes that the agent in question accepts that morality. So relativists with different moralities will, if they have their wits about them, tend not to make this

WHAT IS MORAL RELATIVISM?

37

sort of judgement about the same agents. And, at least as a first approximation, it will turn out that even superficially conflicting judgements cannot both be true of the same agent even if the judgements are made in relation to different moralities, since the judgements will involve conflicting presuppositions about the morality the agent accepts. That is only a ‘first approximation’, because a particular agent will typically accept more than one morality and those moralities can make conflicting demands concerning a particular case. In that case, superficially conflicting moral judgements about that agent (e.g. Antigone) made in relation to different moralities can both be true if the agent accepts both moralities relative to which each of the judgements is made. In any event, the relativist's usage does not permit unrestricted universalization. As a relativist, I cannot accept the principle that, if it would be wrong for me to do something, it would be wrong for anyone else in a similar situation to do that thing. I do not suppose that the principle is true for those who have no reason to accept the relevant aspects of my morality. But the following principle is acceptable. If it is wrong for someone else to do something, it would be wrong for me to do that in a similar situation. In other words, if it is okay for me to do something, it cannot be wrong for anyone else in my situation. (Universalizability is discussed further in Essay 3.)

2.10 Assessing Meta-Ethical Relativism My second and third versions of moral relativism are theses about moral judgements. But we need to specify whose judgements these versions are about: the relativist's or the nonrelativist's? And, moral judgement relativism is clearly a thesis about the relativist's usage; it is not meant to apply to the nonrelativist's usage. What about meta-ethical relativism? This is the claim that really conflicting moral judgements can both be right. Now, superficially conflicting relativistic moral judgements can both be right if made in relation to different moralities, but these judgements do not really conflict. And it does not seem to make sense to suppose that really conflicting relativistic moral judgements can both be right. So meta-ethical relativism does not seem correct as a thesis about relativistic moral judgements. Is it correct as a thesis about nonrelativistic moral judgements? One might suppose that normative moral relativism implies that no nonrelativistic moral judgements can be right, since all such judgements presuppose

38

MORAL RELATIVISM

an unrestricted principle of universalizability. But that would be like denying that any Newtonian judgements concerning the mass of a particular object could be right, on the grounds that these judgements presuppose that mass is invariant from one inertial framework to another. In practice we would say that a Newtonian judgement of mass is right if the corresponding relativistic judgement is correct made in relation to an inertial framework that is salient to the person making the Newtonian judgement. Similarly, we should count a nonrelativistic moral judgement right if the corresponding relativistic moral judgement is right made in relation to the morality accepted by the person making the nonrelativistic moral judgement. The relativist can intelligibly suppose that really conflicting nonrelativistic moral judgements are both right in this sense. I conclude, then, that there is a reasonable form of moral relativism that involves normative moral relativism, moral judgement relativism as a thesis about relativistic but not nonrelativistic moral judgements, and meta-ethical relativism as a thesis about nonrelativistic but not relativistic moral judgements.

3 Relativistic Ethics: Morality as Politics 3.1 The Naive View of Morality Let me begin by saying something about what I will call ‘the naive view’ of morality. In this view, morality involves one or more basic moral demands that everyone accepts, or has reasons to accept, as demands on everyone on which all moral reasons to do things depend. I do not mean to suggest that the naive view is the ordinary view of morality. There is not a single ordinary view of morality. It is possible that most people do not accept the naive view. But many people do accept it, especially teachers of moral philosophy, and people who accept it often take it to be the ordinary view. (See Essay 5 below, especially section 5.2.4) The naive view has three parts. First, it says that there are certain basic moral demands that everyone accepts or at least has reasons to accept. Let me refer to this as the claim that morality is ‘absolute’. Second, these demands are supposed to be accepted as demands on everyone. They are to have universal application. I will refer to this as the claim that morality is ‘universal’. Third, these demands are supposed to be the source of all moral reasons for agents to do things. I will refer to this as the claim that morality is ‘agent-centred’ or, sometimes, that it ‘takes the point of view of an agent’ rather than a critic. In the naive view, then, morality is absolute, universal, and agent-centred. In the naive view, the basic moral demands are or ought to be accepted by everyone, as demands on everyone, providing reasons for everyone. But who is ‘everyone’? Kant (1788) thought every rational being should be included, not only human beings but also angels and rational inhabitants of other planets. A rational Martian would accept or have reasons to accept the basic demands of morality, because according to Kant the basic moral demands are simply demands of rationality that follow from the nature of reason (which he took to be the same in all rational beings).

40

MORAL RELATIVISM

We need not be so strict, however. Some people who accept what I am calling the naive view may not wish to commit themselves about Martians. They may see the source of moral demands in human nature, rather than in an absolutely universal reason. They may suppose, for example, that moral demands derive either from a natural human sympathy that people feel for others, or from some other fact about the human condition. They may therefore wish to restrict the range of ‘everyone’, claiming only that every rational human being accepts or has reasons to accept the basic demands of morality. Some may even want to restrict the range of ‘everyone’ to those who have been exposed to ‘civilized morality’, for they may not wish to claim that so-called ‘ignorant savages’ have any reason to accept the basic demands of morality.

3.2 Against Absolutism For present purposes, it does not matter whether or not the range of ‘everyone’ is restricted in one of these ways. It seems to me highly probable that, whether or not the range of ‘everyone’ is thus restricted, there are no substantive moral demands everyone has a reason to accept. I am inclined to believe that, for any such demand, someone might fail to accept it without being ignorant of relevant facts, without having miscalculated, without having failed to see the consequences of his or her opinions, and without being in any way irrational. You cannot always argue someone into being moral. Much depends on a person's antecedent interests and principles. If these principles and interests diverge sufficiently from yours, it may happen that the other person has no reason to accept your morality. Invaders from outer space who are unaffected by weapons we might use against them may have no reason to concern themselves with this. The fact that their actions are harmful to us may carry no weight at all with them, without their being in any way ignorant or irrational. Certain successful criminals seem to have no reason to be at all concerned about their victims; they simply do not care, seemingly without being irrational or ignorant in not caring. Cynical politicians who are interested only in acquiring and maintaining power lie to the public without necessarily being uninformed, irrational, or stupid. Certain business executives successfully pursue their profits without concern for the rights of others with full information, rationality, and cleverness. In each of these cases, someone would seem not to have reasons to accept moral demands most of us take

RELATIVISTIC ETHICS: MORALITY AS POLITICS

41

to be important. Many other cases might also be cited, including cases of commonplace amorality as well as cases in which other people accept moralities that are radically different from ours. (See Essay 4.)

3.3 Hare's Response Such cases obviously pose a problem for the naive view. Various responses are possible. Consider for example, R. M. Hare's early theory (Hare, 1952, 1963). Recall that, in the naive view, morality involves one or more basic demands that everyone accepts or ought to accept as demands on everyone on which all moral reasons to do things depend; morality is, in other words, absolute, universal, and agent-centred. In his early writings, Hare was attracted to this naive position but also supposed that there was no single substantive set of moral demands that everyone had a reason to accept. His response was to abandon absolutism, while attempting to retain universality and the point of view of the agent. In his view, then, each person accepted certain basic demands as demands on everyone on which all moral reasons to do things depended, but the demands in question varied from one person to another. This meant that Hare took himself to accept certain moral demands as demands on everyone even though he also agreed that some people had no reason at all to act in accordance with those demands. This is puzzling. To accept a demand as a moral demand on George, among others, would seem to involve the thought that George ought to act in accordance with that demand—that there are moral reasons for him to do so. How, then, could Hare have also supposed that George may, after all, have had no reasons to act in accordance with that demand? The answer is that, in Hare's early view, one could accept a moral demand as a demand on everyone without supposing that everyone now had a reason to act in accordance with that demand. Hare argued that it was enough if one intended to try to get everyone else to accept that demand as a demand on him or herself. There were two things wrong with this. First, it did not seem to provide a reasonable interpretation of universality. It seemed that Hare could be only pretending to accept certain demands as demands on everyone. Rather, he would be accepting them merely as demands on himself which he intended to try to get everyone else to accept. Second, the intention to try to get everyone else to accept a given demand seemed itself quite unusual. It was unclear that many people ever had such intentions. So it

42

MORAL RELATIVISM

seemed unlikely that many people would accept moral demands in the way that Hare's early theory said they should. A related difficulty arose for Hare's analysis of the moral ‘ought’. In Hare's early view, ‘George ought not to eat meat’ was roughly equivalent to ‘George's not eating meat is in accordance with the general imperatives I hereby accept’. This allowed a vegetarian to judge that George ought not to eat meat even if the vegetarian also believed that George (who was not a vegetarian) had no reason not to eat meat. But this was surely to misuse the moral ‘ought’, which is normally used to speak of things an agent has moral reasons to do. Indeed, in each of its uses, the word ‘ought’ is used to speak of things for which someone has reasons. The epistemic ‘ought’, for example, is used to speak of things that there are reasons to expect, as when we say that the train ought to be here soon. The evaluative ‘ought’ is used to speak of what there are reasons to hope or wish for or take some other positive attitude towards, as when we say that there ought to be more love in the world or that a knife ought to be sharp. The simple ‘ought’ of rationality is used to speak of something for which an agent has reasons of any sort to do, as when we say that a burglar ought to wear gloves. And the moral ‘ought’ is used to speak of things an agent has moral reasons to do, as when we say that a burglar ought to reform and go straight. In elaborating this connection between the moral ‘ought’ and talk of reasons, I do not mean that the fact that an agent morally ought to do something is itself a moral reason to do it. That would be open to the objection that such a fact gives an agent a reason to do something only if he or she has some reason to care about what he or she ought morally to do. The issue here would be whether the fact that an agent has a reason to do something is itself a reason for him or her to do that thing. Similarly, imagine a dispute as to whether the fact that the train ought to be here soon is a reason to expect the train to be here soon. That would be to dispute whether the fact that there is a reason to expect the train to be here soon is itself a reason to expect the train to be here soon. This sort of technical issue is not directly relevant to our present concerns, and I will not try to resolve it.

3.4 Against Universality The problems mentioned for Hare's early view result from the attempt to retain universality and the point of view of the agent while abandoning

RELATIVISTIC ETHICS: MORALITY AS POLITICS

43

absolutism. Given the point of view of the agent, for any substantive moral demand D, there will be the following argument against the universality of D. (1) There is someone S who fails to adhere to D without being in any relevant way irrational, stupid, or uninformed. So, (2) S has no reason to adhere to D. So, (3) It is not true that S ought morally to adhere to D. We have been discussing the move from (2), ‘S has no reason,’ to (3), ‘It is not true that S ought.’To reject this step of the argument is, I think, to distort the meaning of the moral ‘ought’. Similarly, to reject the step from (1), ‘S can fail to adhere to D without being irrational, etc.,’ to (2), ‘S has no reason to adhere to D,’ is to break the connection between reasons and rationality, in my view thereby draining the term ‘reason’ of its ordinary meaning. It might be objected that, despite the truth of (1), S does have a reason to adhere to D, namely, the fact that he or she ought morally to do so. But, first, as we have seen, it is obscure whether such a fact itself gives S a reason to adhere to D. And, second, it begs the question to assume that this is a fact, because the argument purports to show that it is not a fact. To suppose that it is a fact that S ought morally to adhere to D is simply to suppose that S has a moral reason to adhere to D. But S has no such reason, if he or she can fail to adhere to D without being irrational, stupid, or uninformed.

3.4.1 Reasons A related objection (Frankena, 1976a; Holmes, 1973–4) is that (2), ‘S has no reason to adhere to D,’ is ambiguous. In this view, ‘reason forS to adhere to D’ might mean either of two things. It might mean ‘motivating reason forS to adhere to D’. Or it might mean ‘reason forS to think he or she ought to adhere to D’. The objection says (2) follows from (1) only on the first interpretation of ‘reason’. And (3) follows from (2) only on the second interpretation. Therefore, the argument rests upon equivocation in this view. In fact, the term ‘reason’ is not ambiguous in this way, and there is no equivocation in the argument. ‘Reason for S to adhere to D’ never means either ‘motivating reason’ or ‘reason for S to think he or she ought to

44

MORAL RELATIVISM

adhere to D’. The second interpretation gets things backwards, since ‘S ought to adhere to D’ means roughly that S has a reason to adhere to D. A reason for S to think he or she ought to adhere to D is not directly a reason for S to adhere to D. It is only a reason for S to think he or she has a reason to adhere to D. The question whether this is indirectly a reason for S to adhere to D is the same obscure technical issue previously mentioned as not directly to the point. On the other hand, ‘reason for S to adhere to D’ cannot ever mean motivating reason, since it is always consistent to suppose that such a reason is not a motivating reason because S is unaware of it, or is aware of it but is acting irrationally, and so on. So the objection is simply mistaken. A different objection distinguishes S's having reasons to do D from there being reasons for S to do D. According to this objection, for it to be true that S ought to do D, it is enough that there be reasons for S to do D. It is not required that S actually have those reasons to do D. (S might be quite unaware of the reasons.) So (3), ‘It is not true that S ought,’ does not follow from (2), ‘S has no reason,’ according to this objection. This raises a complex issue. For one thing, we must note an ambiguity in the remark that there are reasons for S to do D. This might refer to reasons for doing D that S has or would have if he or she was aware of them. It might also refer to reasons for our hoping or wanting S to do D. In this second sense, ‘There are reasons for S to do D’ is clearly not equivalent to ‘S ought morally to do D.’ It is equivalent only to ‘S ought to do D,’ meaning, ‘It ought to be the case that S does D.’ So only the first interpretation of ‘There are reasons for S to do D’ is relevant to our present concerns. Now, the distinction between there being reasons for S to do D, in this sense, and S's having reasons to do D, is a subtle and difficult one. I am not exactly sure what the distinction rests on. It is not simply a question of whether S is aware of the reasons, since S may have a reason of which he or she is unaware. The distinction seems to have something to do with how easy it would be for S to become aware of relevant facts or do the relevant reasoning, etc. If it is relatively easy, we might say S has reasons and, if it is less easy, we might say only that there are reasons for S. In any event, the distinction is obviously not crucial in the context of the argument (1)–(3). In premise (1), we suppose that S is fully informed, etc. In this context, then, to speak of the reasons there are is to speak of the reasons S has. If S has no reasons to do D, then in the relevant sense there are no reasons for S to do D. So, one cannot escape the argument (l)–(3) by distinguishing reasons forS to do D from reasons S has to do D.

RELATIVISTIC ETHICS: MORALITY AS POLITICS

45

3.4.2 Ought S to Care? Let me consider one other objection. According to this objection, S ought to care about certain things, whether or not S actually does care. For example, S ought to care about the welfare of other people, even if in fact S does not care. If S did care about others, that would give S a reason to treat others in certain ways. Since S does not care, S fails to have this reason to treat others in that way. But, S ought to treat others in that way, since S ought to care about their welfare, according to this objection. So, according to the objection, it can be true that S ought to do D even if S has no reason to do D. I believe this objection confuses different senses of ‘ought’. If S ought to care about the welfare of others in the sense that he or she has reasons to care, then these reasons are also reasons for S to act in the appropriate way towards others. But, it may be that S ought to care only in the sense that it ought to be the case that S cared; it would be a good thing if S cared. In that case all that follows is that it ought to be the case that S act in that way. It does not follow that S ought morally to act in that way. So this last objection fails too. Since the objections do not go through, it seems that the argument (1)–(3) does show that, if absolutism is denied, universality must also be denied, given the point of view of the agent.

3.5 Reconsidering Absolutism Now, there do not seem to be any very persuasive direct arguments in favour of absolutism. Someone might claim that it is part of the meaning of rational, as the term is ordinarily used, that someone is correctly called rational only if he or she accepts certain basic moral demands. But that is clearly false as a piece of empirical linguistics. There is also Thomas Nagel's (1970) ingenious argument, which seeks to show how the failure to accept certain moral demands involves a failure to appreciate the objective nature of reasons, a failure that amounts to ‘practical solipsism’. But Nagel does not show that any mistake or irrationality is involved in not taking reasons to be objective, in his sense; he does not show that people have reasons to avoid ‘practical solipsism’. So, he does not show that everyone has reasons to accept the relevant moral demands. (I say more about Nagel's argument in Essay 4.)

46

MORAL RELATIVISM

There is, however, a more promising indirect argument in favour of absolutism. Many people take morality to involve universality, so that basic moral demands are demands on everyone. But we have seen in (1)–(3) that the falsity of absolutism implies the falsity of universality, given the point of view of the agent. Putting this the other way round, from the point of view of the agent, universality implies absolutism. So, assuming the truth of universality, (l)–(3) can be used in reverse to argue for absolutism. Rather, we should say this: two plausible ideas are in conflict, the denial of absolutism, on one hand, and universality, on the other. It is difficult to believe in absolutism because it is difficult to believe that there are substantive moral demands that everyone has reasons to accept. But, if absolutism is false, then universality is false, and many people find that hard to believe, because they find it hard to believe that basic moral principles do not apply to everyone. More precisely, if absolutism is false, as seems likely, one common conception of morality is based on a false presupposition. Those who are attracted to that conception find themselves in the position of those who think morality is the law of God and then decide there is no God. Relativism implies that morality so understood is a delusion, a vain and chimerical notion. Fortunately, there is a reasonable substitute. They can reject absolutism and still have a good approximation of morality as conceived by many, if they modify their naive view by relativizing the range of ‘everyone’ throughout to those who accept or have reasons to accept certain basic moral demands. In this relativistic view, there are various moralities, each involving different basic demands, demands that certain people accept or have reasons to accept as demands on those same people, demands on which depend all reasons arising from that morality for those people to do things.

3.6 Morality as Conventional This is a natural view of morality to take if one supposes that morality rests on a tacit agreement or convention among a group of people. Other things being equal, the existence of an agreement or convention gives someone a reason to act in accordance with it only if he or she accepts the agreement or has reasons to do so. Furthermore, the hypothesis that moral demands derive from conventions arrived at through implicit bargaining

RELATIVISTIC ETHICS: MORALITY AS POLITICS

47

and mutual adjustment helps to account for the content of actual moralities. The fact that participants aim at provisions that are in their own interests helps to explain the rough utilitarian character of our own morality. As discussed in Essay 1, the fact that compromises are necessary among people of different powers and resources may account for the relative strictness of the duty to avoid harm to others as compared with the weaker duty to help those who need help. Everyone benefits equally from a duty not to harm others. But the poor and weak benefit much more from a duty of mutual aid than the rich and powerful would. The expected compromise is a fairly strong prohibition against harm and a weaker duty of mutual aid. That is what we have in our morality.

3.6.1 Trammel's Alternative Explanation It is difficult to see how else the distinction we make might be explained. Consider, for example, the suggestion made by Trammel (1975) that three factors are relevant. (1) You can avoid harming anyone but you cannot help all who would benefit from your help, since there are too many for you to help them all. (2) If you harm someone, that person is necessarily harmed; but, if you fail to help someone who needs help, that person is not necessarily unhelped, because someone else may help. (3) If you harm someone, you are necessarily responsible for that person's plight; but, if you fail to help someone who needs help, you are not necessarily responsible for that person's plight. These three factors are real enough, but they fail to explain the moral distinction in question. For (1) and (3) presuppose that very distinction and (2) is a purely verbal point. Let us begin with (2) If you harm someone, that person is necessarily harmed; but, if you fail to help someone who needs help, that person is not necessarily unhelped, because someone else may help. In order to see that (2) makes a purely verbal point, it is sufficient to observe that, if you act so that a particular person will be harmed unless someone else intervenes to save that person, that person is not necessarily harmed, because someone else may intervene. On the other hand, if you fail to help a person needing help whom no one else is going to help, that

48

MORAL RELATIVISM

person is necessarily unhelped. Now, almost any case in which someone violates the duty not to harm someone can he described either as a case of harming someone or as a case of acting in such a way that the person in question will be harmed unless someone intervenes to save that person. Given the one description, the person is necessarily harmed; given the other description of the same act, the person is not necessarily harmed. Similarly, almost any case in which someone violates what we take to be the weaker duty to help someone can be described either as a case of failing to help someone needing help or as a case of failing to help someone needing help whom no one else is going to help. Given the one description, the person is not necessarily unhelped; given the other description of the same act, the person is necessarily unhelped. We do not suppose, however, that the strength of these duties depends on how cases falling under them are described. Let us now consider Trammel's (1) You can avoid harming anyone but you cannot help all who would benefit from your help, since there are too many for you to help them all. To see that this presupposes that we distinguish harming and not helping as we now do, consider what would be true in a society in which people were as strongly motivated morally to help each other as not to harm each other. Such a society would be much more egalitarian and altruistic than ours. There would not be the vast numbers of people needing help in such a society that there are in our society, because the members of the imagined society would be quick to provide help whenever they perceived a need for it. In that society, (1) would no longer be true. You would be able to help all who would benefit from your help to the same extent that you could avoid harming anyone. Proposition (1) is true of our society only because our recognition of a moral distinction between harming and not helping is responsible for there being vast numbers of people who need help who do not get it. Let us turn now to Trammel's (3) If you harm someone, you are necessarily responsible for that person's plight; but, if you fail to help someone who needs help, you are not necessarily responsible for that person's plight. This is true of our society only because what someone is responsible for depends in part on that person's duties and obligations. In a more egalitarian and altruistic society, which did not recognize a moral distinction

RELATIVISTIC ETHICS: MORALITY AS POLITICS

49

between harming and not helping, you would be considered responsible for a person's plight if you failed to give the person help when it was needed. Trammel's account of our intuitive distinction is therefore no real alternative to a tacit agreement theory that explains the intuitive distinction as arising as a result of a compromise between people of varying powers and resources.

3.7 Senses of ‘Ought’ Moral argument can involve not only argument over the consequences of basic demands but also bargaining over the basic demands themselves. Morality is continuous with politics. Furthermore, a person may belong to a number of different groups with different moralities that may sometimes have conflicting implications. When that happens, a person must decide what side he or she wants to be on. As in Hare's analysis, a moral ‘ought’ judgement will be made in relation to moral demands accepted by the person making the judgement. But, since such a judgement says that a certain agent has reasons to do something, the judgement presupposes that the agent also accepts the relevant moral demands, or at least has reasons to accept them. ‘George morally ought not to eat meat’ means roughly that George's not eating meat is in accordance with the moral demands that ‘we’ accept, where ‘we’ includes the speaker, George, and the intended audience, if any. Moral ‘ought’ judgements are therefore a species of what I call ‘inner judgements’. Such judgements are made only about those who are assumed to accept (or have reasons to accept) the moral demands on which the judgements are based. Inner moral judgements include not only moral ‘ought’ judgements but also any other moral judgements that attribute moral reasons to someone—for example, the judgement that it was morally wrong of someone to have done a certain thing. Although all moral judgements are, in this view, relative judgements made in relation to certain moral demands, not all moral judgements are inner judgements in this sense. One can judge that certain outsiders are good or bad or evil from the point of view of one's morality, even if they do not share that morality, just as one can judge that outsiders are friends or enemies. Similarly, one can judge that certain situations are good or bad or right or wrong or that they ought or ought not to be the case, even if these situations involve agents that do not participate in the morality relative to which the judgement is made.

50

MORAL RELATIVISM

As noted in Essay 1, it is important to distinguish the moral ‘ought’, which is used to make inner moral judgements about agents, from the evaluative ‘ought’, which is used to make non-inner judgements about situations. The sentence, ‘George ought not to eat meat,’ is ambiguous. It may be used to express an inner moral judgement of George, which presupposes that he accepts or has reasons to accept the relevant moral demands. It may instead be used to express a favourable evaluation of the possible state of affairs in which George does not eat meat. Such an evaluation does not imply or assume that George himself has any reason to avoid meat. It implies only that we, with our standards, have a reason to hope or wish that George will not eat meat (and therefore perhaps a reason to try to bring it about that George does not eat meat). The first interpretation, as an inner judgement, is more natural if we say George ought morally not to eat meat. The second interpretation, as a non-inner evaluation, is more natural if we say meat ought not to be eaten by George or it ought not to be the case that George eats meat.

3.8 Universalizability Now, philosophers commonly say that moral ‘ought’ judgements are ‘universalizable’, by which they mean in part this: If you judge of someone that he or she ought morally to do a particular thing, then you are committed to the further judgement that anyone else in similar circumstances ought morally to do the same thing. This does not hold in the relativistic conception of morality I have been describing (section 2.9 of Essay 2). A moral ought judgement about someone who accepts, or has reasons to accept, the relevant moral demands has no implications about anyone who neither accepts nor has reasons to accept those demands. A useful special case does hold, however: If you judge of someone that he or she ought morally to do a particular thing, then you are committed to the further judgement that in similar circumstances you ought to do the same thing. This means, in particular, that you cannot complain about him or her unless you agree that if the situation were reversed, yon ought not to act as he or she is acting.

RELATIVISTIC ETHICS: MORALITY AS POLITICS

51

Moral ‘ought’ judgements are also often said to be universalizable in another respect, which is supposed to rule out arbitrary discrimination: If you ought to treat someone in a particular way, you ought to treat any other person in the same way unless there is a morally relevant difference between them. This principle can be counted trivially true, since any difference in the way you ought morally to treat people can be counted as a ‘morally relevant’ difference. Still, the spirit of the principle can be violated in a relativistic conception of morality if the distinction between ‘us’ and ‘them’ is taken to be ‘morally relevant’. In fact, many moralities afford less protection for outsiders than for the primary members of the group, without basing this distinction on any further ground. A morality is, of course, not forced to discriminate in this particular way, but almost every morality does so to some extent.

3.9 Substitutes for the Naive View Of course, a morality may be the product of a tacit convention or agreement without participants realizing it. Participants may in fact accept the naive view, with its presupposition of absolutism. I do not wish to say that the relativistic theory I have been describing, morality as politics, captures everyone's conception of morality. My claim is, first, that morality as conceived by the naive view does not exist. Second, the moralities that do exist derive from convention or tacit agreement. And third, morality as politics is an acceptable substitute for the naive conception, and indeed is the most acceptable substitute for it. Notice that in this conception of morality, which I am calling ‘morality as politics’, a morality is basically a group affair, depending on moral demands jointly accepted by several people after a certain amount of tacit bargaining and adjustment. We might allow also for a purely personal morality in which someone places demands only upon himself or herself. But that is a special and limiting case.

3.9.1 Existentialism and Emotivism There are other possible substitutes for naive morality, substitutes that take a morality to be essentially an individual, personal matter, depending only on the moral demands accepted by a single person. Such conceptions

52

MORAL RELATIVISM

tend to be either extremely agent-centred, as in one form of existentialism, or extremely critic-centred, as in Hare (1952, 1963) and in one version of emotivism (Sartre, 1973). In an extremely agent-centred conception of morality, someone accepts moral demands only as demands on himself or herself. He or she is therefore able to judge his or her own actions morally, but not those of anyone else. In an extremely critic-centred conception, like Hare's, someone accepts certain demands as demands that can be used in morally judging anyone's actions, including those who have no reasons to act in accordance with those demands. The same is true of that version of emotivism which takes moral ‘ought’ judgements merely to express one's favourable approval of someone's doing something (Ayer, 1936)—obviously one can be in favour of someone's doing a particular thing without supposing that he or she has any reasons to do it. Such a conception of morality is best thought of as abandoning the moral ought to do, used to speak of an agent's reasons, in favour of the evaluative ought to be, used to evaluate possible states of affairs. It matters a great deal in practice which substitute for naive morality is adopted. The acceptance of an extremely agentcentred conception of morality would make moral discourse among different people virtually impossible. The acceptance of an extremely critic-centred conception, on the other hand, would tend to lead to the acceptance of unrealistic normative theories like one or another version of utilitarianism or hypothetical contract theory.

3.9.2 Act Utilitarianism Consider act utilitarianism, which says that everyone ought always to do whatever promises to have the best consequences; in other words, everyone ought always to act so as to maximize utility (Smart, 1961). This view has many implausible consequences. It would have you break your solemn promise whenever that would have better consequences than keeping your promise. It forbids you to favour friends and relatives over perfect strangers in distributing the benefits of your actions. It implies that almost everything you do is morally wrong, since there is almost always an alternative action open to you that promises to have better consequences. It treats your failure to help someone as morally equivalent to actually harming someone. One objection to act utilitarianism is that it has such implausible consequences. Another perhaps more revealing objection is that people are not only motivated to maximize utility, but also have other motives which

RELATIVISTIC ETHICS: MORALITY AS POLITICS

53

they take to be morally respectable, in some cases even morally required, without being irrational, stupid, or uninformed. Act utilitarianism is therefore quite unrealistic as a theory about the moral reasons that people actually have. So, it is unrealistic as an account of what people ought, morally, to do. This lack of realism arises, I suggest, from a failure to distinguish the moral ought to do from the evaluative ought to be. Act utilitarianism makes some sense as an account of what ought to be the case. Act utilitarianism makes no sense as an account of what people ought morally to do. It ought to be the case that utility was maximized; that would be wonderful! It does not follow that a person has moral reasons only to maximize expected utility and do nothing else. Now, an extremely critic-centred theory, like emotivism, can make an unrealistic moral theory like act utilitarianism seem quite plausible. For, it may well be that, as critics, we would always be in favour of utility being maximized. This does not mean, however, that we do or should suppose that a person always has moral reasons only to maximize utility.

3.9.3 Rule Utilitarianism Similar remarks apply to other versions of utilitarianism. Ideal rule utilitarianism, for example, says that you should always follow the best rules, namely the rules it would maximize utility for everyone to follow. Philosophers disagree as to whether this version has more plausible implications concerning what people ought to do than act utilitarianism does. But that disagreement is unimportant, since it is anyway obvious that no one has any reason to act as ideal rule utilitarianism recommends. To follow the best rules because it would maximize utility for everyone to follow them is to overlook the fact that other people will not in fact follow those rules. This is the Pacifist Fallacy. Pacifism says that one should never use force against other people, even to defend oneself against a direct attack, because (a) the world would be a much better place if no one ever used force against others, and (b) if no one ever used force against other people, there would be no need to use force to defend oneself against a direct attack by someone else, since no such attack would ever occur. This overlooks the obvious point that in the real world people do sometimes attack other people and, in such a case, the person attacked often has a reason to use force to repel that attack. In endorsing this sort of argument

54

MORAL RELATIVISM

for pacifism, ideal rule utilitarianism confuses how one should act in an ideal world in which everyone followed the best rules, with how one should act in the actual world in which people do not follow those rules. There is a similar lack of realism in Kant's (1785) suggestion that you ought always to act as if you were a member of a Kingdom of Ends in which everyone always did the right thing. The fact that people often do not do the right thing is something that you cannot ignore in deciding what you ought morally to do. The same point applies to hypothetical contract theories that say one ought to act in accordance with the rules that would be accepted under certain ideal conditions rather than the rules that have actually been accepted in the real world. Like act utilitarianism, ideal rule utilitarianism is quite unrealistic as an account of what people ought morally to do—that is, as an account of what they have moral reasons to do—and is more plausible instead as an account of something that would be desirable—that is, as an account of what ought to be the case. It would be wonderful if everyone always acted on the best set of rules—perhaps it ought to be the case that everyone so acted. It does not follow that anyone now has any reason to act in that way. Ideal rule utilitarianism, like act utilitarianism, is critic-centred rather than agent-centred, although there is a difference in what is being judged in the two cases. Whereas a simple form of emotivism can make act utilitarianism look plausible, Hare's theory, with its Kantian overtones, can make ideal rule utilitarianism look plausible. In the first view, what is taken to be relevant is one's attitude towards a particular agent's doing something given the actual circumstances of the case. In Hare's view, what is relevant is one's attitude towards universal adherence to one or another set of principles. Philosophers have sometimes argued that act utilitarianism and ideal rule utilitarianism must coincide (Lyons, 1965)—that they are ‘extensionally equivalent’. The point is that, if everyone were to follow the best set of rules, each person would presumably also always be acting so as to maximize expected utility. What that shows, however, is that act utilitarianism and ideal rule utilitarianism would be ‘extensionally equivalent’ in an ideal world in which everyone acted on the best set of rules. It does not show that these views are ‘extensionally equivalent’ in the actual world. And clearly they are not, since ideal rule utilitarianism recommends total pacifism, even in the real world, whereas act utilitarianism recommends against total pacifism in the real world, on the grounds that there are times when using force has better consequences than not using it. Other versions of utilitarianism can be mentioned. Actual rule utilitarianism

RELATIVISTIC ETHICS: MORALITY AS POLITICS

55

says that a person ought to follow the actual rules accepted by the members of his or her group and that these rules are themselves to be assessed on utilitarian grounds. As an account of what people ought morally to do, this seems more realistic but not completely right, since you should sometimes not follow the rules everyone else follows. This suggests a variant of actual rule utilitarianism: a person ought to follow those actual rules which are justified on utilitarian grounds. This leaves a number of questions. What is it for a rule to be justified on utilitarian grounds? Different views are possible. It might be required that the rules have the best possible consequences, or it might be required only that its consequences be better than those of having no rule at all. What if the relevant actual rule fails this test? What should you do then? Again, there is a range of options. You might be left free to do what you want. Or you might be required in this case to maximize expected utility. Or you might be required to act on the best rule, from a utilitarian point of view. We might try to put various pieces together so as to obtain, almost by brute force, the utilitarian theory with the most plausible implications for what one ought morally to do. But this would hardly be an illuminating exercise. The complications in such a theory would make it difficult to see what implications the theory really had concerning what people ought morally to do. It is likely that the most that could really be said for such a theory would be that there were no clear counter-examples to it. Furthermore, the theory would almost certainly fail to say why anyone would have reasons to act in the way it said they had reasons to act.

3.10 Morality as Politics Consider, by contrast, how easily the conception I have been calling morality as politics can account for utilitarian aspects of the moral reasons people actually have. As we have already noted, if a morality is based on a tacit agreement, it is to be expected that the rules of such a morality will be of some benefit to the people involved. Why else would they accept those rules? This, of course, does not mean that the rules actually accepted are ideal from a utilitarian point of view or even that all the rules accepted really do have good consequences. Compare morality as politics with actual rule utilitarianism. Both say that moral reasons derive from actual rules or principles. But there are important differences. A number of difficulties facing actual rule utilitarianism are easily resolved in the conception of morality as politics. Actual

56

MORAL RELATIVISM

rule utilitarianism says that you should follow the actual rules—the rules that people actually follow. But which people? Different people follow different rules. Which rules should you follow? It might be said that you should follow the principles of your society—not those of some other society. But what makes a society your society? Won't it turn out that there are a number of different groups to which you belong, with different principles? Should you adhere to all of these principles? What if they conflict? What rules do you follow then? And, anyway, why should you follow certain rules simply because other people do? According to the conception of morality as politics, the principles that give you moral reasons to do things are the moral principles that you actually accept. You accept them as principles also accepted by the members of some more or less well-defined group. What makes a society your society is not, say, the proximity of its other members, but rather that you accept the principles of that society as principles for you and other members of that society. You are indeed, in the relevant sense, a member of different groups with different principles: your family, friends, neighbours, colleagues at work, and so on. Since you accept all of the principles of all of these groups, all of the principles give you reasons to do things. This is not to say that once you accept certain principles you must blindly adhere to them ever after. Your acceptance of certain principles can give you reasons to give up or modify other principles you accept. Your other goals and interests can have the same effect. The principles you accept may conflict. In that case, there is typically no easy answer as to what you ought to do. You will have to choose, reasoning somewhat as in any other case in which prior goals, intentions, or principles conflict, making a minimal change in your antecedent principles, etc., which will promote the coherence of your resulting overall view. What reason do you have to follow the relevant principles? You accept them; and your acceptance of them gives you reason to follow them. What reasons did you have to accept the principles in the first place? Reasons of various sorts, for example, reasons of self-interest: for, if you accept certain principles as governing your dealings with others, those others will tend to accept the principles as governing their dealings with you. And, as previously mentioned, once you accept certain moral principles, that may give you reasons to accept other principles. I conclude that morality as politics provides the most reasonable substitute for the naive conception of morality. Recall that in what I am calling the naive view morality is based on certain moral demands which everyone accepts, or at least has reasons to accept, as demands on everyone

RELATIVISTIC ETHICS: MORALITY AS POLITICS

57

and on which all moral reasons depend. Although this seems to capture essential aspects of many people's conception of morality, I have suggested that it must almost certainly be rejected on the grounds that there are no substantive moral demands satisfying those conditions. Morality as politics retains much of the content of the naive view but restricts the references to ‘everyone’ to the members of one or another group, all of whom accept certain moral demands as demands on themselves and other members of the group. I have argued that other conceptions of morality retain less of the content of the naive view and tend toward unrealistic accounts of what morality requires. Extremely agent-centred theories, as in certain forms of existentialism, practically abandon morality as a social enterprise. This is also, though less obviously, true of extremely critic-centred theories, like certain forms of emotivism and like Hare's theory, which are best seen as rejecting the moral ‘ought’ in favour of the might of evaluation. This is, I have suggested, why these theories can make some very unrealistic accounts of what morality requires seem plausible, such as one or another form of utilitarianism.

4 Justice and Moral Bargaining 4.1 Introduction In my view, although there may be universal formal principles of justice (Perelman, 1963; Essay 13, below), substantial principles of justice derive from continual tacit bargaining and adjustment. This is not to say social arrangements are just whenever they are in accordance with the principles of justice accepted in that society. We can use our own principles of justice in judging the institutions of another society, and we can appeal to some principles we accept in order to criticize other principles we accept. To use Hume's (1739) model of the relevant sort of convention, two people rowing a boat continually adjust their rates of rowing to each other so that they come to row at the same rate, a rate that is normally somewhere between the rate at which each would prefer to row. In the same way, the basic principles of justice accepted by people of different powers and resources are the result of a continually changing compromise affecting such things as the relative importance attached to helping others as compared with the importance attached to not harming others. Hume's rowers provide an example of a ‘convention’ that is normally completely tacit. There are other models in which the bargaining can be more explicit, for example, when a seller comes to set prices that are acceptable to customers, when employers reach understandings with employees concerning wages, or when political groups influence legislation. I want eventually to consider the implications for moral reasoning and argument of the thesis that principles of justice result from implicit bargaining and convention in this way. But first, let us review the reasons for thinking that justice is, indeed, conventional.

4.2 Is There a Single True Morality? A first point is that it seems extremely unlikely that there is any substantive moral demand which all rational agents have sufficient reasons to

JUSTICE AND MORAL BARGAINING

59

accept. This seems unlikely because, for any given moral demand, it is possible to find people who do not seem to have sufficient reasons to accept that demand. The people in question may be either professional criminals, or simple egoists, or adherents to a competing moral principle. No matter what the moral demand, there will be people of this or some other sort who fail to accept that demand, whose failure does not seem to be the result of their having overlooked reasons they have to accept that demand. That is, it does not seem that they had to accept the demand in question because they have overlooked or not understood certain reasoning, nor because they are in some way irrational or ignorant of relevant facts, nor because weakness of will leads them to do something they think they really should not do.

4.3 Neutral Versus Normative Reasons It might be said that the people in question are after all overlooking something, namely, the fact that what they are doing is morally wrong. It might be argued that, if they were to realize it is wrong not to follow the moral demand in question, and were to fully understand this, and were not subject to weakness of will, then they would act in accordance with the moral demand. This strikes me as an unsatisfactory and question-begging response. Following a terminological suggestion of Mark Johnston's (personal communication), let us distinguish two conceptions of moral reasons, two senses in which someone might be said to have moral reasons, a ‘neutral’ conception of moral reasons and an ‘evaluative’ conception. In either conception, to say an agent has sufficient moral reasons to do something is to make an evaluative judgement about that person and to express one's moral approval of the agent's doing it. In the neutral conception of moral reasons, it is also to imply that, if the agent doesn't do it, he or she either lacks certain nonmoral information, or fails to appreciate certain reasoning, or has a nonmoral defect like weakness of will or some other sort of nonmoral irrationality. On the other hand, in the evaluative conception of moral reasons, to say an agent has sufficient moral reasons to do something is merely to make a moral evaluation of the agent's doing it, where this allows that the agent could fail to do it without lacking nonmoral knowledge, without failing to appreciate any reasoning (in a neutral sense of ‘appreciate’), and without being subject to any other sort of nonmoral defect.

60

MORAL RELATIVISM

Observe that the evaluative conception of moral reasons is compatible with a purely emotivist analysis of judgements about such reasons. According to such an analysis, for a speaker to judge that an agent has sufficient moral reasons to do something is simply for the speaker to express approval of the agent's doing it. Since the considerations that move the speaker to approve of the agent's following certain rules might be considerations that carry no weight at all with the agent, it is possible for an emotivist to claim the agent has sufficient moral reasons to follow those rules while agreeing there may be no considerations that carry weight with the agent favouring the agent's doing so. Now, when I say that it seems likely there is no single set of moral demands all rational agents have sufficient reasons to follow, I mean, of course, reasons in the neutral sense, reasons of a sort that carry weight with the agent, or would carry weight apart from nonmoral defects in the agent, failure to follow or appreciate (in a neutral sense of this term) some argument or other reasoning, and lack of nonmoral information. I see no point in talking about reasons in the evaluative sense. Why should an agent care about what the speaker takes to be reasons for the agent to accept a given rule, if the relevant considerations carry no weight with the agent? True, the speaker will approve of the agent's caring about these things and so can correctly say the agent ‘should’ care, given an emotivist account of ‘should’. But it is no answer to the question the agent would ask, ‘Why should I care?’ that the speaker cares. And to allow no other account of this question is simply to try to define a legitimate question out of existence. In my opinion, this represents a reductio ad absurdum of an emotivist account of what it is to say another person has sufficient reasons to do something. Similar remarks apply to Nozick's discussion of the ‘value sanction’ (1981: 409–11). Nozick says that an immoral person is ‘worse off ’ than a moral person because his life is not as ‘valuable’. A speaker can believe that another person is worse off in this sense without the relevant considerations carrying any weight with the other person.

4.4 Moral Values It also seems likely that there are no nontrivial universal moral values. It seems unlikely that there is anything of a very substantive sort that everyone values morally, or has reasons to value morally, even if we restrict our consideration to people who have some clearly moral values. It may or

JUSTICE AND MORAL BARGAINING

61

may not be true that all people have a capacity for sympathy with others, and even with animals, which leads to an instinctive or automatic reaction against suffering in people and animals. But this by itself is not to value the absence of suffering, certainly not to value it morally. And even when the absence of suffering is valued morally, it may be assigned various degrees of importance. One's moral values depend on the morality one accepts. Different people attach different relative weights to the things they value. During the war in Vietnam, some people were more upset about the bombing of ancient temples than about the bombing of civilian populations, others were more upset about the bombing of civilians. It is implausible to suppose this difference reflects, say, a difference in the calculation of the human happiness that would be increased as the result of the preservation of the ancient temples. It almost certainly reflects a difference in basic moral values. Similarly, people differ in the extent to which they value individuality and freedom as compared with community spirit and a harmonious society. And this affects the way in which they evaluate their own and other social systems.

4.5 Justice Judgements about the justice of a particular transaction or arrangement are evaluative judgements, and justice seems to be relative in the way that other values are. It may be that everyone has some innate inclination towards ‘fairness’, but it seems likely that, in the absence of actual bargaining, agreement is possible only on a formal conception of justice, not on a substantive conception. There are many competing ideals of justice: as rewarding virtue and punishing vice, as treating people equally, as protecting property and entitlement, as ensuring that basic needs are satisfied, and so on. Many people feel the appeal of several of these ideals but value them differently. I do not mean merely that different societies have different arrangements. I mean that people give different weights to various considerations in judging, say, Chinese society in the 1970s. For some, equality is more important than liberty or entitlement; for others, the reverse is true. Such disagreements would seem to reflect different basic values. It is hard to believe that in the nature of things one side is right about this and the other is wrong. It is hard to believe there is reasoning which, once appreciated, makes all but one weighting of these values irrational.

62

MORAL RELATIVISM

4.6 Morality as a Commitment of a Rational Agency Now, philosophers occasionally argue that there are certain basic moral principles to which all rational agents are committed. Kant (1785) famously argues in this way for his ‘categorical imperative’, and so does Gewirth (1978). But none of these arguments succeed. A typical argument of this sort asks you to consider a given situation from another person's point of view with the aim of showing you that you have a reason to take the other's interests into account in the present unreversed situation. To suppose you do not have sufficient reason to take the other's interests into account in this situation is to suppose the other would not have sufficient reason to take your interests into account in the reversed situation. The argument is that, since you cannot accept the other's not taking your interests into account in the reversed situation, you would be inconsistent not to take the other's interests into account in the actual situation. But this depends upon an equivocation concerning what it is to ‘accept’ a person's doing something, which may be either (a) to want the person to do it (or at least not want the person not to do it) or (b) to think the person has sufficient reasons to do it. The sense in which you cannot ‘accept’ the other person's ignoring your interests is sense (a): you do not want the other person to ignore your interests. From your point of view it ought not to be the case that the other person ignores your interests. It does not follow that you cannot ‘accept’ the other person's ignoring your interests in sense (b). You may very well realize that from the other person's point of view it ought to be the case that he or she ignores your interests. You may see that the other person, indeed, has sufficient reasons to ignore your interests. According to Gewirth (1978), you are committed as a rational agent to thinking of your freedom and well-being as ‘necessary goods’. So, he argues, you are committed to supposing others must not interfere with your obtaining and possessing freedom and well-being. And this, Gewirth claims, is to suppose you have a right to freedom and well-being simply because you are a rational agent. But, he concludes, that supposition commits you to allowing that all other rational agents have the same right and, therefore, that you must not interfere with their obtaining and possessing freedom and well being. Gewirth's argument can be interpreted in many different ways; but I see no way to interpret it that avoids equivocation. For example, it might be

JUSTICE AND MORAL BARGAINING

63

interpreted so that the equivocation occurs in the second step of this argument. The supposition that others must not interfere with your obtaining and possessing freedom and well-being is ambiguous. It can mean either (1) it must not happen that they interfere or (2) there is a requirement on them, giving them reasons not to interfere. The supposition that your freedom and well-being are necessary goods to you as an agent provides grounds for concluding that others must not interfere with you only in sense (1): it must not be the case (from your point of view) that they interfere with you. This supposition does not logically imply that there is any requirement on them, giving them compelling reasons not to interfere with you. That is sense (2). But sense (2) of the supposition that they must not interfere is what is needed to get to the supposition that you have a right to such noninterference and, therefore, that others have a similar right. The supposition that you have such a right does not follow simply from the thought that from your point of view they must not interfere. Alternatively, the equivocation might be located at a later point in the notion of a right. We might say that you have a ‘right’ to noninterference in the sense of a ‘justified claim’ to noninterference because your self-interest justifies you in making such a claim. Then others have a similar right; they have a self-interested justification for a similar claim that they not be interfered with. However, it still does not follow that you have any reason to accept their claim in the sense of being motivated not to interfere with them. That follows only from a different notion of ‘right’. McMahon (1986) gives an elegant account of various other ways to understand Gewirth's argument, all involving some sort of equivocation or other error.

4.7 Defective Agents Some philosophers and psychologists claim that anyone who fails to accept certain basic moral requirements can be shown to be defective as an agent, in a sense of ‘defective’ that does not mean simply ‘morally defective’. For example, Nagel (1970) argues that to think you have no reasons to try to satisfy the interests of others is to be a ‘practical solipsist’, because you deny the reality of others in your practical reasoning even if you accept their reality in your theoretical reasoning. And Kohlberg (1981) argues for stages of moral development leading to a sixth and highest stage at which people agree on basic moral principles. According to Kohlberg, people who do not yet accept these principles have a defective

64

MORAL RELATIVISM

conception of morality. But neither Nagel nor Kohlberg makes a good case.

4.7.1 Nagel and the Avoidance of Practical Solipsism Nagel argues that, as a rational person who is not a solipsist, you must acknowledge that a judgement in the first person, like ‘I am five feet nine inches tall’, has the same content as the corresponding judgement in the third person, ‘That person is five feet nine inches tall’, except for a difference in perspective. Nagel argues further that your judgement that you have certain reasons to do or want something has ‘motivational content’ in that such a judgement is capable of motivating you to do the act in question if no other reasons are relevant. He concludes that you are committed to supposing that the corresponding third-person judgement has motivational content, too; otherwise you would be a ‘practical solipsist’. But to be committed to taking third-person judgements of this sort to have motivational content is, according to Nagel, to be committed to being motivated to help others do things they have reasons to do or want. This argument depends crucially on what we might call the ‘objectivity principle’, namely, that first- and third-person judgements have the same content, apart from considerations of perspective. Nagel argues for this principle by noting that it seems to hold for certain judgements, and by observing that certain violations of the principle are associated with solipsism, namely, cases in which you take judgements about your own feelings to have an entirely different content from judgements about the feelings of others. However, this leaves obscure exactly what is to count as part of the ‘content’ of a judgement for the purposes of the objectivity principle. An analogous argument with respect to pains yields the following bizarre result. We might say that first-person judgements about pains have ‘affective content’, in the sense that if you sincerely judge you are in pain, you normally are in pain. By Nagel's objectivity principle, your third-person judgements should have a similar affective content, so that you can only judge that another person is in pain if it hurts you too. That is clearly absurd. So, the objectivity principle cannot apply to ‘affective content’. But then some further argument is needed to show it should be applied to ‘motivational content’. Since he does not offer any such argument, Nagel does not show that there is anything defective about the practical reasoning of an agent who fails to take the interests of others as providing him or her with reasons.

JUSTICE AND MORAL BARGAINING

65

4.7.2 Kohlberg's Stages of Moral Development Kohlberg's argument combines empirical and philosophical considerations. His theory is that there are six stages of moral development which come in pairs, two ‘preconventional’ stages, two ‘conventional’ stages, and two ‘postconventional’ stages. In the first stage, a child takes doing what is right to be a matter of obeying authority, avoiding punishment, and not harming others. In the second stage, doing what is right also includes making fair deals in terms of concrete exchange and satisfying immediate needs. The third and fourth stages are centred on socially approved virtue and character (third stage), or social rules and principles (fourth stage). At the end of the fourth stage, people begin to think about the whole system of societal morality and its justification. Some people become and remain sceptical or relativistic at this point; others ‘progress’ to the post-conventional fifth stage in which they take people to have rights that are prior to society. Finally, at the sixth stage people come to ‘realize’ that there are universal ethical principles all humanity ought to follow—principles of respect for other persons as ends in themselves. Social rules are judged by considering whether one would find them acceptable if one did not know one's place in society. The principles accepted at the sixth stage are similar to those defended in Rawls (1971). Kohlberg hypothesizes that moral development from one stage to the next occurs as someone acquires a more adequate conception of morality. He suggests that Stage VI is the highest stage reached because it represents the most adequate conception of morality. However, the evidence does not support such a strong hypothesis. Obviously, the key issue concerns what happens after Stage IV, because the development from Stage I (obedience to authority) to Stage IV (acceptance of the authority of the moral rules of your society) involves nothing more than coming to appreciate conventional morality. But Kohlberg's data do not indicate a single direction of development from Stage IV on. Development from that point occurs when people raise questions about the rules of their own society and become aware of alternative possibilities. Kohlberg's data (and common observation as well) show that at this point people in fact go off in various directions. Many people come to accept something like the egoism of the second stage (leading Kohlberg to speak of a Stage 4½). Some people become moral relativists. Some people become utilitarians, taking morality to be concerned with the promotion of the general happiness. Some people decide moral rules are binding only on those who consent to them. Some people try to reach their own arrangements with a ‘counter-culture’.

66

MORAL RELATIVISM

And there are many other options. Kohlberg's fifth stage is not specially privileged. Indeed, it is not a single option at all, but a collection of some of the possibilities. For example, it includes both utilitarianism and also the competing view that moral principles are binding only on those who consent to those principles. Finally, Kohlberg's sixth stage simply seems to represent the sort of thinking Kohlberg himself thinks is most adequate. Nothing in Kohlberg's data supports a privileged position for his fifth and sixth stages.

4.8 Reasons and Morality So, it seems to me that there is no single set of basic moral principles everyone has reasons to follow. From this I infer that there are no basic moral principles that apply to everyone. For it seems to me that to say a moral principle applies to someone is to say he or she ought to follow the principle, which is to say he or she has sufficient reason to follow it. This marks a conspicuous difference between law and morality. A law applies to someone if that person satisfies the conditions of application specified in the relevant legal system, quite apart from whether or not the person ought (i.e. has sufficient reason) to do what the law requires. But morality is not like that. To think a moral principle applies to someone is to think he or she ought morally to follow that principle, and that is to think he or she has sufficient moral reasons to follow it. So, if there are no basic moral principles that everyone has sufficient reasons to follow, then there can be no single morality that applies to everyone, no single basic set of moral values, no uniquely correct principles of justice.

4.9 Hypothetical Agreement Theories The basic moral principles that apply to a given person are principles that person actually accepts as principles regulating the conduct of members of some group to which the person belongs, where these are principles arrived at through a process of mutual adjustment and implicit bargaining. Ignore the special case in which the person is the only member of the relevant group, when personal ideals take on a moral character. The point is that the relevant principles are the principles that actually are accepted, not those that would be agreed to under certain conditions of ‘initial

JUSTICE AND MORAL BARGAINING

67

equality’ (Rawls, 1971) or ‘constrained neutral dialogue’ (Ackerman, 1980). One problem with a hypothetical agreement theory might be that there is no reason to expect agreement to be reached under the stated conditions, or at least no reason to think a unique agreement would be reached under those conditions. Another problem is that, even if a unique agreement would be reached under the specified hypothetical conditions, that by itself would not give someone in actual conditions any reason to follow the rules that would be agreed to. Since to say someone ought to follow certain rules is to say he or she has sufficient reason to follow those rules, the fact that certain rules would be agreed to under those conditions does not imply anyone ought to follow those rules. A qualification is needed here. It may be possible to give a hypothetical, or conditional, analysis of what it is to have sufficient reasons to do something. According to one such analysis, having sufficient reasons to D consists in the situation's being such that one would D if one were fully informed, fully rational, aware of relevant arguments and able to follow and understand these arguments, not suffering from weakness of will, and so forth. Given this analysis, the fact that one would accept certain rules under those conditions does imply one ought to accept those rules. However, there is still the first sort of problem for hypothetical agreement theories. There are no substantive moral principles everyone would accept under such conditions. For any principle one might choose, it is possible to find someone who does not accept it where this does not seem to be because that person is ignorant of something, irrational in some way, unable to follow some argument, suffering from weakness of will, and so forth. There are no moral principles that everyone has sufficient reasons to accept. Something else should be noted here. I said above that the moral principles that apply to a person are the principles the person actually accepts. That is not quite right. The principles a person accepts are the principles that give a person reasons to act in various ways. But this does not mean a person always has sufficient reasons to act in accordance with the principles he or she accepts. Nor does it mean that what a person has sufficient moral reasons to do always falls under some antecedently accepted moral principle. Some of the principles a person accepts may conflict with others and/or with nonmoral goals and policies of the person in question. In the light of such a conflict, it may be that what the person ought to do, all things considered, is to modify some of the moral principles he or she accepts. The moral principles a person accepts ‘apply’ to that person only

68

MORAL RELATIVISM

in the sense that they are the source of that person's moral reasons to do things (Harman, 1976, 1995).

4.10 Moral Conventions Suppose I am right about all this. Suppose there is no single true morality all people have reasons to accept. Suppose the source of moral reasons lies not in reason itself or in universal human nature but in the particular variable principles a person happens to accept. This raises the question what reason anyone has to accept any particular moral principles. There are various reasons, but the main reason why a person accepts the principles he or she does is that it is in his or her interest to do so if others do too. I accept a policy of acting in a certain way towards you so that you will accept that policy in acting towards me. I don't push you around so that you won't push me around. You are nice to me so that I will be nice to you. In other words, we accept the moral principles we do, as useful conventions. This is not to say that people are always aware of the conventional nature of moral principles. The conventions are normally arrived at tacitly, by mutual adjustment of different people's behaviour, without conscious awareness. Nor is it to say that the appeal to self-interest is the only appeal. Once some moral principles and values are accepted, these restrict further appeals to self-interest. But self-interest never becomes completely irrelevant. People can always opt out of the morality they currently accept; and they can always threaten to opt out if their self-interest is not sufficiently taken into account by the current morality. As emphasized in earlier essays, one can begin to appreciate the conventional self-interested source of principles of justice by considering otherwise puzzling aspects of ordinary moral views. For example, the duty not to initiate harm to others is usually taken to be more important and more stringent than the duty to help others avoid the same sort of harm (Foot, 1978, Thomson, 1978). From a utilitarian point of view, this is puzzling, because the same utility is lost when one fails to help someone avoid harm as when one initiates the harm oneself. And, although there have been ingenious attempts to try to give a nonbargaining rationale for the different strengths of these duties, none of these attempts has been successful (see Essay 3, above). In Essay 1, I suggest that the difference in strengths in these duties is a

JUSTICE AND MORAL BARGAINING

69

consequence of the fact that morality represents a compromise between people of different powers and resources, since everyone benefits from a strict duty not to harm others, but only the poor and weak would benefit from an equally strict duty to help others avoid harm. The rich and powerful would not benefit from such a strict duty of mutual aid because, given their resources, they would be called upon to do most of the helping and would have little need of such help for themselves. The rich and powerful would benefit from a strict duty not to harm others, however, since they could be harmed if the weak and poor were to organize against them. Therefore, although everyone will be in favour of a strict duty not to initiate harm against others, an equally strict duty of mutual aid will be favoured only by the poor and weak and not by the rich and powerful. Thus, if, as I suggest, our principles of justice represent a compromise between the rich and powerful on the one hand and the poor and weak on the other, that would explain why our morality contains a strict principle of non-injury and a somewhat weaker principle of mutual aid, the exact relative strengths of each depending on the relative bargaining positions of the rich and powerful versus the poor and weak. I want to consider in a moment whether such an explanation in terms of relative bargaining positions shows that there is no morally relevant difference between harming and not preventing harm. But before I get to that, let me mention a related example, the difference in the way we think we should behave towards animals and towards people. Although most of us feel that some ways of treating animals are wrong, most of us believe that the moral limits on behaviour towards animals are quite different from the moral limits on behaviour towards people. Most people see nothing wrong with the ways in which animals are kept as pets, raised as food, hunted for sport, and used for medical experiments, although they would think it wrong to treat people in this way. From a utilitarian point of view, there is little justification for this difference. Animals are sentient beings with their own lives to lead. Why should they be given such an inferior status as compared with people (P. Singer, 1973)? On the other hand, it is relatively easy to explain the difference in moral status of people and animals if we suppose this has its source in conventions we accept for mainly self-interested reasons. Our moral conventions favour people over animals because they are conventions arrived at by people, for people. The animals do not participate. If at a given time morality discriminates against some part of society, the affected people can try to put pressure on the others to change this part of morality. They can threaten to withhold their own participation in the moral conventions.

70

MORAL RELATIVISM

This can lead to changes in the morality. But the animals cannot get together to put pressure on people in the same way. The convention theory of justice has no problem explaining why animals have a lower moral status than people. If there is a problem, it is the opposite one of explaining why animals have any status at all in our morality. What explains our moral disapproval of cruelty to animals, given that animals do not participate in our conventions? We can say, of course, that this is connected with our caring about animals, with our having some sympathy with them. But then the explanatory work is being done by the assumption about what people care about, not by the assumption that justice is conventional. If we had to explain moral conventions for dealing with people by invoking the assumption that people care about other people, then convention would not play as major a role in morality as I believe it does.

4.11 Explaining Concern for Others There is clearly some connection between morality and concern and respect for others. One possible view is that we are generically constructed so as to feel such concern and respect for others, and that makes morality possible. There is obviously something to this (discussed further in the final three essays in this volume); but I suggest that there is also a conventional basis to our moral concern and respect for other people. One develops concern and respect for others as part of accepting a convention, on the supposition that others are developing or have developed similar concern and respect. Such concern and respect can be developed through practice. One develops a habit of taking an interest in others. This is possible because taking an interest in something or someone is the kind of thing one can do. One watches a game on television and decides to root for one or another of the teams; one does this to make the game more interesting to watch; doing so structures one's perception of the game. Similarly, one takes an interest in a conversation at a party because that makes it less boring to stand there and listen. In playing a game with friends, one takes an interest in winning; otherwise one's friends will be mad. In the same way, I suggest, one takes an interest in the welfare of other people. This is something one is encouraged to do as a child and one develops the habit of doing so. Initially, you tend to feel concern and respect only for certain selected

JUSTICE AND MORAL BARGAINING

71

people, those who participate in the relevant conventions, which at first are the practices of your family and later are the practices of those you came into contact with. Some sort of ‘stimulus generalization’ may occur also. You have a habit of feeling concern and respect for certain people. There will be a tendency for this habit to apply to similar cases, so you may feel some concern and respect for outsiders, too. You may even come, through stimulus generalization. to be disposed to feel a certain concern for animals. So, it is possible to explain our feelings about animals by appeal to the theory that morality is conventional, without supposing a primitive, unexplained sympathy we have with animals.

4.11.1 Implications of This Sort of Explanation One possible reaction to this sort of explanation of our moral views in terms of bargaining positions and power relationships, is to think that the explanation actually undermines our views by showing that they are based on morally irrelevant considerations. Two different reasons ought to be given for this. First, it might be thought that true morality has some other source than convention, and that supposing our views have a conventional source conflicts with supposing their source is true morality. I will not consider this possibility, since I have already explained why I do not believe there is such an alternative source for true morality. Second, it might be thought there is something unjust about agreements whose outcome is affected by the differences in power relationships among the affected parties. It may not seem fair that the poor and weak should have to accept only limited help from the rich and powerful simply because of the bargaining strength of the rich and powerful. It may not seem fair that animals should be placed in an inferior moral status simply because they are not able to participate in the bargaining which leads to the relevant moral conventions. This is the worry I want to consider now. What is the source of these moral judgements about unfairness? If there is no a priori morality, these judgements must be based ultimately on values that are part of the conventional morality accepted by those who make the judgements. This suggests there is a tension or instability between our current moral views and any explanation of those views in terms of differences in bargaining position. This tension might be resolved by giving up that sort of explanation, but I see no other way to explain the moral distinction we make between harming and not helping, or the differences in treatment

72

MORAL RELATIVISM

we feel are appropriate toward animals and people. Suppose, then, that we do not give up these explanations. How else might the tension be resolved? Someone might suggest it is part of our moral view that agreements are null and void if they are reached under circumstances where the affected parties have little or no bargaining power. They might suggest, furthermore, that the explanations I have offered are of that sort; that is, the explanations of the moral distinction we make between harming and not helping, and the difference in moral status of people and animals. It would follow that, according to our ordinary moral view, these aspects of that view are null and void. In fact, I do not think we would normally take even explicit agreements of this sort to be null and void; but suppose this were part of our moral view. What would our options be? One possibility would be to conclude that the relevant agreements are, indeed, null and void. That would presumably mean we had no moral reason to adhere to these agreements; for example, no reason to refrain from harming other people. I hope people would not reach this conclusion. The other possibility would be to give up the idea that such agreements are completely null and void, at least when they are basic moral agreements of this sort. One might go on, also, to give up the idea that there is something unfair about the agreement. Or one could retain one's view that there is some unfairness in such agreements, taking this to provide some reason to modify the results of the agreement so that they come to be more in accord with what might have emerged if the bargaining position of affected parties had been the same. Presumably, those with the superior bargaining position favour giving up the idea that the agreement is unfair, and those with the inferior bargaining position will favour retaining the idea that the agreement is unfair, taking this as a reason to make the outcome fairer. Does either side have the edge here? It might be said that those who conclude that the agreement is unfair have a bit of an edge, since they are simply drawing a consequence from the principles that have been accepted now by all in the relevant group. But the other side can say with equal justice that they too are simply drawing a consequence from principles accepted by all until now, since those principles have involved taking harming to be more seriously wrong than not helping, and have taken animals to have an inferior moral status to humans. When there is a tension or inconsistency in a set of accepted moral principles, given certain factual assumptions, conflicting positions can be derived. The resulting dispute can only be resolved when some new, consistent consensus is reached.

JUSTICE AND MORAL BARGAINING

73

4.12 Explicit Moral Bargaining I have suggested that our current principles of justice are the result of ongoing implicit bargaining and adjustment. What would happen if this view were widely accepted? What if moral bargaining became more explicit and people came to see self-interest as relevant to whether one accepts certain principles of justice in the same way it is relevant to whether one accepts a certain salary? It might be suggested that it would be terrible if the rich and powerful were to come to see self-interest as a legitimate moral consideration of this sort. The rich and powerful already tend to be too concerned to support efforts to attack egalitarian aspects of our current moral and political understanding, as embodied, for example, in progressive taxation. It might be feared that if the rich were to acquire a completely clear conscience about appeals to their own self-interest, they might, then, come to run roughshod over the poorer and weaker members of society with no moral restraint. A similar but opposite worry might be raised about the poor and weak. If they came to think appeals to self-interest were quite in order in moral thinking, they might come to see no moral reason to avoid organizing together to try to get all they can from those better off than they are. One might reply that such worries assume self-interest is the only motive. In fact, rich and poor will have other motives, including those derived from the moral values they have lived with for many years, which they are not going to abandon in a day. However, I am not sure this makes much difference. It is certainly true that self-interest is not the only consideration, as is shown by our moral disapproval of cruelty to animals. But it is also true that self-interest is a particularly strong and stable motivation when compared with other motives, as is shown by the inferior moral status we assign to animals. A better reply is that if it becomes generally believed that justice rests on bargaining in which self-interest plays a major role, then the bargaining between the rich and the powerful, on the one hand, and the poor and weak, on the other, will become more explicit. I see no reason to think the result will be much different from our current moral consensus, which is the result of implicit bargaining. It is already widely believed that self-interest plays a major role in the moral positions that other people adopt. Those who are not rich and powerful have not failed to notice the self-interested motives that can lead the rich and powerful to appeal to liberty and property rights. The rich and

74

MORAL RELATIVISM

powerful have in turn not failed to notice the self-interested motives that can lead the poor and weak to emphasize equality. Each side is conscious of the self-interested reasons the other side has, while being less conscious of the way in which they themselves are motivated by self interest. There are advantages to making such self-interested motives explicit. Doing so would lead to greater clarity and honesty and less hypocrisy in moral argument. This would further values that are already part of our morality. It would also tend to diminish pointless taboos and superstitions and might reduce certain sorts of exploitation.

4.13 Moral Argument Moral argument often involves consideration of how previously accepted principles apply to new cases, as well as the way in which various moral judgements might be accommodated by appeal to newly stated principles. It also happens that new principles and new interpretations of old principles are put forward because they are seen to be in the selfinterest of the people putting them forward. This is often thought to be quite disreputable, but I am saying we ought to count it as a perfectly acceptable form of moral argument: explicit moral bargaining. Moral bargaining is always involved in moral argument because there is the ever-present possibility that those dissatisfied with the outcome will drop out of the moral consensus. To the extent that others have a reason to keep them from dropping out, those others have a reason to accept the suggested reinterpretation of the principles involved. Often what looks like a simple dispute as to the implications of previously accepted moral principles is really an instance of this sort of moral bargaining. Members of disadvantaged groups argue that considerations of fairness support certain conclusions, even if these considerations have not been previously understood as supporting those conclusions. Conservatives reject this sort of argument. I suggest that in such a case the disadvantaged people are usefully interpreted as expressing dissatisfaction with the current consensus, and are threatening to withdraw support unless this consensus is modified in certain ways. The conservatives are threatening not to go along with any such change, and are invoking worries about what it would be like to abandon any sort of moral consensus altogether. Of course, as I have emphasized all along, self-interest is not the only factor that can fuel such disputes. This emerges in the argument over

JUSTICE AND MORAL BARGAINING

75

abortion. Self-interest plays a part in that argument, namely the self-interest of those who see themselves as benefiting from women having the right to abortion. But prior principle is what is most important on the other side. Indeed, prior principle almost always plays an important role in a moral dispute, and not just as a peg on which to hang one's selfinterested claims. What a person has reason to do depends on that person's various desires, plans, intentions, and values. Some of these come from a person's acceptance of one or more moral understandings with others. But acceptance is not forever. A person can have reasons to stop accepting a given moral understanding, reasons that derive from these prior desires, etc. Your prior commitments can give you reasons to modify some of your prior commitments. These reasons can be purely internal to morality, but may also reflect considerations of self-interest if the morality you accept is getting in your way too much. Similarly, a person can have reason to try to change the general moral consensus in some group, because the person does not want simply to opt out of that group and yet thinks things would either be morally better from the point of view of prior moral standards as they have been understood all along, or better from the point of view of self-interest or the interest of someone the person cares about, if the change were made. All this applies whether these considerations are explicit or implicit.

4.14 Tolerance Finally, since the conventionalist theory of justice I am defending is a form of moral relativism and moral relativism is often thought to imply tolerance of other moralities, I want to consider briefly what the implications of explicit moral bargaining would be for tolerance and moral compromise. Consider the question of vegetarianism. Suppose it is agreed on all sides that morality is conventional in the way I have suggested. Moral vegetarians, then, cannot suppose that they have some sort of privileged insight into moral values, and must agree that their view represents values that they but not others accept. These values are, perhaps, accepted as an interpretation of more widely accepted moral conventions, an interpretation not universally accepted, however, but one which can be rejected without mistake of fact or reasoning. This should lead moral vegetarians to a certain tolerance of nonvegetarians.

76

MORAL RELATIVISM

Exactly the same considerations apply to the case of abortion. Once it is agreed by anti-abortionists that their view does not represent the one true morality but arises from values that they but not others accept, perhaps as an interpretation of more widely accepted conventions, then it would seem they should be more tolerant of those who are not opposed to abortion. It should come to seem less appropriate to have laws against abortion, given the large numbers of people who are not opposed to abortion. This is not to say that this sort of relativism leads to tolerance of everything. Moral relativists need not be tolerant of harms committed by criminals who have not accepted conventional morality. There are strong self-interested reasons for us to include in our moral understanding the proviso that we can use force against those who harm others. There are not similarly strong, self-interested reasons for vegetarians to include in their moral understanding the condition that they can use force against those who raise animals for food. I do not mean to suggest self-interested reasons are the only reasons that might play a role, here. Moral vegetarians may value animal life very highly, and that can lead them to favour laws against cruelty to animals, against animal experimentation, against certain methods of treating livestock, and so forth. But moral vegetarians must as a practical matter also reach some understanding with the rest of us; and that is where the pressure of tolerance comes from. On the other hand, in a society in which all but a very few people were moral vegetarians, there would be little or no reason for them to tolerate meat-eating. The question of tolerance of abortions is harder to judge. On the one hand, some sentiment against abortions is derived from certain religious views. It is not easy to say how much opposition to abortion would survive acceptance of moral conventionalism. Much opposition might survive. If there were enough, it might be thought unnecessary to reach a compromise with those who favour allowing abortions. Abortionists might continue to accept principles permitting the use of force to prevent abortions. But as a practical matter, it is hard to believe that this can prevail in the actual situation because of the many people who perceive their own self-interest to lie on the other side. They cannot at this point agree to permitting this use of force. Eventually, some sort of compromise must be reached, one mostly favouring the pro-abortion side, since self-interest plays no direct role on the other side.

5 Is There a Single True Morality? 5.1 Confession I have always been a moral relativist. As far back as I can remember thinking about it, it has always seemed to me obvious that the dictates of morality arise from some sort of convention or understanding among people, that different people arrive at different understandings, and that there are no basic moral demands that apply to everyone. For many years this seemed so obvious to me that I assumed it was everyone's instinctive view, at least everyone who gave the matter any thought. When I first studied philosophical ethics, I was not disabused of this opinion. The main issue at the time seemed to be to determine exactly what form of ‘noncognitivism’ was correct. (According to noncognitivism, moral judgements do not function to describe a moral reality, but do something else—express feelings, prescribe a course of action, and so forth.) It is true that many of the philosophers I studied seemed for some reason to want to avoid calling themselves ‘relativists’. This was usually accomplished by defining moral relativism to be an obviously inconsistent position (M. Singer, 1961; Williams, 1972); for example, the view both that there are no universal moral truths and also that everyone ought to follow the dictates of his or her group, where this last claim is taken to be a universal moral truth. I wasn't sure what this verbal manœuvre was supposed to accomplish. Why would anyone want to give such a definition of moral relativism? Moral relativism was obviously correct, and the philosophers I was studying seemed all to be moral relativists even if they did not want to describe themselves in that way. Later I was distressed to hear from various people teaching ethics that students in their classes tended to proclaim themselves moral relativists until they had been shown how confused they were about ethics. I suspected that the confusions were not confusions of the students, but were confusions of their teachers, due perhaps to a faulty definition of moral relativism. The obvious solution was to show that moral relativism can be

78

MORAL RELATIVISM

consistently defined as a plausible view and that standard objections to moral relativism are mistaken. So, I eventually wrote and published an essay about this (the original version of Essay 1, above), thinking it would clear things up and end worries about moral relativism. I was surprised to discover that this did not happen. I was also startled to find that many students in my own ethics courses resisted my attempt to make clear what I thought they instinctively believed. After some study I concluded that in fact only some of the students in my courses were instinctive moral relativists; a significant number of them were instinctive absolutists. I had known, of course, that there were philosophers and friends of mine who were not moral relativists. For a long time I attributed this to their perversity and love of the bizarre, and attached no significance to it. But then I discovered that some of them thought moral relativism was the perverse view, a kind of philosophical folly like scepticism about other minds or the external world (for example, Nagel, 1980). I was stunned! How could they think that, when they knew so many moral relativists (like me) and no epistemological sceptics (at least none who took such scepticism seriously in ordinary life)? It then occurred to me to wonder how I could think of moral absolutism as such a perverse view, when I knew so many moral absolutists.

5.2 The Issue It turns out, to my surprise, that the question whether there is a single true morality is an unresolved issue in moral philosophy. On one side are relativists, sceptics, nihilists, and noncognitivists. On the other side are those who believe in absolute values and a moral law that applies to everyone. Strangely, only a few people seem to be undecided. Almost everyone seems to be firmly on one side or the other, and almost everyone seems to think his or her side is obviously right, the other side representing a kind of ridiculous folly. This is strange, since everyone knows, or ought to know, that many intelligent people are on each side of this issue.

5.2.1 Two Approaches In this essay I want to suggest that part of the explanation for this mutual incomprehension is that there are two different ways to do moral philosophy. If one approach is taken, moral relativism, noncognitivism, or

IS THERE A SINGLE TRUE MORALITY?

79

scepticism may seem obviously correct and moral absolutism may seem foolish. If the other approach is taken, absolutism may seem clearly right and scepticism, relativism, and noncognitivism may seem foolish. The difference in approaches is, to put it crudely, a difference in attitude toward science. One side says we must concentrate on finding the place of value and obligation in the world of facts as revealed by science. The other side says we must ignore that problem and concentrate on ethics proper. Both sides agree that we must begin at the beginning with our initial beliefs, moral and nonmoral, and consider possible modifications that will make these beliefs more coherent with each other and with plausible generalizations and other explanatory principles. Eventually, we hope to arrive at a ‘reflective equilibrium’ (Rawls, 1971) when no further modifications seem called for, at least for the time being. The process will leave some issues unresolved; in particular, we may find ourselves with no account of the place that value and obligation have in the world of facts. This will not dismay someone who is willing to leave that question unanswered. But it will be disturbing to someone who, on the way to reflective equilibrium, has come to think that the basic issue in moral philosophy is precisely how value and obligation fit into the scientific conception of the world. I will use the term ‘naturalism’ for an approach to ethics that is in this way dominated by a concern with the place of values in the natural world. I will call any approach that is not so dominated an instance of ‘autonomous ethics’, because such an approach allows us to pursue ethics internally. Autonomous ethics allows that science is relevant to ethics in as much as ethical assessment depends on the facts of the case. But unlike naturalism, autonomous ethics does not take the main question of philosophical ethics to be the naturalistic status of values and obligations.

5.2.2 Naturalism I hope the terms ‘naturalism’ and ‘autonomous ethics’ are not misleading. The term ‘naturalism’ is sometimes reserved for the thesis that moral judgements can be analysed into or reduced to factual statements of a sort clearly compatible with the scientific world view. I am using the term ‘naturalism’ more broadly. Naturalism in this sense does not have to lead to naturalistic reduction, although that is one possibility. Another possibility is that there is no way in which ethics could fit into the scientific conception of the world. In that case, naturalism leads to moral nihilism, as in Mackie (1977). Mackie supposes that ethics requires absolute values.

80

MORAL RELATIVISM

Such values would have the property that anyone aware of their existence must necessarily be motivated to act morally. Because our scientific conception of the world has no place for entities of this sort, and because there is no way in which we could become aware of such entities, Mackie concludes that ethics must be rejected as resting on a false presupposition. That is a version of naturalism as I am using the term. Naturalism can also lead one to a noncognitive analysis of moral judgements. In this view, moral judgements do not function to describe the world, but to do something else—to express one's attitudes for and against things, as Stevenson (1963a) argues—or to recommend one or another course of action or general policy, as Hare (1952, 1963) proposes. Or a naturalist may decide that moral judgements do make factual claims that fit in with the claims of science. This can be illustrated by some sort of naturalistic reduction. One example would be an analysis that takes moral claims to be claims about the reactions of a hypothetical impartial observer as in Hume (1739) or Firth (1952). More complex positions are possible. Mackie (1977) argues in chapter 1 that ethics rests on a false presupposition, but then he goes on in later chapters to discuss particular moral issues. It is almost as if he had first demonstrated that God does not exist and had then gone on to consider whether He is wise and loving. Presumably, Mackie believes that ethics as normally conceived must be or can be replaced with something else. But he does not indicate exactly what sort of replacement he has in mind—whether it is an institution of some sort, for example. Nor does he say how moral claims made within this replacement fit in with the claims of science. I suspect he would accept some sort of noncognitivist account of the judgements that are to replace the old moral judgements. It is possible to be both a naturalist and an absolutist, although this is not very common. Firth (1952) defends an absolutist version of the ideal observer theory and Hare (1963) defends an absolutist version of noncognitivism. But I believe that the most plausible versions of naturalism involve a moral relativism that says different agents are subject to different basic moral requirements depending on the moral conventions in which they participate.

5.2.3 Autonomous Ethics Naturalism tends toward relativism. What I am calling autonomous ethics, on the other hand, can have a very different tendency. In this approach, science is relevant, since our moral judgements depend on what

IS THERE A SINGLE TRUE MORALITY?

81

we take the facts to be; but we attach no special importance to saying how obligations and values can be part of the world revealed by science. Rather, we do ethics internally. We begin with our initial moral beliefs and search for general principles. Our initial opinions can be changed to some extent so as to come into agreement with appealing general principles and our beliefs about the facts, but an important aspect of the appeal of such principles will be the way in which they account for what we already accept. The autonomous approach normally (but not always) involves an initial assumption of moral absolutism. In this context, absolutism is of course not the thesis that there are simple moral principles that hold absolutely without exceptions. It is rather the thesis that there are basic moral demands that apply to all moral agents. Autonomous ethics tends to retain that absolutist thesis. It may also involve some sort of intuitionism, claiming that each of us has immediate insight into the truths of certain moral principles. It sometimes leads to a fairly conservative morality, not much different from one's initial starting point. That is not surprising, given the privileged position assigned to our initial moral beliefs. But let me stress that conservatism is not inevitable, and autonomous ethics can and often does lead to more radical moralities, too. It leads some philosophers to a radical utilitarianism. It leads Rawls (1971) to principles of social justice that appear to be considerably more egalitarian than those most people accept. Nozick (1974), using the same general approach, comes out at a very different place, in which he ends up denying that any sort of egalitarian redistribution by governments is ever morally justified. (However, the moral theory in Nozick, 1981, as contrasted with the political theory in Nozick, 1974, insists on the moral requirement of helping others.) Indeed, there are many different ways in which ethics can be pursued as an autonomous discipline with its own principles that are not reducible to the principles of any science. I can illustrate this variety by mentioning a few of the many other contemporary philosophers who accept some form of autonomous ethics: Baier (1958), Darwall (1983), Donagan (1977), Frankena (1976a), Fried (1978), Gewirth (1978), Grice (1967), Nagel (1970, 1980), and Richards (1971). Each of these philosophers has a somewhat different approach, although all are absolutists who rely on some form of autonomous ethics. I should say that it is possible to believe in autonomous ethics without being an absolutist. One might be impressed by the variety of views held by those who accept autonomous ethics and so be led to allow for relativism while continuing to accept the method of autonomous ethics,

82

MORAL RELATIVISM

believing that naturalism must be rejected. But the tendency of autonomism in ethics is toward absolutism. In what follows I will restrict my discussion to absolutist versions of autonomous ethics and to relativistic versions of naturalism.

5.2.4 Teachers of Ethics I might also mention that ethics pursued internally, as in autonomous ethics, is more interesting to many people than ethics as pursued by naturalism. That is because autonomous ethics allows one to spend more of one's time thinking about interesting, complicated moral puzzles than naturalistic ethics does, and many people find moral puzzles more interesting than ‘abstract’ questions about the objectivity of value and its place in nature. Philosophers attracted by naturalism tend not to find ethics as interesting a subject as do philosophers attracted by autonomous ethics. So, relativists tend to be less interested in ethics than absolutists. For example, logicians, philosophers of science, and philosophers of mathematics, who tend toward naturalism, are usually not moral absolutists and are not very interested in ethics as a philosophical subject. Philosophers who are relatively interested in ethics tend to be those who favour autonomous ethics and therefore tend to be absolutists. This is why teachers of ethics tend, more than their students, to be absolutists. It is not merely, as they sometimes suppose, that ethics teachers have seen through confusions that affect their students. A more important factor is that relativists tend not to become teachers of ethics. (I am an exception!)

5.3 Why Do We Believe What We Believe? Autonomous ethics and naturalism represent very different attitudes towards the relation between science and ethics. Consider, for example, the question of what explains our believing what we in fact believe. Naturalists see an important difference between our factual beliefs and our moral beliefs. Our ordinary factual beliefs provide us with evidence that there is an independent world of objects, because our having those beliefs cannot be plausibly explained without assuming we interact with an independent world of objects external to ourselves, objects we perceive and manipulate. But our having the moral beliefs we do can be explained entirely in terms of our upbringing and our psychology, without any appeal to an independent realm of values and obligations. So our moral

IS THERE A SINGLE TRUE MORALITY?

83

beliefs do not provide us with evidence for such an independent realm of values and obligations, and we must choose between scepticism, noncognitivism, and relativism (Harman, 1977: ch. 1). Autonomists disagree with this. They claim we often believe that something is good or right or obligatory in part because it is good or right or obligatory. They accuse naturalists of begging the question. When naturalists say that a belief cannot be explained by virtue of something's being right, unless that thing's being right consists in some psychological or sociological fact, they simply assume that all explanatory factors are part of the world revealed by science. But this is the point at issue. Autonomists argue that it is more obvious that we sometimes recognize what is right than that naturalism is correct. True, we may be unable to say how a given ‘moral fact’ and someone's recognition of it fit into the world of facts as revealed by science. But there are always unanswered questions. To jump from our current inability to answer this question to scepticism, relativism, or noncognitivism is to make a more drastic move than this puzzle warrants, from the point of view of autonomous ethics.

5.3.1 Explanation and Reduction The naturalist seeks to locate the place of value, justice, right, and wrong, and so forth in the world in a way that makes clear how they might explain what we take them to explain. A naturalist cannot understand how value, justice, right, and wrong might figure in explanations without having some sense of their ‘location’ in the world. We can say that this involves ‘naturalistic reduction’, but it need not involve reductive definitions of a serious sort. Indeed, reduction rarely (if ever) involves serious reductive definitions. We identify tables with clusters of atoms in a way that allows us to understand how tables can hold up the things they hold up without having to suppose the word ‘table’ is definable using only the concepts of physics. Similarly, we identify colours with dispositional properties of objects, namely, their tendency to look in certain ways to certain sorts of observers in certain conditions, without having to suppose there is a satisfactory definition in these terms. Similarly for temperatures, genes, and so on. What a naturalist wants is to be able to locate value, justice, right, wrong, and so forth in the world in the way that tables, colours, genes, temperatures, and so on can be located in the world. What is at issue here is understanding how moral facts might explain something, how the badness of someone's character might explain why that person acts in a certain way, to take an example from Sturgeon

84

MORAL RELATIVISM

(1985). It is not sufficient that one be prepared to accept the counterfactual judgement that the person would not have acted in that way if the person had not had a bad character, if one does not see how the badness of the person's character could have such an effect. A naturalist believes one can see that only by locating badness of character in aspects of the world that one sees can have the relevant effect. Notice that a naturalist, as I am here using the term, is not just someone who supposes that all aspects of the world have a naturalistic location in this way, but rather someone who takes it to be of overriding importance in doing moral philosophy actually to attempt to locate moral properties. My claim is that, when one takes this attempt seriously, one will tend to become sceptical or relativistic. Sturgeon is not a naturalist in my sense, despite his insistence that he takes moral facts to be natural facts.

5.4 Moral Absolutism Dened I now want to be more specific about what is to count as moral absolutism. Various things might be meant by the claim that there are absolute values and one true morality. Moral absolutists in one sense might not be moral absolutists in other senses. We must be careful not to mix up real issues with purely verbal issues. So let me stipulate that I will take moral absolutism to be a view about the moral reasons people have to do things and to want or hope for things. I will understand a belief about absolute values to be a belief that there are things that everyone has a reason to hope or wish for. To say that there is a moral law that ‘applies to everyone’ is, I hereby stipulate, to say that everyone has sufficient reasons to follow that law. It is true that many philosophers pursue something that resembles autonomous ethics when they ask what principles an ‘ideal’ moral code of one or another sort would have, quite apart from the question whether people now have any reason to follow that code. Depending on what sort of idealization is being considered, there may or may not be a unique ideal code of that sort. But I am not going to count as a form of moral absolutism the claim that there is a unique ideal moral code of such and such a type. Relativists and absolutists in my sense might very well agree about this claim without that having any effect at all on what I take to be the basic issue that separates them. A claim about ideal codes has no immediate relevance to questions about what reasons people actually have to hope for certain things or to do certain things.

IS THERE A SINGLE TRUE MORALITY?

85

Similarly, I am not going to count as a form of moral absolutism the claim that there is one true morality that applies to everyone in that everyone ought to follow it, if this is not taken to imply that everyone has a sufficient reason to follow it. I am not sure what ‘ought’ is supposed to mean if it is disconnected in this way from reasons to do things. If what is meant is that it ought to be the case that everyone followed the one true morality—in other words, that it would be a good thing if they did—then this is a version of the view that there is a unique ideal moral code. I am not sure what else might be meant, although a great deal more could be said here (see Essay 4). Rather than try to say it, however, I simply stipulate that this sort of claim is not a version of what I am counting as moral absolutism. I should note that, of the contemporary philosophers I have identified as absolutists, Baier, Darwall, Donagan, Frankena, Gewirth, Grice, Nagel, and Richards clearly advocate moral absolutism in this sense. They all think that there are basic moral demands that in some sense every competent adult has reasons to adhere to. I believe the others I mentioned—namely, Rawls, Nozick, and Fried—also agree with this, although they do not explicitly say so in the works I have cited.

5.5 Does a Single Moral Law Apply to Everyone? Consider the issue between absolutism and relativism concerning reasons people have for doing things. According to moral absolutism about this, there is a single moral law that applies to everyone; in other words, there are moral demands that everyone has sufficient reasons to follow, and these demands are the source of all moral reasons. Moral relativism denies that there are universal basic moral demands, and says different people are subject to different basic moral demands depending on the social customs, practices, conventions, values, and principles that they accept. For example, a moral absolutist might suppose there is a basic moral prohibition on causing harm or injury to other people. This prohibition is in one sense not absolute, since it can be overridden by more compelling considerations and since it allows exceptions in order to punish criminals, for instance. But the prohibition is supposed to be universal in the sense that it applies to absolutely all agents and not just to those who happen to participate in certain conventions. The absolutist claims that absolutely everyone has sufficient reasons to observe this prohibition and to act as it and other basic moral requirements dictate.

86

MORAL RELATIVISM

A moral relativist denies this, and claims that many people have no reasons to observe this prohibition. Many people participate in moralities that sharply distinguish insiders and outsiders and do not prohibit harm or injury to outsiders, except perhaps as this is likely to lead to retaliation against insiders. A person participating in such a morality has no reason to avoid harm or injury to outsiders, according to the relativist, and so the general prohibition does not apply to that person. The person may be a member of some primitive tribal group, but he or she need not be. He or she might also be part of contemporary society, a successful professional criminal who recognizes various obligations to other members of a criminal organization but not to those on the outside. According to the moral relativist, the successful criminal may well have no reason at all not to harm his or her victims.

5.5.1 An Argument for Relativism Let us concentrate on this case. The moral absolutist says the demands of the one true morality apply as much to this successful criminal as to anyone else, so this criminal does have a reason not to harm a given victim. The relativist denies the criminal has any such reason, and so denies the relevant moral demand is a universal demand that applies to everyone. Here naturalism tends to support relativism in the following way. Consider what it is for someone to have a sufficient reason to do something. Naturalism requires that this should be explained in terms congenial to science. We cannot simply treat this as irreducibly normative, saying, for example, that someone has a sufficient reason to do something if and only if he or she ought to do it. Now, presumably, someone has a sufficient reason to do something if and only if there is warranted reasoning that person could do that would lead him or her to decide to do that thing. A naturalist will suppose that a person with a sufficient reason to do something might fail to reason in this way to such a decision only because of some sort of empirically discoverable failure. Such a failure might be inattention, lack of time, failure to consider or appreciate certain arguments, ignorance of certain available evidence, an error in reasoning, some sort of irrationality or unreasonableness, or weakness of will. If the person does not intend to do something and that is not because he or she has failed in some such empirically discoverable way to reason to a decision to do that thing, then according to the naturalist the person cannot have a sufficient reason to do that thing. This is the first premise in a naturalistic argument in support of the relativist.

IS THERE A SINGLE TRUE MORALITY?

87

The other premise is that there are people, such as certain professional criminals, who do not act in accordance with the alleged requirement not to harm or injure others, where this is not due to any these failings. The argument for this is simply that there clearly are people who do not adhere to the requirement in question and who do not seem to have failed in any of the ways mentioned. So, in the absence of special theoretical reasons, for example, deriving from psychology, to think these people must have failed in one of the specified ways, we can conclude they have not done so. From these two premises it follows that there are people who do not have sufficient reasons, and therefore do not have sufficient moral reasons, to adhere to the general prohibition against harming or injuring others. In particular, a successful criminal may not have a sufficient reason not to harm his or her victims. The moral prohibition against harming others may simply fail to apply to such a person. It may fail to apply in the relevant sense, which is of course not to say that the principle makes an explicit exception for criminals, allowing them but not others to injure and harm people without restraint. Rather, the principle may fail to apply in the sense that the criminal in question may fail to have sufficient reason to act in accordance with the principle.

5.5.2 An Absolutist Reply Moral absolutism must reject this argument. It can do so by invoking autonomous ethics at the place at which moral relativism invokes naturalism. Autonomous ethics does not suppose that we must give some sort of naturalistic account of having a sufficient reason to do something. Nor does it suppose that only a science like psychology can discover the conditions under which someone has failed to reason in a certain way because of inattention, irrationality, unreasonableness, or any of the other causes of failure mentioned in the relativistic argument. Autonomous ethics approaches this issue in the following way. We begin with certain beliefs. Presumably these imply that everyone has a sufficient reason to observe the prohibition against harm to others, including the successful criminal who does not participate in or accept any practice of observing this general prohibition. At the start we therefore believe that the criminal does have sufficient reason not to harm his or her victims. Following autonomous ethics, then, we should continue to believe this unless such continued belief conflicts with generalizations or other theoretical principles internal to ethics that we find attractive because they do a better job at making sense of most of the things we originally believe.

88

MORAL RELATIVISM

Taking this approach, the absolutist must claim that the relativistic argument does not provide sufficient reason to abandon our original absolutism. It is more plausible, according to the absolutist, that at least one of the premises of the relativistic argument is false, than that its conclusion is true.

5.5.3 Assessing the First Premise The first premise of the relativistic argument is that for someone to have a sufficient reason to do something, there must be warranted reasoning available to that person that leads to a decision to do that thing. So, if the person fails to intend to do that thing, this must be because of inattention, lack of time, failure to consider or appreciate certain arguments, ignorance of relevant evidence, an error in reasoning, irrationality, unreasonableness, or weakness of will. The absolutist might object that this is oversimplified. If a person with sufficient reason to do something does not do it, then something has gone wrong. It might be one of the things the relativist mentions, but it might be something else as well. There might just be something wrong with the person in question. That person might be immoral. The failure might simply be a failure not to care enough about other people. A person ought to care about others and there is something wrong with a person who does not care, even if that person is not inattentive, ignorant, rushed, or defective in any other of the particular ways the relativist mentions. So, even if some people fail to observe the prohibition against harming others not because of inattention, lack of time, and so forth, but simply because of lack of concern and respect for others, such people still do have sufficient reason not to harm others. (Thomas M. Scanlon suggested this response on behalf of absolutism to me.) This response to the relativistic argument is a response within autonomous ethics. It does not explain having a sufficient reason to do something in terms that are acceptably factual from a naturalistic perspective. It also appeals to the notion of something's being wrong with someone, where what might be wrong is simply that the person is immoral. It is like saying one has a sufficient reason to do something if and only if one ought to do it, or if and only if it would be wrong not to do it. The relativist claims that the only plausible accounts of these normative notions are relativistic ones. There is no prohibition on harm to outsiders, in the criminals' morality. There is such a prohibition only in some other morality. In that other morality, something is wrong with a

IS THERE A SINGLE TRUE MORALITY?

89

person who has no compunction about injuring someone else; but nothing is wrong with such a person with respect to the criminal morality, as long as those injured are outsiders. But how can it be a sufficient reason for the criminal not to harm his or her victims that this is prohibited by somebody else's morality? How can its being bad, immoral, or wrong in this other morality not to care about and respect others give the criminal, who does not accept that morality, a sufficient reason to do anything? The absolutist responds that failure to respect others is not just wrong according to some morality the criminal does not accept; it is also wrong, period. Something is really wrong with lack of respect and concern for others. It is not just wrong in relation to one or another morality. Of course, the relativist will not be satisfied with this answer and, appealing to naturalism, will ask what it is for something to be wrong in this way. The absolutist supposes that the failure to care about and respect others does involve something the absolutist points to by saying this failure is wrong. But what is this thing that is true of such a failure to care and that can give the criminal a sufficient reason not to harm and injure others? The relativist can see no aspect of such a failure that could provide such a reason. This is because the relativist, as a naturalist, considers only aspects of the failure that are clearly compatible with a scientific world view. The relativist disregards putative aspects that can be specified only in normative terms. But the absolutist, as an autonomist, can specify the relevant aspect of such a failure to care about others: it is bad, immoral, wrong not to care. The criminal ought to have this concern and respect and so ought not to harm and injure others, and therefore has a sufficient reason not to harm and injure them.

5.5.4 Assessing the Second Premise We have been discussing an argument for relativism concerning moral reasons. We have seen that naturalism supports the first premise of this argument and that autonomous ethics allows the rejection of this premise. The same thing is true of the second premise, which says that there are people, such as the successful criminal, who do not observe the alleged requirement not to harm or injure others, and this is not due to inattention, failure to consider or appreciate certain arguments, ignorance of relevant evidence, errors in reasoning, irrationality, unreasonableness, or weakness of will. Naturalism supports this, because there do seem to be such people, and no scientifically acceptable grounds exist for thinking this is an illusion. On the other hand, autonomous ethics allows other

90

MORAL RELATIVISM

grounds, not reducible to scientific grounds, for thinking this is an illusion. In autonomous ethics we begin by supposing that we recognize the wrongness of harming others, where this is to recognize a sufficient reason not to harm others. If that is something we recognize, then it must be there to be recognized, so the successful criminal in question must be failing to recognize and appreciate something that is there. The absolutist might argue that the criminal must be irrational, or at least unreasonable. Seeing that a proposed course of action will probably cause serious injury to some outsider, the criminal does not treat this as a reason not to undertake that course of action. This must be irrational or unreasonable, because such a consideration simply is such a reason and indeed is an obvious reason, a basic reason, not one that has to be derived in some complex way through arcane reasoning. But then it must be irrational, or at least unreasonable, for the criminal not to care sufficiently about others. The criminal's lack of concern for others is what is responsible for the criminal's not taking the likelihood of harm to an outsider to be a reason against a proposed course of action. This is one way an absolutist might argue. The relativist's reply to such an argument is that, on any plausible characterization of reasonableness and unreasonableness (or rationality and irrationality) as notions that can be part of the scientific conception of the world, the absolutist's claim is just false. Someone can be completely rational without feeling concern and respect for outsiders. But, of course, this reply appeals to naturalism. The absolutist who rejects naturalism in favour of autonomous ethics relies on an unreduced normative characterization of rationality and irrationality (or reasonableness and unreasonableness). Now the argument continues as before. The relativist argues that, if rationality and irrationality (or reasonableness and unreasonableness) are conceived normatively, they become relative notions. What one morality counts as irrational or unreasonable, another does not. The criminal is not irrational or unreasonable in relation to criminal morality, but only in relation to a morality the criminal rejects. But the fact that it is irrational or unreasonable in relation to this other morality not to have concern and respect for others, does not give the criminal who rejects that morality any reason to avoid harming or injuring others. The absolutist replies that relative irrationality or unreasonableness is not what is in question. The criminal is irrational, or at least unreasonable, period; not just irrational or unreasonable in relation to a morality he or she does accept. Since it is irrational or unreasonable for anyone not to care sufficiently about others,

IS THERE A SINGLE TRUE MORALITY?

91

everyone has a sufficient reason not to injure others. This is so, whether he or she recognizes this reason or, through irrationality or unreasonableness, does not recognize it. The naturalist is unconvinced by this because the naturalist can find no aspect of the criminal the absolutist might be referring to in saying the criminal is ‘irrational’ or ‘unreasonable’, if this aspect is to give the criminal any reason to care about others. This, of course, is because the naturalist is considering only naturalistic aspects of the criminal, whereas the absolutist, as an autonomist, is thinking about an unreduced normative aspect, something the naturalist cannot appeal to. So, as was true of the first premise of the relativistic argument about reasons, the second premise depends on an assumption of naturalism. By appealing to autonomous ethics, an absolutist can reject this premise. An absolutist may in fact actually accept one or the other of the premises of the relativistic argument (although, of course, not both). A given absolutist might reject either the first premise, or the second, or both premises. An absolutist might even be undecided, holding merely that one or the other premise must be rejected, without saying which. There is nothing wrong with being undecided about this. Reflective equilibrium leaves many issues unresolved.

5.6 Are There Absolute Moral Values? The situation is similar in the theory of value. Naturalism tends to support the conclusion that all value is relative and that something is always good for one or another person or group of people or in relation to a specified set of purposes or interests or aims. Autonomous ethics allows also for absolute values, things that are good, period, and not just good for someone or some group or for some purpose. The issue here concerns the goodness or value of a possible state of affairs, not the goodness or value of something as a thing of a given sort. The issue is not what it is for something to be a good thing of a kind, a good knife, a good watch, a good backswing, a good apple, a good farmer, a good poem. The issue is rather what it is for an event or situation to be a good thing; what is it, for example, to be a good thing that it is raining or that Egypt and Israel signed a peace treaty. It is uncontroversial that this sort of goodness is sometimes relational. A situation is good for someone or some group of people, good from a certain point of view, in relation to certain purposes or interests. That it is

92

MORAL RELATIVISM

raining is a good thing for the farmer, but not for the vacationer. That Egypt and Israel signed a peace treaty might be good from their point of view, but not from the point of view of the PLO. Given a fixed point of reference, we can evaluate states of affairs as better or worse. The value of a state of affairs in relation to that reference point represents the degree to which someone with the relevant purposes and interests has a reason to try to bring about, or want, or at least hope for that state of affairs. Now, it can be argued that there is also a kind of absolute value. (However, one can be a moral absolutist without believing in absolute values. Thomson, 1997, is an example.) The claim is that states of affairs can be good or bad, period, and not merely good or bad for someone or in relation to given purposes or interests. On hearing of pointless painful experiments on laboratory animals, for example, one immediately reacts with the thought that this is bad and it would be good to eliminate such practices. Clearly, one does not simply mean that these tortures are bad for the animals involved and that these animals would benefit if such experiments were ended. A heartless experimenter might agree that what he does is bad for the animals without having to agree that it would be a good thing to eliminate this sort of experimentation. Similarly, it seems intelligible to suppose that it would be better if there were no inequalities of wealth and income in the world even though this would not be better for everyone, not for those who are now relatively wealthy, for instance. And this seems to say more, for example, than that the average person would be better off if there were no such inequalities, since an elitist might agree with that but not agree that the envisioned state of affairs would be better, period, than our present situation. Again, we can consider which of various population policies would lead to the best resulting state of affairs even though these policies would result in different populations, so that we cannot be simply considering the interests and purposes of some fixed group. It may seem, then, that we can consider the absolute value of a possible state of affairs.

5.6.1 Scepticism About Absolute Values The relative value of a possible state of affairs in relation to given purposes and interests is a measure of the extent to which someone with those purposes and interests has a reason to try to bring about, or want, or hope for that state of affairs. The absolute value of a possible state of affairs is a measure of the extent to which anyone, apart from having a personal stake in the matter, has a reason to try to bring about, or want, or hope for that

IS THERE A SINGLE TRUE MORALITY?

93

state of affairs. Naturalism leads to scepticism at this point. How could we ever be aware of absolute values? How could we ever know that everyone has a reason to want a certain possible state of affairs? Further reflection along naturalistic lines suggests that apparent absolute values are often illusory projections of one's personal values onto the world. Sometimes this sort of projection yields plausible results, but usually it does not. To begin with the most plausible sort of case: in hearing about the pain involved in animal experimentation, our sympathies are immediately and vividly engaged. We immediately side with the animals against the experimenters. In saying, ‘That is awful!’ we are not just saying, ‘That is awful for the animals,’ because our remark expresses our sympathetic identification with the point of view of the animals. We do not merely state a fact; we express our feelings and we expect an awareness of this state of affairs to call forth the same feelings of dismay in everyone. This expectation seems reasonable enough in this case, because it may well be, as Brandt (1976) argues, that everyone has a sympathetic reaction to suffering. But plausibility vanishes as soon as the case becomes even a little complex. Suppose the animal experiments are not pointless. Suppose they are an essential part of a kind of medical research that promises to alleviate a certain amount of human suffering. Alternatively, suppose that, although the experiments promise no practical benefit of this sort, they are relevant to a theoretical issue in psychology. Someone may still feel that it is bad that the experiments should occur and that it would be good if they were not done, the gain not being worth the cost. Again, this is not just to say that the experiments are bad for the animals, something to which everyone would agree. The remark expresses overall disapproval of the experiments, with the expectation that others will also disapprove, at least if they consider the issue in an impartial way. The trouble is that people react differently to these cases. Consider the question whether it is good or bad to experiment painfully on animals in order to resolve certain theoretical issues in psychology. The extent to which this is (absolutely) good is the extent to which everyone (apart from any personal stake in the matter) has a reason to try to bring it about that such experiments are done, or to want them to be done, or hope that they are done. The extent to which this is (absolutely) bad is the extent to which everyone (apart from any personal stake) has a reason to try to end the experiments, or want them to end, or hope they end. But naturalism suggests that there is no unique answer here. What a person has a reason to want will depend on the relative value he or she

94

MORAL RELATIVISM

attaches to animal suffering, to using animals as means, and to theoretical progress in psychology. Different people can attach different values to these things without having overlooked something, and without being irrational or unreasonable. So it seems that some people will have reason to be in favour of the experiments and others will have reason to be opposed to the experiments, where this is determined by the personal values of those people. If we suppose that our answer is the right answer, we are merely projecting our own values onto the world.

5.6.2 The Issue Joined Of course, autonomous ethics sees nothing wrong with projecting our own values onto the world, holding that such projection is exactly what is needed, at least at the beginning. We begin with our initial valuations and modify them only in the interests of theoretical simplicity. If we start out believing in absolute values, we should continue believing this until forced to believe otherwise. Clearly the controversy over absolute values parallels the controversy about reasons to do things. The argument against absolute values has the same structure as the relativistic argument about reasons to do things. Its first premise is that a person has a reason to want, hope for, or try to bring about a particular state of affairs, only to the extent that he or she would be irrational or unreasonable not to want that state of affairs, unless he or she was unaware of some relevant consideration, was confused, or had some other specified defect. Its second premise is that, except for the simplest cases, a person can fail to want a given state of affairs without being irrational or unreasonable or ignorant or whatever. The conclusion is that, except possibly for simple cases where, for example, the only thing relevant is that a creature suffers, there are no reasons everyone has to want or hope for or try to bring about a given state of affairs. So there are no nontrivial absolute values. As before, the two premises are defended in each case by an appeal to naturalism: We must give a naturalistic account of reasons and we must give empirical grounds for supposing someone to be irrational or unreasonable. The absolutist rejects the argument as before by invoking autonomous ethics, perhaps by rejecting the naturalistic account of reasons, perhaps by rejecting the requirement that scientific grounds must be given for a judgement of irrationality or unreasonableness, possibly remaining undecided between these alternatives.

IS THERE A SINGLE TRUE MORALITY?

95

5.7 Naturalism Versus Autonomous Ethics So the issue between relativism and absolutism comes down to the dispute between naturalism and autonomous ethics. Which is the best approach in moral philosophy? Should we concentrate on the place of values and reasons in the world of scientific fact, as naturalism recommends? Or should we start with our initial moral beliefs and look for general principles and moral theories that will eventually yield a reflective equilibrium, not putting too much weight on the question of the place of value in the world of facts?

5.7.1 Religious Beliefs In thinking of the issue between naturalism and autonomous ethics, it is useful to consider analogous issues that arise in other areas. Consider religious beliefs. Our scientific conception of the world has no place for gods, angels, demons, or devils. Naturalists hold that there is no empirical evidence for the existence of such beings, nor for any sort of divine intervention in human history. Naturalists say that people's religious beliefs can be explained in terms of their upbringing and psychology without any supernatural assumptions, so these beliefs provide no evidence whatsoever for the truth of religious claims. Naturalists therefore incline toward scepticism and atheism, although naturalism might also lead to a kind of religious noncognitivism holding that religious language makes no factual claims about a supernatural realm but has a different function, for example, in religious ritual. Another approach to religion is for a believer to start with his or her initial religious beliefs, including beliefs in the authority of certain writings, and then to develop general principles and theories that would accommodate these beliefs, allowing modifications in the interest of more plausible general principles. This will continue until no further modifications seem useful in improving the organization and coherence of that person's views. Inevitably, many questions will remain unanswered, and these may include issues concerning the relation between that person's religious views and his or her scientific views, for example, as regards creation. But this is not a serious worry for autonomous religion, which will say this shows merely that science is not everything, or at least that there are things we do not know and perhaps never will understand. Naturalists say there is no reason to accept religious claims, because the

96

MORAL RELATIVISM

fact that people have the religious beliefs they have can be explained without any supernatural assumptions. Religious autonomists say there is reason to accept religious claims, at least for someone who begins with religious beliefs, since the process of generalization, systematization, and theory construction internal to religion will give that person no reason to abandon more than a few, if any, of those religious beliefs. Furthermore, certain supernatural events might be part of the correct explanation of the appearance of sacred texts, the occurrence of miracles, and particular religious experiences. There is at present no way to say how these religious explanations mesh with ordinary scientific conceptions, but that by itself is no more an objection to religion than it is an objection to science. Naturalists in ethics might urge this religious analogy as an ad hominem argument against those defenders of autonomous ethics who are not willing to take the same line with respect to religion.

5.7.2 Beliefs About the Mind There is another sort of issue in which an autonomous position comes off looking rather good, namely, the so-called mind–body problem. Here the naturalistic position corresponds to the thesis of physicalism, according to which all real aspects of mind must be features of the physical brain and central nervous system, its atomic or neural structure, or some more complex structure that the brain and nervous system instantiate. This may involve behaviourism or some sort of functionalism that treats the brain as an information-processing system like a computer in a robot. A few defenders of this approach, like Skinner (1974), conclude that there are no mental events, no mind, no consciousness, no sensation. (Rorty, 1965, sympathetically describes a similar view, ‘eliminative materialism’.) But most physicalists suppose that mental events and other aspects of mind do exist and can be identified with certain physical or structural or functional aspects of the brain and central nervous system. On the other side is autonomous mentalism, which holds that the physicalist hypothesis leaves something out. In this view, we know we are conscious, can initiate action, and have experiences of a distinctive phenomenological character and feeling. The physicalist hypothesis does not account for this knowledge. A computer or robot is not conscious. Although a robot can move, it does not act in the way people can act. And a robot has no sensuous experience. Indeed, something could have exactly the functional structure of the human brain and nervous system without

IS THERE A SINGLE TRUE MORALITY?

97

being conscious. Block (1978) describes a case in which one billion people in radio communication with each other model a particular brain for an hour, each person corresponding to a particular neuron in the brain. Block takes it to be absurd to suppose that this vast collection of people would have a group consciousness that was phenomenologically the same as the consciousness of the person whose brain and central nervous system was being modelled. Nagel (1974) observes that we might know everything there was to know about the neurophysiological structure and functioning of the brain and central nervous system of a bat without knowing what the experience of the bat was like. Defenders of autonomous mentalism agree that this leaves a mind–body problem. They agree that they are unable to say how consciousness, free will, and sensory experience can be part of the world described by physics. But they deny that this means we must stop believing in consciousness or must identify it with some aspect of physical or functional structure. For they claim, with considerable plausibility, that it is much more reasonable to believe in consciousness, free will, and sensory experience, and to believe that these are not aspects of neurophysiological functional structure, than it is to believe in physicalism. I do not say that autonomous mentalism is more plausible than physicalism. After all is said and done, I find a physicalistic functionalism more plausible than autonomous mentalism. My point is that autonomous mentalism is a perfectly respectable philosophical position. A defender of autonomous ethics might even argue that naturalism in ethics loses much of its plausibility once autonomous mentalism is recognized as plausible. For that casts doubt on the universal applicability of the naturalistic approach and therefore casts doubt on the naturalist's argument that a belief that something is right cannot be explained by that thing's being actually right unless that thing's being right consists in some psychological or sociological fact. The naturalist's only argument for this, it might be said, depends on accepting the general applicability of naturalism. But it is not obvious that this approach, is generally applicable, since it is not obviously correct as compared with autonomous mentalism. There is at least some plausibility to the claim that one's awareness of what red looks like is to be explained by appeal to an experience of redness that does not consist entirely in some neurophysiological event. It might be said that the naturalist has no argument against autonomous ethics, since the naturalist cannot take for granted the general applicability of naturalism.

98

MORAL RELATIVISM

5.7.3 Ethics Defenders of autonomous ethics argue that their approach represents the only undogmatic way to proceed. They say that naturalism begs the question in supposing that everything true must fit into a scientific account of the world and by supposing that the central question about morality is how, if at all, morality fits into such a scientific account. Defenders of naturalism reply that naturalism itself is the result of following the method of reflective equilibrium. Autonomous ethics begs the question by assigning a specially protected status to initial moral beliefs as compared, say, with initial beliefs about the flatness of the earth or the influence of the stars on human history. Naturalists say that, starting with our initial beliefs, we are led to develop a scientific conception of the world as an account of everything there is. In doing so, we also acquire beliefs about how we learn about the world and about how errors can arise in our thinking. We come to see how superstition arises. We begin to worry about our moral views: are they mere superstitions? We note certain sorts of disagreement in morality and extreme differences in moral customs. We observe that some people are not much influenced by what we consider important moral considerations. All this leads us to raise as a central question about morality how morality fits in with our scientific conception of the world. Naturalism is no mere prejudice in favour of science; it is an inevitable consequence of intelligent thought. This, at least, is what a defender of naturalism will say. A defender of autonomous ethics will reply that moral disagreements, differences in custom, and the behaviour of criminals prove nothing. All these things are compatible with moral absolutism. The naturalist retorts that any view can be made compatible with the evidence; astrology, for example, is perfectly compatible with the evidence. The issue is not what is compatible with the evidence, but what best accounts for it. The naturalist argues that relativism accounts for the evidence better than absolutism does, since relativism is able to say how reasons and values are part of the world science describes, whereas absolutism is not able to do that. The defender of autonomous ethics replies that such an argument is no better than the corresponding argument for behaviourism. Behaviourism is able to say how mental states (as it conceives them) are part of the world physics describes, and autonomous mentalism is not able to say how mental states (as it conceives them) are part of the world physics describes. But one should not for this reason alone abandon one's initial view that one is

IS THERE A SINGLE TRUE MORALITY?

99

conscious, makes decisions, has feelings, and so on, where this is not just being disposed to act in various ways. Something could have the dispositions without being conscious and could be conscious without having the dispositions. Similarly, one should not accept the naturalistic argument and give up one's belief in absolute values and universal moral reasons. I see no knockdown argument for either side. A question of judgement is involved: ‘Which view is more plausible, all things considered?’ To me, the relativistic, naturalist position seems more plausible. Others find the absolutist position of autonomous ethics more plausible. In this essay, I have not tried to argue that one side is correct. I have tried to bring out the central issue.

This page intentionally left blank

Part II Intrinsic Value

This page intentionally left blank

6 Toward a Theory of Intrinsic Value A normative theory of value might include an account of when something's being the case would or would not be a ‘good thing’, a ‘worthwhile thing’, or a ‘desirable thing’. Some philosophers believe that in order to give such an account, one need (and need only) say when something's being the case would or would not be intrinsically good, worthwhile, or desirable. To do this would be to present a theory of intrinsic value. In this essay, I examine what was once a standard account of intrinsic value as it appeared in certain textbooks (Hospers, 1961: 104–5; Frankena, 1963: 67; Brandt, 1959: 32, 1963: 18–20). I argue: (a) it is not clear whether a theory of intrinsic value can be developed along the lines of that standard account; (b) if one is to develop such a theory, one will need to introduce a notion of ‘basic intrinsic value’ in addition to the notion of ‘intrinsic value’; and (c) several different theories of intrinsic value may account for the same judgements concerning desirability, and it will be arbitrary to choose one of these theories over another. I discuss other ways of looking at intrinsic value in Essays 7 and 8.

6.1 Preliminary Remarks For the purposes of this essay, I consider the following expressions to be equivalent: ‘X is (or would be) a good thing,’ ‘X is (or would be) a worthwhile thing,’ ‘X is (or would be) a desirable thing,’ ‘X is (or would be) good,’ ‘X is (or would be) desirable,’ ‘X is (or would be) worthwhile,’ ‘X has (or would have) value.’ I am interested only in cases in which the X of these sentence forms is replaced with the nominalized form of some sentence. For example, taking ‘The cat is on the mat’ as the sentence to be nominalized, the first of these sentence forms could be realized in any of the following ways: ‘That the cat is on the mat is a good thing,’ ‘That the

104

INTRINSIC VALUE

cat would be on the mat would be a good thing,’ ‘For the cat to be on the mat would be a good thing,’ ‘The cat's being on the mat is a good thing,’ etc. Again, for the purposes of this essay, I consider all these sentences to be equivalent. (I will not be concerned with whether such nominalised sentences refer to anything or, if they do refer, with whether they refer to states of affairs, possible events, sentences, etc.) So, I am not concerned with cases in which the X is replaced by something that refers to an object in a restricted sense of ‘object’. For example, Beardsley (1965: 8) discusses the value of a rare postage stamp. The standard theory of intrinsic value does not have any application to such a case. It makes sense to ask whether my having such a stamp would be intrinsically worthwhile or whether such a stamp's existing would be intrinsically desirable. But, since it does not make sense to ask whether such a stamp is itself intrinsically valuable, apart from one or another of these possibilities, I will not be interested in discussing the value of objects like stamps, per se. I provisionally accept a distinction between the normative theory of value and a metaethical theory of value. I understand a normative theory of value to be a systematic account of when something's being the case would be a good thing. A metaethical theory is a theory about the meaning of value judgements in ordinary language and, possibly, a theory about the way in which these judgements are to be ‘proved’ or ‘validated’. I take the notion of intrinsic value to be a technical notion introduced to simplify the normative theory of value: once one has specified what has intrinsic value, one can then say that anything has value to the extent that it ‘leads to’ something having intrinsic value. I doubt that the notion of intrinsic value plays an important role in metaethics. For example, I do not think that it is intuitively obvious what things have intrinsic value; nor do I believe one ordinarily decides whether something's being the case would have value by discovering whether it would ‘lead to’ intrinsically desirable things. Frankena (1963: 67) disagrees. In the following passage he says we ordinarily determine what things have value on the basis of what things have intrinsic value: But it still remains true that in order to come to a judgement about whether something is good on the whole or good in any of the other senses, we must first determine what its intrinsic value is, what the intrinsic value of its consequences or of the experience of contemplating it is, or how much it contributes to the intrinsically good life.

TOWARD A THEORY OF INTRINSIC VALUE

105

This very strong claim about intrinsic value contrary to Beardsley's (1965: 7) claim that ‘The paradoxical feature of our value knowledge is just that we have a good deal of sound knowledge about instrumental values, but are in considerable doubt about intrinsic values’. Brandt (1959: 302) is rightly more cautious than Frankena: It would seem, however, that if anything is desirable at all, it must be because some facts or events (perhaps the thing itself) are intrinsically desirable. But a person would not be contradicting himself if he denied this statement, and it is an interesting question for reflection what one should say to a person who did deny it. In this essay I attempt to explain why it can seem that what is desirable depends upon what is intrinsically desirable; and I try to specify what would be involved in denying this claim.

6.2 The Standard Account Hospers (1961: 104–5) introduces the notion of intrinsic value in a standard way: Our question, ‘What things are good?’ is ambiguous. There are some things which we consider good (or desirable) only for their results—for what they lead to. There are other things which we consider good not because of what they lead to but because of what they are in themselves: we consider them worth having or pursuing not merely as ways of getting other things but because of their own intrinsic nature. The first kind of good is called instrumental good because the goodness or worthwhileness of these things lies in their being instruments toward the attainment of the other things which are considered good not merely as instruments. The second kind of good is called intrinsic good because we value these things (whatever they may turn out to be) not for what they lead to but for what they are. Brandt (1959: 302) says, ‘To say that something is intrinsically desirable is to say that it is desirable, taken just for itself, viewed abstractly, and in particular, viewed without respect to any consequences its existence will or may produce’. Brandt (1963: 18) offers what he calls ‘a rough definition of “intrinsic value” ’: ‘Something is intrinsically desirable if it would be properly valued or desired or chosen for its own sake, when considered independently of its consequences.’ His more precise explanation (p. 19) is: Some facts or events, then, are desirable for no reason beyond their being what they are; they are solely intrinsically desirable. Other facts or events are desirable

106

INTRINSIC VALUE

solely for reasons beyond themselves, usually their effects; they are not intrinsically desirable at all. Still other things are desirable partly just because of what they are but partly because of other things, pre-eminently their effects; they are in part intrinsically desirable but in part desirable for other reasons. Each account obscurely distinguishes two sorts of reasons for which something may be desirable. Hospers says the one sort of reason mentions what things ‘lead to’; the other sort mentions ‘what they are’. But how are we to tell when a reason mentions what S is rather than what S leads to? Similarly, how are we to understand Brandt's distinction between reasons for things' being desirable which are ‘their being what they are’ and reasons for things' being desirable which are ‘reasons beyond themselves’? Brandt (1963: 19) answers these questions by saying, of the ‘facts’ that might be cited as reasons: Now such facts that make an event desirable may or may not be different from the very fact or event originally said to be desirable and may be different from anything included in, or logically required or necessitated by, this fact or event. I think this remark can be interpreted as follows: A reason (for S's being desirable) refers to ‘what S is’ and is, therefore, ‘included in S’, if and only if the reason is that something is logically required or necessitated by S; so S is intrinsically desirable to the extent that what makes S desirable is that S entails something, T. S fails to be intrinsically desirable and is instrumentally desirable to the extent that what makes S desirable is something other than S's entailing something. (This interpretation does not quite fit what Brandt says. Instead of saying that S is intrinsically desirable if what makes S desirable is that S entails something, Brandt seems to say that S is intrinsically desirable if S entails what makes S desirable. However, this last remark is obscure when taken literally: and I think Brandt meant to say what I interpret him as saying.) It still remains to be explained how a reason why S is desirable can be what makes S desirable in the required sense. There are reasons why S is desirable that are not the sort of reason Brandt must have in mind. For example, if something is a reason why S is desirable in that it is the causal reason why S will have desirable consequences this time, then this reason would not be what Brandt has in mind when he speaks of what makes S desirable. Again, if something is a reason why S is desirable in that it provides a reason for inferring that S is desirable (e.g. the reason is that someone reliable has said that S is desirable), then such a reason would also not be what Brandt has in mind. Brandt suggests something of the nature of the relevant sort of reason when he equates ‘desirable’ with ‘properly

TOWARD A THEORY OF INTRINSIC VALUE

107

desired or chosen’ and remarks: ‘Something is intrinsically desirable if it would be properly desired or chosen for its own sake’ (p. 18). This indicates that the relevant sort of reason must be a reason for desiring or choosing the thing in question. If S is not already desired, reasons for desiring S typically cite S's connection with something that is already desired. The connection may be that S entails something already desired or that S would ‘lead to’ something already desired. We want to adapt these remarks to desirability; although we cannot of course make use of a literal distinction between what is already desirable and what is not already desirable (to parallel the distinction between what is already desired and what is not already desired). In effect, the standard account identifies what has intrinsic value with what is already properly desired. This works well for instrumental desirability: if S does not entail but ‘leads to’ something intrinsically desirable, then S is instrumentally desirable. However, if S entails something having intrinsic value, we want to be able to say that S is itself intrinsically desirable. Perhaps the best way to handle this case is to say that, if S entails something already properly desired, then S is already implicitly properly desired and is therefore intrinsically desirable. (The awkwardness of this manœuvre arises from the fact that the real parallel is not between what is already desired and what is intrinsically desirable, but is rather between what is already desired and what is basically intrinsically desirable. The notion of basic intrinsic value is discussed in section 6.5 below.) The standard account says: (1) S is intrinsically desirable to the extent that S entails something that is intrinsically desirable. (2) S is instrumentally desirable to the extent that S does not entail but ‘leads to’ something intrinsically desirable. Notice that neither of these propositions permits a noncircular definition of ‘intrinsic value’ or of ‘instrumental value’.

6.3 An Improvement in the Standard Account Hospers's introduction of the term ‘intrinsic value’, quoted above, suggests that the second type of reason for properly desiring S must be that S would ‘lead to’ things of intrinsic value. It is natural to interpret ‘S would lead to T’ to mean ‘S would cause or result in T’ or ‘T would be an effect

108

INTRINSIC VALUE

of S’. Brandt is nearer the truth. He says the reasons for which things are instrumentally desirable are only ‘preeminently their effects’ (p. 19). (I understand him to mean that the reasons are pre-eminently that these things have certain effects.) Even Brandt's claim is too strong. An important way in which T may provide a reason for properly desiring S is that S increases the likelihood of T, where T is desirable. S may be a sign of something desirable or of something undesirable, without necessarily being a cause of something desirable or undesirable. For example, whether or not it would be a good thing for me to receive one hundred thousand dollars sometime this year depends heavily on how it happens that I receive this money. If the only probable way in which this could happen requires the death of some member of my family so that I receive insurance, then I cannot think of receiving this money as a desirable thing (Jeffrey, 1965: 198). So, the desirability of S will be affected by the desirability of those things which S makes more or less probable. Furthermore, just how the desirability of S will be affected depends upon just how desirable these things are and just how S would affect their likelihood. To spell this out will require our being able to speak meaningfully of assigning some measure to desirability. Even if we were to restrict our attention to S's effects, it would be an oversimplification to say that S is desirable because of what it would lead to. Instead, we must say that S is desirable because of what it is likely to lead to. S may happen in various ways having different consequences. For example, your giving me ten dollars will have varying consequences depending on whether you give me the money as a gift, to hold for the purposes of a bet, or to give to someone else. Its consequences will also depend upon whether you give me the money in pennies, in dollar bills, or in the form of a cheque (Jeffrey, 1965: 190–2, 196–9). S's being desirable because of what it would lead to must be taken as a special case of S's being desirable because of what it would make more likely. Judgements about what has already happened are only apparent counterinstances to this claim. It is true that, when we judge Caesar's crossing the Rubicon to be a good thing, we base this judgement on the actual consequences of his crossing the Rubicon and not on the probable consequences of this crossing. It is also true that, relative to our present knowledge of those consequences, we cannot say that Caesar's crossing made or would make these consequences probable. However, the judgement that Caesar's crossing the Rubicon was a good thing amounts to the judgement that it was better than his not crossing the Rubicon. To evaluate this latter claim involves an estimate of what the consequences of his

TOWARD A THEORY OF INTRINSIC VALUE

109

not crossing would have been; and, since there are various ways in which it might have happened that he did not cross, one would have to estimate the probabilities of various consequences in a way similar to how one estimates the probabilities of consequences of something that has not yet happened. Although past events are not judged desirable or undesirable because of what consequences they made likely, they are judged desirable or undesirable because of what consequences alternatives would have made more or less likely. In what follows, I will not make use of the special formulations needed because of the indirect and counterfactual character of judgements about the past. Therefore, the improved standard account of intrinsic value says: (1) S is intrinsically desirable to the extent that S entails something that is intrinsically desirable. (2′) S is instrumentally desirable to the extent that S does not entail but increases the probability of something intrinsically desirable.

6.4 An Unexpected Problem We are closer to being able to say just what is involved in a theory of intrinsic value. According to our new version, the value or desirability of something S depends upon S's intrinsic value and also upon the intrinsic value of those things S makes more or less probable. Given (1) and (2′), this amounts to the claim that all value is either intrinsic or instrumental or both. In order to specify the theory further, we would like to be precise about what contribution something else can make to the value of S. Suppose that we are able to make sense of the notion of a definite amount of value. Then it would seem natural to say that T's contribution to the value of S is the intrinsic value of T multiplied by the amount that S increases the probability of T. But this will not do. Let V(X) be the overall value of anything X. Let I(X) be the intrinsic value of X. Let P(X) be the (unconditional) probability of X and let (P(X | Y) be the conditional probability of X, given Y. Then MP (X | Y) = P(X | Y) – P(X) is the amount that Y increases the probability of X. For any function F whose values are real numbers, positive or negative, let ∑XF(X)

110

INTRINSIC VALUE

be the sum of the values of F for every possible value of the argument X. Suppose also that S and T are identical if and only if they entail each other. Then the proposal under discussion is (3) For any (3) For any S,V(S)=∑T[I(T)×MP(T|S) The difficulty with the proposal is that, according to (1), something may have intrinsic value because it entails something else of intrinsic value. For example, suppose that S has intrinsic value. Then so does S & T, since S & T entails S. But, once we have given S & T credit for increasing the probability of something that has intrinsic value in its own right, we do not want to give S & T further credit for increasing the probability of something else, itself, that has intrinsic value just because it entails the first thing. For example, suppose that someone's getting pleasure at noon today would have an intrinsic value of ten units and has an initial probability of 0.50. Then my getting pleasure at noon today would also have an intrinsic value of ten units, and an initial probability of 0.50 or less, because it entails someone's getting pleasure. And my getting pleasure at noon today in this room would also have an intrinsic value of ten units, and an initial probability of 0.50 or less, for the same reason. My getting pleasure at noon in this room increases the probability by at least 0.50 of at least three things, each possessing ten units of intrinsic value: itself, my getting pleasure at noon, and someone's getting pleasure at noon. Using (3), my getting pleasure at noon would have to be worth at least fifteen units of value. But this result must be wrong.

6.5 Basic Intrinsic Value The example suggests that something may have intrinsic value in a more or less basic way. Someone's getting pleasure at noon today might have intrinsic value in the most basic way, whereas my getting pleasure at noon today in this room may have intrinsic value only in a derivative way, because it entails something that has intrinsic value in the most basic way. Something made more or less probable by S contributes to the value of S only if it possesses intrinsic value in the most basic way. If S makes probable my getting pleasure at noon today in this room and if this has intrinsic value only because it entails someone's getting pleasure at noon today, then we want to give S credit only for increasing the probability of someone's getting pleasure at noon today. It would be an error to give S further

TOWARD A THEORY OF INTRINSIC VALUE

111

credit for increasing the probability of my getting pleasure at noon today in this room. Therefore the theory of intrinsic value requires the introduction of the notion of basic intrinsic value and a basic intrinsic value function B satisfying the following two requirements. (Here I am indebted to Robert Nozick.) First, for any function F, let ∑T:S⇒TF(T)be the sum of the values of F for all arguments T such that S entails T, Then the intrinsic value of S is given by (4) I(S)=∑S⇒TB(T) The total value of S is given by (5) (5) V(S)=∑T[B(T)×MP(T|S)] So, the following statement represents a necessary condition of the truth of any theory of intrinsic value: (6) There is a function B, the basic intrinsic value function, such that the intrinsic value of any S is given by (4) and the value of S is given by (5). Statement (6) is a strong thesis. First, it presupposes that there is a measure of desirability susceptible to the sort of addition and multiplication envisioned in (4) and (5). Second, it also claims that there is at least one function, B, determining the values of everything by satisfying (5). Third, it claims that, of the functions satisfying (5), at least one represents basic intrinsic value in that it satisfies (4). None of these claims is obviously true. So it cannot be obvious that anything has intrinsic value.

6.6 A First-Cause Argument There is a familiar argument for the claim that some things must be intrinsically desirable. (Beardsley (1965: 6) calls this argument, ‘The Dialectical Demonstration’.) It goes as follows: Consider something A that is desirable. Either A is desirable for itself, and therefore something is intrinsically desirable, or A is desirable only because A leads to something else, B, that is desirable. Then either B is intrinsically desirable or B is desirable only because B

112

INTRINSIC VALUE

leads to something else, C, that is desirable. And so on. But if nothing is intrinsically desirable, how can A ever be desirable, since A is desirable only because B is desirable, B is desirable only because C is desirable, and so on ad infinitum, with no explanation why any of the members of this series is desirable? As it stands, this argument is no more compelling than the first-cause argument for the existence of God. There is nothing inconsistent in the notion of an infinite series A, B, C, etc., each member of which is desirable only because the next is. However, it is by now easy to see why one can think it obvious that something must have intrinsic value and why one might be tempted to argue for this using the ‘first-cause argument’ just outlined, if one identifies intrinsic value with basic intrinsic value. Suppose there is a basic intrinsic value function B satisfying (5). If nothing had basic intrinsic value, then B(T) = 0 for all T, and V(S)=∑T0×MP(T|S)=0,for arbitrary S. So, if nothing had basic intrinsic value, then (as the proponents of the ‘first-cause’ argument maintain) nothing would have any value at all. Since something does have value, it would follow that something must have basic intrinsic value. However, this conclusion follows only because one has assumed that there exists a basic intrinsic value function B, satisfying (5). Since the assumption is not obviously true, it is not obviously true that something has basic intrinsic value.

6.7 An Argument Against the Theory of Intrinsic Value Consideration of the following arguments helps to clarify what is involved in the theory of intrinsic value. The first is a familiar argument against the theory that all pleasure and only pleasure is of intrinsic value: Suppose that during a war a prisoner, Smith, is being tortured in an attempt to extract information from him. His torturer, Jones, is someone who has never done this sort of thing before. In a few minutes, Jones and Smith are going to die when a bomb destroys their building. Call these circumstances C. Then compare the relative desirabilities of S and S′, where S is that in C Jones enjoys himself, and where S′ is that in C Jones does not enjoy himself. S and S′ are quite similar in what they make more or less probable, except that S

TOWARD A THEORY OF INTRINSIC VALUE

113

entails that Jones will have pleasure that S′ entails Jones will not have. On the assumption that all and only pleasure is of intrinsic worth, S ought to be more desirable than S′. But S is less desirable than S′. (Suppose for the sake of argument that this premise is true.) Therefore, the assumption that all and only pleasure is of intrinsic worth must be false. A similar argument against the possibility of any basic intrinsic value function B continues as follows: Suppose there is a function B satisfying (5). Then,V(S)=∑T[B(T)×PM(T|S)]andV(S′)=∑T[B(T)×MP(T|S′)]The difference between these sums must result from the difference between S and S′. Since this difference is solely that S entails Jones will have pleasure that S′ entails Jones will not have, it follows that the respective sums will be exactly the same, except that the first sum will contain terms representing the value of Jones's pleasure, whereas the second sum will not contain these terms. V(S)=∑T[B(T)×PM(T|S)]=∑T[B(T)×MP(T|S′)]+(terms representing the value of Jones's pleasure)=V(S′)+(terms representing the value of Jones's pleasure).. The terms representing the value of Jones's pleasure are those terms in the sum for the value of P, where P is that Jones enjoys himself during a certain interval of time; so V(S) = V(S′) + V(P). Since V(P) > 0, it follows thatV(S) > V(S′). But, by hypothesis, V(S) < V(S′). The original assumption that there is a basic intrinsic value function B has led us into contradiction; so that assumption must be incorrect. The first argument, against the view that all and only pleasure is intrinsically good, seems valid. (The truth of the premise that S is less desirable than S′ is another matter.) The second argument, against the view that there is a basic intrinsic value function B, goes wrong in claiming that the terms representing the value of Jones's pleasure are simply the terms in the sum for V(P). In fact, there need be no simple relationship between the sums for V(S) and V(S′). The slight difference between S and S′ may make a great deal of difference in what S and S′ make more or less probable. S′ makes it certain that Jones is not taking pleasure in Smith's suffering, whereas P (i.e. Jones's getting pleasure over a certain interval of time) does not, taken by itself, make it highly probable that Jones is taking pleasure in

114

INTRINSIC VALUE

Smith's suffering; yet S does make it highly probable that Jones is taking pleasure in Smith's suffering. Although S makes it certain that Jones is getting pleasure, something to which the basic intrinsic value function will perhaps assign a positive value, S also makes it highly probable that Jones is getting sadistic pleasure, and the basic intrinsic value function might assign this a low enough intrinsic value so that, as a result, S ends up with a lower total value than S′. Therefore, the above argument cannot be used to show that there is no basic intrinsic value function B.

6.7 Possible Arbitrariness in the Theory of Intrinsic Value Beardsley (1965) offers a different type of argument against the theory of intrinsic value. He claims that ‘the concept of intrinsic value is inapplicable’—even if something has intrinsic value, we could not know it, and therefore that it can play no role in ethical or aesthetic reasoning (pp. 12–13). His argument is that, since we cannot discover the intrinsic desirability of S by discovering that S would lead to other desirable things, the only other way to discover S′s intrinsic desirability would be via immediate intuition. He takes this consequence to discredit the theory of intrinsic value. But it is not true that the only way to show that something is desirable is to show that it would lead to (i.e. make more or less probable) something else. Another way to show that something is desirable is to show how the assumption that it is desirable would explain why other things are desirable. In other words, we might use in ethics the same sort of argument we would use elsewhere, an argument that something is the best explanation of our data. And that is just the way in which one would want to defend a theory of intrinsic value. One would want to argue that it is possible to simplify and thus explain the theory of value by assuming, for example, that someone's getting pleasure is always of basic intrinsic value. Beardsley has not shown that such an argument could not be given for some theory of intrinsic value. Therefore, he has not shown that ‘even if something has intrinsic value, we could not know it’. Nevertheless, it is not obvious that the theory of intrinsic value can be defended in the way just suggested. Beardsley points out (personal communication) that there may be more than one function B, satisfying (5). If so, then these different B's may define different intrinsic value functions I, by virtue of (4), and one would need to have some way of deciding

TOWARD A THEORY OF INTRINSIC VALUE

115

which such function was the real intrinsic value function. (Of course, there may be several functions B, satisfying (5), each defining what turns out to be the same intrinsic value function by virtue of (4).) For example, suppose one wants to defend the view that getting pleasure is always and only of intrinsic value. It would seem that the same values of V will be determined if we take someone's getting pleasure to be of basic intrinsic value, or if we take it that, for any person P, P's getting pleasure has basic intrinsic value. This would seem to be a case where different B's might satisfy (5). But (4) will define different intrinsic value functions depending on which B is selected. If we say that, for any P, P's getting pleasure is of basic intrinsic value and is the only sort of thing that is, then it will not follow that either Jack's getting pleasure or Bill's getting pleasure will be of intrinsic value, since, for any P, this does not entail P's getting pleasure. On the other hand, if we say that someone's getting pleasure is of basic intrinsic value, then either Jack's getting pleasure or Bill's getting pleasure will be of intrinsic value, since it entails someone's getting pleasure. Therefore, there may be several intrinsic value functions satisfying (4) and (5). But if there are several such functions, nothing said so far can choose among them, since all the restrictions stated so far have been encapsulated in (4) and (5). This suggests that we ought to give up the idea of a unique intrinsic value function giving the real intrinsic value of various things. Instead we ought to take the choice among the functions satisfying (4) and (5) to be arbitrary in the way that the choice of axioms for a formalised theory is arbitrary. A particular choice of axioms is made for reasons of simplicity and economy; and one would want to choose the basic intrinsic value function for similar reasons.

6.9 Conclusion There are, no doubt, other problems with the conception of intrinsic value outlined in this essay. For example, if an infinite number of things have basic intrinsic value, then the sum in (5) may not be well defined. This problem would arise if, following a suggestion in Jeffrey (1965: 196–9), we took the things having basic intrinsic value to be represented by what Jeffrey refers to as ‘novels’. These are infinite sets of sentences such that, for every S, either S or not-S is included in any particular set. Since, in general, S will increase the probability of any novel by an infinitesimal

116

INTRINSIC VALUE

amount, the sum in (5) cannot be used to give V(S) on this proposal. On the other hand, there is no reason why we must accept this proposal. I cannot think of a way to show there is a function B, satisfying (5); nor can I think of a way to show there is no such function. It would be useful to discover such a function B, or prove that there can be no such function. Until this has been done, it is difficult to evaluate the claim made by Frankena, Brandt, and Hospers, and possibly denied by Beardsley, that the notion of intrinsic value enables us to simplify the theory of value. Alternatively, we might use Brandt's equation: desirable = properly desired. For example, one might distinguish intrinsic desires from instrumental desires, and equate intrinsically desirable with properly desired intrinsically. Then one might concentrate on developing a theory of intrinsic desires. This idea is explored further in Essays 7 and 8.

7 Desired Desires A number of philosophers have appealed to ‘second-order desires’ (desires about desires) in order to explain basic moral notions. Frankfurt (1971) suggests that freedom of the will might be identified with the ability to satisfy a certain sort of second-order desire; Jeffrey (1974) makes a similar proposal. Watson (1975) objects that the resulting account of freedom of the will does not work and that the appeals made to second-order desires should be replaced with references to what an agent values. Watson's view can be reconciled with Frankfurt's if valuing is identified with a kind of second-order desire. Lewis (1989) offers such an identification and Copp (1993) suggests that we might identify a person's values with preferences for which the person has a certain second-order preference. However, these appeals to second-order desires in order to explain freedom of the will, valuing, or values are all unsuccessful. In arguing for this negative conclusion, I defend a number of subsidiary points: (1) Positive intentions are reflexive or self-referential; (2) Intrinsic desires are not just noninstrumental desires; (3) To value something is in part to want it; (4) There is sometimes a difference between believing that something is good and valuing it; (5) There is a difference between valuing something and having it as one of your values.

7.1 Frankfurt on Freedom of the Will Frankfurt motivates his view by criticizing Strawson's (1959) account of the concept of a person. Strawson holds that the concept of a person is ‘the concept of a type of entity such that both predicates ascribing states of consciousness and predicates ascribing corporeal characteristics . . . are equally applicable to a single individual of that single type’ (pp. 101–2). Frankfurt points out that Strawson's suggestion fails to distinguish people

118

INTRINSIC VALUE

from many other animate beings and announces his own view that ‘one essential difference between persons and other creatures is to be found in the structure of a person's will. . . . [Persons] are able to form what I shall call “second-order desires” or “desires of the second order” . . . No animal other than man . . . appears to have the capacity for reflective self-evaluation that is manifested in the formation of second-order desires.’ Frankfurt explains that he means ‘desire’ loosely to include anything a person can be said ‘to want’, where this covers ‘a rather broad range of possibilities’ (pp. 6, 7, 12–13). Frankfurt says, ‘it is having second-order volitions, and not having second-order desires generally, that I regard as essential to being a person’, where a second-order volition is wanting a certain desire to be one's will, that is, wanting a certain desire to be ‘effective in action’. He uses the term ‘wanton’ to refer to agents who have first-order desires but who are not persons because, whether or not they have desires of the second order, they have no second-order volitions. . . . The class of wantons includes all nonhuman animals that have desires and all very young children. Perhaps it also includes some adult human beings as well. In any case, adult humans may be more or less wanton; they may act wantonly, in response to first-order desires concerning which they have no volitions of the second order, more or less frequently. (1971: 10–11) To illustrate his thesis, Frankfurt distinguishes between two kinds of drug addict: an unwilling addict, who wants his desire for the drug not to be effective, and a wanton, who is not ‘concerned whether the desires that move him to act are desires by which he wants to be moved to act’. Frankfurt claims that the wanton addict may deliberate and make plans to obtain the drug he desires. ‘But it never occurs to him to consider whether he wants the relations among his desires to result in his having the will he has’ (p. 12). Frankfurt connects the capacity for forming second-order volitions and the capacity ‘both of enjoying and of lacking freedom of the will’, where freedom of the will is not just ‘freedom of action’. He says: freedom of action is (roughly, at least) the freedom to do what one wants to do. Analogously, then, the statement that a person enjoys freedom of the will means (also roughly) that he is free to want what he wants to want. More precisely, it means that he is free to will what he wants to will, or to have the will he wants. . . . It is in securing the conformity of his will to his second-order volitions, then, that a person exercises freedom of the will. (1971: 14, 15)

DESIRED DESIRES

119

7.1.1 Wanting Someone to Do Something for You One way to begin to explore Frankfurt's proposals is to consider the different ways in which you might want someone to do something for you. In one case, you want someone to do something for you because that person knows that you want it. For example, you want a friend to bring you flowers as a result of your friend's recognition that you want flowers. In a different case, brought to my attention by Claudia Mills, you want someone to do something but not because of your desire. For example, you want a doctor to say you are going to recover because he thinks you are going to recover, not merely because he realizes you want him to say that you are going to recover. Often when you want someone P to do something for you, your hope is that P will recognize your desire and be motivated by that recognition to do what you want. In such cases, you have a second-order desire that your first-order desire will be effective in getting you what you want, via its recognition by P. Second-order desires also figure in H. P. Grice's (1957) theory of communicative meaning. According to Grice, a speaker S communicating with an audience A means that M only if S intends A to think that M by virtue of A's recognition of S's intention that A should think that M. Now, intentions count as desires for Frankfurt because, if you intend to do something, there is a sense in which that is what you want to do (even if there are other respects in which you do not want to do it). So, wherever Grice's analysis of communicative intentions applies, such communication will involve a second-order desire that a first-order desire should be effective. To have such a second-order desire is not yet to have a second-order ‘volition’ in Frankfurt's sense, because it is not the same as having the second-order desire that your first-order desire to do something should be effective in action. To have the second-order desire that your first-order desire for flowers will be effective in getting you what you want via your friend's recognition of that desire is not quite the same, for example, as having the second-order desire that your first-order desire to say, ‘I'd love some flowers,’ should be effective in leading you to say, ‘I'd love some flowers.’ Everyone, or almost everyone, has second-order nonvolitional desires, because everyone, or almost everyone, has communicative intentions. In particular, almost any actual human addict will have communicative intentions in trying to purchase the needed drug. Almost any addict will be capable of wanting a friend to provide some of the drug on recognizing

120

INTRINSIC VALUE

the addict's desire for it. Furthermore, it is likely that some animals can have communicative intentions.

7.1.2 Positive Intentions I have argued earlier (1976; 1986: ch. 8) that intentions are typically self-referential and therefore are not only firstorder, but second-order and every higher-order as well. In particular, an intention is typically the intention that that very intention will result in your doing what you intend to do. Searle (1983: ch. 3) defends a similar conclusion by way of a different argument. A ‘positive intention’ is an intention to do something that you suppose you will do in consequence of your adopting that intention. For example, consider the intention to clean up your room. Normally, this would be a positive intention, because when you form such an intention you suppose that you will clean up your room only if your having that intention leads to your doing so. Positive intentions contrast with negative intentions, as for example when you decide not to clean up your room, where you suppose that what you intend, namely that you not clean up your room, will be so whether or not you form the negative intention not to clean up your room, as long as you do not form the positive intention to clean up your room. There are also conditional intentions, like the intention to clean up your room if asked to do so, where you do not unconditionally suppose that the intention will lead to success but suppose only that it will have an effect if you are asked to clean up your room. There is a sense of ‘doing something’ in which cleaning up your room is doing something and not cleaning up your room is not doing something. The positive intention to clean up your room is an intention to do something. The negative intention not to clean up your room is not an intention to do something in this sense. (The distinction is not merely a grammatical distinction. The intention to stay home can be a negative intention that is not an intention to do something in the relevant sense even though the intention is not negatively expressed.) Doing something is active, an action. What makes an event your action in this sense is in part that it is controlled or ‘guided’ by your intentions. The term ‘guided’ is used in Frankfurt (1978). Mele (1987) also takes an action to be guided by an intention or plan. It may be that usually a plan guides an action by virtue of causing aspects of the action, but, as Frankfurt points out, there are also cases in which there is merely a conditional dependence between plan and action, a readiness to intervene if

DESIRED DESIRES

121

things go off track. A positive intention to run someone down blends into a merely conditional intention to turn the wheel if needed to adjust the aim of your car. To intend to do something that is an action is therefore to intend that there be an event of a certain sort that is controlled or guided by your intentions. So, a positive intention is an intention that something will happen in a way that is controlled or guided by your intentions. In particular, your intention is that something will happen that is controlled or guided by that very intention. So, a positive intention is reflexive or self-referential—it refers to itself. The practical reasoning that culminates in a positive intention has to be sensitive to whether the conclusion of the reasoning will make a relevant difference and lead to your doing what you intend to do. Negative and conditional intentions aside, you can't rationally decide to do something if you believe that your decision will make no difference. It is not enough to suppose that what you propose to do will make a difference; you must also suppose that your decision will make a difference. If you fall off a cliff and realize that you will continue to fall whatever your intentions are, then you cannot rationally decide to stop falling. What you would propose to do, namely, to stop falling, would make a difference to what happens: things would work out differently if you could stop falling. The trouble is that a decision to stop falling won't make a difference. A decision to stop falling can be a reasonable practical conclusion only if you can suppose that your practical conclusion can influence what happens. If you can't conclude that your intention will lead to your doing what you intend, then you can't reach that conclusion as the result of practical reasoning. And that is just what would be expected if the intention (that is, the conclusion of your practical reasoning) includes the thought that it leads (in a more or less specific way) to an intended result. Furthermore, I can't think of any other natural way to capture this point without appeal to self-referential intentions. You can rationally form a positive intention to do something only if you can see how your intention might lead you to do it. You do not have to have all the details worked out if your plan is that you will work out the details later, as needed. But you do need to be able to suppose that whether you do it or not might depend on whether you have the intention to do it, at least in the sense that your intention might contribute to your doing it. If you cannot see your intention as a possible means, you have no reason to form the intention. So, in forming that intention, you have to intend that your intention will contribute to the intended effect. Your full plan is

122

INTRINSIC VALUE

that your intention will lead in a certain (more or less clearly specified) way to your acting in the way you intend. There are two arguments here for the conclusion that positive intentions are self-referential. First, when an intention is the result of practical reasoning, the intention is the conclusion of that reasoning. You can rationally reach a positive intention through practical reasoning only if you can conclude (C) that the intention will make a difference by leading to your doing as you intend. The conclusion (C) that your intention will lead to your doing as you intend has to be part of your practical conclusion. (C cannot follow your practical conclusion P because P can be reached only if C has already been reached. C cannot precede P either, since C has to be a conclusion about that particular P: C makes sense only if P exists.) So, the conclusion (C) that your intention will lead to your doing as you intend has to be part of your intention. Second, if an intention to do A is a positive intention, you suppose you will do A only if you intend to do A and you suppose that intending to do A promises to lead to your doing A. To make such suppositions about your intention is to view your intention as part of your means for doing A. If you form the intention of doing A, supposing that M will be part of your means for doing A, then your intention to do A includes an intention to do A by virtue of M. So, your positive intention to do A is in part the intention to do A by virtue of having that intention.

7.1.3 Ends and Means Frankfurt (personal communication) offers the following objection: [O]ne usually selects a means in the light of an end one has adopted; but since the formation of an intention is itself the adoption of an end, how can it also be the selection or provision of a means to that end? it would be as though every end automatically came provided with one useful means to attaining it—namely, the intention to attain it! . . . In deciding upon what means to use in seeking a certain end, I will have reasons for regarding M as an appropriate means or for selecting it over other possible means. My reasons for forming an intention are not like that. They are simply my reasons for adopting the end; they are not reasons, given that the end has been adopted, for using certain means to achieve it. It is true that I cannot rationally adopt an end unless I believe that I might be able to achieve it. But that is not the same as believing that my intention to achieve it might lead me to achieve it.

DESIRED DESIRES

123

My response is that we must distinguish the narrow notion of intention from the wider notion of desire. All intentions are desires, in Frankfurt's wide sense of ‘desire’, but not all desires are intentions. So we have to distinguish intended ends from merely desired and not intended ends. One can desire an end without any idea of how it might be obtained; but one cannot intend to obtain an end without some idea of how it might be obtained. I agree that one usually selects a means in the light of an end one has adopted. But that does not mean one must first select an end and only then select a means to that end. My claim is that, if selecting an end is forming a positive intention, one can select that end only if one has some idea of how to achieve it. Such an idea is an idea of what one might do to achieve that end. But an idea of what one might do is an idea of one or more events that are controlled or guided by one's intention. So one can select an end only if one has a conception of how one's intention might lead to or guide or control one's achieving that end. In accepting that end, what one accepts is a plan that includes an account of how one's acceptance of that plan leads to one's achieving that end. So I am arguing that every such end does come ‘provided with one useful means to attaining it—namely, the intention to attain it!’ One's reasons for forming an intention are not just one's reasons for adopting the intended end rather than some other end, since one's reasons for forming an intention also include one's reasons for forming an intention as opposed to merely desiring or hoping for that end. One's acceptance of a given end or goal is part of a larger practical conclusion. The reasons why some parts of that larger conclusion are there are not the same as the reasons why other parts are there. One's reasons for including a given end or goal are not the same as one's reasons for including a particular means to that goal. One selects the means in order to achieve the goal, not the other way round. But one's whole practical conclusion is what one intends. One intends the whole plan, not just the final end. One's plan is that one's intention will lead in a certain way to a certain result. In other words, one's whole practical conclusion involves as a part the thought that that whole conclusion will lead in a certain way to a certain result. The relevant sense in which you cannot rationally adopt an end unless you believe you might be able to achieve it is this: you cannot rationally adopt an end as part of a positive intention unless at the same time you can rationally conclude (as part of your plan) that your having that intention

124

INTRINSIC VALUE

will lead to your achieving that end. Whether or not you believe that your intention will lead you to achieve it, your plan must be that your having that intention will lead you to achieve it.

7.1.4 Innite Regress ‘If positive intentions are self-referential, isn't a bad infinite regress generated?’ To have a positive intention to do D is then to have an intention that your intention will lead you to do D. ‘So do you not also have an intention concerning your intention that your intention will lead you to do D? And so forth?’ (Frankfurt, personal communication, 1990). My reply is that the regress is not a ‘bad infinite regress’. There is only a single intention that refers to itself. It is the intention that that very intention will lead in such and such a more or less specified way to such and such a result. In words: my acceptance of this very conclusion will lead to such and such a result. I discuss issues raised by self-reference further in Harman (1986: 87–8).

7.1.5 A Baby's Intentions Mele (1987) objects that his then eight-month-old daughter can intend to crawl towards him without having any concept of intention and, he argues, if she does not have the concept of an intention, her intention in crawling towards him cannot be in part that her intention will lead her to get to him. He concludes that we must reject the thesis that positive intentions are always self-referential. I reply that the baby does not need a full-fledged concept of intention in order to have self-referential intentions. The question is whether she acts as she does because of a complex planning procedure that is sensitive to whether and how the conclusion of this procedure might affect what happens. It is irrelevant whether the baby has a concept of intention beyond that. For example, the baby does not need to have the concept of a state of intention that others might be in. It is also irrelevant whether the baby has an explicit mental representation of her intention. (Wilson, 1989: 277–8, expresses scepticism about such explicit mental representation.) Perry (1990) observes that a child can have the thought that ‘it is raining’ without having any concepts of places or times and without any inner mental representations of particular places and times, even though the content of the child's thought concerns rain at a particular place and a particular time. The child can have thoughts that are about particular

DESIRED DESIRES

125

places and times without having concepts or representations of those places and times. Similarly, the content of a child's intention can be in part about that intention without the child having any distinctive concept or mental representation of that intention. Furthermore, it is important to distinguish purely verbal questions about how we use the word ‘intention’ from more substantive questions concerning what goes on in full-fledged practical reasoning leading to decisions and plans. As Mele observes, it is an empirical question what leads his eight-month-old daughter to crawl towards him as she does. It may very well be true that his daughter acts as she does because of a complex internal planning procedure that is highly sensitive to whether and how the conclusion of this procedure might affect what happens. Or it may be true that there is no such procedure going on when his daughter crawls towards him. The purely verbal issue is whether it would be correct to describe his daughter as ‘intending’ to crawl towards him on the assumption that there was no such inner procedure. Suppose, for example, that she was simply disposed to crawl in the direction in which she saw him, whether or not there was an obstacle, with no planning at all concerning how to get to him. If so, would it be correct to describe her as ‘intending’ to crawl towards her father? Mele himself may want to say this. Then such ‘intentions’ are (presumably) not about themselves. But this does not touch the substantive point that positive intentions that are the conclusions of full-fledged practical reasoning have to be about themselves, since full-fledged practical reasoning is sensitive to the effects of accepting the practical conclusions to which it leads.

7.1.5 Autonomy and Second-Order Volitions Recall Frankfurt's proposal: the difference between an autonomous agent and a wanton is that an autonomous agent but not a wanton has second-order volitions, where a second-order volition is a second-order desire that a first-order desire should be effective in action. Frankfurt uses the term ‘desire’ loosely to include any case in which a person might be said to ‘want’ to do something. Therefore, intentions count as desires for Frankfurt. So any positive intention is a second-order volition of the relevant sort. A positive intention is in part the intention (desire) that one's intention (desire) should be effective. This implies that any creature with positive intentions is an autonomous agent and not a mere wanton.

126

INTRINSIC VALUE

But then Frankfurt must be wrong about animals being nonautonomous wantons. Animals are capable of avoiding obstacles and of sufficient (even if limited) planning, which means that they must be capable of the sort of practical reasoning that is sensitive to which practical conclusions might be effective in action. So animals must have secondorder volitions in Frankfurt's sense. Consider also Frankfurt's distinction between the willing addict, the wanton, and the unwilling addict. Frankfurt says that all three addicts want to take the drug. The unwilling addict desires not to desire to take the drug and desires that his desire to take the drug not be effective. But his first-order desire to take the drug is stronger than his desire that the first-order desire not be effective. So he is unfree. He acts against his will even though he does what he wants. By contrast, the willing addict is quite happy that his desire for the drug should be effective and the wanton has no desire one way or the other as to whether his desire for the drug should be effective, according to Frankfurt. The trouble is that both the unwilling addict and the alleged wanton intend to take the drug. This intention is a positive intention, namely, the intention that that very intention will lead him to take the drug. Given Frankfurt's loose sense of ‘desire’, this counts as a second-order volition in Frankfurt's sense. So, Frankfurt has to count all three addicts as autonomous agents with free will. I conclude that Frankfurt's appeal to second-order volitions is not the key to distinguishing autonomy from nonautonomy and does not allow an account of freedom of the will.

7.2 Lewis on Valuing I now turn to Lewis's (1989) proposal to identify valuing something with desiring to desire it. Lewis's stated reasons for this identification are perfunctory; he is mainly concerned to argue for a dispositional theory of value, a theory with which I will not be concerned. Lewis motivates his account of valuing as follows. He first suggests that valuing something ‘might be a feeling, or a belief, or a desire. Or a combination of these; or something that is two or three of them at once; or some fourth thing. But let us set these hypotheses aside, and hope to get by with some thing simpler’ (p. 14). Notice, by the way, that Lewis must here be using the term ‘desire’ broadly, in the way Frankfurt does, to include anything

DESIRED DESIRES

127

a person might be said to want. Otherwise the threefold classification would not be plausible. So, in particular, Lewis too must count an intention as a desire. After ruling out the hypothesis that valuing is just a feeling or just a belief, Lewis is left with the conclusion that valuing is a desire. ‘But’, he adds, we'd better not say that valuing something is just the same as desiring it. That may do for some of us: those who manage, by strength of will or by good luck, to desire exactly as they desire to desire. But not all of us are so fortunate. The thoughtful addict may desire his euphoric daze, but not value it. Even apart from all the costs and risks, he may hate himself for desiring something he values not at all. It is a desire he wants very much to get rid of. He desires his high, but he does not desire to desire it, and in fact he desires not to desire it. He does not desire an unaltered, mundane state of consciousness, but he does desire to desire it. We conclude that he does not value what he desires, but rather he values what he desires to desire. (1989: 115, with references to Frankfurt and Jeffrey) After saying that it is hard to imagine more complicated cases involving desires of a higher order and also that we have to allow for ‘someone who desires to value differently than he does’, Lewis ‘hesitantly’ concludes, ‘we do better to stop on the second rung: valuing is just desiring to desire’. However, there is an important footnote: It is comparatively easy to imagine instrumental third-order desires. Maybe our addict wishes he could like himself better than he does; and not by doing away with his addiction, which he takes to be impossible, but by becoming reconciled to it and accepting himself as he is. Or maybe he just fears that his second-order desire not to be addicted will someday lead him to suffer the pains of withdrawal. Either way, he wants to be rid of his second-order desire not to be addicted, but he wants it not for itself but as a means to some end. This is irrelevant: presumably it is intrinsic, not instrumental desiring that is relevant to what someone values. This footnote is puzzling. You can value something instrumentally. For example, you may value someone's opinion without having an intrinsic desire to know that person's opinion and without having an intrinsic second-order desire to want to know that opinion. Why then does Lewis presume that only intrinsic desiring is relevant to what someone values? Perhaps he means that his theory only works for valuing something intrinsically. If so, Lewis's view is, more specifically, that to value something intrinsically is to have an intrinsic desire that you have an (intrinsic) desire for it.

128

INTRINSIC VALUE

7.2.1 Communicative Intentions and Values One way to motivate Lewis's footnoted restriction to intrinsic value and intrinsic desires is to recall our earlier discussion of communicative intentions and second-order desires that first-order desires should be effective in getting you what you want. Consider what the theory would look like without this restriction and, in particular, consider the following scenario. You want flowers, you want John to recognize that you want flowers, and you want this to lead John to give you flowers. You want to do something that will indicate to John that you want flowers. You want to say to John, ‘I would love to have some flowers.’ So, you want to form the intention of saying it. Here you want to have the intention of saying, ‘I would love to have some flowers.’ Since an intention counts as a desire for Lewis, you have a desire for a certain sort of desire that you say this. So Lewis's theory without any restriction to intrinsic desires would imply that you value saying this. But normally in this situation this would not be something you value. By restricting the relevant desires to intrinsic desires, Lewis avoids the conclusion that you value saying, ‘I would love to have some flowers.’ In the scenario given, only the flowers are wanted intrinsically. Your desire to say, ‘I would love to have some flowers,’ is an instrumental desire, since you want to say it because of hoped-for results of saying it. Furthermore, you have only an instrumental desire to intend to say it, because of results you expect from having that intention. At this point, we need to say more about the distinction between intrinsic and other sorts of desires.

7.2.2 Instrumental and Intrinsic Desires Your desire for D is instrumental if you desire D because you think getting D will get you something else E that you desire. You see D as a means to E. It might be proposed that we define intrinsic desires by contrast with instrumental desires. If you desire D but not as a means to anything else, then your desire for D would be an intrinsic desire, according to this criterion. But, consider a desire to hear good news—for example, the desire to hear the doctor say you will recover. This is not an instrumental desire; you do not suppose that the doctor's saying this would promote any end of yours. So it would be an intrinsic desire, by the proposed criterion.

DESIRED DESIRES

129

You might believe that your having a certain desire is good news in this sense. Here is an example from the causal decision literature (Lewis, 1981: 9; Jeffrey, 1983: 15). You might believe that a certain bodily condition makes you resistant to cancer and also causes you to have an intrinsic desire to exercise. An intrinsic desire to exercise would be a sign that you have that bodily condition. Reflecting on this leads you to desire that you have an intrinsic desire to exercise. This latter second-order desire would count as intrinsic by the suggested criterion, because it is not an instrumental desire. It is not that you want the effects of a desire to exercise; you want to desire to exercise, because that would be a sign of your having a desirable bodily condition. On the assumption that a desire is intrinsic to the extent that it is not instrumental, Lewis's account of valuing would imply incorrectly that to have this sort of desire to have an intrinsic desire to exercise would be to value exercise. The implication would be incorrect, because you can have such a second-order intrinsic desire without valuing exercise at all. The best way to avoid this unwelcome result is to define what is to count as an intrinsic desire in such a way that this particular second-order desire does not count. For example, we might stipulate that your desire for D is not an intrinsic desire to the extent that you desire D only because you desire something else E and you believe that having D would make it more likely that you have E, either because D might bring about E or because D might be a sign of E. Now consider the following case. Someone tells you that the experience of listening to Mozart is intrinsically desirable. You believe this. Since you want to do things that are intrinsically desirable, you come to desire that experience. Is this last desire to have the experience of listening to Mozart an intrinsic desire? No, because (as Lewis pointed out to me in discussion) it depends on your desire to do things that are intrinsically desirable. One moral is that to want something only because you think it is good is not to want it intrinsically. This may bear on Quinn (1993). Quinn argues that desire is always based on the thought that something is good. His view would seem to have the implausible consequence that there are no intrinsic desires. However, the issue is too complex for discussion here.

7.2.3 Valuing and Believing Valuable Notice that, quite apart from considerations of intrinsic desire and intrinsic value, you can believe that something is valuable without valuing it. In the Mozart example just given, you think it would be valuable to

130

INTRINSIC VALUE

listen to Mozart, although you do not yet value listening to Mozart. There are instances in which you take something to be valuable that you once valued but out of weariness no longer value, as Stocker (1979) points out. Through spiritual or physical tiredness, through accidie, through weakness of body, through illness, through general apathy, through despair, through inability to concentrate, through a feeling of uselessness or futility, and so on, one may feel less and less motivated to seek what is good. One's lessened desire need not signal, much less be the product of, the fact that, or one's belief that, there is less good to be obtained or produced, as in the case of a universal Weltschmertz. Indeed, a frequent added defect of being in such ‘depressions’ is that one sees all the good to be won or saved and one lacks the will, interest, desire or strength. (1979: 749) In these examples, one ‘sees all the good to be won’, so, as in the Mozart example, one believes that something is of value, but without valuing it. Commenting on this passage from Stocker, Smith (1992) wrongly takes it to show that ‘we may value something without desiring it’. But the examples given are examples in which a person has an evaluative belief without valuing what he takes to be good. We may think that something is good without desiring it, but we cannot value something without desiring it. Just as it is important to distinguish believing that something is good from valuing it, it is also important to distinguish the belief that something is good from the belief that it is valuable. A good baseball bat need not be a valuable baseball bat, a good discussion is not necessarily the same thing as a valuable discussion, a good life is not the same thing as a valuable life, and honesty is good whether or not it is also valuable.

7.2.4 Valuing and Desiring I now want to describe some apparent counterexamples to David Lewis's proposal that to value something intrinsically is to have a second-order intrinsic desire to have an intrinsic desire. First, I describe a case in which you have the relevant second-order desire without valuing the thing in question. Second, I describe a case in which someone values something without having the relevant second-order desire. Consider, first, what is involved in valuing listening to Mozart. Simply desiring to desire listening to Mozart is clearly insufficient. You may wish you did desire to listen to Mozart, because it would make your life more pleasant to have such a desire, given all the Mozart concerts you have to

DESIRED DESIRES

131

attend. In that case, your desire to desire listening to Mozart would be an instrumental desire for an intrinsic desire. Clearly in such a case you still do not value listening to Mozart, Furthermore, it is not enough that you should also acquire an intrinsic second-order desire to want to listen to Mozart, if you continue to lack any first-order desire to listen to Mozart. To me, this is as clear as the case in which you have merely an instrumental desire to want to listen to Mozart. If you do not yet want to listen to Mozart, then you do not yet value listening to Mozart. Furthermore, someone who likes to listen to Mozart can value listening to Mozart without having any second-order desire to have (or to continue to have) that first-order desire. A person can value listening to Mozart and be quite neutral in his attitude about his desire to listen to Mozart. He might realize that his musical interests are easily changed. For all he knows, he might at any point stop being interested in listening to Mozart, and come to be interested in listening to Mahler instead. He realizes that he might stop valuing listening to Mozart and instead start valuing listening to Mahler. This person might have no desire at all that this not happen—no desire at all that he should continue to want to listen to Mozart. He can value listening to Mozart without wanting to want it. I find it difficult to imagine an actual case in which a person would acquire an intrinsic second-order desire to desire to listen to Mozart. It would not be enough for the person to have a second-order desire to want to listen to Mozart that arises out of the belief that it is an intrinsically good thing to desire to listen to Mozart. As we have seen, the resulting desire would not be an intrinsic desire, since it would depend on the existence of the person's desire to have intrinsically desirable desires. By the way, although it seems to me to be generally true that, when a person values something, the person does not have an intrinsic second-order desire of the relevant sort, it is not always easy to show this. Consider valuing such things as (1) being an honest person, (2) accomplishing something significant in your life, and (3) the well-being of particular people, especially members of your family. I am inclined to believe that, although I have second-order desires to have first-order desires for these things, my second-order desires are almost certainly not intrinsic desires. My desire to desire the well-being of members of my family, for example, seems to me to be an instrumental desire, arising in part from the thought that my desiring this will promote their well-being. Similarly for the other values mentioned. I concede that my having instrumental reasons for such second-order

132

INTRINSIC VALUE

desires does not guarantee my not also having intrinsic second-order desires of this sort. Furthermore, I find it unclear how to test for the possibility that these desires might be at least partly intrinsic. But I see no reason to suppose that these second-order desires of mine are at least partly intrinsic. Suppose I were to be placed in a situation in which I could no longer have any effect on the welfare of members of my family. I would be distressed at the further prospect of no longer caring about them. Does that show that I have an intrinsic desire to (continue to) care about them? No, because I don't want to be the sort of person who does not care for his family. My distress would be due at least in part to my instrumental second-order desire to continue to care about my family so as not to become the sort of person who does not care about his family. Also, I would view not caring about my family as ‘bad news’, indicating that I was a bad kind of person, so I still have a noninstrumental, nonintrinsic reason to want to care about my family even if I am no longer able to do anything for them. I conclude that I do not have an intrinsic desire to care about my family. My reasons for this conclusion are (1) that I don't feel as if I have any such intrinsic desire over and above the nonintrinsic desires I have in this regard, and (2) I don't see what there is about my pattern of desiring that needs to be explained by an intrinsic second-order desire.

7.2.5 The Unwilling Addict Revisited Recall Lewis's version of Frankfurt's unwilling addict. He desires his high, but he does not desire to desire it, and in fact he desires not to desire it. He does not desire an unaltered, mundane state of consciousness, but he does desire to desire it. We conclude that he does not value what he desires, but rather he values what he desires to desire. (1989: 115) Two points. First, it seems to me to be absolutely clear that the unwilling addict does desire an ‘unaltered, mundane state of consciousness’. The problem is that this desire is overcome by the addict's other desire, to take the drug. Second, Lewis's theory requires more than that the unwilling addict should have a second-order desire to desire an ‘unaltered, mundane state of consciousness’ if the addict is to value such a state. The theory also requires that the unwilling addict should have an intrinsic second-order desire of this sort. But Lewis does not argue (or even claim) there is such

DESIRED DESIRES

133

an intrinsic second-order desire in this case, and it seems clear to me that the addict would not normally have an intrinsic second-order desire of this sort. Clearly, the addict has an instrumental second-order desire—he wants to have a stronger first-order desire for an ‘unaltered, mundane state of consciousness’ as a means of overcoming his other first-order desire for the altered state of consciousness induced by the drug. But there is no reason to attribute to the addict any additional intrinsic desire of the relevant sort.

7.3 Copp's Analysis of Values Copp suggests the following ‘conative account’ of a person's values: One's ‘values’ at time t are preferences one has at t about the course of one's life (a) that are stable over a period in one's life that includes t, and (b) that one is at t content to have, and (c) that one would not be content at t to anticipate losing in the future, where (d) the attitudes indicated in (b) and (c) are themselves stable. (1993: 125) Clauses (b) and (c) refer to second-order preferences, so this account resembles Lewis's account. But there are several differences. One important difference is that Lewis's account is presented as an account of what it is for a person to value something, whereas Copp's account is presented as an account of what a person's values are. The difference is that between ‘value’ as a verb and ‘value’ as a noun, in one sense of the noun ‘value’. (Copp is not concerned with a person's values in the sense of things that are valuable about the person.) It might seem at first that a person's values in the relevant sense are the same as what the person values, but this is not so. Jack can value Jane's opinion of him without one of Jack's values being Jane's opinion of him. Jack can value listening to Mozart without it being true that one of his values in the relevant sense is listening to Mozart. ‘Jack values honesty’ does seem to be equivalent to ‘One of Jack's values is honesty’ (where this is not to be interpreted as saying that one of Jack's good qualities is honesty). On the other hand, although it clearly makes sense to say, ‘Jack values happiness’, it does not clearly make sense to say, ‘One of Jack's values is happiness.’ The noun ‘value’ in this sense is not closely related to the verb. The noun is similar in meaning to ‘principle’. For example, it makes sense to say, ‘One of Jack's values is that a person should always tell the truth,’

134

INTRINSIC VALUE

which is close to saying, ‘One of Jack's principles is that a person should always tell the truth.’ On the other hand, we would not normally say, ‘One of Jack's values is the principle that a person should always tell the truth,’ whereas we might say, ‘Jack values the principle that a person should always tell the truth.’ (Here I am indebted to Katherine Miller and Christiane Fellbaum.) In any event, the differences in our use of the noun and the verb could mean that Copp and Lewis are not discussing the same concept. So we might consider whether Copp's and Lewis's respective analyses are plausible either for a person's valuing something or for a person's values in the relevant sense. One difference between Copp and Lewis is that Copp identifies a person's values with first-order preferences that satisfy certain further conditions (including conditions involving second-order desires), whereas Lewis identifies a person's valuings with certain second-order desires. This means that Copp is not subject to one of my objections to Lewis, namely, that (contrary to what Lewis claims) a person does not value something unless he already wants it. On the other hand, Copp is subject to the opposite objection: a person might fail to desire X even though X is one of that person's values. Michael Stocker's despairing person ‘sees all the good to be won or saved’ but ‘lacks the will, interest, desire or strength’. One of this person's values might be to promote the cause of human justice, even though he no longer values promoting human justice, no longer has any desire to do so, and lacks any desire to have a desire to do so. Copp requires that the relevant first-order preferences have a certain content: they are preferences about the course of one's life. Lewis has no restriction of this sort on content. Copp's restriction doesn't work for certain cases of valuing, such as Jack's valuing Jane's opinion of him. That is a case of instrumental valuing, of course. It also seems to fail for certain cases of intrinsic valuing, such as Jack's valuing the happiness of his children. Copp's restriction also seems to fail for a person's values in the relevant sense, since a person's values in this sense are not restricted to preferences about that person's life. A person can have values concerning how people in general should live their lives. For example, if one of Jack's values is honesty, that is not just a preference about the course of Jack's life. The second-order aspect of Copp's analysis is contained in his requirement that one be content to have the preferences in question and not content to anticipate losing those preference in the future. I have already

DESIRED DESIRES

135

argued that this fails for certain cases of valuing: a person who values listening to Mozart might be quite complacent about the prospect of coming to value listening to Mahler instead. It is more plausible as a requirement on a person's values. I can't think of any counterexamples. If your values include honesty, I would expect you not to be content to anticipate losing your preference for honesty. There are counterexamples in the other direction, however. A person's preference can satisfy Copp's definition without the thing preferred being one of the person's values or one of the things that the person values. For example, Jack may prefer to part his hair on the left, where this preference is a preference (a) that is stable and secure, (b) that Jack is content to have, and (c) that Jack, because of a certain rigidity, would not be content to anticipate losing in the future; and (d) where the latter preferences are stable and secure. These conditions can all be satisfied without Jack's values including his parting his hair on the left and without it being true that he values parting his hair on the left.

7.3.1 Valuing What is it, then, to value something? It seems to me that valuing something involves desiring it—having a first-order desire. It seems clear to me that if you don't have a first-order desire for something, you do not value it. (I am indebted to Alfred Mele for pointing out to me that exception must be made for the case in which the something in question itself involves your having a desire. Valuing that sort of something involves desiring it, and such a desire is by definition not a first-order desire. But this is no objection to the claim that to value something you must desire it.) Furthermore, as I have argued, you can value something without having a second-order desire to have a first-order desire for it. On the other hand, not all desiring is valuing. You can want to go to the movies tonight without valuing going to the movies tonight. I conclude that valuing is a particular kind of desiring, but I am unable to say more about what kind. As for a person's values, I find that an even more perplexing notion.

7.4 Conclusion I have looked at Frankfurt's account of freedom of the will, Lewis's account of valuing, and Copp's account of a person's values. All three

136

INTRINSIC VALUE

accounts appeal to second-order desires in somewhat different ways. Each faces specific counterexamples. In considering the fate of these accounts, it looks unlikely to me that any appeal to second-order desires can provide the key to an adequate philosophical account of freedom of the will or of what it is to value something.

8 Intrinsic Value 8.1 Introduction Many theories of value are theories of intrinsic value. For example, hedonism says that pleasure is the only thing with positive intrinsic value and pain the only thing with negative intrinsic value. Critics of hedonism reply either that some pleasures are not intrinsically worthwhile—e.g. malicious pleasures—or that things other than pleasure are intrinsically worthwhile—e.g. knowledge and justice. In this case, the disputants agree that all value is either intrinsic or derivative from intrinsic value. Indeed, agreement on this point is sometimes even built into the definitions of key terms. According to an entry in the Macmillan Encyclopedia of Philosophy, ‘An intrinsic good is something valuable in and of itself; a nonintrinsic good is something valuable by virtue of its relationship to an intrinsic good’ (Olson, 1967: 367). Similar accounts occur in Hospers (1961: 104–5), Frankena (1963: 67), and Brandt (1963: 18–20, 1959: 302). Notice that this rules out by stipulation the possibility of a nonintrinsic good that is valuable by virtue of its relation to something else, but not by virtue of any relation direct or indirect, to an intrinsic good. Many writers distinguish intrinsic value from instrumental value, the value something has because it may prove useful in obtaining other things of value. Others allow also for contributory value. ‘Something, such as a dissonant chord in a symphony, whose value depends upon being a part of a whole, is frequently called a contributory good’ (Olson, 1967). Olson holds that the value of a contributory good derives from the intrinsic value of the whole to which it contributes. Similarly, Brink (1989) explains, ‘Intrinsic goods are to be contrasted with things that are extrinsically valuable and things that are necessary conditions of realizing intrinsic value.’ In these views, intrinsic value is the source of all other value, so, if nothing were of intrinsic value, nothing could have any value at all. But it is also possible to hold that all value is instrumental and that there is no such thing as intrinsic value (Beardsley, 1965; Dewey, 1922, 1939).

138

INTRINSIC VALUE

In Essay 6, I discuss the idea that intrinsic value serves as the source of all other value. I begin with the hypothesis that there is a way of assigning intrinsic value to states of affairs that allow value to be equated with ‘expected utility’ as defined in standard decision theory, with intrinsic value being the ‘utility’ being expected. (Actually, as I will explain, ‘utility’ is a poor word for the relevant concept. However, ‘utility’ is the word used in decision theory.) I note that this first hypothesis must fail because it leads to overcounting. Whatever intrinsic value is possessed by a state of affairs S will also be possessed by any ‘larger’ state of affairs T that includes S. Suppose that an event E brings about T (and so brings about S). If an event is credited with the intrinsic value of all the states of affairs that event serves to bring about, E must be credited with the value of T and the value of S. But the value of T derives entirely from its containing S, so the value of S is incorrectly here counted twice. (Indeed, it is evident that there will be infinite overcounting, not just double counting.) I then note Robert Nozick's suggestion (personal communication) that the difficulty might be met by appealing to a notion of ‘basic value’, or ‘basic intrinsic value’. If T contains S and S has basic value, T has intrinsic value by virtue of containing S but T does not inherit basic value from S. Decision-theoretic ‘utility’ is equated with basic intrinsic value and then value is identified with ‘expected utility’, as before. The intrinsic value of a situation is taken to be equal to the total basic value contained in the situation. The proposal is that there is such a thing as intrinsic value if and only if there is a way to assign basic value to situations in such a way that everything comes out right in this way. Quinn (1974) observes that this proposal gives incorrect results about the intrinsic value of disjunctive situations. Suppose that S and T are situations with their own distinctive basic values. For example, let S be that a certain person, Jack, is happy to exactly degree D and let T be that a different person, Jill, is happy to exactly degree D. Then consider the disjunctive situation that obtains if and only if either S or T obtains. This situation, S/T, in which either Jack or Jill is happy to degree D, contains (entails) neither S nor T nor (let us suppose) any other situation with basic value. Then S/T has no intrinsic value, according to the suggestion—a counterintuitive result, for surely there is intrinsic value in any situation in which either Jack or Jill is happy to a given degree. This last way of putting things points to what is basically Quinn's own proposal, which I will describe rather loosely. The idea is that our judgements of intrinsic value are tied to actual situations. In any actual situation

INTRINSIC VALUE

139

in which S/T holds, either S holds or T holds, so there will be intrinsic value in any such situation. When we judge that there is intrinsic value in the state of affairs, S/T, we are (in this view) thinking that there is intrinsic value in a realization of that state of affairs. Things get tricky because any state of affairs can be treated as a disjunction of more finely grained states of affairs. In the extreme, a state of affairs might be treated as an infinite disjunction of the possible worlds in which that state of affairs would hold. In that case, we would have to identify the intrinsic value of the state of affairs with all of the intrinsic value that actually obtains, which is also counterintuitive. Quinn sketches a method for avoiding this result, but it is not clear that his method can be made to work. In any event, Quinn's point about disjunctive situations indicates one way in which the proposal in Essay 6 is inadequate as an account of intrinsic value. But Quinn agrees that the important notion is that of basic intrinsic value. Traditional theories of intrinsic value are best interpreted as theories of basic intrinsic value. A second weakness in Essay 6 arises from its treatment of expected utility. I will postpone discussion of that for the time being. A third inadequacy in Essay 6 (and in discussions by others) is a terminological laxity that treats as equivalent various expressions that are not equivalent—x is valuable, x is of value, x is worthwhile, x is good, x is desirable, x has utility, etc. As indicated in Essay 7, this will not do. A good painting is not always a valuable painting and a valuable painting is not always a good painting. Furthermore, and more important, what one values is not the same thing as what one takes to be desirable. Value may well be always either intrinsic, instrumental, or contributory, but (as we shall see) desirability is not. Something can be desirable without being desirable for itself, as a means to something else, or as part of a larger desirable whole. But before we say more about the concept of (basic) intrinsic value, we need to remind ourselves what role this concept plays in actual philosophical arguments. In particular, let us briefly review some of the disputes in the theory of value with respect to hedonism.

8.2 Malicious Pleasure One familiar objection to hedonism claims that hedonism cannot give a good account of what is wrong with torture. The hedonist must treat it as

140

INTRINSIC VALUE

an open question whether it would be worthwhile for Jack to torture Jill, just for the fun of it. On the one hand, Jill suffers; on the other hand, Jack gets pleasure. Hedonism must conclude that torture is worthwhile whenever the pleasure Jack gets is greater than the pain Jill suffers. But that would seem to be an objectionable result. It would seem that Jack's torturing Jill just for the fun of it can never be worthwhile, no matter how much pleasure Jack would get from it. Indeed, any pleasure Jack gets from torturing Jill only makes things worse, not better. Suppose Jack tortures Jill in order to extract information from her about some vast criminal conspiracy. It may or may not be true that circumstances justify this torture, if the information is sufficiently vital. But one thing that won't make things better (according to this objection) is that Jack should enjoy what he does to Jill in order to get her to reveal her information. Any pleasure Jack gets from torturing Jill under these conditions only makes things worse, in this view. We can extend the objection by supposing that Jill is tortured, not for the sake of information she can supply to authorities, but simply in order to make a film. Suppose that the resulting film is watched by George, Herbert, and others, all of whom enjoy watching this depiction of Jill's suffering. It seems the hedonist must agree that some good has come out of Jill's torture, namely, the enjoyment that so many people get from the film. It will be objected that such a conclusion is simply wrong: the pleasure received from watching films of torture is never of any value. A standard hedonistic reply agrees that malicious pleasure is a bad thing but denies that this shows malicious pleasure to have no intrinsic worth. The hedonist argues that the badness of malicious pleasure is extrinsic to the pleasure, having to do, for example, with the undesirable behaviour that might be expected from a person who takes pleasure in another person's suffering. Such a person is likely to treat other people badly, and that expected badness is one of the things what makes their malicious pleasure bad. Furthermore, the fact that people like George and Herbert enjoy watching films of torture creates a market for such films that may encourage the making of such films, involving the torture of other women, in addition to Jill. (This possibility is discussed in MacKinnon, 1987.) If so, that is a further way in which malicious pleasure can be bad without being intrinsically bad. Nozick (1974: 42–5) raises a different sort of objection to hedonism. Consider two possible situations in which a person has the same subjective experiences, and so the same pleasures and pains. In the one situation, a person is a respected member of society, loved by his or her spouse and

INTRINSIC VALUE

141

children, with accomplishments in business and civic life. In the other situation, the person is laughed at behind his or her back, despised by his or her spouse and children, with what seem to him or her to be accomplishments but are really the work of others. Which life would you prefer for yourself? If the first, why, if all that matters is subjective pleasure and pain? Why not a life on drugs? Or hooked up to an ‘experience machine’? Most people would rather not have such lives even if pleasant experiences could be guaranteed. Nozick argues that this shows that most people assign basic intrinsic value to more than just experience. In response, the hedonist might argue that we value a life of activity, close personal relationships, and solid accomplishments only because such things are regular means of obtaining pleasure and avoiding pain. Our reaction to Nozick's examples is similar to the reaction of a miser who continues to want additional money even after he has more than he can ever use in life.

8.3 Dening Intrinsic Value With these sample arguments in mind, let us return to the question of defining intrinsic value. Something x has intrinsic value to the extent that x has value that is due entirely to what x is intrinsically and apart from x's relations to other things. Certain issues immediately arise. First, there is a serious obscurity in the notion of what is intrinsic to x (D. Lewis 1983a, 1983b). And even if the notion of an intrinsic property is fully intelligible, it is debatable whether objects have any intrinsic properties. Secondary qualities, like colour, are not clearly properties that an object has apart from its relations to perceivers. Whether so-called primary qualities are intrinsic to an object would seem to be a scientific issue without an obvious answer. An object's position and shape are treated as relative to a spatiotemporal framework in contemporary physics. An object's mass may or may not be intrinsic; it may, for example, have something to do with the object's relation to fixed stars. If x is a state of affairs, situation, or fact, what is intrinsic to x is (I suppose) whatever is entailed or necessitated by x. If so, scepticism about entailment or necessity carries over to scepticism about intrinsic aspects of states of affairs, situations, and facts. Physical necessity does not seem to be sufficient. The fact that x, together with the laws of physics, implies

142

INTRINSIC VALUE

that y happens in one hundred years, does not mean that y is intrinsic to x. But then the problem arises of distinguishing the right sort of necessity from physical necessity, and I myself am doubtful such a distinction can really be made out (Harman, 1973, 1986). So that is one preliminary issue, concerning what makes something intrinsic to something else. I won't press the issue further here. A second issue is raised by the familiar suggestion that something has value only to the extent that it is or would be valued by someone, as in, for example, impartial spectator theories of value (Essay 11). Should we understand the suggestion to imply that nothing has intrinsic value, since what value it has depends on its relation to an actual or possible valuer? If some version of an impartial spectator theory is accepted, we can agree that there is an important respect in which value is never intrinsic. But we can nevertheless go on to consider whether there are things that are or would be valued intrinsically, that is, valued for the way they are intrinsically and not for their relations to other things. If so, we can still allow for a kind of intrinsic value, namely, value that is due to intrinsic valuing. Something has intrinsic value in this second respect to the extent that it is (or would be) appropriately valued intrinsically. A third worry is that it is unclear what is meant in saying that value ‘depends on’ or is ‘due to’ something that is or is not intrinsic to the object or state. What sort of dependence is this? We might consider a counterfactual test for determining whether a certain part of the value of x depends on something else y: consider whether x would have that value if x were there but y were not there. But any change like this might affect the overall value of x in various ways. How can we determine whether that would involve a change in the particular part of the value of x? If value derives from actual or possible valuing, it is relevant why x is or would be valued. Whatever it is about x that is or would be valued is what x's value ‘depends on’ in the relevant sense.

8.4 Instrumental Value We can suppose that x has instrumental value to the extent that x has value that is due to x's being possibly instrumental in bringing about something else. Or, in terms of valuing, x is valued instrumentally to the extent that x is valued because x is (or would be) instrumental in bringing about something else. This definition does not require that what is brought about have intrinsic

INTRINSIC VALUE

143

value. Money has instrumental value because it can be used to purchase things; we can suppose this without having any particular purchases in mind and without supposing that the items that may be purchased are valued intrinsically. Many of these items—food, shelter, medical care, transportation, and clothing—are themselves highly valued; but it would seem that they themselves are valued instrumentally rather than intrinsically. Now, food is valued in part because it tastes good, and it is plausible that the experience of eating tasty food is intrinsically good. If so, money leads indirectly to something of intrinsic value. As we have seen, many philosophers assume that instrumental value is always in this way derivative of the expected intrinsic value to which something might lead. In what follows we will consider whether this is a defensible assumption.

8.5 Good News and the Value of Information You want very much for the doctor to say that you are in good health. This is not something you want intrinsically, but it also does not have to be something you want for its expected effects, either. It need not be an instrumental desire on your part. You want the doctor to say you are in good health because you want to be in good health, but you do not expect the doctor's saying so to have a beneficial effect on your health. Rather: you take what he says as a sign or indication of the state of your health. So, something can be desirable not only intrinsically, for itself, and instrumentally, for what it might lead to, but also evidentially, because of what it indicates. Many writers call this sort of desirability a kind of ‘utility.’ That can be misleading, given that the ordinary meaning of ‘utility’ is usefulness. Good news is not the same thing as useful news. Its goodness or desirability as news is not a kind of utility in any ordinary sense of ‘utility.’ Contrary to what I assume in Essay 6, this sort of desirability is not a kind of value. You want the doctor to say that you are in good health, but you do not exactly value the doctor's saying that. What you value is the doctor's telling you the truth. You value: the doctor's saying that you are in good health only if in fact it is true that you are in good health. The doctor's telling you that you are in good health is good news, desirable news, news you want to hear, but it need not be valuable news. Now, suppose that the doctor is busy. Patients are given tickets when

144

INTRINSIC VALUE

they arrive at the office and are seen in the order of their tickets. When you arrive at the office, all the tickets for today have been distributed. One of the patients ahead of you offers to sell you her ticket. You purchase the ticket for fifteen dollars. It would seem that you value having that ticket and, furthermore, that the value of the ticket to you is instrumental. But is this because you anticipate that the ticket may enable you to obtain something of intrinsic value? Well, the ticket might enable you to hear the doctor say that you are in good health. But, although hearing that is something you desire, it is not something you value. The value of the ticket arises from its making it possible for you to get information about whether you are in good health. You do value getting this information, even if you do not expect to act on the information. Suppose you get information when the doctor says, ‘You are in good health.’ What makes his saying it information that you value is the connection between his saying it and the state of your health. What you value in his saying it is not intrinsic to the event of his saying it; you value it only because of the way it is related to the state of your health. So, his saying it is not of intrinsic value to you. Nor is it of value because it might lead to something else that you value. Does that mean it is not of instrumental value, either? No. You value finding out that you are in good health. The doctor's saying that you are in good health brings it about that you find out that you are in good health. So it does have instrumental value, after all. You value finding out that you are in good health even if this information is not useful to you. Is finding out that you are in good health then something you value intrinsically? No. For consider what is it to find out something. One familiar answer is that to find out something is to acquire a belief that is connected in the right way to what is actually the case. In this instance, the doctor says, ‘You are in good health,’ and that leads you to believe that you are in good health. Your belief has the right sort of connection (via the doctor's investigation) with your health, so it counts as knowledge. What you value in this occurrence (coming to believe that you are in good health) is not completely intrinsic to that occurrence. You value the occurrence in part because it is related in a certain way to the state of your health (although you do not intrinsically value this other thing, your health). It is plausible that your forming that belief is of contributory value to you; it is part of a larger situation or event that is of value, finding out that you are in good health, a situation which includes your actually being in

INTRINSIC VALUE

145

good health and a certain connection holding between that situation and your belief.

8.6 Valuing Retributive Punishment Here is a related example. Consider someone K who values retributive punishment. K values punishing people for certain crimes quite apart from deterrent or other effects. In a particular case, punishment is not intrinsically valuable; its value derives from an appropriate relation to a past crime. But the value of the punishment does not derive from the value of the crime. What makes the punishment worthwhile is not that it has an appropriate relation to a past event that was worthwhile. On the contrary, what makes the punishment worthwhile is that it bears an appropriate relation to something in the past that was not worthwhile! The punishment has contributory value. It contributes to a larger event with a certain value, namely, the complex event C•P, where C is the past crime and P the present punishment. K values P because of its involvement in C•P. At this point it may be useful to make the obvious point that K can take P to be of value because it is part of a valuable larger whole, C•P, without supposing that C•P is of positive value. When K says that a certain punishment would be a good thing, K does not mean that the whole situation is a good thing, the crime and then the punishment. It would have been better not to have had either. All K means is that, given that the crime has occurred, it would be better that the punishment also occur. The value of the punishment in this circumstance derives from the intrinsic value of one larger whole when compared with another. When K considers the value of P, K supposes that C has already occurred. K is therefore comparing the two larger events C•P and C•∼P, where ∼P is the situation in which there is no punishment. In thinking that the value of P depends on its relation to C, K is thinking that C•P is better than C•∼P. On the other hand, K does not suppose that the event C•P is worthwhile. K only supposes that the P part of this event is worthwhile in the sense that it makes the resulting complex event better than it would otherwise have been. In considering instrumental value, K will allow for the value of punishment, that is, the value of aiding punishment. By bringing about a condition of punishment, K does something that is instrumental in seeing to it that C•P obtains rather than C•∼P, so this action is instrumental in

146

INTRINSIC VALUE

increasing the intrinsic value (or decreasing the intrinsic disvalue), according to K. There are many other examples of this sort of contributory value. Tom values sitting in a certain chair. Why? Because Freud once sat in that chair. Tom attaches basic intrinsic value to sitting in the chair that Freud sat in. That is, if F is Freud's having sat in this chair and T is Tom's now sitting in the chair, Tom attaches more basic intrinsic value to F•T than to F•∼T. Here T has contributory value, since it makes the difference between F•T and F•∼T. What Dworkin (1993) calls ‘sacred value’ is a kind of contributory value. Dworkin observes that to believe in the value of human life is not to be committed to thinking that it would be better if there were more human life. If E is the present existence of a particular human life and F is the future existence of that same human life, this is to say that E•F is intrinsically better than E•∼F, which does not commit one to saying that E is better than ∼E. Classical hedonistic utilitarians can avoid at least one familiar objection if they take happiness to have contributory value rather than simple intrinsic value. They hold that we ought to maximize happiness. ‘Objection: should we maximize total happiness or average happiness? If total happiness is the goal, we should greatly increase the size of the population. If average happiness is wanted, we should eliminate those who are temperamentally unhappy. Neither alternative is plausible.’ Reply: given that someone exists, it is better if that person is happy. But that is not to say it is good to bring new happy people into existence. In this version, the utilitarian goal is to maximize the happiness of those people who exist.

8.7 Desirability Without Value Now let us go back to the case of good news and remove the value of information. Suppose that a question has arisen as to whether you have a dread disease D. If you have the disease, nothing can be done for you and you will be dead in a couple of weeks. If you don't have the disease, no treatment is required. A highly reliable test has been performed to determine whether you have the disease, but you are absolutely not going to be told the results of the test. (The test was performed simply to satisfy the idle curiosity of your doctor.) Consider your attitude about how the test comes out. You hope very much that the test comes out negative, which means that you do not have

INTRINSIC VALUE

147

the disease. That outcome of the test is a highly desirable outcome for you. But you can attach no utility to the test's coming out one way or another. Nor do you value the tests coming out negative. How the test comes out is of no value to you at all. Since you will not find out how the test comes out, it will have no value to you as information. It may have value to your doctor, but (let us suppose) you attach no value to your doctor's learning a couple of week's early whether or not you have D (especially since the doctor is not going to tell you how the test came out). If a payment of a small amount of money were needed in order to complete the test (for the purchase of litmus paper), you would have no interest in paying it. You care about the outcome of the test only because of the indicative connection between the test and whether you have the disease. You want the test to come out negative, because you want it to be the case that the test comes out negative. You take the test's coming out negative to be a desirable thing, even though you are not going to learn how the test came out (except indirectly, if you learn that you have survived for more than two weeks). But you do not take the test's coming out negative to be in any way a valuable thing. Here, then, is a clear example in which value and desirability diverge.

8.8 Conclusion In Essay 6, I argue against the standard view that all value is either intrinsic, instrumental, or contributory. There I see a need also for a kind of evidential value. But this is a mistake. I confuse value with desirability. Something can be desirable because of what it indicates without having any value, as is illustrated by the doctor's test just described. Here is an argument for the standard view. Suppose a state S is of value, where this is neither intrinsic or instrumental value. Then S must be of value in part because of a relation to something else T but not because S tends to produce T. It is hard to see how this could be unless the larger event S•T were of greater value, and indeed of greater intrinsic value, than S•∼T. (To the extent that S•T is merely of greater instrumental value than S•∼T, the value of T in contributing to S•T is itself instrumental and not contributory.) This argument is not conclusive. If there were such a thing as purely evidential or indicative value, that would provide a way to avoid the conclusion. Let H be that you are healthy and N be that the test comes out

148

INTRINSIC VALUE

negative. H•N is of no more value than H•∼N, so, N does not have contributory value. If there were pure evidential or indicative value, N would have it, and N would then be a counterexample to the argument. Since there is no pure evidential or indicative value, this N is not a counterexample. But there may yet be some further unknown example.

Part III Virtue

This page intentionally left blank

9 Human Flourishing, Ethics, and Liberty What kind of ethics do we get, if we begin with a conception of human flourishing and attempt to derive the rest of ethics from that conception? A number of writers have expressed sympathy for such an approach to ethics, although they disagree about details: Veatch (1964, 1971, 1985), Nozick (1981), MacIntyre (1981), Finnis (1980), Norton (1976), Foot (1978), Machan (1975), Anscombe (1958), Rand (1957, 1963), and Maslow (1962).

9.1 Functionalism It is characteristic of this approach to take the basic form of evaluation to occur when something is assessed with respect to the way in which it fulfils its function or satisfies certain interests. Let me begin by summarizing this idea as I understand it.

9.1.1 Artefacts A bread knife has a certain purpose: it is used to slice bread. A ‘good’ bread knife is one that is easy to use to slice bread smoothly. A ‘bad’ bread knife has ‘defects’ of one or another sort that make it not well suited for this purpose. A bread knife ‘ought’ to be easy to use in this way; it ‘ought’ to be sharp enough, well balanced, not too heavy, and so forth. Something is ‘wrong’ with it if it is too dull to cut bread. A dull bread knife is not a ‘good specimen’ of a bread knife. A ‘real’ or ‘true’ bread knife functions well. Here ‘real’ and ‘true’ mean something like ‘ideal’. There are, of course, other, irrelevant uses of these words. For example, a dull bread knife is still a real bread knife in the sense that it is not a fake one, carved out of soap. So, a number of evaluations can be made of a bread knife with respect

152

VIRTUE

to its associated function. Furthermore, the very phrase ‘bread knife’ expresses a functional notion, since to understand what a bread knife is, is to understand the purpose of a bread knife. A bread knife is ‘by definition’ a knife used for cutting bread. Similar remarks can be made about other categories of artefact. For example, a clock is by definition something people use to tell the time and is to be evaluated as a clock in terms of its fittingness for this function. A clock ‘ought’ to keep good time; something is ‘wrong with it’ if it runs slowly; and so forth.

9.1.2 Bodily Organs Artefacts are not the only sorts of things that can be evaluated functionally in this way. Bodily organs are also associated with functions. A heart is something that functions to pump a creature's blood through its circulatory system. To understand what a heart is, to understand its ‘nature’, is to understand that this is its function. True, it was an empirical discovery that the heart has this function, but in making that discovery people learned what a heart is. A creature's heart is not whatever bodily part has a certain shape; it is whatever bodily part has the function of pumping the creature's blood. We evaluate organs with respect to their natures, that is, their functions. A ‘good’ heart is one that functions well, pumping blood with just the right pressure through the circulatory system. A ‘bad’ heart has ‘something wrong with it’. A ‘real’ or ‘true’ heart, a ‘good specimen’ of a heart, is one that functions well. Similarly for other organs.

9.1.3 Plants and Animals A similar sort of evaluation applies to whole organisms. Associated with a particular type of plant or animal is what might be called a condition of health or ‘flourishing’. We evaluate organisms with respect to this condition. A ‘good specimen’ of an oak tree is an oak tree that is flourishing, not one that is stunted or diseased. That is a ‘true’ or ‘real’ oak tree. Something is ‘wrong’ with an oak tree that does not flourish, perhaps its root system is inadequate or there are not enough nutrients in the soil. An oak tree ‘needs’ the necessary conditions of its flourishing. It ‘needs’ a good root system, adequate water and nutrients, light, air, and so forth. Similarly, animals may or may not flourish. A ‘true’ lion is a healthy,

HUMAN FLOURISHING, ETHICS, AND LIBERTY

153

happy one. The lion ‘needs’ adequate food, air, and water, and perhaps also exercise and companions. People may or may not flourish in this sense. They may or may not be healthy and happy. Of course, happiness is connected not only with the satisfaction of bodily needs but also with the satisfaction of a person's incidental desires and interests. Things that answer to the needs and interests of plants, animals, and people can be evaluated in terms of their function in answering to those needs and interests. ‘Good’ roots are roots that are adequate for the flourishing of the plant. A ‘good’ diet is a nutritious one. The phrase ‘good food’ is ambiguous. It may mean food that is nutritious (satisfying bodily needs) or food that is delicious (satisfying certain interests or desires connected with food).

9.1.4 Flourishing as Basic It would seem that the more basic evaluation concerns what constitutes a creature's flourishing and that functional evaluations of artefacts and organisms are derivative from that. The function of a bodily organ lies in its doing something which normally contributes to the flourishing of an animal with that organ. The function of a root system lies in its serving to do something that normally contributes to the flourishing of a plant. The function of an artefact lies in its contribution to the satisfaction of certain desires or interests, where the satisfaction of desires and interests is normally part of the flourishing of a creature with interests. This sort of connection between functional evaluations and flourishing is indirect, mediated by ‘normally’. It can happen that a plant fails to flourish under certain conditions precisely because of its very good root system (if it does not have enough foliage, for example). A good root system is a root system of a sort that normally contributes to the flourishing of a plant. It may not so contribute in every possible situation. Furthermore, something might be an excellent device for extracting confessions from suspects, without ever contributing to the flourishing of any creature. Something is good in relation to certain interests if it can be used to satisfy those interests, where a creature's flourishing normally involves the satisfaction of (many of) its interests. It is possible that in fact no one has the interests in question, and, even if someone has such interests, it may be that the satisfaction of those interests in this instance will not contribute to the flourishing of that or any other person.

154

VIRTUE

9.1.5 People With Certain Roles A kind of functional evaluation can be made of people with certain roles. Farmers, teachers, tailors, bank presidents, and janitors have certain roles or functions and can be evaluated with respect to their satisfaction of that function. A ‘good’ farmer is one who farms well, who grows crops or raises livestock successfully. Otherwise he or she is ‘not much’ of a farmer, ‘not a real’ farmer, and ‘something is wrong’ with him or her as a farmer. Similarly, a ‘true’ teacher is one who teaches well, so that students come away with proper wisdom about the subject taught. Similar functional judgements can be made concerning roles that do not contribute overall to human flourishing. A ‘good’ burglar is one that burgles well and does not get caught. Otherwise, he or she is ‘not much’ of a burglar.

9.1.6 People as People We can, of course, evaluate people as people and not just as teachers, doctors, farmers, and so forth, where this is also not just evaluating their health and well-being. We have a sense that there are certain ways to live that are desirable ways to live and that a person ought to live in one of those ways. A ‘good’ person lives in such a way, such a person is a ‘true’ or ‘real’ person. Otherwise a person is ‘not much of a person’, ‘not a good specimen’ of a person.

9.1.7 Lives Let us use the term ‘flourishing’ in a somewhat extended sense to refer to living in one of these desirable ways. Flourishing in this sense would seem to involve happiness, virtue, and accomplishment. To flourish is to lead the sort of life it is good to lead, by which is meant the sort of life you want your children to lead, as well as the sort of life you want to lead yourself. Such a life is not just good in the sense that it is good that someone should lead such a life—it may be good that someone leads a life of self-sacrifice without that person's life being the sort of life it is good to lead in the relevant sense, good for him or her, as it were. But a life that is good for a person to lead is also not just a life that is good for that person in a purely egoistic or self-interested sense. It is rather the sort of life one wants to have led, all things considered, in a calm moment of summing up. Such a life is a ‘real’ and ‘true’ life. Something is ‘wrong with’ a life that

HUMAN FLOURISHING, ETHICS, AND LIBERTY

155

falls short of this. What is ‘wrong’ might be something about the person living the life; something might be wrong with him or her. Or it might be something about the circumstances, in which case we speak of ‘misfortune’. So, it is not quite accurate to say, as I said above, that a person is to be evaluated entirely in terms of the life he or she leads. A person might be a ‘good’ person, a ‘true’ and ‘real’ person even though he or she does not flourish in the relevant sense, if the cause of this failure to flourish is not some defect in the person, but rather some misfortune. (By a defect ‘in the person’ I mean something like a defect in motivation. There might be a defect literally ‘in’ the person that was really a misfortune, a defect in the body of the person, some sort of disease, for example.) The connection between the goodness or realness of a person as a person and his or her flourishing is indirect: a person is a ‘good’, ‘true’, or ‘real’ person to the extent that the person's actions and character are of the sort that normally contribute to a flourishing life, in the absence of one or another misfortune. (However, a ‘real’ or ‘true’ person is also one who reacts or would react in certain ways to misfortune, so this is not stated quite right.)

9.1.8 Societies and Laws Finally, we can assess societies with reference to the extent of human flourishing within them; and we can assess systems of laws in terms of the extent to which they promote or would promote human flourishing. A ‘good’ society is one in which people flourish in the sense of living desirable lives. That is what a society ‘ought’ to be like. Otherwise it is ‘not much’ of a society, not a ‘real’ society; something is ‘wrong’ with it. ‘Good’ laws are laws that promote a good society, in which people flourish. ‘Bad’ laws have the opposite effect, in which case there is ‘something wrong’ with the laws. But can we say that such ‘bad’ laws are ‘not true laws’, ‘not really laws’? Maybe, maybe not. The point is controversial, much more controversial than what I have been saying so far, I think.

9.2 A Basis for Ethics? I now want to consider the extent to which moral or ethical assessments fit into the functionalist framework we have been considering. To repeat the question with which I began: what kind of ethics do we get if we begin

156

VIRTUE

with a conception of human flourishing and attempt to derive the rest of ethics from that conception?

9.2.1 Relativism One implication would seem to be moral relativism, since what counts as ‘flourishing’ seems inevitably relative to one or another set of values. People with different values have different conceptions of ‘flourishing’, of the ‘good life’. For some, the good life includes the discriminating enjoyment of good meat and wine, others hold that no life can count as a good life if it involves the exploitation of animals raised for food. Some say the good life involves at its core the pursuit of an individual project of excellence; some say it involves service to others. Some people stress the importance of elaborate social rituals of politeness; for others, such rituals are trivialities of no importance at all to the good life. People put different weights on the joys of combat and competition as against the benefits of cooperation and shared undertakings. They disagree on the relative importance of knowledge and culture as compared with pleasure and simple happiness. Of course, the mere fact of disagreement over values, and therefore over what constitutes flourishing, is not sufficient to show there is no absolute flourishing, as opposed to flourishing in relation to one rather than another set of values. But it is difficult to see how one rather than another conception of flourishing is to be validated simply in ‘the nature of things’ or in the ‘nature’ of persons—except in the sense in which different sets of values yield different conceptions of nature or of the nature of a person (Phillips and Mounce, 1965).

9.2.2 Consequentialism A second feature of this approach to ethics is that it tends toward utilitarianism or consequentialism. The basic value is human flourishing. Actions, character traits, laws, and so on are to be assessed with reference to their contribution to human flourishing. A number of different theories about what one ought to do are compatible with the basic approach. First, there is the egoistic suggestion that one should always act so as to maximize one's own (expectation) of flourishing. Second, there is the act-utilitarian suggestion that one should always act so as to maximize total human flourishing. Third, there is the rule-utilitarian suggestion that one should follow rules designed to promote

HUMAN FLOURISHING, ETHICS, AND LIBERTY

157

human flourishing. This idea has a number of variants, depending on whether one is supposed to follow the rules actually current in one's group or one is supposed to follow the ideal rules which would best promote human flourishing if they were widely followed, or widely accepted. Fourth, there is the ‘virtue-utilitarian’ suggestion that one should act in accordance with traits of character which promote human flourishing. Again, there are variants, depending on whether one is to act from one's actual character traits or in accordance with ideal character traits. All these suggestions are subject to well-known difficulties. For example, egoism is either quite counterintuitive in its implication that the moral thing to do is to concentrate on one's own flourishing and otherwise ignore others, or circular if it argues that flourishing involves acting morally toward others. This last idea is circular because it explains what it is to act morally in terms of promoting one's own flourishing, but then flourishing is said to involve acting morally, where acting morally is not simply acting so as to promote one's own nonmoral flourishing. Act utilitarianism has various counterintuitive consequences. It implies one is morally required to sacrifice some people for others, if that increases overall human flourishing, even if it involves framing an innocent person on a charge of murder. It implies that a doctor is morally permitted, indeed required, to murder a relatively healthy patient (or hospital visitor) if the doctor can use that person's vital organs to save the lives of at least two people who would otherwise die. It implies that one should devote one's life to the service of others, because under current conditions that promises a greater improvement in human flourishing than other things one might do with one's life. Actual rule utilitarianism, which says one should follow the actual rules of one's society, conflicts with the commonsensical view that sometimes one should not follow these rules, if they are unjust, for example. On the other hand, ideal rule utilitarianism, which says one should act on the best set of rules whether or not others are doing so, implies one should ignore the fact that others are not acting on the best set of rules. For example, rule utilitarianism implies one should follow the rule of never using force against other people, since things would be much better if everyone were to follow that rule. Perhaps one should follow such a rule if everyone else follows it, but it is folly to act as if all people are following such a rule when one knows they are not. Virtue utilitarianism shares the problems of rule utilitarianism, although it is harder to demonstrate this since virtue utilitarianism is a vaguer view. Indeed, one method of defending utilitarianism is to devise

158

VIRTUE

a relatively complicated version whose implications for practice are obscure. But all such complex forms of utilitarianism face the problem of explaining what reason anyone could have to follow this or that complicated utilitarian formula. Given that human flourishing is the ultimate value, it is possible to see why someone might have a reason to follow the advice of egoism or of act utilitarianism, since these views say one should promote this basic value, either for oneself or for everyone. But what reason could anyone have to follow the dictates of some form of rule utilitarianism, or some even more complicated version, if these dictates conflict with those of egoism and of act utilitarianism? In the absence of an answer to this question, it is impossible to take these complicated forms of utilitarianism seriously.

9.2.3 Imitation of Excellence I can think of only one other way to try to base ethics on the value of human flourishing. The idea is to take the life of someone who flourishes as an ideal. To flourish is not just to be healthy and happy but to have good character and act rightly. The right thing to do in any particular case is therefore the same as what someone who flourishes would do in that case. So, what one ought to do in any particular case is exactly what someone who flourishes would do in that case. One problem with this suggestion is that, if one is not already an excellent person who is flourishing, one's situation may well be of a sort which a flourishing person could never be in. For example, one may have done someone a wrong, and the question is what one should do now. It may be that a flourishing person could not have done that sort of wrong to anyone. In such a case, the suggestion does not indicate what one should do now. Furthermore, one must allow for one's current lack of character. It may be that a flourishing person in this sort of situation could promise to undertake something with the assurance that he or she would succeed in the undertaking, an assurance one does not have because of one's weak character. In that case, one should not act in the way the flourishing person would act; that is, one should not promise. A deeper worry is this. Can we really suppose that our conception of what it is to flourish in the relevant sense, that is, our conception of what it is to lead the good life, is more basic than our conception of what someone ought morally to do on one or another occasion, given that flourishing in this sense involves having good character and acting in the right

HUMAN FLOURISHING, ETHICS, AND LIBERTY

159

way? Is it really true that we first see what is involved in a good life and that that tells us what a person ought to do on a particular occasion? Can one really determine what one ought to do on a particular occasion by asking what an ideal person would do on that occasion? Isn't it rather the other way, namely, we figure out what an ideal person would do by first figuring out what a person ought to do in that situation? If so, then the suggestion we are considering involves the sort of vicious circle that arises in certain forms of egoism. In order to determine what one should do, one must determine what a person with a flourishing life would do. But in order to determine what is involved in a flourishing life, one must determine what a person should do in various situations. So, it seems to me that, if ethics is to be based on the value of human flourishing, this will have to be within a consequentialist theory that interprets human flourishing in an ethically neutral way in terms of health and happiness, despite the difficulties consequentialism faces. (I will say more about these difficulties in what follows.)

9.2.4 Rights What are the implications of the approach we are considering for rights? Let us distinguish various sorts of rights, or perhaps various senses of the term ‘right’ used as a noun. First, to say someone has ‘a right’ to something may be to say only that it is ‘all right’ for him to do it; he is ‘permitted’ to do it; it is not wrong of him to do it. In this sense, a person ‘has no right’ to promise to meet you for lunch if he or she has no intentions of being there. The person ought not to do that; it would be wrong of him or her to do it. Second, a person may have ‘a right’ to something in the sense of a particular claim to it. An author may have a right, based on contract, to a regular reporting of the sales of the author's books. A divorced father may have a right to see his children on a regular basis. Third, a person has certain basic lights which we think ought to be protected by society—a right to life, to noninjury, to be able to own personal property, to various sorts of liberty including political and religious freedom, to nondiscrimination on the basis of sex or race. Let us concentrate on this third sort of right, basic rights. What does the approach we are considering imply about these? Presumably it must involve a consequentialist theory of such rights. One suggestion would be that to say people have a basic right to something is to say people in general need that something if they are to flourish. The laws of a society must

160

VIRTUE

therefore be designed to ensure that these rights are protected. This follows from the purpose of law, namely, to promote human flourishing. I do not think this suggestion can be adequate by itself. It ties rights too closely to what people in general need. But it is not clear that people in general need freedom of speech, or political or religious liberty, in order to flourish. Some people would flourish best in a society with a single religion or political party, if they belonged to that religion or party. No, a consequentialist argument for such liberties has to be more along the lines Mill (1959) gave: the protection of freedom of speech and political and religious liberty tends in the long run best to promote human flourishing. I will explain shortly why I think that this sort of consequentialism is not the real source of these basic rights.

9.3 Reasons It is very important to distinguish between (1) the idea that it ‘ought to be’ that a particular agent does something, in the sense that it would be desirable, a good thing, if that agent were to do that thing and (2) the idea that the agent in question ‘ought to do’ that thing, in the sense that the agent has sufficient reasons to do it, reasons which are stronger than reasons he or she has not to do it. The first thought is the thought that there are reasons for wanting the agent to do it, which for us, the thinkers of the thought, the critics, are reasons that we have to want this to happen. The second thought is the thought that there are reasons the agent in question has to do that thing. These are quite independent thoughts. We can suppose an agent has sufficient reasons to do something without supposing we have sufficient reasons to be in favour of his or her doing it, since we might not care one way or the other and be perfectly justified in not caring. Similarly, we can suppose we have reasons to favour that agent's doing something without supposing the agent has sufficient reasons to do that thing. These points continue to hold when moral oughts and reasons are in question. There is a distinction between the thought that it ought morally to be the case that a given agent do something and the thought that the agent ought morally to do it, just as there is a distinction between the thought that it is wrong that the agent does as he does and the thought that it iswrong of the agent to do as he does. The one thought concerns moral value, the other concerns moral duty or obligation. We might say that the one thought was evaluative, the other normative. To say that, from a moral

HUMAN FLOURISHING, ETHICS, AND LIBERTY

161

point of view, it ought to be that the labourer, Albert, made as much money as the bank president, Bertrand, or to say that it is morally wrong that Albert makes so much less money, is clearly not to say that Albert ought morally to make more money in the sense that it is wrong of Albert that he does not make more money. It is equally true that this is also not to say that Bertrand ought morally to give some of his money to Albert, that it is wrong of Bertrand not to do so. Even though someone might take the former, evaluative judgement (that it is wrong that Bertrand makes more than Albert) to provide a forceful reason for the latter, normative judgement (that it is wrong of Bertrand not to give away his extra income to Albert and others similarly situated), the judgements are clearly distinct.

9.3.1 Equivocation in Utilitarianism Utilitarianism notoriously blurs this distinction by simply identifying or confusing what a person ought morally to do with what it ought to be that the person does. Of course, from a moral point of view, it would be desirable, a good thing, wonderful, if the agent would act so as to maximize expected human flourishing. There is a sense, then, in which it ought to be that the agent acts in that way, a thought that can be expressed in the words, ‘The agent ought to maximize human flourishing.’ But these words also have another, perhaps more ordinary, interpretation in which they imply that the agent has sufficient reasons to so act, reasons that outweigh the agent's reasons for not so acting or for doing something else. And the fact that it would be wonderful for humanity if the agent were to act in a certain way does not by itself establish that the agent has sufficient reasons to act in that way. For the agent will have concerns other than the concern to further human flourishing in general, and the course of action supported by the weight of reasons, given all of his or her concerns, may be something quite different from the act that would maximize human flourishing in general. Act utilitarianism involves an equivocation between ‘ought to be’ and ‘ought to do’. Other forms of utilitarianism can involve the same equivocation. For example, ideal rule utilitarianism, with its endorsement of the pacifist's fallacy, equivocates between saying, on the one hand, ‘From a moral point of view, it ought to be the case that everyone followed this rule (that is, this would be wonderful),’ and saying, on the other hand, ‘Everyone ought morally to follow this rule (in the sense of having a sufficient reason to do so).’ The fact that both propositions might be expressed in the same

162

VIRTUE

words, namely, ‘Everyone ought to follow this rule,’ makes the equivocation possible, despite its absurd consequences.

9.3.2 What Is It to Have a Reason to Do Something? To a first approximation, action is the result of desire plus belief. This is Hume's view. The agent has various desires and believes that a particular act promises to contribute to the satisfaction of these desires; this leads the agent to desire to do that action. If the action is immediately within the agent's power, the desire to do it leads to its being done. In this view, an agent has a reason to do something to the extent that the agent's beliefs imply that doing that thing will contribute to the satisfaction of the agent's desires. A somewhat more adequate theory allows that the agent may have reasons to modify his or her beliefs in various ways, by addition and subtraction. In this more adequate view, what an agent has reasons to do depends upon the implications of these rationally modified beliefs for the satisfaction of the agent's desires. A still more adequate theory allows for other concerns over and above an agent's desires, including intentions, plans, policies, and values. Reasoning leads to changes, by addition and subtraction, in these concerns, just as it leads to changes in beliefs. In this theory it is still true that the reasons an agent has to do things always derive from current concerns, so that an agent may fail to have a reason to do something if he or she fails to have the relevant concern, if he or she simply does not care. It seems to me that some such theory must be correct. Notice that the theory is not the view that an agent only has reasons to do what the agent wants to do. Even the Humean ‘first approximation’ is not absurd in that way.

9.4.3 Relativism Again Given what seem to be the facts about human concerns, this theory of practical reasons implies a form of moral relativism, namely that there are no basic moral demands everyone has sufficient reason to accept as providing reasons for action. People have different concerns, different values. And some people simply do not care enough about others for their concerns to give them reasons to act in the way most people consider morally right. There is a possible functionalist reply to this: even though there are

HUMAN FLOURISHING, ETHICS, AND LIBERTY

163

people who do not care enough about others in this way, they ought to care, and there is something wrong with them that they do not care. If they do not care enough about others, they will not flourish. To be sure, they may have healthy, pleasurable lives, full of a rich sense of accomplishment, but flourishing involves more than that; it involves having a good character and acting rightly. It involves having the proper concerns. Lacking that, these characters will not flourish. Furthermore, in this view, the reasons an agent has to do things do not always have to derive from the agent's current actual concerns; also relevant are the concerns the agent ought to have if the agent is to flourish. So, in this view, there can be basic moral demands that everyone has sufficient reasons to accept as guides to action, even if there are people who lack the sorts of concerns for others that would give them reasons of a Humean sort to accept these demands. The trouble with this functionalist defence of moral absolutism is that it would allow me to use my values in deciding what reasons I think you have to do things, since it is with respect to my own values that I will decide what it would be for you to flourish. This is quite implausible. I cannot seriously think my values might supply you with reasons to act if you do not share those values.

9.5 The Real Source of Values and Rights I argue in Essays 1–4 that the source of morality lies not in the nature of things but in human arrangements. People come to accept certain rules and values in order to get along with each other. The moral reasons a particular person has to act in one or another way depend ultimately on that person's values. To repeat, on that person's values. People in a given society or group of people will tend to share certain values, but they do not always, and the reasons an agent has to do things derive from his or her values and not from the values of those in the agent's group or society. If we want to make a judgement concerning that agent's reasons, we must do so in relation to the agent's values and other concerns and not in relation to our own values and concerns, if ours are different. This does not mean we cannot condemn other people as evil, bad, or dangerous by our lights, or take them to be our enemies. Nothing prevents us from using our values to judge other people and other moralities. But we only fool ourselves if we think our values give reasons to others who do not accept those values.

164

VIRTUE

The basic protections of morality have arisen as a result of bargaining and compromise, sometimes after serious conflict and even war. To return to an example discussed in Essay 1, ordinary morality draws an important distinction between the weighty ‘negative’ duty not to harm others and the less weighty ‘positive’ duty to intervene to help prevent others from being harmed. So, a doctor may not seize and cut up a healthy visitor to the hospital in order to save the lives of several patients by distributing the visitor's organs where they would do the most good. The harm to the visitor is not permitted morally even in order to prevent more harm to others. The healthy visitor has a ‘right’ not to be harmed by the doctor. The patients who might be saved by this organ distribution do not have an equally strong ‘right’ to have a corresponding harm to them from disease prevented. I suggest that this distinction in ordinary morality is a consequence of the fact that morality is a compromise between people of different powers and resources. Whereas everyone benefits from the general recognition of a duty not to harm others, the rich and strong would lose, in comparison with the poor and weak, from a general recognition of an equally strong duty to prevent others from being harmed, because (a) the strong and rich are better able to protect themselves from being harmed as long as there is general recognition of a duty not to harm, and therefore do not benefit as much as the poor and weak from a general policy of helping others avoid harm, and (b) since the strong and rich have considerably more resources than the poor and weak for the prevention of harm, the rich and strong would be called upon to do much more of the preventing than the poor and weak would. So, sometimes we recognize certain things as basic rights because they are immediately important to all (or almost all) people, as in the right not to be injured by others. But there are other basic rights which we recognize for different reasons, among them the right to liberty, including freedom of speech and religious and political freedom. These rights are not of immediate importance to everyone in the same way in which the right to life and noninjury is immediately important. They may, indeed, be indirectly important to everyone for the sorts of reasons Mill stresses, but that does not seem to be why we recognize these rights as basic. What seems to be the crucial factor here is that often, when these rights have been denied, there have been many people who found the resulting situation intolerable, and some of these people have been willing to fight the issue out. The basic rights of liberty have come to be recognized only as the result of these bitter conflicts.

10 Moral Philosophy Meets Social Psychology: Virtue Ethics and the Fundamental Attribution Error 10.1 Folk Physics and Folk Morality Ordinary untrained physical intuitions are often in error. Ordinary people expect that something dropped from a moving vehicle or aeroplane will fall straight down to the point on earth directly underneath the place from which it was released. In fact, the dropped object will fall in a parabolic arc in the direction of the movement of the vehicle or aeroplane from which it was dropped. This means, among other things, that bombardiers need to be trained to go against their own physical intuitions. There are many similar examples (McCloskey, 1983; Holland, Holyoak, Nisbett, and Thagard, 1986). Considering the inadequacies of ordinary physical intuitions, it is natural to wonder whether ordinary moral intuitions might be similarly inadequate. And, while many moral philosophers seem to put great confidence at least in their own moral intuitions, others argue for revisions. Consequentialism may be put forward not as an attempt to capture intuitive folk morality but rather as a critique of ordinary intuitions (Kagan, 1989). Similarly, moral relativism might be defended as the truth about morality, whether or not moral relativism accords with everyone's intuitions (Harman, 1996). On this occasion I discuss a different kind of rejection of folk morality, one that derives from contemporary social psychology. It seems that ordinary attributions of character traits to people are often deeply misguided, and it may even be the case that there is no such thing as character, no ordinary character traits of the sort people think there are, none of the usual moral virtues and vices.

166

VIRTUE

In attempting to characterize and explain the movements of a body, folk physics places too much emphasis on assumed internal characteristics of the body, ignoring external forces. It happens that ‘[t]he relation between lay personology and a more correct theory of personality is analogous to the relation between lay and scientific physics’ (Ross and Nisbett, 1991: 161, citing earlier work including Lewin, 1935). In trying to characterize and explain a distinctive action, ordinary thinking tends to focus on a corresponding hypothetical distinctive characteristic of the agent and tends to overlook the relevant details of the agent's perceived situation. Because of this tendency, folk social psychology and more specifically folk morality are subject to what Ross (1977) calls ‘the fundamental attribution error’. Empirical studies designed to test whether people behave differently in ways that might reflect their having different character traits have failed to find relevant differences. It is true that studies of this sort are very difficult to carry out and there have been very few such studies. Nevertheless, the existing studies have had negative results. Since it is possible to explain our ordinary belief in character traits as deriving from certain illusions, we must conclude that there is no empirical basis for belief in the existence of character traits.

10.2 Character Character traits must be distinguished from psychological disorders like schizophrenia, mania, and depression, and from innate aspects of temperament such as shyness or being basically a happy or sad person. Character traits include virtues and vices like courage, cowardice, honesty, dishonesty, benevolence, malevolence, as well as certain other traits like friendliness or talkativeness. Aristotle (1985) describes the ordinary conception of such character traits. They are relatively long-term, stable dispositions to act in distinctive ways. An honest person is disposed to act honestly. A kind person is disposed to act kindly. The relevant dispositions must involve habits and not just skills, including habits of desiring. To be sure, as we normally conceive of certain character traits or virtues, they may involve certain strengths or skills, as in courage or strength of will (Brandt, 1988). But they involve more than simply having relevant skills or know-how. A person with the relevant character trait has a long-term, stable disposition to use the relevant skills in the relevant way. Similarly, the virtue of benevolence

MORAL PHILOSOPHY MEETS SOCIAL PSYCHOLOGY

167

may involve practical knowledge concerning how to benefit people; but mere possession of that knowledge with no disposition to use it to benefit people would be insufficient for possession of a benevolent character. In ordinary conceptions of character traits and virtues, people differ in their possession of such traits and virtues. A particular character trait fits into one or more ranges of ways of behaving. In some cases, the relevant virtue can be seen as a mean between extremes (Aristotle, 1985). Courage is a mean between rashness and timidity, for example. Proper benevolence is a mean between stinginess and profligacy. Where some people have a given virtue, others have one or another corresponding vice. Different ways in which people behave on different occasions are sometimes due to their having such different character traits. Finding a wallet on the sidewalk, an honest person tries to locate the owner; a dishonest person pockets the contents and throws the rest of the wallet away. How a stranger reacts to you depends on whether the stranger is basically friendly or unfriendly. We ordinarily suppose that a person's character traits help to explain at least some things that the person does. The honest person tries to return the wallet because he or she is honest. The person who pockets the contents of the wallet and throws the rest of the wallet away does so because he or she is dishonest. The fact that two people regularly behave in different ways does not establish that they have different character traits. The differences may be due to their different situations rather than to differences in their characters. To have different character traits, they must be disposed to act differently in the same circumstances (as they perceive those circumstances). Furthermore, character traits arebroad-based dispositions that help to explain what they are dispositions to do. Narrow dispositions do not count. If fifteen-year-old Herbert is disposed to refuse to ride any roller coaster, but is not cowardly or fearful in other ways, his particular disposition is not an instance of cowardice, and indeed may fail to be an instance of any character trait at all. If Herbert also acquires a disposition to refrain from speaking up in history class (but not in other subjects) and the explanation of this latter reluctance is quite different from the explanation of his avoidance of roller coaster rides, then these two dispositions are not special cases of a single character trait. Nor can cowardice or fearfulness be constructed out of a collection of quite separable dispositions of this sort, if there is no common explanation of the resulting behaviours.

168

VIRTUE

10.3 Virtue Ethics Some theorists suppose that proper moral development requires moral instruction in virtue. In this view, moral instruction involves teaching relevant habits of action, perhaps habits of desire, in some cases also relevant skills. If a learner's dispositions fall more towards one of the extremes in one or another relevant range of behaviour, moral educators should encourage the learner to aim more towards the opposite extreme until the right balance is achieved. It is occasionally remarked that one thing wrong with contemporary American society is that too little attention is being paid to this sort of character development (e.g., Bennett, 1993). Some philosophers argue, further, that morality or perhaps the ordinary conception of morality is best analysed by beginning with a conception of virtue and character and then explaining other aspects of morality in terms of them (Taylor, 1991; Hursthouse, 1996). In this view, we determine what a person ought morally to do in a particular situation by considering what a person of good character would do in that situation. An act is morally right to the extent that it is the result of the agent's good character and morally wrong to the extent that it is the result of the agent's bad character. Perhaps we can also say that a situation or state of affairs is morally good to the extent that it would be favoured by a good person. Some versions of virtue ethics connect virtues with human flourishing. In one version, a virtue is a character trait that contributes to the flourishing of the agent. In another version, the virtues are character traits that contribute to the flourishing of people in general. In either version, as argued in Essay 9, it is not easy to provide a noncircular account of human flourishing that leaves the resulting view sounding plausible. The details of how virtue ethics might be developed are interesting, but I do not want to get into them on this occasion. For present purposes, the main point is that this sort of virtue ethics presupposes that there are character traits of the relevant sort, that people differ in what character traits they have, and that these traits help to explain differences in the way people behave. (Near the end of this essay I will briefly mention two versions of ‘virtue ethics’ that do not treat virtues as traits of character.)

10.4 Social Psychology Philosophers have begun to notice that recent social psychology challenges ordinary and philosophical views about character traits. Flanagan

MORAL PHILOSOPHY MEETS SOCIAL PSYCHOLOGY

169

(1991) discusses the challenge at length, arguing that it is not as radical as it may seem. Railton (1997) thinks the challenge is more serious, as does Doris (forthcoming) in an important book length study. Let me begin my own account by emphasizing that the empirical results of contemporary social psychology can seem extremely counterintuitive on first acquaintance. When students of mine return from Fall break after reading parts of Ross and Nisbett's useful textbook (Ross and Nisbett, 1991), they report that their parents express dismay at the ‘nonsense’ they are being taught at Princeton. Flanagan (1991), who is a pioneer in discussing the philosophical implications of the relevant social-psychological literature, does not seem to me fully to appreciate its radical import. He mentions what he calls the ‘extreme view’, according to which, ‘Good behaviour is not the result of good character. It is the result of a certain kind of dominating environment. Take away the powerful external props, and what seems to be a consistently good character will evaporate into thin air.’ He continues, ‘Almost no one holds such an extreme view.’ Contrary to this remark of Flanagan's, the ‘extreme view’ is widespread among social psychologists. Ross and Nisbett (1991) report that ‘[t]he experience of serious graduate students, who, over the course of four or five years, are immersed in the problems and the orientation of the field [of social psychology] . . . is an intellectually wrenching one. Their most basic assumptions about the nature and the causes of human behaviour . . . are challenged’ (p. 1). At one point, Ross and Nisbett ‘seriously entertained the hypothesis that most of [the] seeming order [in ordinary human behaviour] was a kind of cognitive illusion. We believed that human beings are adept at seeing things as they believe them to be, at explaining away contradictions and, in particular, at perceiving people as more consistent than they really are.’ Ross and Nisbett now think that there are at least regularities in human behaviour and that lay personality may work in the sense of enabling people to manage in ordinary life, just as lay physics works for many ordinary situations. ‘That is, people often make correct predictions on the basis of erroneous beliefs and defective prediction strategies’ (pp. 7–8). [I]n everyday experience the characteristics of actors and those of the situations they face are typically confounded—in ways that contribute to precisely the consistency that we perceive and count on in our social dealings. People often choose the situations to which they are exposed; and people often are chosen for situations on the basis of their manifest or presumed abilities and dispositions. Thus, clerics and criminals rarely face an identical or equivalent set of situational challenges.

170

VIRTUE

Rather they place themselves, and are placed by others, in situations that differ precisely in ways that induce clergy to look, act, feel, and think rather consistently like clergy and that induce criminals to look, act, feel, and think like criminals. (1991: 19) In addition, ‘individuals may behave in consistent ways that distinguish them from their peers not because of their enduring predispositions to be friendly, dependent, aggressive, or the like, but rather because they are pursuing consistent goals using consistent strategies, in the light of consistent ways of interpreting their social world’ (p. 20). And ‘people sometimes feel obliged, even committed to act consistently. This may be because of their social roles, because of the real-world incentives’ (p. 19).

10.5 Two Experiments Social psychologists have shown many different ways in which ordinary observers wrongly infer that actions are due to distinctive character traits of an agent rather than relevant aspects of the situation. Here I briefly review two wellknown experiments, one by Milgram and one by Darley and Batson.

10.5.1 Obedience to Authority Milgram (1963) describes an experiment in which a subject was given the task of administering an increasingly intense electric shock to a second person, the ‘learner’, whenever the learner gave the wrong answer. (Subjects were also told to treat a failure to answer as a wrong answer.) The shocks started at 15 volts and increased in 15-volt intervals to the highest level of 450 volts. The device used had labels at various points indicating ‘Slight Shock’, ‘Moderate Shock’, ‘Strong Shock’, ‘Very Strong Shock’, ‘Intense Shock’, ‘Extreme Intensity Shock’, ‘Danger: Severe Shock’, and ‘XXX’. At the 300-volt level the learner pounded loudly on the wall of the room but did not answer the question. This is repeated at the 315-volt level. At higher levels there was no further response from the learner. Whenever the subject asked the experimenter for advice or the subject said he did not want to continue, the experimenter had a list of four things to say, which would be said only if needed and only in sequence: (1) ‘Please continue’ or ‘Please go on,’ (2) ‘The experiment requires that you continue,’ (3) ‘It is absolutely essential that you continue,’ and (4) ‘You

MORAL PHILOSOPHY MEETS SOCIAL PSYCHOLOGY

171

have no other choice, you must go on.’ If the subject persisted in asking to stop after being told these four things, he or she would then be excused. The experiment was designed to test how far subjects would go in administering shock under these conditions. The experimenters had expected that few subjects would go beyond the designation ‘Very Strong Shock’ (150 volts). But in fact, of the forty subjects in one (typical) early experiment, all went past that point. Five stopped at the 300-volt level right before the label ‘Extremely Intense Shock’ and the point at which the learner pounded on the wall. Four more stopped at the next stage, 315 volts, when the learner pounded the wall again. Two stopped at 330 volts, when the learner made no response at all. One stopped at 345 volts and another at 360 volts. The twenty-six remaining subjects, 65 per cent of the total, continued on to 450 volts. In other words, most of the forty subjects went all the way to give the maximum shock. To repeat an important point, the experimenters (and others whom they questioned both before and after) did not at all expect this sort of result. They expected almost everyone to stop well before 300 volts, by 150 volts. In addition, people who have had the experiment described to them in detail, tend to be quite confident that, if they had participated in the original experiment, they would have stopped administering shocks at or before that relatively early point (150 volts), much earlier than anyone did in the actual experiment. Now consider any one of the subjects who went all the way to 450 volts, past the label ‘Danger: Severe Shock’ and well past the point at which the learner had stopped responding in any way. It is hard not to think there is something terribly wrong with that subject. It is extremely tempting to attribute the subject's performance to a character defect in the subject rather than to details of the situation. But can we really attribute a 2 to 1 majority response to a character defect? And what about the fact that all subjects were willing to go at least to the 300-volt level? Does everyone have this character defect? Is that really the right way to explain Milgram's results? A different kind of explanation (Ross & Nisbett, 1991: 56–8) invokes relevant features of the situation. First, there is ‘the step-wise character of the shift from relatively unobjectionable behaviour to complicity in a pointless, cruel, and dangerous ordeal’, making it difficult to find a rationale to stop at one point rather than another. Second, ‘the difficulty in moving from the intention to discontinue to the actual termination of their participation’, given the experimenter's refusal to accept a simple announcement that the subject is quitting—‘The experiment requires

172

VIRTUE

that you continue.’ Third, as the experiment went on, ‘the events that unfolded did not “make sense” or “add up” . . . The subjects' task was that of administering severe electric shocks to a learner who was no longer attempting to learn anything . . . [T]here was simply no way for [subjects] to arrive at a stable “definition of the situation”.’ The fundamental attribution error in this case consists in ‘how readily the observer makes erroneous inferences about the actor's destructive obedience (or foolish conformity) by taking the behaviour at face value and presuming that extreme personal dispositions are at fault’.

10.5.2 Good Samaritans The second experiment that I will mention derives from the parable of the Good Samaritan, which goes like this. ‘And who is my neighbour?’ Jesus replied. ‘A man was going down from Jerusalem to Jericho, and he fell among robbers, who stripped him and beat him, and departed, leaving him half dead. Now by chance a priest was going down the road; and when he saw him he passed by on the other side. So likewise a Levite, when he came to the place and saw him, passed by on the other side. [Levites were important participants in temple ceremonies.] But a Samaritan [a religious outcast], as he journeyed, came to where he was; and when he saw him, he had compassion and went to him and bound his wounds, pouring on oil and wine; then he set him on his own beast and brought him to an inn, and took care of him. And the next day he took out two dennarii and gave them to the innkeeper, saying, “Take care of him; and whatever more you spend, I will repay you when I come back.” Which of these three, do you think, proved neighbour to him who fell among the robbers?’ He said, ‘The one who showed mercy on him’. And Jesus said to him, ‘Go and do likewise’. (Luke 10: 29–37, Revised Standard Version) Darley and Batson (1973) observe that people can envision various differences between the priest and Levite on the one hand and the Samaritan on the other hand. The priest and Levite might have well have had their minds on religious matters, whereas the Samaritan probably did not. The priest and Levite were probably hurrying along to various appointments, whereas the Samaritan was probably less in a hurry. The parable also suggests that there is a difference in type of religiosity or morality. The priest and Levite act virtuously in order to please God, where the Samaritan responds more directly to the needs of another person. The standard interpretation of the parable focuses on the third of these variables, the type of religious or moral character of the agent. Darley and Batson designed an experiment aimed at uncovering which

MORAL PHILOSOPHY MEETS SOCIAL PSYCHOLOGY

173

of these differences might be most relevant to explaining the differences in behaviour. Subjects in this experiment were students at Princeton Theological Seminary. As each subject arrived, he was informed that he was to give a talk that would be recorded in another building. Along the way to the place for the talk, the subject encountered a ‘victim’ slumped in a doorway. The question was under what conditions would a subject stop to help the victim. Half of the subjects were assigned to talk on the Good Samaritan Parable; the others were assigned a different topic. Some of the subjects were told they were late and should hurry; some were told they had just enough time to get to the recording room; and some were told they would arrive early. Judging by their responses to a questionnaire, they had different religious and moral orientations. The only one of these variables that made a difference was how much of a hurry the subjects were in. 63 per cent of subjects that were in no hurry stopped to help, 45 per cent of those in a moderate hurry stopped, and 10 per cent of those that were in a great hurry stopped. It made no difference whether the students were assigned to talk on the Good Samaritan Parable, nor did it matter what their religious outlook was. Standard interpretations of the Good Samaritan Parable commit the fundamental attribution error of overlooking the situational factors, in this case overlooking how much of a hurry the various agents might be in.

10.6 Direct Empirical Challenges to Character Traits But don't we know from ordinary experience that people differ in character traits? Here it is useful to consider related issues. Psychoanalysts acquire a considerable experience in treating patients and can cite many instances in which psychoanalytic treatment is successful. However, empirical studies of psychoanalytic treatment as compared with no treatment have found no objective benefit. (Dawes, 1994) Some diagnosticians have used Rorschach inkblot tests to make psychological diagnoses. It seemed to those using these tests that they had abundant evidence that certain characteristics of the test results were diagnostic of certain disorders. Empirical studies showed there was no correlation between those characteristics and the test results (Nisbett and Ross, 1980: 93–7).

174

VIRTUE

Many employers are convinced that useful information can be gained from interviewing potential employees. However, for the most part, interviews simply add noise to the decision process. Empirical studies indicate that decisions made on information available apart from an interview are more reliable than decisions made when an interview is added (Ross and Nisbett, 1991: 136–8). Discovery of such errors in reasoning has encouraged research into why people are subject to such errors (Tversky and Kahneman, 1974, Nisbett and Ross, 1980). One suggested reason is confirmation bias. Given a hypothesis, one tends to look for confirming evidence. Finding such evidence, one takes it to support the hypothesis. Evidence against the hypothesis tends to be ignored or downplayed. Ross and Nisbett suggest that the initial source of the fundamental attribution error may have to do with Gestalt considerations of figure and ground. Where we distinguish figure from ground, we pay more attention to figure and less to ground and we try to explain what happens in terms of features of the figure rather than in terms of features of the ground. Typically the actor is figure and the situation is ground, so we seek an explanation of the action in features of the actor in the foreground rather than in features of the background situation. The suggested explanation is then subject to confirmation bias. Additional support comes from the fact that other people give explanations in terms of dispositional features of agents rather than in terms of aspects of their situations. When investigators have looked for objective evidence that people differ in character traits, the results have been much as with psychoanalysis, Rorschach tests, and interviews. People take themselves to have lots of evidence that agents differ in character traits. Yet empirical studies have not found any objective basis for this confidence. Summarizing a number of studies, Ross and Nisbett (1991: 95) report that the ‘average correlation between different behavioural measures designed to tap the same personality trait (for example, impulsivity, honesty, dependency, or the like) was typically in the range between 0.10 and 0.20, and often was even lower’. These are very low correlations, below the level which people can detect. Using such correlations to make predictions yields hardly any improvement over guessing. Even if predictions are limited to people one takes to be quite high on a particular trait, the correlations are still very low. Ross and Nisbett observe that people have some appreciation of the role of situation in the way they understand such stories as The Prince and

MORAL PHILOSOPHY MEETS SOCIAL PSYCHOLOGY

175

the Pauper or the movie Trading Places. But for the most part, people are quick to infer from specific actions to character traits. It is true that there are better correlations for very specific situations. Hartshorne and May (1928) found that the tendency to copy from an answer key on a general information test on one occasion was correlated .79 with copying from an answer key on a similar test six months later. Newcomb (1929) found that talkativeness at lunch was a highly stable attribute; it just was not very highly correlated with talkativeness on other occasions. (Ross and Nisbett, 1991: 101) Surprisingly, Flanagan (1991) argues that this shows there really are character traits, ‘albeit not traits of unrestricted globality or totally context-independent ones’. I guess he means such character traits as ‘being disposed to copy from an answer key on a certain sort of test’ and ‘being talkative at lunch’. But, first, no reason has been given for thinking that these specific narrow regularities in behaviour reflect dispositions or habits rather than, for example, skills or strategies that have worked in the past. Second, and more importantly for our purposes, ordinary thinking about personality and character attributes is concerned with more global traits like honesty and talkativeness. Flanagan concludes: ‘Yes, there are character traits. The language of character traits picks out psychologically real phenomena.’ But I do not see that he has cited any empirical evidence for this claim. Flanagan also seems to think that it is inconsistent to argue against character traits by appeal to the fundamental attribution error. He says, ‘It is telling against the situationalist who is also an eliminativist that he will have extreme difficulty (indeed he courts inconsistency) in positing attributional biases of any sort if by these he means to refer to what he must be taken to want to refer to, namely, dispositions to think in certain ways’ (1991: 305). But this is true only if a ‘situationalist’ is someone who denies that there are any dispositions at all, or who (perhaps like Skinner, 1974) denies that it is useful to explain anything in terms of dispositions. The issue we have been concerned with is whether people differ in certain particular dispositions—character traits. To deny that people differ significantly in character traits is not to deny that they have any dispositions at all. People might well all share certain dispositions, such as a disposition to make the fundamental attribution error. Secondly, they might differ in various dispositions that do not constitute character traits, such as personality disorders and other mental illnesses. (So, for example, to deny

176

VIRTUE

that there are character traits is not to accept the view in Laing, 1960, that schizophrenia is simply a rational response to a difficult family situation.)

10.7 Benets of Appreciating the Fundamental Attribution Error There are various benefits to a proper appreciation of ways in which ordinary moral thinking rest on the fundamental attribution error.

10.7.1 Philosophy 10.7.1.1 Virtue Ethics Character-based virtue ethics may offer a reasonable account of ordinary moral views. But to that extent, these ordinary views rest on error. It is worth mentioning that there are variants of virtue ethics that do not require character traits in the ordinary sense. For example, Thomson (1996) tries to explicate moral thinking by appeal to judgements about whether particular actions are just or courageous or whatever. To the extent that such judgements are concerned entirely with the action and not with any presumed underlying trait of character, Thomson's enterprise is unaffected by my discussion. Maria Merritt (1999) has been developing a version of virtue ethics that emphasizes the role of the situation in maintaining relevant regularities in behaviour.

10.7.1.2 Better Understanding of Moral Luck. Adam Smith (1759) wrote about the influence of fortune on our moral judgements, giving nice examples. Someone carelessly throws a brick over a wall. His companion may complain about this even if no harm is done. But if the brick does hit someone, much greater condemnation of the agent ensues. Nagel (1979) gives a similar example of a driver who takes his eyes off the road for a second. That's bad, but suppose in that second a child darts into the street and is hit. Then much worse condemnation of the agent seems appropriate. Smith and Nagel note that, from a certain point of view, our moral judgement of the agent should be based entirely on the motives of the agent and the agent's epistemic situation, so that from that point of view

MORAL PHILOSOPHY MEETS SOCIAL PSYCHOLOGY

177

there should be no difference between two cases that are the same in those respects in one of which someone is hit by the brick (or car) and in the other of which no one is hit. Yet it is clear that we will judge the cases differently. Perhaps these are simply further instances of the fundamental attribution error. This bad thing has happened and we attribute it to the bad character of the agent in the foreground.

10.7.2 Real Life 10.7.2.1 Moral Education If there is no such thing as character, then there is no such thing as character building.

10.7.2.2 Tolerance When things go wrong, we typically blame the agent, attributing the bad results to the agent's bad character. Even when things do not go bad, we are quick to interpret actions as expressive of character traits, often hostile traits. For example, a person with poor vision may fail to recognize an acquaintance, who then attributes this to coldness in that person. A greater understanding of the agent's situation and how it contributed to the action can lead to a greater tolerance and understanding of others.

10.7.2.3 Better Understanding of Ethnic Hatred Internal conflicts in the 1990s in the former Yugoslavia are often attributed to historical ‘ethnic hatreds’. Yet it is possible to explain these events in rational terms (Hardin, 1995). Suppose there are limited resources and a successful coalition will benefit its members more than those excluded from the coalition. Such a coalition is possible only if insiders can be distinguished from excluded outsiders and only if it is possible to keep members from defecting to other groups. Coalitions formed around ethnic or religious lines might succeed. The threat that one such coalition may form can lead other groups to form competing coalitions and to struggle against each other. If stakes are high enough, such struggles can become violent. If we attribute the resulting violence to ethnic hatred, we may very well doubt that there is anything we can do. If we understand the way the violence arises from the situation, we may see more opportunities to end the conflict.

178

VIRTUE

10.8 Summary We very confidently attribute character traits to other people in order to explain their behaviour. But our attributions tend to be wildly incorrect and, in fact, there is no evidence that people differ in character traits. They differ in their situations and in their perceptions of their situations. They differ in their goals, strategies, neuroses, optimism, etc. But character traits do not explain what differences there are. Our ordinary views about character traits can be explained without supposing that there are such traits. In trying to explain why someone has acted in a certain way, we concentrate on the figure and ignore the ground. We look at the agent and ignore the situation. We are naive in our understanding of the way others view a given situation. We suffer from a confirmation bias that leads us to ignore evidence against our attributions of character. It is very hard to do studies that might indicate whether or not people differ in character traits, but the few studies that have been done do not support this idea. We must conclude that, despite appearances, there is no empirical support for the existence of character traits. Furthermore, it is clear that ordinary thinking about character traits has deplorable results, leading to massive understanding of other people, promoting unnecessary hostility between individuals and groups, distorting discussions of law and public policy, and preventing the implementation of situational changes that could have useful results.

Part IV Explaining Value

This page intentionally left blank

11 Moral Agent and Impartial Spectator One important type of ethical theory treats moral properties as analogous in certain respects to ‘secondary qualities’ like colours. According to this sort of theory, whether something is right or wrong depends on how impartial spectators would react to it. In the eighteenth century, the Scottish philosophers Francis Hutcheson (1725, 1728), David Hume (1739, 1748), and Adam Smith (1976) explored theories of this type. In the twentieth century, this sort of theory was sometimes discussed under the name ‘ideal observer theory’ (Sharp, 1950; Firth, 1952; Brandt, 1959). In the latter quarter of the twentieth century, especially in Britain, there was renewed interest in this sort of ethical theory and its comparison between moral properties and secondary qualities. David Wiggins (1976), John McDowell (1985), and Thomas L. Carson (1984) argued in favour of a view of this sort. Colin McGinn (1983) and Simon Blackburn (1984) argued against it. One possible objection to an impartial spectator theory is that it seems to require an overly aesthetic conception of morality to take the primary point of view in ethics to be that of a spectator rather than that of the agent (Hampshire, 1972). If the spectator is taken to be primary, then the agent's aim would seem to be to produce something that will or would please the spectator. But that is just wrong. Such an aim is too ‘outer directed’ to count as a moral motive. Morality is more agent-centred than that. It is much more plausible to take the agent's point of view as primary. In the first instance, morality is a matter of the moral reasons an agent has to act in one way or another, where these reasons derive from the relevant moral rules rather than from a desire to gain the approval of spectators. I pressed this objection myself when I discussed the ideal observer theory in a textbook of ethics (Harman, 1977). But I find I was too hasty. In reading Adam Smith's Theory of the Moral Sentiments, I discovered that Smith explicitly considers this issue and provides a plausible reply to the objection.

182

EXPLAINING VALUE

Because recent discussions have tended not to consider such ‘psychological’ questions as why agents might be motivated to act in ways that impartial spectators would approve (or, for that matter, why impartial spectators would care about anything), I will in this essay ignore recent discussion and return to the three great versions of the theory that were developed in the eighteenth century by Hutcheson, Hume, and Smith, indicating why I think Smith's version of the theory is superior to the others.

11.1 Analogy With Secondary Qualities Hutcheson and Hume each model an impartial spectator theory of moral right and wrong after a normal perceiver theory of secondary qualities like colour, taste, sound, etc. The theory of colour, for example, takes facts about colours to be facts about the reactions of normal perceivers under ideal lighting conditions. For an object to be red is, in this view, for the object to be such that it would look red to normal observers in good light. Similarly, according to Hutcheson and Hume, for an action to be wrong is for the action to be such that it would displease normal observers under conditions ideal for reacting to actions. This sort of account explains a property by appeal to the psychological experience of an observer. Redness is explained in terms of an observer's visual experience—how things look to an observer. Wrongness is explained in terms of moral experience—how things feel to an observer, what attitude (positive or negative) the observer takes. The relevant experience does not have to be actual. It is the experience an observer would have under certain ideal conditions. In yellow light, an object may look orange to observers but actually be red—if it would look red to normal observers in good white light. An action can seem hateful to actual biased or uninformed observers but really be morally right—if the action would be favoured by impartial spectators who knew all the facts. In fact, the distinction between actual and merely hypothetical reactions was not so clear in the eighteenth century. Philosophers who adopted this general approach were apt to worry over the question whether a falling tree makes any sound if no one hears it. The correct answer should have been that the tree does make a loud sound, because, if someone had been there, he or she would have heard something. But philosophers did not always see the possibility of this response. Berkeley thought that the

MORAL AGENT AND IMPARTIAL SPECTATOR

183

only way to allow for the sound of the falling tree was to have the sound heard by God. Indeed, God was needed even for the tree, since without God's perception, not only would there be no sound, but the tree itself would not be there to fall, according to Berkeley. Hume rejected God as a way of saving the falling tree and its sound and concluded that the unperceived tree was a confused fiction we postulate to give order to our experience of the forest. But, although eighteenth-century proponents of theories of this sort do not clearly distinguish between actual and hypothetical reactions of spectators, once the distinction is recognized, it is clear that theories of this sort should refer to hypothetical reactions rather than actual reactions of observers (Firth, 1952). Even so there are problems (Shope, 1978). It seems possible that there might be a red object that would turn green if placed in good light. This red object would look green to normal observers if placed in good light. How can that be, if for the object to be red is for it to be such that it would look red to normal observers if placed in good light? One possible answer is that the object would indeed look red if placed in good light and the object were not to change colour. But then, what is it in this view for something to change colour? Similarly, it seems possible that there could be a wrong action that would have been right if only the act had been considered by impartial spectators. Its bad consequences might outweigh its good consequences, but, if it had been considered by impartial spectators as an example, it would have had enough additional good consequences to turn the act into the right thing to do. Then we seem to have a wrong action that would be approved by impartial spectators if they considered it, again violating the claim of this sort of spectator theory. A possible response is to say that an act is wrong if the act would be disapproved of by impartial spectators who considered the act as it actually was, ignoring any features or consequences attaching to the act through its being considered by the spectators. But let us forget about these (admittedly serious) problems to return to the basic idea behind impartial spectator theories, namely, that the rightness or wrongness of actions is dependent on the actual or hypothetical reactions of impartial spectators in a way that is analogous to the way in which what colour something is depends on how it looks or would look to normal observers in good light. Of course, the analogy can only be partial. One important difference is that an impartial moral spectator does not have to perceive the act being

184

EXPLAINING VALUE

judged. It is enough for the ‘spectator’ to be given a sufficiently full description of the act. Indeed, the act itself may well be merely hypothetical. The agent may be considering whether to do it or not. If the act is wrong, let us hope that the agent does not carry it out. In that case, the act does not exist. The important point of the analogy is that, just as the theory of colour takes colour judgements to be about the hypothetical reactions of normal perceivers in ideal situations, an impartial spectator theory takes moral judgements to be judgements about the hypothetical reactions of impartial, knowledgeable people. So, in this view, to say that something is morally right is, roughly speaking, to say that impartial spectators are or would be in favour of it. To say that something is wrong is to say that impartial spectators are or would be opposed to it.

11.2 Key Issues for Impartial Spectator Theories I now want to consider the following two issues for impartial spectator theories in ethics. First, why should an agent care about the reactions of spectators? Second, why should an impartial spectator care about what an agent does? The first issue lies behind the objection to impartial spectator theories that I have already mentioned. According to such a theory, the desire to do what is right is the desire to act in a way that spectators will approve. But that is too ‘outer directed’. Such a desire is precisely not a desire to do something simply because it is right. Impartial spectator theories might try to avoid this objection by arguing that agents are not motivated directly to do what spectators would approve of. A moral agent's intention is not of the form, ‘Let me do what would gain an impartial spectator's approval.’ Rather, the agent's intention has the form, ‘Let me do D,’ where in fact doing D is something that impartial spectators would approve, but that fact is not the agent's reason for doing D. Of course, an adequate impartial spectator theory cannot treat it as a mere coincidence that moral agents are motivated to act in ways that impartial spectators would approve. So this leaves the problem of explaining how moral agents come to act in ways that impartial spectators would approve of without the agents' having the specific intention to act in that way.

MORAL AGENT AND IMPARTIAL SPECTATOR

185

The second issue for an impartial spectator theory concerns why the spectator cares about what the agent does. This issue lies behind the question whether the reason an act is right is that impartial spectators favour it, or, rather, the reason that impartial spectators favour the act is that the act is right. In the dialogue Euthyphro, Plato has Socrates ask a similar question of the view that something is good if and only if it is beloved by the gods: are actions good because they are loved by the gods or are the relevant actions loved by the gods because the actions are good? Socrates raises the question in order to insinuate that the correct answer is the second one—good actions are loved by the gods because the actions are good—whereas the theory he is discussing must argue that the correct answer is the first one—good actions are good because they are loved by the gods. Both answers cannot be right. It cannot be true both (1) that for an action to be good is for it to be loved by the gods, and also (2) that what the gods love about the action is that the action is good. For that would imply that what the gods love about the action is that they love it, which is perverse. Similarly, it might be argued against an impartial spectator theory that the second option is the correct one: truly impartial spectators favour actions that are right because the actions are right. Impartial spectator theories are necessarily committed to taking the first option here: acts are right because they would be favoured by impartial spectators, who must therefore favour these acts for other reasons than that the actions are right. For it would be perverse to suppose that what impartial spectators would favour in these acts is that the acts would be favoured by impartial spectators. This leaves the problem of saying what it is that would lead impartial spectators to favour one or another course of action. To summarize, any impartial spectator theory in ethics needs to say what explains an agent's moral motivation, what explains a spectator's reaction, and why these two things, agent's motivation and spectator's reaction, should be correlated with each other.

11.3 Hutcheson's Theory Francis Hutcheson (1694–1746) appeals to benevolence to answer these questions. In his view, people are innately disposed to like other people's being happy and to dislike others' being unhappy. On the one hand, this tendency motivates agents to act so as to make other people happier. On

186

EXPLAINING VALUE

the other hand, it leads spectators to favour such actions and to oppose actions that have the opposite tendency. In Hutcheson's view, an agent acts morally in order to make other people happier. The agent is not motivated to act so as to gain the approval of spectators. To be sure, the agent is motivated to act in a way that spectators would approve. Although this is not an intended aspect of the agent's action, it is also no accident, because the same sort of benevolence that leads the agent to act as he or she does also leads the spectator to approve of so acting. The agent's act is right because impartial spectators would favour the agent's acting in that way. Spectators would not in the first instance favour the act because the act is right but would favour the act because they want people to be happy and the act makes people happy. This view yields classical utilitarianism. In Hutcheson's words, ‘that Action is best, which procures the greatest Happiness for the greatest Numbers; and that, worst, which, in like manner, occasions Misery’ (Hutcheson, 1725: III. viii). This is an elegant theory, but it faces a serious objection. The account of moral motivation is implausible. Benevolence is too weak a motive to be identified with the motive to do what is right. People are sometimes motivated by benevolence to try to improve the lot of other people, but this motivation is normally quite feeble when compared with ordinary people's aversions to murder, injuring others, stealing, lying, and failing to keep their promises or pay their debts. Generalized benevolence is normally a much weaker motive than self-interest. But the moral motives just mentioned—to avoid killing others, and so forth—are often just as strong as and sometimes stronger than selfinterest. Generalized benevolence cannot be the whole story about moral motivation.

11.4 Hume's Theory Impressed by these facts about moral motivation, Hume, like Hobbes (1957) and Mandeville (1714), sees self-interest as an important aspect of the motivation of certain moral acts, namely acts of ‘justice’ such as telling the truth and keeping promises. But, unlike Hobbes and Mandeville and like Hutcheson, Hume argues that benevolence (based on what he calls ‘sympathy’) is another aspect. Self-interest is involved because, if you cannot be trusted to tell the truth, keep your promises, or avoid injuring your associates, people will not engage with you in common enterprises and

MORAL AGENT AND IMPARTIAL SPECTATOR

187

you will lose out in comparison with other people who do tell the truth, keep their promises, and avoid injury to associates. So you have a strong self-interested reason to keep your obligations. In this view, self-interest leads people to enter into cooperation with others. Over time, co-operation becomes more formalised into a system of cooperation. The overall system of co-operation makes people better off, so benevolent spectators approve of this self-interested motivation. It therefore counts as moral motivation. Hume allows for benevolent motivation in addition to self-interest. Benevolent feelings in the agent may reinforce selfinterested behaviour, leading to even stronger motivation. But benevolence by itself is a weak motive. So the motive to be charitable to others is weaker than the motive to keep your obligations. Charity is not in your interest in the way that promise-keeping is. If benevolence plays only a small role in moral motivation, it plays more of a role in explaining the reactions of spectators, in Hume's view. Impartial spectators are (by definition) not personally affected by the agent's act, so selfinterest does not favour or oppose the act. Benevolence is therefore the sole source of a spectator's approval or disapproval. Hume offers an explanation of benevolence in terms of what he calls ‘sympathy’. This contrasts with Hutcheson, who simply postulates that God has made us benevolent. Hume's account of sympathy goes like this. To think of someone else as in pain is to have a painful feeling yourself, but located in your image of the other person. To think of another person as pleased is to have a pleasurable feeling located in your image of that person. So, it is pleasant to think that others are happy and unpleasant to think that others are unhappy. That is why impartial spectators prefer agents to do things that tend to make people happier. For Hume, the association of ideas plays an important role in determining what spectators will approve of. When the spectator thinks of the agent acting in a certain way, association of ideas leads the spectator to think of the typical effects of such an action. This thought makes the spectator happy to the extent that these envisioned typical effects involve happiness. So, to the extent that Hume's theory is utilitarian, it tends toward rule utilitarianism rather than act utilitarianism. Roughly speaking, act utilitarianism says that an act is right to the extent that that very act makes people happier or less unhappy. Rule utilitarianism says that an act is right to the extent that acts of that sort tend to make people happier or less unhappy. Now, any given particular act is of many different sorts of act, so a crucial question for rule utilitarianism

188

EXPLAINING VALUE

is to decide which are the relevant sorts of act for the purposes of moral evaluation. In Hume's version of the theory, this is a psychological question: in thinking about a particular act, a spectator will associate that act with various other acts; this psychological association determines what the relevant sorts of act are for the purposes of moral evaluation. Hume's theory has little difficulty with the second problem facing an impartial spectator theory, the problem of accounting for the spectator's approval of right actions without supposing the spectator approves of them because they are right. In Hume's view, a spectator's approval arises simply from the spectator's sympathetic thought of the pleasures and pains produced by the agent's act and any associated acts, not from any judgement that the act is right. The act is right because it is favoured, not favoured because it is right. But Hume's account of moral motivation is still not very plausible. In his view, a moral agent acts largely out of selfinterest or out of habits for which there is a self-interested justification. This is an outer-directed motivation—the agent is concerned with the reactions of others because he or she wants them to continue dealing with him or her. This seems wrong. As Kant (1785) objects, a shopkeeper who gives children the correct change because it would be bad for his business if people were to think he or she cheated children is doing the right thing, but not acting from a moral motive. The shopkeeper's act has no particular moral worth. Hutcheson identifies the moral motive with benevolence. That seems a motive of the right sort to be a moral motive—it is aimed in the right direction—but it is not a strong enough motive. (Kant famously disagrees, holding that benevolent motivation too is of no moral worth. But here Kant's view is counterintuitive.) Moral motivation is much stronger than benevolence. Hume takes the moral motive to be mainly self-interest, which is a strong enough motive, but a motive that does not seem to be aimed in the right direction.

11.5 Adam Smith's Theory Adam Smith (1976) bases his positive theory on an insightful criticism of Hume's account of sympathy. Smith points out that Hume is wrong to suppose that merely knowing what another person feels is sufficient for sympathy with that person. You might very well not sympathize with

MORAL AGENT AND IMPARTIAL SPECTATOR

189

another person's feelings if you thought those feelings were inappropriate to the situation. Consider someone who is very upset over a minor scratch on his knee from a fall, for example. You can understand perfectly well how upset the person is without at all sympathizing with the person's extreme agitation. Smith observes, further, that Hume is also wrong in taking sympathy with another person's pain to be always entirely unpleasant. It is more unpleasant to have dealings with someone who is inappropriately unhappy, too upset about something the person ought to treat as a trifle, than it is to have dealings with someone who is upset to the same extent but for an appropriate reason. In the first case, you do not sympathize with the person; in the second case you do sympathize with the person. The other person's pain is easier to take if you can sympathize with that person. Smith thinks Hume is right about the importance of sympathy in ethics but wrong about what sympathy is and why it is important. For Smith, the key point is that sympathy is desirable. Not only do spectators want to be able to sympathize with agents, but agents also want the sympathy of spectators. This gives agents a motive to try to have reactions of the sort that spectators can sympathize with. Furthermore, in Smith's view, ‘Sympathy . . . does not arise so much from the view of the passion, as from that of the situation which excites it’ (1976: I. i. 1. 10). This gives a spectator a standard for judging the other agent. The spectator imagines himself or herself in the circumstances of the agent and imagines how he or she would react in those circumstances. If the agent's reaction is similar to the reaction the spectator imagines having, the spectator sympathizes with the agent. If the agent's reaction is more extreme than the spectator's imagined reaction, the spectator does not sympathize with the agent. Spectators approve of reactions they can sympathize with and disapprove of reactions they cannot sympathize with. In Smith's words, ‘To approve or disapprove . . . of the opinions of others is acknowledged, by every body, to mean no more than to observe their agreement or disagreement with our own. But this is equally the case with regard to our approbation or disapprobation of the sentiments or passions of others’ (1976: I. i. 3. 2). Smith's approach leads to a very different conception of the content of ethics. Hutcheson's and Hume's theories imply utilitarianism. The aim is to maximize happiness and minimize unhappiness. Smith's theory has no such implication. Spectators do not just add up pleasures and pains. Smith's normative theory is more stoical than utilitarian. His approach

190

EXPLAINING VALUE

puts considerable importance on self-control. According to Smith, the feelings an ordinary nonideal spectator imagines having in a given situation are rarely as intense as the feelings someone else would have in that situation. What an actual spectator imagines an agent feeling is rarely as intense as what the agent actually is feeling. So, the reaction the spectator imagines having in the agent's circumstances tends not to be as extreme as the agent's actual reaction would be if that reaction were not influenced by the thought of how the reaction might appear to spectators. Since a person wants sympathy after stubbing a toe, the person will not for long yell and shout and moan, but will try to restrain his or her feelings so as not to appear ridiculous, Of course, a more knowledgeable impartial spectator would not underestimate the intensity of an agent's pain. But such a spectator would still favour restraint in the agent's reactions, because the reaction that the spectator (correctly) envisions having in the agent's circumstances would be restrained. The spectator would by now have acquired a habit of restraint in reactions as a way of obtaining the sympathy of people who do underestimate the intensity of pains in others. In Smith's view, the spectator's reactions are heavily influenced by convention. A spectator tries to imagine how he or she would react in the agent's circumstances. But, if the spectator were in the agent's circumstances, he or she would try to modify an otherwise natural reaction so as to accord with the imagined reactions of other, not necessarily ideal, spectators. This means that a spectator will be influenced strongly by his or her expectations of how people ordinarily act and react. Conventional ways of acting and reacting serve as evidence about the feelings of other impartial spectators. When a given spectator is imagining how he or she would react, since the spectator will imagine acting in ways that other spectators will sympathize with, what he or she imagines will be skewed in the direction of the conventional reactions. So, Smith's theory is much more conventionalistic than Hutcheson's or even Hume's. Hume takes convention to be important because conventions are useful: people are happier because of what they can accomplish when they adhere to conventions. But, for Smith, conventions have a more direct effect. The conventions a spectator participates in determine how the spectator will react and so determine what reactions the spectator will sympathize with. How does Smith handle the problems with which we began? His response to the second problem concerning spectators' reasons for favouring certain actions is the same as the response made by Hutcheson and Hume. The spectators approve or disapprove of actions depending on

MORAL AGENT AND IMPARTIAL SPECTATOR

191

whether or not they can sympathize with them. Acts are wrong because spectators disapprove of them, that is, because they do not sympathize with them. It is not that the spectators disapprove of the acts because the acts are wrong. But what about the first problem concerning the motivation of moral agents? Hutcheson's appeal to benevolence did not work because of the weakness of benevolence. Hume's appeal to self-interest refers to a strong enough motive, but one that is too outer directed if it is involved in the way that Hume thinks it is involved. It may seem that Smith's theory resembles Hume's in this respect. Here again it looks as if the envisioned source of moral motivation is strong enough but too outer directed. Smith seems to be saying that moral motivation is motivation to act so as to obtain the sympathy of spectators. That seems quite wrong. But in fact, Smith explicitly denies that moral motivation is motivation so as to obtain the sympathy of others. An agent is motivated to be worthy of praise: The love of praise-worthiness is by no means derived altogether from the love of praise . . . The love and admiration which we naturally conceive for those whose character and conduct we approve of, necessarily dispose us to desire to become ourselves the objects of the like agreeable sentiments, and to be as amiable and as admirable as those whom we love and admire the most. . . . Emulation, the anxious desire that we ourselves should excel, is originally founded in our admiration of the excellence of others. Neither can we be satisfied with being merely admired for what other people are admired. We must at least believe ourselves to be admirable for what they are admirable. . . . [If others praise us, their] praise necessarily strengthens our own sense of our own praiseworthiness. In this case, so far is the love of praise-worthiness from being derived altogether from that of praise; that the love of praise seems, at least in a great measure, to be derived from that of praise-worthiness. (III. 2. 2–3; cf McDowell, 1985: 117–20, and Wiggins, 1987: 189) In order to accommodate this observation, Smith postulates a primitive desire to be worthy of praise: Nature, when she formed man for society, endowed him with an original desire to please, and an original aversion to offend his brethren. She taught him to feel pleasure in their favourable, and pain in their unfavourable regard. . . . But this desire of the approbation, and this aversion to the disapprobation of his brethren, would not alone have rendered him fit for that society for which he was made. Nature, accordingly, has endowed him, not only with a desire of being

192

EXPLAINING VALUE

approved of, but with a desire of being what ought to be approved of, or of being what he himself approves of in other men. (III. 2. 6–7) This passage is difficult to interpret, since a desire ‘of being what ought to be approved or’ is not quite the same as a desire ‘of being what [one] approves of in other men’. And simply postulating a desire to be ‘what ought to be approved of ’ would not eliminate the difficulty. According to Smith's impartial spectator theory, a desire to be ‘what ought to be approved of ’ is precisely a desire to be such that impartial spectators would approve of oneself. But, as Smith emphasizes at length, that desire is not yet of the right sort. A desire to be ‘what he himself approves of in other men’ is more to the point, if it means a desire to be ‘what he himself would approve of in others from an impartial perspective’. But Smith does not really postulate any such desire as a basic unexplained fact about people. Instead, he offers a more complex account of moral motivation that anticipates certain elements of Freud's theory of the superego. Smith holds that, at first, a child is motivated to restrain its reactions so as to have the sympathy of parents and other spectators. As an aid in doing this, the child tries to view itself as seen by others. Eventually, it acquires a habit of doing this—a habit of pretending to be an impartial spectator of its own actions and reactions to see which actions it approves and which it disapproves. The child then tries to restrain its reactions so as to be approved by this pretend spectator—the little person within who now serves as a kind of inner conscience. When I endeavour to examine my own conduct, when I endeavour to pass sentence upon it, and either to approve or condemn it, it is evident that, in all such cases, I divide myself, as it were, into two persons; and that I, the examiner and judge, represent a different character from that other I, the person whose conduct is examined into and judged of. The first is the spectator, whose sentiments with regard to my own conduct I endeavour to enter into, by placing myself in his situation, and by considering how it would appear to me, when seen from that particular point of view. The second is the agent, the person whom I properly call myself, and of whose conduct, under the character of a spectator, I was endeavouring to form some opinion. The first is the judge; the second the person judged of. But that the judge should, in every respect, be the same with the person judged of, is as impossible as that the cause should, in every respect, be the same with the effect. (III. 1. 6). This might seem to be even worse than purely outer-directed motivation. It may seem that Smith is saying that the agent is motivated to obtain the sympathy and approval of a pretend person (Harman, 1977: 61–2).

MORAL AGENT AND IMPARTIAL SPECTATOR

193

But this objection to Smith is a mistake. Despite the division into what Smith calls ‘two persons’, the ‘examiner and judge’ is not really any different from ‘the person I call myself ’. The ‘examiner and judge’ is the agent himself or herself, viewing things from a certain perspective. When the agent pretends to be an impartial spectator, he or she ends up approving and disapproving from an impartial point of view. Viewing things in that way can then influence the agent's motives and feelings, since it is the agent who is doing the viewing and the approving and disapproving and the approval or disapproval is of the agent himself or herself. Consequently, the agent's actual motives will become more moral, because they are in part the result of the agent's looking at things from a moral point of view. To the extent that the agent views things impartially, the agent will genuinely not want to cheat and injure others. What starts as a strategy for knowing what to do to obtain sympathy ends up giving the agent a genuinely moral outlook that can motivate moral behaviour. The truly moral agent does not aim at getting the sympathy of impartial observers. Instead, the moral agent is motivated to act morally by virtue of motives acquired by viewing things from an impartial standpoint. The moral agent cares about the things that an impartial spectator cares about because the agent has in part become an impartial spectator.

11.5 Concluding Summary Hutcheson, Hume, and Smith all put forward versions of the impartial spectator theory. All three agree that right acts are right because they would be favoured by impartial spectators who favour these acts for other reasons than that the acts are right. But they disagree about what explains spectators' reactions and what accounts for moral motivation. They also disagree as to whether the same thing explains agents' motivation and spectators' approval. Hutcheson and Hume agree that spectators' reactions derive from benevolence. Hutcheson believes that benevolence is also the source of moral motivation. Hume thinks this is too weak a motivation; he thinks moral motivation is also based in self-interest. Where Hutcheson simply treats benevolence as a God-given motive, Hume offers an explanation of benevolence in terms of what he calls sympathy. Smith disagrees with Hume as to what sympathy is. Smith also disagrees with both Hume and Hutcheson about the relevance of benevolence. Smith argues that neither agents nor spectators are much

194

EXPLAINING VALUE

influenced by benevolence, although the desire for sympathy is important. Smith agrees with Hutcheson and disagrees with Hume in holding that moral motivation is of a piece with what it is that leads spectators to approve or disapprove of various actions. What is important, according to Smith, is the taking of an impartial view. Smith works this theory out with a mass of detail which I cannot try to summarize. The book in which he works it out, his Theory of the Moral Sentiments, is one of the great works of moral philosophy. I conclude that the impartial spectator theory has an answer to the objection that it offers an overly aesthetic conception of morality, that it overemphasizes the point of view of the spectator over the point of view of the agent, and that it cannot account for the motivation to do what is right. This is not to say that I am convinced that this is the best theory. One alternative is to try to develop what might be called an ideal agent or ideal practical reasoner theory, as opposed to an ideal observer or impartial spectator theory. This alternative would place primary emphasis on moral reasons for acting, on the viewpoint of the agent. In one version of this theory, the agent has moral principles that the agent intends to follow. The agent's moral reasons typically derive from principles the agent accepts as a member of a group (Harman, 1977: ch. 9). I am not sure how to decide between the impartial spectator theory and the ideal reasoner theory. So let me conclude with a couple of further comments on the history of moral philosophy in the eighteenth century.

11.5.1 Postscript 1 There is an interesting irony in the way in which Hume's use of the term ‘sympathy’ leads Smith to his own very different theory, a theory that in my view is much better than Hume's at accounting for moral phenomenology. Smith's criticism of Hume's use of the term ‘sympathy’ is not a serious one. It is of no importance whatsoever whether the meaning that Hume gives to the term ‘sympathy’ is the ordinary one and it in no way damages Hume's view for him to acknowledge that his use is nonstandard. Hume can use the term however he wants. If he wants to use it in a special sense in order to develop his own view, there is nothing wrong with that. If someone really objects, the term can be replaced with another. Nothing in Hume's view depends on his having captured the ordinary meaning of the term ‘sympathy’. The irony is that taking Hume's term

MORAL AGENT AND IMPARTIAL SPECTATOR

195

seriously leads Smith to a more accurate account of morality. A purely verbal point yields a powerful substantive theory.

11.5.2 Postscript 2 Finally, it is perplexing that Adam Smith's ethics should be so relatively unread as compared with Hume's ethics when there is so much of value in Smith. What I have talked about here only scratches the surface. Why should Smith's ethics be so neglected? Is it that Hume also had a metaphysics and an epistemology and that Smith did not? Or is it that Smith was a more important economist than Hume? And why should that matter? I do not know.

12 Explaining Value I am concerned with values in the descriptive rather than in the normative sense. I am interested in theories that seek to explain one or another aspect of people's moral psychology. Why do people value what they value? Why do they have other moral reactions? What accounts for their feelings, their motivations to act morally, and their opinions about obligation, duty, rights, justice, and what people ought to do? I believe that philosophers can profitably join forces with social psychologists and other scientists. To some extent, that has already happened. Darley and Schultz (1990) ‘have found theoretical inspiration in philosophical analyses of morality or law’ by John Austin, H. L. A. Hart, and others. The writings of Piaget (1956), Kohlberg (1981), and Gilligan (1982) have influenced philosophers like Rawls (1971), contributors to Kittay and Meyers (1987), and Flanagan (1991). The journal Biology and Philosophy is one of several in which there is mutual discussion of morality among philosophers and scientists. One issue is whether philosophical speculation might be useful for social psychology, perhaps both in uncovering data of a sort not often considered in social psychology and in the explanations of the data that it offers. A good example of new data comes in philosophical discussion of the trolley problem. As a trolley speeds down the track, five people are spotted on the track ahead, with no way for them to get off the track in time. The brakes on the trolley fail, but it can be turned onto a side track where there is only one person. Most people believe it would be morally permissible to turn the trolley onto the side track. But in many other cases, most people think it would be seriously wrong to sacrifice one person to save five. For example, five patients are dying, but can be saved by a surgeon cutting up a sixth patient and distributing his organs to the others. Almost everyone believes it would not be morally permissible to cut up the sixth patient for this purpose. Again, most people believe that it would be morally impermissible to push a fat person onto the track in order to stop

EXPLAINING VALUE

197

the trolley, even though this would still be sacrificing one in order to save five. Several principles have been proposed to account for people's reactions to such cases, principles I will come back to later, including the Catholic principle of double effect, Foot's (1978) principle that negative duties are stricter than positive duties, and Thomson's deflection principle (1986: chs. 6 and 7). The philosophical literature (some collected in Fischer and Ravizza, 1992) discusses at length whether these ordinary distinctions are really justified and asks what really is the right or wrong thing to do in these and other cases discussed in the literature. I propose to ignore normative questions about what really is right or wrong in order to concentrate on finding explanations of people's reactions to these cases. Let me add that I do not mean to rule out explanations that appeal to the actual rightness or wrongness of various courses of action, although I will not be discussing such explanations in this essay. I mean only that I am concerned with explaining people's reactions. This is what I mean by ‘explaining values’. Explaining why something is of value is to be distinguished from explaining why people value something. There can be significant relations between these two kinds of explanation, including these: (1) explaining why something is of value might help to explain why people value it. (2) Given an impartial spectator theory of morality of the sort discussed in Essay 11, to explain why people value something may be to explain why it is of value. (3) Certain explanations of why people value what they value may undermine the conviction that the things valued are of value.

12.1 Method The enterprise of explaining values requires an adherence to somewhat stricter procedures than philosophers may be used to. First, it is important to indicate what data are to be explained, and it is important to establish that the data really are data. Data have to be objective. The data in the trolley problem might be that certain subjects have expressed the opinion that it is morally permissible to turn the trolley. It would be relevant to ask a group of subjects for their opinions on this and other questions to see how many say that it is okay to turn the trolley but not to cut up the sixth patient. It might turn out that in a class of fifty students, forty-five

198

EXPLAINING VALUE

thought it was okay to turn the trolley and five were unsure, forty-eight thought it was not okay to cut up the sixth patient and two were unsure. Here it is important to specify what it is that is to be explained. An explanation might be offered for all responses or, alternatively, only certain features of the responses might be accounted for. Opinions expressed about one or another version of the trolley problem can depend on the way questions are posed, how issues have been discussed up to that point, what other opinions have been expressed, whether opinions are solicited in private or in public, etc. So, reports of data should make explicit exactly how the data have been obtained, what questions were asked, and what other things were told to subjects. Second, when an explanation for certain data is proposed, the explanation should be as clear and explicit as possible. Of course, an explanation may be more or less ambitious. A relatively unambitious explanation might be of the sort that merely seeks to bring the data under a generalization like, ‘Subjects will almost always think that it is not permissible to sacrifice one person to save five others, unless the five are saved by deflecting a threat to them onto a smaller number of people.’ A more ambitious explanation would try to say something about the processes that lead subjects to react as they do—saying, for example, such things as, ‘Subjects tend to look at situations from the point of view of the people involved. . . . ’ In that case, it is important to indicate how exactly the process is envisioned. Third, it is necessary to see how the suggested explanation fits in with other plausible accounts, including general theories in psychology, sociology, economics, and biology. Is the proposed explanation plausible in the light of other views we have about people? Finally, it is necessary to consider competing explanations to determine whether either the given explanation or one of its competitors has more going for it than the others. If not, there is no reason to accept one rather than another explanation. Not all explanations compete. Some help to fill out others. An explanation of Jack's death as due to poisoning need not compete with an explanation of his death as due to lack of oxygen getting to the brain, if that's part of the way the poisoning might work. An evolutionary explanation of certain moral attitudes in terms of natural selection may fit together with a more psychological explanation in terms of sympathy, if the evolutionary explanation allows for the possibility that natural selection has brought about the desired moral attitudes by encouraging sympathy.

EXPLAINING VALUE

199

12.2 Hume on Chastity and Sexual Promiscuity In order to bring out some of the issues involved in explaining values, consider Hume's (1739: Bk. 3, Pt. 2, sect. 12) account of the different attitudes taken toward chastity and promiscuity in men and in women. Hume observes that it is normally thought to be worse for a woman to be sexually promiscuous than it is for a man to be sexually promiscuous. His explanation of the difference appeals to the fact that women rather than men bear children. Hume's explanation is roughly this. Children need to be cared for and people are more likely to care for children toward whom they feel love and affection. People are more likely to feel love and affection for children they believe to be their own than for other children. Now, because women rather than men bear children, it is more certain who the mother of a child is than who the father is. A key point is that a husband will tend to believe that his wife's children are his children to the extent that he believes his wife is chaste and not promiscuous. Therefore, permitting promiscuity in women would have bad consequences that would not be consequences of promiscuity in men. Hume says that this difference explains why we have different attitudes toward chastity and promiscuity in men and women. This explanation may seem straightforward enough, and even plausible. But how does it fit in with our tests? First, what exactly are the data to be explained? Do people accept the explicit general view that it is worse for women to be promiscuous than it is for men? We can test this by asking subjects whether they accept the general proposition. We may find that few people hold that explicit general view in America today. There have been some changes in attitude since Hume's day (especially since 1960). Whether this is one, I am not sure. Maybe in Hume's time people had the general view. But I don't think many do now. Nevertheless, I believe that further investigation will show that, whatever their explicit general views, people do react differently to men and to women believed to be promiscuous. We might investigate what subjects say in particular cases about the character of individuals described as promiscuous. Even if subjects do not accept as an explicit general claim that it shows worse character in a woman to be promiscuous than it does in a man, they may very well make more severe particular judgements

200

EXPLAINING VALUE

about women they believe to be promiscuous than about men they believe to be promiscuous. In other words, the data may show that subjects' judgements about people are affected by whether they take the people to be promiscuous, that subjects tend to rank promiscuous people lower than other people, and that subjects tend to give more negative weight to promiscuity in women than to promiscuity in men. I do not know whether anyone has actually done relevant experiments, but I believe that data of the suggested sort will be obtained. (My own relatively informal questioning of Princeton undergraduates strongly supports this conclusion.) Suppose we had such data. Let us compare Hume's explanation of that data with a competing explanation in terms of power relations. The idea behind the competing explanation is that men are taken to be in charge, so men are taken to have the right to act as they please; women are supposed to find men to serve. For a woman to act promiscuously is for her to act as if she had the rights of a man, and such behaviour is therefore treated as a kind of sedition. Hume's explanation involves two claims: (1) that a woman's having a reputation for promiscuity tends to have worse consequences than a man's having such a reputation, and (2) that this difference explains why we have different attitudes toward promiscuity in men and women. But how exactly does the explanation work? Is the idea that the condemnation of a particular woman's promiscuity is based on the inference that her promiscuity puts her children at risk? If so, Hume's hypothesis would predict that subjects will condemn promiscuity in women only where there is the possibility that there will be children, and will therefore not condemn promiscuity in women who are past the childbearing age, whereas the power-relations hypothesis predicts that this will not matter, since it is sedition whether or not children are involved. But Hume himself says that in fact people generally continue to condemn promiscuity in women past the childbearing age more than they condemn promiscuity in men of the same age. So, it may seem that Hume's own data favour a competing theory over his own. In fact, Hume does not intend his explanation to be such an intellectualistic explanation. That is, he is not proposing that different reactions to promiscuity in men and women derive from explicit conscious reasoning about the effects such promiscuity may have on children. Hume's actual explanation goes more like this (in my own words). In some cases, a man refuses to take care of children because of his doubts as

EXPLAINING VALUE

201

to whether he is the father, doubts that arise because of promiscuity on the part of the mother. Consequently, the children suffer and observers come to blame the mother for the child's suffering. Psychological generalization occurs, leading observers to condemn promiscuity in other women. Other people go along with this condemnation without necessarily understanding its basis. On the other hand, there are no cases in which a woman refuses to take care of children because she doubts whether she is the mother because of promiscuity on the part of the father! So, men do not get blamed in the way that women do. Why do observers blame the woman rather than the man with whom she was involved? One reason is that normally only she is identifiable as one of the people responsible for the birth of the child unwanted by her husband. Not knowing who the actual father of the child is, observers are not able to blame any particular identifiable man. Hume's explanation appeals to certain psychological principles. One is that, if something happens that is bad for someone and an action of an identifiable person can be seen to be responsible for the badness, then that can lead an observer to think badly of the person responsible. Second, if an observer thinks badly of a person because of something the person did, that can lead the observer to think badly of others who do similar things. (This leaves unexplained what counts as doing ‘similar things’; that is, it leaves unexplained why the application of the generalization is limited in this case to other women rather than people in general.) Third, there can be a herd effect. If some people condemn behaviour of a certain sort, other people are likely to think badly of it also. Here, then, is a brief development of Hume's explanation. A similar development is needed for the power-relations hypothesis, but I will not try to work out the details. It remains to be seen which hypothesis looks best, all things considered. The two hypotheses do seem to make different predictions for a situation in which power relations are reversed between men and women, so that women are socially and financially dominant over men. The power-relations hypothesis predicts that in such a situation it will be worse for a man to be thought promiscuous than for a woman to be thought promiscuous, whereas Hume's more biological explanation predicts the reverse. I do not know whether there is evidence that favours one or the other of these predictions. Of course, it is possible that both factors play a role in explaining current

202

EXPLAINING VALUE

attitudes. They might reinforce each other. It is even possible that Hume's more biological explanation somehow plays a role in the explanation of male dominance. We have been considering how Hume tries to explain a very specific and particular aspect of ordinary moral views. Let us now consider an attempt to explain a much larger part of ordinary morality.

12.3 Utilitarianism and Sympathetic Observers Utilitarianism holds that certain moral decisions rest entirely on the utility (or expected utility) of one or another option, where the utility of an option might be measured by the total net happiness that would result from that option, that is, by the amount of happiness the option would lead to, minus the amount of unhappiness that it would lead to. (Expected utility is measured by the sum of the utility of each potential consequence of an option multiplied by the probability that the option will lead to that consequence. I will suppress further mention of expected utility.) Versions of utilitarianism are usually discussed as normative views. I want to consider related explanatory hypotheses. For example, there is the somewhat implausible intellectualistic claim that people normally reach moral decisions by explicitly calculating the amount of utility that various options would involve in order to favour the option with maximum utility. A more plausible explanatory hypothesis would follow Hume's lead in his discussion of chastity and promiscuity, hypothesizing that moral reactions are sensitive to the extent to which pleasure is produced and pain is avoided, but allowing also for the influence of psychological generalization, the association of ideas, and the influence of custom. Why might someone be sensitive to utility? Hume appeals to what he calls the ‘sympathy’ that he thinks an observer will feel when considering people who might be affected by the options being considered. For Hume, sympathy is like the sympathetic vibration in a piano string when a note of similar frequency is sounding. When an observer becomes aware of another person's pleasure or pain, the observer tends to vibrate sympathetically, feeling pleasure at the other's pleasure and pain at the other's pain. According to Hume, the mere awareness of another's pain or pleasure is enough to set one vibrating in a similar way. Hume offers an impartial-sympathetic-spectator account of moral judgements. He hypothesizes that such judgements purport to represent

EXPLAINING VALUE

203

the reactions you have when you view matters impartially. The reactions you have when you adopt an impartial point of view derive from the sympathy you feel with people affected by a given option. In Hume's view, these reactions have a strong utilitarian component: you are happy with options that make others happy and diminish their unhappiness, and you are unhappy with options that do the opposite. But, as I have already indicated, Hume thinks that other factors will affect these reactions also: various associations, psychological generalization, habit, and custom (Hume, 1739: Bk. 2, Pt. 1, sect. 11, and Bk. 2, Pt. 2, sect. 7; 1748: Bk. 2, sect. 5). As noted in Essay 11, Smith (1976: Pt. I, sect. 1), who also accepts an impartial sympathetic spectator account of moral judgements, rejects Hume's account of sympathy. Smith observes that you might know very well that someone J is in pain, yet fail to sympathize with J if J's reaction is out of proportion to his or her injury. Hume cannot easily account for this aspect of sympathy, nor for the fact that the sympathy of others makes a sufferer feel better. Suppose J is unhappy, and a spectator K is aware of J's unhappiness, and vibrates sympathetically and so unhappily. If perception of K's sympathetic unhappiness leads J to vibrate sympathetically pathetically with K, the result should be even more unhappy vibrations. But in fact K's sympathy can make J feel better, contrary to what Hume's theory predicts. In Smith's view, feeling sympathy towards someone else is not just sympathetically vibrating with that person's experience. It involves a judgement of the appropriateness of the way the other person is reacting to his or her circumstances. To be sympathetic is to judge that the other person's reaction is appropriate. People want sympathy because they want approval. According to Smith, observers judge appropriateness by considering how they would react in a similar situation. The observer K imagines what it would be like to be in J's position and how he or she would react. If K envisions reacting as J does, K is sympathetic. If J is reacting more extremely than K envisions reacting, K will not be sympathetic and will disapprove of J. Observers favour muted reactions. An imagined injury, for example, does not hurt as much as a real injury. Wanting sympathy, agents are motivated towards muted reactions. So, Smith's impartial spectator theory tends more toward stoicism than utilitarianism. Both Hume and Smith stress the importance of custom and convention in moral thinking. For Hume, there is a utilitarian element to this importance, since it is often useful to have customs and conventions regarding

204

EXPLAINING VALUE

exactly what the rules of property are or what the rules of succession are. For Smith, customs are more directly connected to moral judgement. Given a customary way of acting, people will sympathize with behaviour that fits in with that way of acting, because, at least in the first instance, people sympathize with behaviour only when they imagine that they would behave in the same way in the same circumstances. If there is a custom, that is the way most people behave, so it is the way most people imagine themselves behaving, so it is the standard by which others are judged.

12.4 Moral Development Adam Smith offers an account of moral development that in some ways is quite like Freud's (1930: chs. 8–9). It goes something like this. How others react to what you do affects you. So you have an interest in anticipating how others will react to you. This requires trying to imagine how others see you. You will eventually acquire a habit of imagining this. This habit of viewing yourself from the viewpoint of others amounts to looking at your situation from a more impartial or moral perspective. Having such a habit is like having an internal critic. Your internal critic tends to get idealized. Your actions become influenced by a desire to please the internalized critic. (For a contemporary example, compare the way in which athletes talk to themselves in developing an internalized coach.) As a result you become capable of guilt and you acquire the ability to make moral judgements. Such an account of moral development might explain those elements of morality that are explainable by an impartial spectator theory. It might also help to explain aspects of moral motivation. According to Smith, moral motivation begins with a desire to avoid the displeasure of others. It becomes a desire to avoid the displeasure of your internalized critic. You are that critic, so it becomes a desire to avoid your own displeasure when you view your actions from a moral or impartial point of view. In Smith's theory, therefore, viewing things from an impartial standpoint has two aspects. It provides a way of finding out what is right, fair, just, etc. It also is a way of acquiring moral motivation to do what is right and promote what is fair and just. There are other accounts of moral development. Piaget (1956) hypothesizes general stages of cognitive development, including moral development; Kohlberg (1981) and Gilligan (1982) make proposals within this

EXPLAINING VALUE

205

framework; and there has been considerable philosophical interest in these ideas, although the hypothesis of unified cognitive stages has not stood up well to experimental testing (Carey, 1990). Darley and Schultz (1990) point out that the principles followed when a child thinks about a situation morally may not be the same as the principles expressed when the child defends one or another judgement. The principles the child actually uses in his or her own reasoning may not be easily accessible to him or her. The approach pursued by Piaget, Kohlberg, and Gilligan looks at the principles used to rationalize or defend decisions. But when other investigators have studied the principles children actually follow, they find children using fairly sophisticated moral reasoning at relatively young ages. Darley and Schultz also point out that since Kohlberg discusses fairly complex situations, his results may have little to do with children's moral reasoning and more to do with their ability to report on the reasoning they do. ‘Social constructionist’ theories of learning are concerned with how children learn to respond to challenges by adults and peers. There is current research, for example, about the extent to which individual differences are traceable to differences in parents' moral rules. Darley and Schultz cite the following example: people differ in ‘how they integrate the intent of a harm-doer with the degree of harm caused to arrive at a punishment’, and ‘in a majority of cases children's integration schemes matched those of their mothers’ (1990: 544).

12.5 Evolutionary Psychology Moral evolutionary psychology appeals to natural selection to explain aspects of morality. Natural selection is possible when three conditions are met: (1) characteristics have to be heritable; (2) differences in characteristics have to make a difference to reproduction; and (3) there has to be some random variation in characteristics despite heritability. Parents' concern to take care of their children could have a source in natural selection. Such concern is likely to make a difference in whether children will survive to be able to have their own children and so in whether parents' genes will be transmitted to later generations. Children whose parents want to take care of them are more likely to survive than children whose parents do not want to take care of them. If there is some random variation in parental concern of this sort, and if it is heritable,

206

EXPLAINING VALUE

then natural selection will tend to favour the existence of people who inherit concern for their children over other people who do not. Siblings' concern for each other might be similarly based, since there is the same genetic connection between siblings as between parent and child. Any particular gene is acquired either from father or mother (barring mutation). Thus, there is something like a fifty per cent connection between mother and child, and the same fifty per cent connection between two siblings (Hamilton, 1964). Exactly how this works depends on exactly what sort of thing is heritable. There might be an inborn tendency to be more altruistic towards others the more they resembled oneself. Or people might inherit a disposition to be altruistic towards others unless there was some reason to think that the others would not reciprocate (Trivers, 1971). Other aspects of morality possibly explained via evolutionary psychology might include incest prohibitions, a willingness to abide by rules, and even a sense of justice. How are such potential explanations to be evaluated? First, we need to be sure about the data. Second, we need to have a coherent explanation and we have to make sure we understand just what the explanation involves. Third, we need to consider what competing explanations there might be of the data. Fourth, we need to try to assess whether any of the competing explanations is sufficiently better than the others to be acceptable. This may involve getting more data. Consider the following (presumed) data about altruism. (1) Mothers have special concern for their children that leads them to sacrifice their own interests for those of the children. (2) Fathers do too, although perhaps not to the same extent. (3) People seem more concerned in general for members of their family than for others. (4) More generally, people seem more concerned for their friends and acquaintances than for others—and similarly for people from one's own school, or country. An explanation in terms of natural selection might appeal to a mixture of kin selection and reciprocal altruism. This explanation assumes, for example, (a) that whether a mother is specially concerned for her child has a biological basis (and similarly for other altruistic concerns); (b) that this is a heritable feature; and (c) that it arose in a context where it was adaptive. There are various alternative explanations. (1) There might be group selection in favour of groups containing altruists (Sober and Wilson, 1998). Although altruists may not do as well as nonaltruists within any given group, groups with larger percentages of altruists in them are more

EXPLAINING VALUE

207

likely to reproduce themselves than groups with smaller percentages of altruists in them. (2) Altruism arises from conditioning (Brandt, 1976). Unhappiness leads a child to cry. Classical conditioning leads the child to associate the sound of crying with being unhappy. So, in particular, hearing other children cry makes the child unhappy, so he or she is motivated to do things that will stop others from crying. (3) There is a straightforward egoistic advantage to reciprocal altruism. Tit for tat is an excellent strategy in repeated Prisoner's Dilemmas (Axelrod, 1984). This is something children can learn. (4) One can have selfish reasons to be concerned that other people's lives go well, if the other people are potential benefactors. Such concern develops over a period of time into direct concern for others, just as a miser tends to come to love money for its own sake. (5) There is random variation with respect to altruism: some people have more of it, others have less. The advantages and disadvantages tend to cancel each other out. At the moment, as far as I know, none of these explanations is clearly preferable to the others.

12.6 Posner on Wealth Maximization Where utilitarianism attempts to account for aspects of our moral judgements as resting on considerations of general happiness and unhappiness, Posner (1981) defends the hypothesis that many aspects of our moral and legal views reflect judgements about whether options will increase or decrease society's net wealth. Wealth is measured economically; it is connected with market price, although it is not to be crudely identified with market price. The value of something, in Posner's account, is determined by the price for which the owner would be willing to sell it, apart from bargaining considerations, and supposing that the owner really is the rightful owner. Suppose that A sells something to B. Then both A and B will normally gain in wealth. A values the money received more than the object sold; B values the object received more than the money given up. The net wealth of A and B increases by the difference between the minimum for which A was willing to sell the object and the higher minimum for which B is now willing to sell it. Posner argues that the goal of maximizing wealth gives a better account of our judgements of social practices than the goal of maximizing happiness. In particular, it gives a better account of our views about the sort of

208

EXPLAINING VALUE

right that amounts to a veto power. For example, if A owns something, then A has a right against other people using it. Here is a sketch of Posner's analysis. (1) Suppose A wants to engage in an activity that B does not want A to engage in. Suppose B has veto power. And suppose that transaction costs between A and B are insignificant. If the activity is worth enough to A, A can reach a deal with B and pay B to allow A to engage in that activity. (2) Alternatively, if nothing prevents A from proceeding, and it is worth a lot to B not to have A proceed, B can pay A not to proceed. So, whether or not B gets a veto power, the more valuable outcome will still ensue, if transaction costs can be ignored. Posner argues that the only reason to assign veto power or rights one way rather than another has to do with the costs of transactions (a kind of friction). Initial rights should be assigned to the party who cares more by this test, that is, the one who would otherwise pay to get his or her way. Theft is ruled out. Since a thief T is not normally willing to buy what T steals, T acquires something that is worth less to T than it is to the owner. Theft diminishes wealth, even when it may maximize utility! Indeed, Posner argues that the usual Calvinist virtues (keeping promises, telling the truth, etc.) are supported by the wealth-maximization principle, because they facilitate transactions. A serious study of Posner's hypothesis must consider the standard issues: (1) what are the data? Are we sure about them? (2) How does the explanation go in detail? How does it fit in with psychology, biology, and what is known about social evolution? (3) What competing explanations are there? Utilitarianism is one, of course. (4) How do these competing explanations compare, and is any sufficiently better than the others that it can be believed? With respect to (4), Posner argues that standard worries about utilitarianism are not problems for his hypothesis. (a) Utility monsters are people who get so much utility from resources that utilitarianism implies that all resources should go to them (Nozick, 1974: 41). Wealth maximization has no such implication. (b) Utilitarianism implies that the happiness of animals should be counted as much as the happiness of people. Wealth maximization does not imply this. (c) Utilitarianism has a problem with population size. Classical utilitarianism implies that the population should be increased to the point at which any further increase fails to add

EXPLAINING VALUE

209

to the total happiness. If instead we are to maximize average utility, then utilitarianism seems to imply we should kill off (or allow to die) people whose happiness is sufficiently below average that their existence drags the average down. Again, wealth maximization has neither implication. (However, as discussed in section 8.6 in Essay 8, above, utilitarianism can avoid this problem if it takes individual happiness to have contributory value rather than simple intrinsic value.) With respect to (2)—how does the explanation fit in with other things we believe?—Dworkin (1985: 240) objects that ‘it is unclear why social wealth is a worthy goal’. Posner agrees that wealth maximization is not a basic value but gives the following hypothetical case in reply. Suppose we discover the following pattern in the decisions of courts of appeal in the United States: the number of cases that are reversed is approximately equal to the square of the number of cases that are affirmed during any sufficiently long time period (e.g. one year or more). This would be an interesting result with real predictive and in some sense explanatory value, even though we did not know why the pattern held. Similarly, when we find that common-law decisions in courts fall under the pattern suggested by wealth maximization, that is an interesting result with real explanatory power, even if we do not have a deeper account of this pattern.

12.7 Double Effect Posner's proposal strikes many people as bizarre, because they have trouble understanding how wealth maximization could be a relevant principle. Here utilitarianism would seem to have an edge, because utilitarianism makes sense if explained, for example, by appeal to Humean sympathy. But utilitarianism has its problems. In addition to the difficulties Posner mentions, utilitarianism notoriously has trouble accounting for ordinary judgements about the surgeon case in a way that is also compatible with ordinary judgements about the trolley case, and Posner's principle of wealth maximization would seem to have the same trouble. Since sacrificing one to save five would normally maximize both expected utility and social wealth, both hypotheses would seem to treat the surgeon and trolley cases in the same way, which conflicts with the judgements most people make about these cases. Principles have been proposed to account for the different ways people think about these cases, principles like double effect, the relative strictness

210

EXPLAINING VALUE

of negative as compared with positive duties, and Thomson's deflection principle. It is interesting that these principles are often defended in the way in which Posner defends wealth maximization, by indicating how the principles fit cases and without any attempt to explain why ordinary judgements might be sensitive to such principles. According to the principle of double effect, while it is an objection to a proposed course of action that it will harm someone, the objection is considerably more weighty if the harm is intended, either as one's aim or as a means to achieving one's end, than if the harm is merely foreseen as a side effect of one's action. The difference between the surgeon case and the trolley case is that the harm to the patient, whose organs are to be distributed to the other five, is an intended means of saving those five, whereas the harm to the person on the side track that results from switching the trolley is not a means of saving the five on the main track but merely a side effect of action taken to save them. This proposed explanation of the difference between these cases is like Posner's; it brings the cases under a generalization without indicating why that generalization holds. Various hypotheses are possible as to why double effect is as successful as it seems to be in accounting for cases. One clearly false hypothesis would be that the principle of double effect is consciously adopted by people and used in reaching moral conclusions. That hypothesis is ruled out by the fact that few people are aware of the principle, and those who are aware of the principle dispute whether the principle is correct and what its correct formulation might be. A weaker hypothesis is that the principle is tacitly accepted in something like the way in which, say, principles of grammar are tacitly accepted, or the principles by which one does simple arithmetic (Chomsky, 1988; Van Lehn, 1989). These principles might represent what is often called ‘procedural knowledge’ rather than ‘declarative knowledge’. This hypothesis would raise the question why one has such tacit principles. They might somehow be innate, in which case we might turn to evolutionary psychology for an explanation. They might be learned, in which case we might turn to sociology and the theory of convention. Or there might be some other explanation of the sort Hume appeals to in his hypothesis concerning attitudes about chastity and promiscuity. Similar possibilities arise for Posner's hypothesis. It is clear that people do not consciously reach moral conclusions using Posner's principle of wealth maximization, so if the people accept the principle, they do so tacitly. Questions arise about how they come to accept it. There might be a Humean type of explanation. For example, when something happens that

EXPLAINING VALUE

211

decreases wealth, people may notice and disapprove of that event and, if this sort of thing happens enough, their disapproval could generalize to other events of a similar sort.

12.8 Positive and Negative Duties Foot (1978) rejects the double-effect hypothesis in favour of a distinction between positive duties, such as the duty to help other people, and negative duties, such as the duty not to harm other people. Foot's hypothesis is that we take negative duties to be stricter than positive duties. If the surgeon were to distribute the organs of one person to five others, he or she would violate a negative duty not to harm the one person in order to uphold a positive duty to help the other five, and the relative strength of the negative duty rules this out. On the other hand, the trolley driver would violate a negative duty not to harm people, no matter what he or she does, so in his case the trolley driver is allowed to minimize the damage. (We will not be concerned with further questions about how much stricter negative duties are supposed to be than positive duties. We will not consider, for example, whether in this or any other view the surgeon could cut up one person to save millions of other people.) Foot's hypothesis and double effect make different predictions concerning certain cases. Foot considers a surgeon who can save five patients by manufacturing a special medicine using a serum machine that produces fumes that will kill a sixth patient in the room next door, where neither the machine nor the sixth patient can be moved. Double effect would not keep the surgeon from using the machine, since the harm to the sixth patient is not part of the surgeon's means of saving the other five patients but is merely a side effect. On the other hand, Foot's hypothesis implies that the surgeon cannot use the machine to save the five, because that would be a case of violating the negative duty not to harm people simply in order to avoid violating the positive duty to save people. In other words, Foot's hypothesis takes the serum-machine case to be like the original surgeon case, whereas double effect takes the serum-machine case to be like the original trolley case. Remember that we are interpreting these hypotheses as offering explanations of people's ordinary moral judgements. Controlled experiments are needed to determine how people actually react to the serum-machine case. My own preliminary investigations of student reactions in classes suggests that most people do react as Foot's hypothesis predicts rather

212

EXPLAINING VALUE

than as double effect predicts. (But I find that student judgements about cases do not always coincide with published claims about the cases.) Again this leaves open the question why there should be this difference in strength. Is it that people consciously believe that negative duties are stronger than positive duties? That seems wrong. Students in my classes do not seem to have had a view about this before the issue explicitly comes up. In any event, it remains unclear what it is about the duties involved that makes people attach more weight to negative duties (if indeed they do).

12.9 Sociological Explanations of Aspects of Morality Some aspects of morality are conventional—for example, what counts as a binding agreement, what the detailed principles of property are, etc. Different sorts of explanation are possible for conventional aspects of morality. For example, specific historical events may be involved. Or something about the ‘logic’ of social interaction may play a role—or class conflict, etc. Even things that are not conventional (because they are true of all moralities) might receive a sociological explanation of this sort. A sociological explanation appeals to certain general sociological principles, principles about how groups develop, function, and evolve. Sociological explanations tend to have a logic like natural selection. They do not require that people are acting in concert with the explicit goal of reaching a particular result. Many economic explanations are of this sort—for example, explanations of why there are lines at gas stations during some oil shortages but not others, depending on whether price controls are in place. Many evolutionary explanations have sociological competitors. For example, the facts to be explained by kin selection might alternatively be explained by selection for better social conventions, if groups in which parents take care of their young have done better than other groups in the struggle for social survival. Sociological selection does not yield a biological bias toward relatives, only a conventional bias. (And situations may change. Groups may emerge that do better by having children brought up in other ways.) It is unclear what sort of evidence might be relevant to deciding

EXPLAINING VALUE

213

between biological and social explanations. It is also unclear that one or the other explanation is preferable on grounds of parsimony. Essay 1 above suggests a possible sociological explanation for the pattern that Foot claims to find, namely, that normally killing is considered worse than letting someone die, that the negative duty of not harming others is normally treated as considerably stronger than the positive duty of helping others. My hypothesis is that morality is a compromise between rich and poor. Everyone benefits from the negative prohibition against harm, but the poor are the main beneficiaries of any general positive requirement to help others. So, we expect a strong negative prohibition against harm and a weaker positive requirement to help those in need. An explanation of this sort relies on something like Hume's (1739: Bk. 3, Pt. 2, sect. 2) account of implicit convention: people adjust their behaviour to conform to the behaviour of others; there is pressure to conform; things work out better for each person involved if he or she conforms. What explanations compete with the sociological account? Is there a competing sociobiological explanation? There is at least one: this is random drift! The sociological explanation is a ‘better explanation’ in some sense. It says the distinction is ‘no accident’. But sometimes it is more likely that something is due to accident than that it is ‘no accident’ (Sober, 1990).

12.10 History Historical explanations seem clearly appropriate where there are aspects of law or morality that differ from one society to another. Various aspects of the United States Constitution for which historical explanation is appropriate would include the bicameral legislature; the division between the presidency and Congress, as compared with parliamentary systems in which the prime minister is chosen by the legislature; the specific terms (four years for president, six for the Senate, two for the House); and certain rights, recognized in the Bill of Rights, especially the right to bear arms, which is not to be found in other constitutions. It is interesting to consider whether there are aspects of contemporary morality, as opposed to law, that have historical explanations of this sort. Of course, people's beliefs have changed in various ways over the last two centuries, and we can try to explain those changes—for example, views about slavery, the relations between the sexes, and so on. But in these cases, at least, it is plausible that the changes are due to a general appreciation

214

EXPLAINING VALUE

of certain disadvantaged people and their situation. Are there other cases in which aspects of contemporary views seem more ‘accidental’? To mention one possible case, it is sometimes argued that the usual distinction in the United States between acceptable recreational drugs and unacceptable recreational drugs is to be explained by appeal to historical accident. The upper classes used alcohol and did not use marijuana. Might similar ‘accidental’ events be responsible for some of the complexities in present-day opinions in the United States about freedom of speech and freedom of religion? Consider the legal situation. The First Amendment to the U.S. Constitution reads: Congress shall make no law respecting an establishment of religion, or prohibiting the free exercise thereof; or abridging the freedom of speech, or of the press; or the right of the people peaceably to assemble, and to petition the Government for a redress of grievances. This seemingly strong statement is normally deemed compatible with various restrictions on speech. There are currently laws against libel and slander, false medical claims and other false advertising, invasions of privacy, revealing classified information, and child pornography. On the other hand, under current law there are limits to the extent that public figures can sue for libel; racial insults are constitutionally protected; and most pornography is constitutionally protected except where it involves children. The laws in other countries differ from those in the United States especially with respect to pornography and libel. It is possible that the best way to understand the current legal situation in the United States (and therefore a certain amount of moral opinion) is historically, explaining just how certain events led to the current legal situation. It is interesting to try to develop a Herculean rationale for current law (Dworkin, 1986), but that is probably not the best way to explain current law.

12.11 Deection Thomson's deflection principle (interpreted as a hypothesis about ordinary moral judgements) competes with double effect and with Foot's claim that negative duties are stricter than positive duties (Thomson, 1986: chs. 6 and 7). The deflection principle assumes that people normally will judge that an agent may not bring harm to one person in order to save several others, but allows an exception if the harm threatening several

EXPLAINING VALUE

215

others can be deflected onto a smaller number of people; in that case, people may think it permissible to deflect the harm. But people will not normally approve of initiating a new process that will harm someone in order to prevent harm to others. The deflection principle clearly makes the correct distinction between the original trolley case and the original surgeon case. It also agrees with Foot's hypothesis as regards the serum machine that produces noxious gases that will kill a patient who cannot be moved. But it disagrees with Foot's hypothesis about one of Thomson's versions of the trolley case, in which the trolley is turned by a bystander who happens to be near a switch in the track, rather than by the original driver. If the bystander B does not turn the trolley, B does not kill the five people in the main track, and so B does not violate a negative duty not to kill them. Instead, B merely refrains from saving them and so violates a positive duty. If B turns the trolley, on the other hand, B does violate a negative duty not to kill someone merely for the sake of the positive duty to save those on the track ahead. So, Foot's hypothesis classifies this case with the original surgeon case. Thomson's deflection hypothesis classifies it with the original trolley case, since in both trolley cases a threat to five people is deflected to one person. The two hypotheses therefore make different predictions about people's judgements about the bystander-at-the-switch version of the trolley case. A poll of students at Princeton suggests that the prediction made by Foot's hypothesis is falsified and the one made by Thomson's deflection principle is correct. But it would be useful to have data more rigorously collected. If Thomson's deflection principle accounts well for other similar cases, it captures a pattern in our moral judgements, a pattern that may seem as puzzling as Posner's wealth-maximization pattern. Why exactly the deflection pattern should be found remains an interesting and relatively unexplored question.

12.12 Structures Almost all moral principles seem to have exceptions—for example, the usual principles about not lying, not stealing, not harming others. How are we to account for such complications? (Nozick, 1968). Principles like ‘Do not lie’ and ‘Do not steal’ are sometimes called ‘prima facie principles’, and the corresponding duties not to lie and steal are sometimes called ‘prima facie duties’. These are like principles that

216

EXPLAINING VALUE

have elsewhere been called ‘default principles’, which are studied as part of the general subject of nonmonotonic implication (Ginsberg, 1987). Several principles may apply to a given situation, giving conflicting advice about what to do. In ‘production systems’ developed in artificial intelligence, conflict is resolved, for example, by preferring a more specific rule to a more general rule. A different idea is to take certain specific examples as paradigmatic and then choose the solution for the present case that fits the paradigm closest to the present case. This is the method used in constraint-satisfaction systems, including neural networks. (Is this what is meant by ‘existential ethics’?) A somewhat different and more traditional idea in moral philosophy is to assign right-making and wrong-making features to an act based on certain principles and then try to balance these against each other. One difficulty with this approach is arriving at relevant weights to assign. Furthermore, as Nozick observes, features that are relevant to what to do are not always simply pro or con considerations. The obligation to keep a promise can be eliminated if it becomes impossible to keep it, if there is no point in keeping it, if the promise was extracted by force or by fraud, etc. It would be useful to be able to reduce moral reasoning to a more general theory of practical reasoning, but this has not yet been accomplished.

12.13 Final Remarks Moral philosophers might join forces with social psychologists in attempting to explain people's values. Classical British moral philosophers like Hume and Smith present accounts of moral attitudes that are easily interpreted as social psychology. Hume's account of different attitudes toward chastity in men and in women is one example. Hume's and Smith's contrasting discussions of sympathy are two others. Recent philosophical discussions of so-called ‘trolley problems’ can also be understood as concerned with explaining people's actual moral attitudes. These discussions benefit through the methodological discipline of having to be clearer about objective data. The discussions are also useful in suggesting empirical hypotheses to account for certain data about people's moral attitudes.

13 Moral Philosophy and Linguistics 13.1 Introduction Morality is in many ways like language. Both admit of great apparent variation. Both appear at first to be group phenomena but are probably best studied as instantiated in particular individuals. Both are acquired without explicit instruction through appropriate experience of the world. Such similarities between morality and language suggest there may be other similarities. Perhaps, as Rawls (1971: 47; also Mikhail, 1996) has suggested, morality can be studied in the way that language is studied in linguistics. In fact, linguistics appears to uncover a ‘universal grammar’ behind the surface diversity between languages. Might moral philosophers discover that, behind the surface diversity of moral views and customs, there is something like a universal grammar of morality? In what follows I begin with a brief description of the extent of apparent moral diversity. I then sketch how linguists are able to accommodate linguistic diversity within universal grammar. Finally, I consider how there might be an analogous study of universal morality.

13.2 Apparent Moral Diversity There are two kinds of apparent moral diversity, societal and individual.

13.2.1 Societal Differences Apparent differences in the moralities associated with different societies range from trivial differences in etiquette (burping after dinner being polite in some places, impolite in others), to differences concerning cannibalism, economic systems, forms of marriage, and caste systems. It might be thought that there is a universally recognized central core of morality involving at least prohibitions against killing and harming

218

EXPLAINING VALUE

others, against stealing, and against lying to others. But whether this thought is correct would seem to depend on what counts as acceptance of the same principle in different societies. Consider theft. Societies that do not recognize private property cannot have rules against stealing, because without property, there is no such thing as theft and therefore there are no rules against theft. On the other hand, it is trivial to say that all societies recognizing private property have rules against stealing, because having rules against stealing is a necessary condition of recognizing private property. Consider rules against murder and harming others. Typically, a morality does not prohibit all killing or harm. For example, in many societies a master is thought to have an absolute right with respect to slaves, including arbitrarily beating and killing them. In many societies there are no limitations on what a husband can do to his wife or on what a father can do to his young children. Furthermore, where moral prohibitions on harming and killing do exist, they typically apply only within some privileged group and not with respect to outsiders. Needless to say, different moralities privilege different groups in this way. The morality of group A says it is wrong to harm members of group A but allows harm to members of group B. The morality of group B says the reverse: it is wrong to harm members of group B but acceptable to harm members of group A. Prohibitions on lying and cheating are similarly often limited to the local group and not to outsiders. Someone who can successfully cheat outsiders may be treated as an admirable person. Similarly for someone who is able to harm and kill outsiders. It may be true that rules against killing, harm, lying, and cheating occur in all societies. But there is a sense in which these are not exactly the same rules, since the protected group changes from one society to another.

13.2.2 Individual Differences An even more significant aspect of apparent moral diversity is that it occurs not just between societies but also within societies, and in a way that often leads to seemingly intractable moral disagreements that rest on irreconcilable differences in basic values and not just on differences in opinion about the nonmoral facts. Moral vegetarians, who believe that it is wrong to raise animals for food, exist in the same society (even the same family) as nonvegetarians. A disagreement between moral vegetarians and nonvegetarians can seem to survive full discussion and agreement on the

MORAL PHILOSOPHY AND LINGUISTICS

219

facts and often appears to depend on a basic disagreement about the significance assigned to animals as compared with humans. In a similar way, disagreements about the moral acceptability of abortion or euthanasia seem to survive extensive discussion and awareness of all relevant information about abortion. Such disagreements appear to depend on basic differences concerning the degree to which human life has intrinsic value as compared with the degree to which those things that life makes possible are of intrinsic value, things such as pleasurable experience and fulfilling activity.

13.3 Language Before comparing the study of morality to the study of language, I need to say something about language and linguistics.

13.3.1 Linguistic Variation At least on the surface, linguistic diversity is even greater than moral diversity. Languages differ in vocabulary and in what sounds are used to form vocabulary. Where the English say ‘white’ the French say ‘blanc’, which sounds like the English word ‘black’. Languages also differ in syntax or word order. The English ‘white hat’, in which the adjective precedes the noun, contrasts with the French ‘chapeau blanc’, in which the adjective follows the noun. Languages also differ in whether thematic relations like subject and object are normally revealed through word order, as in English, or through case, as in Latin. There are many subtler differences between languages as well.

13.3.2 Language, Dialect, Idiolect It might be supposed that a language is always the language of a group and is based on linguistic conventions of that group, but this seemingly plausible idea is hard to reconcile with dialectical differences within what are normally called languages. Dialects blend into one another in ways that do not fit national or geographical boundaries. Even for speakers of what we normally consider the ‘same language’, it is unlikely that any two people speak exactly the same language. Your English differs somewhat from mine; her French differs from his. There will be some difference in vocabulary, pronunciation, or grammar. The

220

EXPLAINING VALUE

dialect of a particular person that is in certain respects unique to that person is that person's ‘idiolect’. Different people speak at least slightly different idiolects. The serious study of language therefore takes an idiolect to be the unit of language. The more popular, nonscientific notion of a language, like French or German or Chinese, is a political notion. ‘A language is a dialect with an army and a navy.’ Even the linguist's notion of an idiolect is an abstraction. An individual may make use of several different idiolects, speaking one way at home and another way at work, for example. So-called bilinguals are a relatively extreme example.

13.3.3 Competence, Performance, and I-Grammar An individual's linguistic performance depends in part on his or her linguistic competence, perhaps as specified by an internal grammar, or I-grammar, but also depends in part on non-linguistic factors, including limited memory, nonlinguistic goals and beliefs, functioning vocal tract and hearing, parsing strategies, etc. Although linguistic intuitions are the primary source of evidence for linguistic theories, such intuitions are influenced by more than linguistic competence. Sentences that seem ill formed may be grammatical but hard to process. Possible examples are (A) ‘The horse raced past the barn fell’ as a variant of ‘The horse that was raced past the barn fell’. And (B) ‘The malt that the rat that the cat that the dog bothered chased ate lay in the house that Jack built’. (‘The malt that the rat ate lay in the house that Jack built’ is fine. ‘The malt that the rat that the cat chased ate lay in the house that Jack built’ is less good. Sentence (B) is worse. It seems to be a matter of degree.) On the other hand, sentences that seem well formed to the person may actually be ungrammatical in his or her idiolect. One possible example is ‘The not unhappy student finished his term paper’. ‘The’ + ‘not’ + adjective is not usually okay. Compare, ‘The not happy student finished his paper.’ The evidence of linguistic intuition bears only indirectly on linguistic theory.

13.3.4 Language Acquisition and Universal Grammar All normal children acquire language. The only exceptions are children with specific brain deficits and children brought up in terribly isolated conditions.

MORAL PHILOSOPHY AND LINGUISTICS

221

How do children first acquire language? It may seem that children are normally taught language by their parents. But explicit instruction in language is not necessary for language learning and may not be very useful. Children who interact with people speaking a local dialect acquire (a version of) that dialect, whether or not their parents ever do. Important aspects of language are not learned in any explicit way, as in the following somewhat oversimplified example. In dialects of English, questions can be formed by moving a question word to the beginning of the sentence and following it with the first auxiliary verb, as in ‘Who is Alfred asking Bob to talk to?’ which comes from ‘Alfred is asking Bob to talk to who?’ But this rule cannot apply if the question word occurs in grammatical conjunction with another noun phrase, as in ‘Alfred is asking Bob to talk to who and Mary?’ because the result, ‘Who is Alfred asking Bob to talk to and Mary?’ is not a well-formed question. Now it is very difficult to see how a child might learn such a constraint (the ‘coordinate structure constraint’) given the available evidence. Indeed, the coordinate structure constraint was unknown to linguists before 1950. But, if children do not learn the constraint, that can only be because it is somehow built into language acquisition ahead of time. So it should apply no matter what language is learned, which is to say it must be a universal constraint, one that holds in every language that is acquired as a first language in the normal way. Reasoning in this manner, linguists use facts about a single dialect of English to make predictions about all dialects that are acquired in the normal way as first languages. And, indeed, the prediction about the constraint on question-word movement has been confirmed through the study of a great variety of languages, as have other similar predictions. Considerations of learnability lead linguists to universal grammar. Notice that such universal principles of language must hold not only for already existing languages but for newly created natural languages as well. For example, deaf children of hearing parents sometimes invent their own new sign language. Such an invented sign language has all the complexity of spoken natural languages and admits of the same linguistic universals, despite being newly invented. Furthermore, when people who speak different languages come together, they may for business purposes invent what linguists call a ‘pidgin’ mixture of their languages. A pidgin language does not satisfy the principles of universal grammar. That is not a refutation of universal grammar, since a pidgin language is not a language that has been learned by children as a first natural language. However, consider children born to speakers of a

222

EXPLAINING VALUE

such a pidgin language. The children acquire what linguists call a ‘creole’, corresponding to the pidgin spoken by their parents, except that the creole language satisfies the principles of universal grammar. Pidgins are not natural languages in the relevant sense, even though they arise in a natural way when people speaking a variety of languages need to do business together. Pidgins are not ‘natural languages’ because they cannot be acquired in the natural way in which children acquire a first language. I said that my discussion of the constraint on question-word movement is oversimplified. A rule that is not learned might be built into universal grammar merely as a default case, so that there might be exceptions to it. If a child had evidence that the rule did not hold in the local dialect, the child would learn a dialect that is not subject to the rule, even though in other circumstances a child would learn a language that is subject to the rule without having to learn the rule. Further refinements of this approach have led to a ‘principles and parameters’ theory of language, according to which languages are basically the same except for vocabulary and a few parameters whose values are picked up during early language acquisition, parameters that determine, for example, whether the head of a phrase goes at the beginning or end of the phrase. The task of universal grammar is then to specify the relevant principles and parameters.

13.4 Linguistics and Morality 13.4.1 Moral Idiolects, I-Morality Now, as with language, it might at first be supposed that a morality is always the morality of a culture or group of people. But, just as different people in the culture or group have different idiolects, they also have at least somewhat different moral understandings, and the moral understandings of different people vary continuously in ways that do not respect national, cultural, or group distinctions. Two people in the same family might have slightly different (or even very different) moralities. We must therefore concentrate on the characterization of an individual person's morality—that person's moral idiolect, as it were. As with I-grammars, individuals often have available several I-moralities, sometimes one applying at home, another at work; one with old school mates, another with new friends. There may even be moral

MORAL PHILOSOPHY AND LINGUISTICS

223

‘bimorals’ who possess two radically different I-moralities—the moral version of bilinguals. Tamar Gendler argues (personal communication) that comparison of bilingualism and bimoralism points to a disanalogy between language and morality. A person with two I-moralities suffers from a divided moral mind in a way for which there is no parallel for linguistic bilinguals. A person with two I-moralities has conflicting opinions about right and wrong. A bilingual need not thereby have conflicting views about anything. I am not sure what to think about Gendler's observation. In any event, as with language, we must recognize that a person's actions and conscious judgements about right and wrong depend not only on that person's moral competence or I-morality but also on many other factors, including various beliefs, goals, limits of information, memory, and processing strategies. Although moral intuitions are the primary source of evidence for moral theories, such intuitions can be influenced by more than just the individual's moral competence (Horowitz, 1998; Kamm, 1998). The evidence of moral intuition bears only indirectly on the correct characterization of an individual's I-morality.

13.4.2 Moral Education Despite rather large apparent differences in moral idiolects, all normal people acquire a moral sense. The only exceptions are people with certain sorts of brain damage and people whose upbringing is unusually deprived. Although many parents explicitly teach morality to their children, such teaching is probably not needed and may not be particularly effective. Just as children acquire a version of a local dialect simply through being exposed to others who already speak it, without any need for explicit instruction, it is likely that this is true for morality also. But, if morality is universal in the sense that all normal people develop it, how much of the content of morality is universal?

13.4.3 Universals of Language and Morality Despite obvious surface differences between languages (English and Japanese, for example), there are universals of language, characteristics that all languages share. There is universal grammar. Individual languages

224

EXPLAINING VALUE

are actually very similar except for vocabulary and a few choices about syntax. Is the same true of morality? Despite large obvious surface differences in moralities, are there underlying universal features, so that differences in moralities depend on a small number of parameters? If so, what are the universals and how can we find them? Where should we look for them?

13.4.4 Possible Moral Universals The I-grammars of all languages are based on categories like phoneme, morpheme, phrase, clause, etc., with nouns, verbs, prepositions, adjectives, etc. Similarly, we might expect that all I-moralities make use of basic notions like right, wrong, permitted, not permitted, good, bad, fair, unfair, etc. Questions might arise about some notions: do all I-moralities contain a notion of individual rights in anything like the modern sense? Is some relation to guilt crucial to I-morality? Whatever moral principles there are in a given I-morality, it is likely that they do not hold absolutely but are themselves default rules. There are situations in which an I-morality permits one to kill or injure or interfere with someone in the relevant protected group. The need for a systematic way of dealing with default rules complicates I-morality (Nozick, 1968). Or maybe some other part of the mind figures out what to do when there are conflicts among the requirements of I-morality.

13.4.5 Unlearned Moral Principles, Moral Pidgins, and Moral Creoles That part of moral philosophy that looks most like linguistics tries to find principles that would characterize intuitive judgements about trolley cases and related examples that derive from Foot's (1978). Suggested principles include the principle of double effect (which says that it can be worse to bring about something wrong either as an end or means than to bring it about merely as a foreseen side effect of something else one does), the principle that negative duties are stricter than positive duties, and Thomson's (1978) deflection principle, which specifies an exception to the general rule that one is not permitted to harm someone to save others, the exception being when one deflects an existing harm from a larger group of people to a smaller group of people. (Thomson, 1990, abandons the deflection principle.) It is interesting that the suggested principles are not

MORAL PHILOSOPHY AND LINGUISTICS

225

principles that ordinary people are aware of accepting. Indeed, philosophers disagree about which principles are appropriate and how they are to be formulated. If any of these principles should turn out to be adequate to an ordinary person's I-morality, the situation would be like that with respect to the constraint on question-word movement discussed earlier. An ordinary person was never taught the principle of double effect or the deflection principle, and it is unclear how such a principle might have been acquired from the examples available to the ordinary person. This suggests that the relevant principle is built into Imorality ahead of time, in which case we should expect it to occur in all I-moralities (or be a default case, or something of the sort). In other words, the principles should be part of universal moral grammar. (Here I am indebted to unpublished work by John Mikhail.) A further implication is that we might be able to distinguish between, as it were, pidgin and creole moralities. A pidgin morality would be an invented morality such as act utilitarianism that did not adhere to such universal principles as double effect or the deflection principle. The corresponding creole morality would be the I-morality acquired in the normal natural way by children of utilitarians.

13.4.6 Principles and Parameters It may be that all I-moralities contain certain principles which involve parameters that differ from one I-morality to another. Perhaps all I-moralities contain a rule prohibiting killing of members of a protected group G, where G may vary. Similarly, I-moralities may contain a rule prohibiting harm to members of a protected group H, where H may or may not be the same as G. Is it true that, where there is a rule against harming Hs, there is also a rule against killing Hs? Not obviously, because many people seem to accept a rule against cruelty to animals while allowing animals to be killed for food. There may be a universal principle of noninterference with the activities of the members of protected group H. Or perhaps all moralities accept some sort of Golden Rule or variant of Kant's categorical imperative: one thinks about cases by imagining the situation reversed. Or one does this at least for situations involving members of some favoured group G. Notice that disputes about the extent of the protected group G yield some of the most compelling examples of intractable moral disagreement.

226

EXPLAINING VALUE

Are animals included? Are foetuses? Such disputes may reflect genuine parametric variation!

13.5 Conclusion An account of moral diversity might be modelled on linguistics. The plan would be to give a relatively precise account of particular moral idiolects and to consider whether the principles involved could be learned in order to see what universal constraints there are on morality. Such an approach promises new ways of looking at morality.

Bibliography Ackerman, Bruce (1980). Social Justice in the Liberal State. New Haven: Yale University Press. Anscombe, G. E. M. (1958). ‘Modern Moral Philosophy.’ Philosophy, 33: 1–19. Aristotle (1985). Nicomachean Ethics, trans. T. Irwin. Indianapolis, Ind.: Hackett. Axelrod, R. (1984). The Evolution of Cooperation. New York: Basic Books. Ayer, A. J. (1936). Language, Truth, and Logic. London: Gollancz. Baier, K. (1958). The Moral Point of View. Ithaca, N.Y.: Cornell University Press. Beardsley, M. C. (1965). ‘Intrinsic Value.’ Philosophy and Phenomenological Research, 26: 1–17. Bennett, W. J. (1993). The Book of Virtues. New York: Simon & Schuster. Blackburn, S. (1984). Spreading the Word. Oxford: Oxford University Press. —— (1985). ‘Errors and the Phenomenology of Value.’ In T. Honderich (ed.), Morality and Objectivity: A Tribute to J. L. Mackie. London: Routledge & Kegan Paul. Block, N. (1978). ‘Troubles with Functionalism.’ In C. W. Savage (ed.), Perception and Cognition: Issues in the Foundations of Psychology. Minnesota Studies in the Philosophy of Science, 9: 261–325. Minneapolis, Minnesota: University of Minnesota Press. Brandt, R. B. (1954). Hopi Ethics. Chicago: University of Chicago Press. —— (1959). Ethical Theory. Englewood Cliffs, N.J.: Prentice-Hall. —— (1963). Value and Obligation. New York: Harcourt, Brace, & World. —— (1967). ‘Ethical Relativism.’ In P. Edwards (ed.), Encyclopedia of Philosophy. New York: Macmillan and Free Press. —— (1976). ‘The Psychology of Benevolence and its Implications for Philosophy.’ Journal of Philosophy, 73: 429–53. —— (1988). ‘The Structure of Virtue.’ Midwest Studies in Philosophy, 13: 64–82. Brink, D. (1989). Moral Realism and the Foundations of Ethics. Cambridge: Cambridge University Press. Carey, S. (1990). ‘Cognitive Development.’ In D. N. Osherson and E. E. Smith (eds.), Thinking: An Invitation to Cognitive Science, 1st edn., vol. 1. Cambridge, Mass.: MIT Press. Carson, T. L. (1984). The Status of Morality. Dordrecht: Reidel. Chomsky, N. (1988). Language and Problems of Knowledge: The Managua Lectures. Cambridge, Mass.: MIT Press. Copp, D. (1993). ‘Reasons and Needs’. In R. Frey and C. Morris (eds.), Value, Welfare, and Morality. Cambridge: Cambridge University Press.

228

BIBLIOGRAPHY

Darley, J. M., and Batson, C. D. (1973). ‘From Jerusalem to Jericho: A Study of Situational and Dispositional Variables in Helping Behavior.’ Journal of Personality and Social Psychology, 27. Darley, J. M., and Schultz, T. (1990). ‘Moral Rules: Their Content and Acquisition.’ Annual Review of Psychology, 41: 523–56. Darwall, S. L. (1983). Impartial Reason. Ithaca, N.Y.: Cornell University Press. Davidson, D. (1969). ‘How Is Weakness of Will Possible?’ In J. Feinberg (ed.), Moral Concepts. Oxford: Oxford University Press. Dawes, R. M. (1994). House of Cards. New York: Free Press. Dewey, J. (1922). Human Nature and Conduct. New York: Holt. —— (1939). Theory of Valuation. Chicago: University of Chicago Press. Donagan, A. (1977). The Theory of Morality. Chicago: University of Chicago Press. Doris, J. M. (forthcoming). Lack of Character: Personality and Moral Behaviour. New York: Cambridge University Press. Dworkin, R. M. (1985). ‘Is Wealth a Value?’ In his Matter of Principle. Cambridge, Mass.: Harvard University Press. —— (1986). Law's Empire. Cambridge, Mass.: Belknap Press. —— (1993). Life's Dominion : An Argument about Abortion, Euthanasia, and Individual Freedom. New York: Knopf. Finnis, J. (1980). Natural Law and Natural Rights. Oxford: Oxford University Press. Firth, R. (1952). ‘Ethical Absolutism and the Ideal Observer.’ Philosophy and Phenomenological Research, 12: 317–45. Fischer, J. M., and Ravizza, M. (eds.) (1992). Ethics: Problems and Principles. Fort Worth: Harcourt Brace Jovanovich. Flanagan, O. (1991). Varieties of Moral Personality. Cambridge, Mass.: Harvard University Press. Foot, P. (1974–5). ‘ “Is Morality a System of Hypothetical Imperatives?” A Reply to Mr. Holmes.’ Analysis, 35: 53–6. —— (1978). ‘The Problem of Abortion and the Doctrine of the Double Effect.’ In her Virtues and Vices and Other Essays in Moral Philosophy. Oxford: Basil Blackwell. Frankena, W. (1963). Ethics. Englewood Cliffs, N.J.: Prentice-Hall. —— (1973). Ethics, 2nd edn. Englewood Cliffs, N.J.: Prentice-Hall. —— (1976a). ‘Obligation and Motivation in Recent Moral Philosophy.’ In K. E. Goodpaster (ed.), Perspectives on Morality: Essays by William Frankena. Notre Dame, Ind.: University of Notre Dame Press. —— (1976b). ‘On Saying the Ethical Thing.’ In K. E. Goodpaster (ed.), Perspectives on Morality: Essays by William Frankena. Notre Dame, Ind.: University of Notre Dame Press. Frankfurt, H. G. (1971). ‘Freedom of the Will and the Concept of a Person.’ Journal of Philosophy, 68: 5–20.

BIBLIOGRAPHY

229

—— (1978). ‘The Problem of Action.’ American Philosophical Quarterly, 15: 157–62. Freud, S. (1930). Civilization and Its Discontents, trans. J. Riviere. New York: Jonathan Cape and H. Smith. Fried, C. (1978). Right and Wrong. Cambridge, Mass.: Harvard University Press. Gewirth, A. (1978). Reason and Morality. Chicago: University of Chicago Press. Gilligan, C. (1982). In a Different Voice: Psychological Theory and Women's Development. Cambridge, Mass.: Harvard University Press. Ginsberg, M. L. (ed.) (1987). Readings in Nonmonotonic Reasoning. Los Altos, Calif.: Morgan Kaufmann. Grice, G. R. (1967). Grounds of Moral Judgment. Cambridge: Cambridge University Press. Grice, H. P. (1957). ‘Meaning.’ Philosophical Review, 66: 377–88. Hamilton, W. D. (1964). ‘The Genetic Evolution of Social Behavior.’ Journal of Theoretical Biology, 7: 1–16. Hampshire, S. (1972). ‘Fallacies in Moral Philosophy.’ In his Freedom of Mind and Other Essays. Oxford: Oxford University Press. Hardin, R. (1995). One for All: The Logic of Group Conflict. Princeton, N.J.: Princeton University Press. Hare, R. M. (1952). The Language of Morals. Oxford: Oxford University Press. —— (1963). Freedom and Reason. Oxford: Oxford University Press. Harman, G. (1973). Thought. Princeton, N.J.: Princeton University Press. —— (1976). ‘Practical Reasoning’. Review of Metaphysics, 29: 431–63. —— (1977). The Nature of Morality: An Introduction to Ethics. New York: Oxford University Press. —— (1986). Change in View: Principles of Reasoning. Cambridge, Mass.: MIT Press. —— (1995). ‘Rationality’. In E. E. Smith and D. N. Osherson (eds.), Thinking: Invitation to Cognitive Science, vol. 3. Cambridge, Mass.: MIT Press. —— (1996). ‘Moral Relativism.’ In G. Harman and J. J. Thomson, Moral Relativism and Moral Objectivity. Oxford: Blackwell. —— (1999). ‘Rationality.’ In his Reasoning, Meaning, and Mind. Oxford: Clarendon Press. Harman, G., and Thomson, J. J. (1996). Moral Relativism and Moral Objectivity. Oxford: Blackwell. Hartshorne, H., and May, M. A. (1928). Studies in the Nature of Character, I: Studies in Deceit. New York: Macmillan. Hobbes, Thomas (1957). Leviathan. Oxford: Oxford University Press. Hoffman, M. L. (1970). ‘The Development of Ethical Values in the Individual.’ In P. H. Mussen (ed.), Carmichael's Manual of Psychology, 3rd edn., vol. 2. New York: Wiley. Holland, J. H., Holyoak, K. J., Nisbett, R. E., and Thagard, P. R. (1986). Induction: Processes of Inference, Learning and Discovery. Cambridge, Mass.: Bradford Books/MIT Press.

230

BIBLIOGRAPHY

Holmes, R. L. (1973–4). ‘Is Morality a System of Hypothetical Imperatives?’. Analysis, 34: 96–101. Horowitz, T. (1998). ‘Philosophical Intuitions and Psychological Theory.’ Ethics, 108: 367–85. Hospers, J. (1961). Human Conduct. New York: Harcourt Brace & World. Humberstone, I. L. (1971–2). ‘Two Sorts of “Oughts”.’ Analysis, 32: 8–11. Hume, D. (1739). Treatise of Human Nature. —— (1748). Enquiries concerning Human Understanding and the Principles of Morals. Hursthouse, R. (1996). ‘Normative Virtue Ethics.’ In R. Crisp (ed.), How Should One Live? Essays on the Virtues. Oxford: Oxford University Press. Hutcheson, F. (1725). An Inquiry concerning Moral Good and Evil. —— (1728). Illustrations upon the Moral Sense. Jeffrey, R. C. (1965). The Logic of Decision. New York: McGraw-Hill. —— (1974). ‘Preferences among Preferences.’ Journal of Philosophy, 71: 377–91. —— (1983). The Logic of Decision, 2nd edn. Chicago: University of Chicago Press. Kagan, S. (1989). The Limits of Morality. Oxford: Oxford University Press. Kamm, F. M. (1998). ‘Moral Intuitions, Cognitive Psychology, and the Harming-Versus-Not-Aiding Distinction.’ Ethics, 108: 463–88. Kant, I. (1785). Grundlegung zur Metaphysik der Sitten. —— (1788). Kritik der praktischen Vernunft. Kittay, E. F., and Meyers, D. T. (1987). Women and Moral Theory. Totowa, N.J.: Rowman & Littlefield. Kohlberg, L. (1981). Essays on Moral Development, vol. 1: The Philosophy of Moral Development: Moral Stages and the Idea of Justice. San Francisco: Harper & Row. Laing, R. D. (1960). The Divided Self. Chicago: Quadrangle. Lewin, K. (1935). Dynamic Theory of Personality. New York: McGraw-Hill. Lewis, D. (1981). ‘Causal Decision Theory.’ Australasian Journal of Philosophy, 59: 5–30. —— (1983a). ‘Extrinsic Properties’. Philosophical Studies, 44: 197–200. —— (1983b). ‘New Work for a Theory of Universals.’ Australasian Journal of Philosophy, 61: 355–8. —— (1989). ‘Dispositional Theories of Value.’ Proceedings of the Aristotelian Society, suppl. vol. 63: 113–37. Lyons, David (1965). The Forms and Limits of Utilitarianism. Oxford: Clarendon Press. Machan, T. (1975). Human Rights and Human Liberties. Chicago: Nelson Hall. McCloskey, M. (1983). ‘Intuitive Physics.’ Scientific American, 248: 122–30. McDowell, J. (1985). ‘Values and Secondary Qualities.’ In T. Honderich (ed.), Morality and Objectivity: A Tribute to J. L. Mackie. London: Routledge & Kegan Paul.

BIBLIOGRAPHY

231

McGinn, C. (1983). The Subjective View. Oxford: Oxford University Press. MacIntyre, A. (1981). Beyond Virtue. Notre Dame, Ind.: Notre Dame University Press. Mackie, J. L. (1977). Ethics: Inventing Right and Wrong. Harmondsworth: Penguin. MacKinnon, C. (1987). Feminism Unmodified. Cambridge, Mass.: Harvard University Press. McMahon, Christopher (1986). ‘Gewirth's Justification of Morality.’ Philosophical Studies, 50: 261–81. Mandeville, B. de (1714). The Fable of the Bees, or Private Vices Publick Benefits. Maslow, A. (1962). Toward a Psychology of Being. New York: Van Nostrand. Mele, A. (1987). ‘Are Intentions Self-Referential?’ Philosophical Studies, 52: 309–29. Merritt, M. (1999). ‘Virtue Ethics and the Social Psychology of Character.’ Ph.D. diss., University of California at Berkeley. Mikhail, J. (1996). ‘The Moral Faculty.’ Unpublished workshop presentation at MIT. Milgram, S. (1963). ‘Behavioural Study of Obedience.’ Journal of Abnormal and Social Psychology, 67: 371–8. Mill, J. S. (1959). On Liberty. Chicago: Regnery. Nagel, T. (1970). The Possibility of Altruism. Oxford: Oxford University Press. —— (1974). ‘What is it Like to Be a Bat?’ Philosophical Review, 83: 435–50. —— (1979). ‘Moral Luck.’ In his Mortal Questions. Cambridge: Cambridge University Press. —— (1980). ‘The Limits of Objectivity.’ In S. M. McMurrin (ed.), Tanner Lectures on Human Values. Salt Lake City: University of Utah Press. Newcomb, T. M. (1929). The Consistency of Certain Extrovert-Introvert Behaviour Patterns in 51 Problem Boys. New York: Columbia University Teachers College Bureau of Publications. Nisbett, R. E., and Ross, L. (1980). Human Inference: Strategies and Shortcomings of Social Judgement. Englewood Cliffs, N.J.: Prentice-Hall. Norton, D. (1976). Personal Destinies: A Philosophy of Ethical Individualism. Princeton, N.J.: Princeton University Press. Nozick, R. (1968). ‘Moral Complictions and Moral Structures.’ Natural Law Forum, 13: 1–50. —— (1974). Anarchy, State and Utopia. New York: Basic Books. —— (1981). Philosophical Explanations. Cambridge, Mass.: Harvard University Press. Olson, R. G. (1967). ‘The Good.’ In P. Edwards (ed.), Encyclopedia of Philosophy, vol. 3. New York: MacMillan and Free Press. Perelman, C. (1963). The Idea of Justice and the Problem of Argument, transl. J. Petrie. London: Routledge & Kegan Paul. Perry, J. (1990). Hempel Lectures. Princeton, N.J.: Princeton University.

232

BIBLIOGRAPHY

Phillips, D. Z., and Mounce, H. O. (1965). ‘On Morality's Having a Point.’ Philosophy, 40: 308–19. Piaget, J. (1956). The Moral Judgment of the Child. New York: Free Press. Posner, R. A. (1981). The Economics of Justice. Cambridge, Mass.: Harvard University Press. Quinn, W. (1974). ‘Theories of Intrinsic Value.’ American Philosophical Quarterly, 11: 123–32. —— (1993). ‘Putting Rationality in its Place.’ In R. Frey and C. Morris (eds.), Value, Welfare, and Morality. Cambridge: Cambridge University Press. Railton, P. (1997). ‘Made in the Shade: Moral Compatibilism and the Aims of Moral Theory.’ Canadian Journal of Philosophy, suppl. vol. 21: 79–106. Rand, A. (1957). Atlas Shrugged. New York: Random House. —— (1963). For the New Intellectual. New York: New American Library. Rawls, J. (1971). A Theory of Justice. Cambridge, Mass.: MIT Press. Richards, D. A. J. (1971). A Theory of Reasons for Action. Oxford: Oxford University Press. Rorty, R. (1965). ‘Mind-Body Identity, Privacy, and Categories.’ Review of Metaphysics, 19: 24–54. Ross, L. (1977). ‘The Intuitive Psychologist and his Shortcomings’. In L. Berkowitz (ed.), Advances in Experimental Social Psychology, vol. 10. New York: Academic Press. Ross, L., and Nisbett, R. (1991). The Person and the Situation: Perspectives of Social Psychology. New York: McGraw-Hill. Sartre, J. P. (1973). Existentialism and Humanism. London: Eyre Methuen. Searle, J. (1983). Intentionality: An Essay in the Philosophy of Mind. Cambridge: Cambridge University Press. Sharp, F. C. (1950). Good Will and Ill Will. Chicago: Chicago University Press. Shope, R. (1978). ‘The Conditional Fallacy.’ Journal of Philosophy, 75: 397–413. Singer, M. (1961). Generalization in Ethics. New York: Knopf. Singer, P. (1973). Animal Liberation. New York: Random House. Skinner, B. F. (1974). About Behaviorism. New York: Knopf. Smart, J. J. C. (1961). An Outline of a System of Utilitarian Ethics. Melbourne, Vic.: Melbourne University Press. Smith, A. (1759). ‘Of the Influence of Fortune upon the Sentiments of Mankind, with regard to the Merit or Demerit of Actions.’ In his Theory of the Moral Sentiments, Pt. II, sect. III. —— (1976). Theory of the Moral Sentiments (Glasgow edition). Oxford: Oxford University Press. Smith, M. (1992). ‘Valuing: Desiring or Believing?’ In D. Charles and K. Lennon (eds.), Reduction, Explanation, and Realism. Oxford: Oxford University Press. Sober, E. (1990). ‘Let's Razor Ockham's Razor.’ In D. Knowles (ed.), The Limits of Explanation. New York: Cambridge University Press. Sober, E., and Wilson, D. S. (1999). Unto Others. Cambridge, Mass.: Harvard University Press.

BIBLIOGRAPHY

233

Stevenson, C. L. (1963a). ‘The Nature of Ethical Disagreements.’ In his Facts and Values. New Haven: Yale University Press. —— (1963b). ‘Relativism and Nonrelativism in the Theory of Value.’ In his Facts and Values. New Haven: Yale University Press. Stocker, M. (1979). ‘Desiring the Bad: An Essay in Moral Psychology.’ Journal of Philosophy, 76: 738–53. Strawson, P. F. (1959). Individuals. London: Methuen. Sturgeon, N. (1985). ‘Moral Explanations.’ In D. Copp and D. Zimmerman (eds.), Morality, Reason, and Truth. Totowa, N.J.: Rowan & Allanheld. Taylor, R. (1991). Virtue Ethics. Interlaken, N.Y.: Linden Books. Thomson, Judith Jarvis (1978). ‘Killing, Letting Die, and the Trolley Problem.’ Monist, 59: 204–17. —— (1986). Rights, Restitution, and Risk: Essays in Moral Theory, ed. William Parent. Cambridge, Mass.: Harvard University Press. —— (1990). The Realm of Rights. Cambridge, Mass.: Harvard University Press. —— (1996). ‘Evaluatives and Directives.’ In G. Harman and J. J. Thomson, Moral Relativism and Moral Objectivity. Oxford: Blackwell. —— (1997). ‘The Right and the Good.’ Journal of Philosophy, 94: 273–98. Trammel, R. L. (1975). ‘Saving Life and Taking Life.’ Journal of Philosophy, 72: 131–7. Trivers, R. L. (1971). ‘The Evolution of Reciprocal Altruism’. Quarterly Review of Biology, 46: 35–57. Tversky, A., and Kahneman, D. (1974). ‘Judgement Under Uncertainty: Heuristics and Biases.’ Science, 185: 1124–31. Van Lehn, K. (1989). ‘Problem Solving and Cognitive Skill Acquisition.’ In M. Posner (ed.), Foundations of Cognitive Science. Cambridge, Mass.: MIT Press. Veatch, H. B. (1964). Rational Man: A Modern Interpretation of Aristotelian Ethics. Bloomington, Ind.: Indiana University Press. —— (1971). For an Ontology of Morals: A Critique of Contemporary Ethical Theory. Evanston, Illinois: Northwestern University Press. —— (1985). Human Rights: Fact or Fancy? Baton Rouge, La.: Louisiana State University Press. Watson, G. (1975). ‘Free Agency.’ Journal of Philosophy, 72: 205–20. Wiggins, D. (1976). ‘Truth, Invention, and the Meaning of Life.’ Proceedings of the British Academy, 62: 331–78. —— (1987). ‘A Sensible Subjectivism.’ In his Needs, Values, Truth. Oxford: Basil Blackwell. Williams, B. (1972). Morality: An Introduction to Ethics. New York: Harper & Row. Wilson, G. M. (1989). The Intentionality of Human Action. Stanford, Calif.: Stanford University Press.

This page intentionally left blank

Index of Names Anscombe, G. E. M. 151 Aristotle 166–7 Austin, J. 196 Ayer, A. J. 52 Baier, K. 81, 85 Batson, C. D. 170, 172–3 Beardsley, M. 104–5, 111–12, 114, 116, 137 Bennett, W. J. 168 Berkeley, G. 182–3 Blackburn, S. 181 Block, N. 97 Brandt, R. 22, 24–5, 27–8, 32–3, 96, 103–8, 116, 137, 166, 181, 207 Brink, D. 137 Carey, S. 205 Carson, T. 181 Chomsky, N. 210 Copp, D. 117, 133–6 Darley, J. M. 170, 172–3, 196, 205 Darwall, S. 81, 85 Dawes, R. M. 173 Dewey, J. 137 Donagon, A. 81, 85 Doris, J. 169 Dworkin, R. 146, 209, 214 Fellbaum, C. 134 Finnis, J. 151 Firth, R. 23, 80, 181, 183 Fischer, J. M. 197 Flanagan, O. 168–9, 175, 196 Foot, P. 11, 151, 197, 211, 215, 224 Frankena, W. 21, 24–5, 30–31, 35, 43, 81, 85, 103–5, 116, 137 Frankfurt, H. 117–20, 122, 124–7, 132, 135–6 Freud, S. 192, 204 Fried, C. 81, 85 Gendler, T. 223 Gewirth, A. 62–3, 81, 85 Gilligan, C. 196, 204–5 Ginsburg, M. L. 216 Grice, G. R. 81, 85 Grice, H. P. 119 Hamilton, W. D. 206 Hampshire, S. 181 Hardin, R. 177 Hare, R. M. 10, 12, 41–2, 49, 52, 54, 80 Hart, H. L. A. 196 Hartshorne, H. 175

Hobbes, T. 13, 186 Hoffman, M. 33 Holland, J. H. 165 Holmes, R. L. 43 Holyoak, K. J. 165 Horowitz, T. 223 Hospers, J. 103, 105–7, 116, 137 Humberstone, I. L. 6 Hume, D. 58, 80, 162, 181–3, 185–90, 193–5, 199–203, 210, 213, 216 Hursthouse, R. 168 Hutcheson, F. 181–2, 185–90, 193–4 Jeffrey, R. C. 108, 115, 117, 127, 129 Johnston, M. 59 Kagan, S. 165 Kahneman, D. 174 Kamm, F. 223 Kant, I. 39, 54, 62, 188, 225 Kittay, E. F. 196 Kohlberg, L. 63–6, 196, 204–5 Laing, R. D. 176 Lewin, K. 166 Lewis, D. 117, 126–36, 141 Lyons, D. 54 McCloskey, M. 165 McDowell, J. 181 McGinn, C. 181 Machan, T. 151 MacIntyre, A. 151 Mackie, J. L. 79–80

236

INDEX OF NAMES

MacKinnon, C. 140 McMahon, C. 63 Mandeville, B. 186 Maslow, A. 151 May, M. A. 175 Mele, A. 120, 124–5, 135 Merritt, M. 176 Meyers, D. T. 196 Mikhail, J. 217, 225 Milgram, S. 170 Miller, K. 134 Mills, C. 119 Mounce, H. O. 156 Nagel, T. 6, 8, 45, 63–4, 78, 81, 85, 97, 176 Nisbett, R. 165–6, 169–7 Norton, D. 151 Nozick, R. 60, 81, 85, 111, 138, 140–1, 151, 208, 215–16, 224 Olson, R. G. 137 Perry, J. 124 Phillips, D. Z. 156 Piaget, J. 196, 204–5 Plato 185 Posner, R. 207–10, 215 Quinn, W. 129, 138–9 Railton, P. 169 Rand, A. 151 Ravizza, M. 197 Rawls, J. 65, 81, 85, 196, 217 Richards, D. A. J. 81, 85 Rorty, R. 96 Ross, L. 166, 169–7 Sartre, J.-P. 52 Scanlon, T. M. 88 Schultz, T. R. 196, 205 Searle, J. 120 Sharp, F. C. 181 Shope, R. 183 Singer, M. 3, 77 Singer, P. 69 Skinner, B. F. 96, 175 Smart, J. J. C. 52 Smith, A. 176, 181–2, 188–95, 203–4, 216 Sober, E. 206, 213 Stevenson, C. L. 23, 25–6, 28–9 Strawson, P. F. 117 Sturgeon, N. 84–5 Taylor, R. 168 Thagard, P. 165 Thomson, J. J. 92, 176, 196, 210, 215, 224 Trammel, R. L. 47–9 Trivers, R. L. 206 Tversky, A. 174

Van Lehn, K. 210 Veatch, H. 151 Watson, G. 117 Wiggins, D. 181 Williams, B. 3, 77 Wilson, D. S. 206 Wilson, G. 124

Index of Subjects abortion 76, 219 absolutism, see moral absolutism agreement, hypothetical 54, 66; implicit 3–4, 13–17; see also bargaining, moral animals, treatment of 17–18, 32–3, 69–71, 73, 92–4 autonomous ethics 79–99 bargaining, moral 10–13, 15, 46–9, 58, 66, 68–7, 71–5, 164 character traits 165–78 chastity 199–202 coherence 17–18, 34 consequentialism, see utilitarianism contract, see agreement conventionalism 190; see also agreement, implicit deflection, Thomson's principle of 214–15, 224 desirability, compared with value 143, 146–7 desires, intrinsic 117, 127–9; second-order 117–36 double effect 197, 209–10, 224 emotivism 25–6, 52–3, 60 evolution by natural selection 198, 205–7, 212 existentialism 52 flourishing 151–64, 168 freedom of the will 117–18, 126 functionalism, evaluative 151–5, 162–3 fundamental attribution error 166, 172–3, 175 good or bad news 108, 128–9, 143, 146–7 harming versus not helping 11–12, 47–9, 68–9, 164, 211–12 ideal observer, see spectator, impartial insiders versus outsiders 51, 86 intentions 117, 119–26 internalism about reasons 31 interview illusion 174 judgements, inner moral 4–10, 13, 36 justice 4, 58, 61, 66, 70–6 linguistics 217–26 moral absolutism 39–41, 78–9, 81, 84–99 moral development 63, 65–66, 204–5 moral luck 176–7 moral motivation 184–94 moral relativism 3–99, 156, 162–3, 165; normative 20–2, 25–7, 29–35, 38; meta-ethical 20, 24–6, 28–30, 37–8; moral judgment 20, 22–3, 25–30, 35–8, 50 morality as politics, see bargaining naive view of morality 39–57 naturalism 79–80, 82–99 noncognitivism 77, 80, 95; see also emotivism ought 12, 42–45, 49–51, 160–2, 168; see also judgments, inner moral

pacifism 18–19, 53–4 punishment, retributive 145–6 reasoning, practical 17–18, 30, 34 reasons 8, 30–4, 39–45, 62–8, 75, 84, 160, 163; neutral versus normative 59–60, 86–91 reflective equilibrium, see automous ethics relativism, see moral relativism religious belief 95–6 rights 62–3, 159–60, 163–4 self-defence 12–13 self-interest 186–8 slavery 15–16, 18

238

INDEX OF SUBJECTS

spectator, impartial 142, 181–94, 197, 202–4 stoicism 189–90 sympathy 70, 186–95, 198, 202–4 tolerance 75–6, 177 trolley problem 196–7 224; see also harming versus not helping universalizability 35–7, 50–1 universals, moral 21, 32 utilitarianism 12, 47, 52–6, 66, 81, 146, 156–8, 161–2, 165, 186–9, 202–3, 208–9 value 60–1, 91–4, 117; basic 61, 103, 107, 110–16, 138–9; contributory 137, 144–8; instrumental 107, 137, 142–5, 147; intrinsic 103–16, 137–48; see also desires, intrinsic; sacred 146 values 133–6 valuing 117, 126–32, 135–6, 142 virtue, see character traits virtue ethics 176; see also flourishing wealth maximization 207–9 wrong 6–7, 160–1, 168