WORLD, MIND, and ETHICS Essays on the Ethical Philosophy of Bernard Williams

Home , Bernard Williams

Bernard Williams is one of the most influential figures in recent ethical theory, where he has set a considerable part of the current agenda. In this collection a distinguished international team of philosophers who have been stimulated by Williams' work give new responses to it. The topics covered include equality; consistency; comparisons between science and ethics; integrity; moral reasons; the moral system; and moral knowledge. Williams then provides a substantial reply, which shows both the current directions of his own thought and also his present view of earlier work of his which has been extensively discussed for twenty years (such as that on utilitarianism). This volume will be indispensable reading for all those interested in current ethical theory.

WORLD, MIND, AND ETHICS

WORLD, MIND, AND ETHICS Essays on the ethical philosophy of Bernard Williams

EDITED BY J. E. J. ALTHAM University Lecturer in Philosophy and Fellow ofGonville and Caius College, Cambridge and ROSS HARRISON Reader in Philosophy, University of Cambridge and Fellow of King's College, Cambridge

US CAMBRIDGE UNIVERSITY PRESS CAMBRIDGE university press Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, São Paulo, Delhi, Dubai, Tokyo, Mexico City

Cambridge University Press The Edinburgh Building, Cambridge CB2 8RU, UK

Published in the United States of America by Cambridge University Press, New York

www.cambridge.org Information on this title: www.cambridge.org/9780521479301

This publication is in copyright. Subject to statutory exception and to the provisions of relevant collective licensing agreements, no reproduction of any part may take place without the written permission of Cambridge University Press.

First published 1995

A catalogue record for this publication is available from the British Library

Library of Congress Cataloguing in Publication Data World, mind, and ethics: essays on the ethical philosophy of Bernard Williams/ edited by J. E. J. Altham and Ross Harrison. p. cm. ISBN o 521 36024 2 (hardback) isbn 0 521 47930 4 (paperback) 1. Williams, Bernard Arthur Owen — Ethics. 2. Ethics. 3. Ethics, Modern - 20th century. 1. Altham, J. E. J. (James Edward John) 11. Harrison, Ross. bj604.w55w67 1995 I7o'.92-dc2o 94-17269 CIP

ISBN 978-0-521-3 6024-1 Hardback ISBN 978-0-521-47930-1 Paperback

Cambridge University Press has no responsibility for the persistence or accuracy of URLs for external or third-party internet websites referred to in this publication, and does not guarantee that any content on such websites is, or will remain, accurate or appropriate. Information regarding prices, travel timetables, and other factual information given in this work is correct at the time of first printing but Cambridge University Press does not guarantee the accuracy of such information thereafter. Contents

Motes on editors and contributors page viii

Introduction i 1 The idea of equality revisited 4 Jon Elster 2 Is the idea of purely internal consistency of choice bizarre? 19 Amartya Sen 3 Science, ethics, and objectivity 32 Nicholas Jardine 4 Fallibilism and objectivity: science and ethics 46 Christopher Hookway 5 Might there be external reasons? 68 John McDowell 6 Aristotle on human nature and the foundations of ethics 86 Martha C. Nussbaum 7 A most peculiar institution 132 Charles Taylor 8 Reflection and confidence 156 J. E. J. Altham 9 The shape of a life 170 Martin Hollis 1 o Replies 185 Bernard Williams Index 225

Vll Editors and contributors

j. E. j. ALTHAM is a Lecturer in Philosophy at the University of Cambridge.

JON ELSTER is Professor of Political Science at the University of Chicago and Research Director, Institute for Social Research, Oslo.

ROSS HARRISON is a Reader in Philosophy at the University of Cambridge.

MARTIN HOLLIS is Professor of Philosophy at the University of East Anglia.

CHRISTOPHER HOOKWAY is Professor of Philosophy at the Uni- versity of Birmingham. NICHOLAS JARDINE is a Fellow of Darwin College and Professor of the History and Philosophy of the Sciences at Cambridge.

JOHN MCDOWELL is University Professor and Professor of Phil- osophy at the University of Pittsburgh. MARTHA NUSSBAUM is University Professor and Professor of Phil- osophy, Classics, and Comparative Literature at Brown Uni- versity.

AMARTYA SEN is Lamont University Professor and Professor of Economics and Philosophy at Harvard. CHARLES TAYLOR is Professor of Philosophy and Political Science at McGill University. BERNARD WILLIAMS is White's Professor of Moral Philosophy at the University of Oxford.

vm Introduction J. E. J. Altham and Ross Harrison

For many decades Bernard Williams has enlivened philosophy in general and ethics in particular. The editors, like many other philosophers, have been stimulated and helped in their own thought by his incisive and wide-ranging work. This stimulus has, perhaps, been more into opposition than discipleship; but numerous philosophers have worked out their own thought in part by having to meet the problems which he has posed. In ethics Williams has set a considerable part of the current agenda. We, working in the field, have had to take account of his insights and have been forced into much richer thought in consequence. The editors, therefore, thought that it would be both useful and interesting to have a collection of new papers devoted to the study of aspects of Williams' work in ethics. We invited a distinguished team of contributors known to be interested in, or influenced by, Williams' work. We are pleased that so many of them have been able to provide new papers, and that Bernard Williams has been able to contribute a substantial piece of his own, linked to some of the contributions. The way in which Williams has set the ethical agenda can be seen from several specific issues which were either started off, or significantly reformulated, by him decades ago, and which are still running with unabated energy. These issues have been extensively discussed, both by colleagues and also in innumerable student essays. One example is Williams' seminal paper on equality, originally published in 1962 and most easily available in his first collection of papers, Problems of the Self, Cambridge University Press, 1973. This is the topic of the first paper in the present collection, Jon Elster's 'The Idea of Equality Revisited'. Another example of a specific contribution of Williams which has set the agenda and produced decades of subsequent discussion is his 2 WORLD, MIND, AND ETHICS introduction of the idea of integrity to the debate about utilitarianism, particularly by means of his case study of Jim, the bandit chief, and the Indians. This appeared in his contribution to Utilitarianism for and Against, (J. C. C. Smart and Bernard Williams, Cambridge University Press, 1973). Jim and the Indians is the topic of the chapter in the collection by Martin Hollis, and this much discussed question is clarified by Williams' reply to Hollis. Another topic to which Williams gave promotion and redirection, and which has stimulated much discussion over the years, is the importance of consistency in ethics. The particular question of the possibility of consistency in choice is the topic of Amartya Sen's contribution. However, the topic of consistency connects in Williams' thought with wider themes about realism and, more generally, the difference between ethical and scientific thought. This central theme in his work reaches its culmination (at least so far) in his Ethics and the Limits of Philosophy (London: Fontana/ Collins, 1985). It is also the focus of the papers by Hookway and Jardine, where the supposed disanalogy between ethical and scientific thought is examined and criticized. Ethics and the Limits of Philosophy sums up and synthesises many of Williams' distinctive themes. Another topic which occurs in it (in chapter five, 'Right reason') and builds on earlier discussion is the possibility of basing right moral judgment, Kant style, on reason alone. In Williams' language this becomes the possibility of what he calls 'external reasons' (that is, reasons not based on antecedent desire or other pre-existing psychological states). John McDowell's contribution (and Williams' reply) take further a debate about this topic which has been running since Williams' seminal 'Internal and external reasons' paper (originally in Ross Harrison (ed.), Rational Action, Cambridge University Press, 1979; reprinted in Bernard Williams, Moral Luck, Cambridge University Press, 1981). This is another example of where a particular contribution of Williams has set a large subsequent agenda. Other central themes in Ethics and the Limits of Philosophy are treated here. Charles Taylor writes about his treatment of morality, and J. E. J. Altham discusses his distinctive claim that ethical reflection should destroy ethical knowledge. The increasingly Greek-related aspect of Williams' later work also gets prominence in the contributions, particularly in the papers by John McDowell and Martha Nussbaum, as well as in Williams' own contribution. The Introduction 3 first three chapters of Ethics concern ancient thought, as does his latest book, Shame and Necessity (Berkeley, CA: University of Cali- fornia Press, 1993). When we originally conceived the idea of this collection, we thought that it would be greatly strengthened if it were possible to include a paper by Williams himself. We are pleased that this has proved to be possible. The collection ends with a long chapter by Williams, in part of which he replies to some of the earlier chapters. We should make clear that our idea, as we put it to Williams, was never that he should provide a systematic, or formal, reply to the earlier chapters. We hoped for, and are pleased to have received, a chapter from Williams, connected with one or more of the previous contributions, on topics where he particularly wished to add or clarify something. Various factors have led to the assembling of this body of papers taking longer than we would have wished. We apologize to our contributors (particularly the prompt ones) for this; as well as to our readers. Now that it has been assembled, we hope that the stimu- lation which we and many others have received from Williams' work will not only have been celebrated, but also taken further. CHAPTER I The idea of equality revisited Jon Elster

Bernard Williams published "The idea of equality" in 1962.1 In the almost thirty years that have passed, the study of distributive justice has expanded vastly. The work of John Rawls and Amartya Sen, among that of many others, has enabled us to progress further in stating the problems and canvassing plausible solutions. Williams himself has returned to issues of political philosophy on a number of occasions. Yet I believe the early article retains its claims on our attention, partly because it makes some interesting and insufficiently criticized mistakes, partly because it contains some valuable and insufficiently explored suggestions. In this chapter I try to remedy both deficiencies. I shall not summarize Williams' argument in "The idea of equality," but only state some of his main distinctions and conclusions. He gives considerable attention to the notion of equality of persons, i.e. their claim to what Ronald Dworkin has called "equal concern and respect." And he is concerned with showing that this ideal has some bite, in that it suggests specific distributive conclusions. He also considers situations in which there is a prima-facie argument for unequal distribution, on the basis of need and merit. Need is taken as constituting unambiguous and unproblematic grounds for access to scarce goods; I shall argue that it is both ambiguous and problematic. Merit, in Williams' exposition, is doubly questionable as a criterion for the allocation of scarce goods, such as access to high-quality education. In the first place, it is inconsistent with the ideal of equality of persons. In the second place, more disturbingly, attempts to redress this flaw may themselves conflict with that ideal. The only references Williams makes to actual distributive issues are to health care and education, with a very brief mention of unemployment benefits. None of these references have any empirical The idea of equality revisited 5 flesh and blood; they are presented in a very cursory and stylized manner. I am convinced that conceptual and theoretical progress on these issues can be made only if more attention is given to actual distributive issues. Actual cases differ; differences suggest distinctions; distinctions suggest concepts and more general propositions. This is not a plea for an inductive approach to distributive justice, but for something like Rawls' method of reflective equilibrium. Unlike Rawls, however, I believe that, in our search for the data that partly constrains a theory of justice, we should go beyond our own intuitions.2 The actual allocative choices made by institutions, even if impure in being the product of a large number of determinants, many of which are obviously irrelevant from the point of view of justice, can serve as a more independent check on our judgments of fairness.3 Be that as it may, in evaluating Williams' arguments I shall draw heavily on allocative decision-making in practice. Following his lead, I shall consider the allocation of scarce medical resources, notably hearts, livers, and kidneys for transplantation, and the allocation of educational resources, notably admission to institutions of higher education. I begin by considering his influential statement that "the proper ground of distribution of medical care is ill health: this is a necessary truth."4 I shall first point out a counter-intuitive implication of the statement, and then discuss some additional arguments against a more attractive (even if textually less plausible) reading of the statement. I read the statement as saying that there is a monotonic relationship between degree of ill health and quantity of medical resources: the worse your health, the more and the better medical treatment you should get. The rich should not be able to buy scarce medical resources to treat their small ailments when the poor have serious illnesses that go untreated. Thus stated, the principle is appealing. Stated in a more fine-grained way, however, it is far from obviously true. Before I proceed to show this, I need to make some general points that apply to many allocative situations. Let us suppose that there is some scarce good X that is to be allocated, and some quantifiable property of people which that good tends to increase. Thus X could be livers or hearts for transplantation, and Y could be either total lifespan or probability of short-term survival; X could be the assign- ment of rehabilitation officers, and Y the probability of rehabili- 6 JON ELSTER tation of prisoners; X could be educational resources, and Y could be extent of knowledge. Most generally, X could be any good whatsoever, and Y could be utility or welfare in some suitably broad sense. Let us suppose, moreover, that there is a choice between giving one unit of the scarce good to individual I and giving it to individual II, whose situations can be described by the following parameters. If not given the good, the individuals would be at levels Yj and Yn respectively. If given the good, they would attain levels Y'I and Y'n respectively. Two allocative principles suggest themselves. Following the level principle, one should give the good to I if and only ifYi < Yn. Following the increment principle, one should give the good to I if and only if (Y'I-YI) > (Y'n-Yn). The former corresponds to the view that we should give priority to the worst-off, the latter to the idea that we should prefer those who can use the good best.5 Consider first the case of welfare or utility. For the sake of argument I shall make a number of questionable assumptions: utility allows for full (unit and level) comparison across individuals; the utility functions of different individuals are roughly similar; marginal utility is decreasing. Under these assumptions, the level and increment principles always dictate the same solution. The worst off should receive the scarce good because they can use it more efficiently, i.e. derive a larger utility increment from it. Something like this idea underlies the traditional utilitarian argument for the redistributive welfare state.6 Consider next the increase of knowledge by educational resources. Williams does not say that the proper ground of distribution of education is lack of knowledge, nor that the proper ground is the assimilation of knowledge. Both ideas, however, are an important part of the educational philosophy. In this case, however, it is less clear that they point in the same direction. It might be the case — this was actually argued by Leibniz7 - that, even if individuals have identical intellectual capacities, the efficient assimilation of knowledge requires that educational resources be concentrated in a few (perhaps randomly selected) individuals, rather than spread thinly over many. Consider finally the rehabilitation of prisoners. Here, again, the two principles might diverge, albeit for a different reason. Some of those who leave prison will have a spontaneous recidivism rate very close to ioo percent. At the other extreme are those who are certain The idea of equality revisited 7 to get and keep a regular job. The intermediate category is made up of those who may or may not manage to stay out of trouble without any assistance. Clearly, the rehabilitation officer should not spend his time on those who will do well without his assistance, although doing so would make for an easier life. Nor, more controversially, should he concentrate on those in the first category. Enhardened criminals are unlikely to be swayed by efforts to keep them away from crime. Instead, the officer should concentrate on the intermediate category, where his work could actually make a difference. In the allocation of medical resources the two principles sometimes coincide. Let us consider organ transplantation, and make the unrealistic assumption that we are comparing cases in which the graft is certain to succeed. We might then have two reasons for giving the organ to a young man rather than to an old man: the young man has a shorter life behind him and he will, if treated, have a longer life before him. In a different set of choice situations, however, the level and increment principles point in opposite directions. When the candidates for transplantation differ in probability of spontaneous remission rather than in age, the dilemma is more similar to that facing the prison rehabilitation officer.8 The relationship looks roughly as in Figure 1.1.

A B C 100 % of survival without treatment

Figure 1.1 8 JON ELSTER In these cases, the level principle will tell doctors to give priority to patients at A, whereas the increment principle tells them to prefer patients at B. (If, above all, they want a good record or a quiet life, they will prefer patients at C.) I understand Williams to be espousing the level principle, at least by implication. Many doctors do the same, because their professional training emphasizes norms of compassion and of thoroughness. Now, an unrestricted version of that principle is obviously indefensible. It would amount to giving priority to patients who are so ill that they are sure to die in any case. On grounds of fairness, one might, nevertheless, argue that even the severely ill should have some chance of being selected for transplantation. I return to that issue later. Here I simply want to note that Williams' unqualified emphasis on "ill health" is misplaced. Some account must be taken of the extent to which that ill health can be improved by medical care. I am quite confident that Williams would agree. On a more charitable, if less literal, reading of his statement, the notion of ill health would include some degree of improvability. Let me, therefore, see whether the statement, thus interpreted, is more defensible. I believe it runs into at least three difficulties, which can be briefly summarized as incentive problems, paternalism, and envy.9 Sometimes, ill health is the predictable outcome of earlier behaviour. In such cases, one might want to limit the extent of free treatment provided. Such restrictions could be justified by backward-looking arguments, in terms of merit. They could also, and more convincingly, be justified on rule-utilitarian grounds. If people knew that society will not bail them out when their health fails as the predictable result of their own behaviour, they might abstain from behaving in that way. Or again, they might not. Not everybody is equally open to incentive arguments. More to the point, those who might be reached by such arguments are mainly the well educated and affluent. At this point, we can draw on what Williams has to say about merit as a ground of access to educational resources. If access to merit is shaped in part by a "curable environment," the basic equality of persons demand that this access be itself equalized to that extent. Similarly, if medical need justifies treatment only if that need is not knowingly self-inflicted, one might impose the additional condition that the ability to be swayed by incentive arguments be itself equalized, to the extent that it can be affected by a curable The idea of equality revisited 9 environment. Before gross poverty has been eliminated, incentive arguments are inacceptable in health care. This seems right. Some might want to go further, and argue that incentive-effect arguments are never appropriate in this area. There will always be some persons, in all classes of education and income, who are too reckless or thoughtless to take proper account of the risks they are running. After all, most of the predictions one can make in this are statistical ones, which do not reach people as directly as do unqualified warnings. It would be callous to leave a car accident victim to die just because he has thoughtlessly failed to use a safety belt and to take out private insurance. This seems right too, but only because the example is so dramatic. Dental care provides a more instructive example. This treatment is reimbursed by social security in the UK and France, but not in Norway. The Norwegian practice might be justified on the grounds that most people will take care of their teeth when they know they will have to pay the dentist's bill themselves, and that it is more acceptable to say "you have made your bed, so you can lie in it" to those who do not, when the discomforts of the bed are minor and the costs of repairing it small. It goes against Williams' "necessary truth" if rich people who neglect their teeth get better dental care than other negligent people, but could not that inconsistency count against his assertion, rather than against the practice?10 Secondly, one might object that Williams' "necessary truth" is a form of disguised, and possibly misguided, paternalism. To be sure, once an individual needs a kidney, heart, or liver transplantation, he would want to have one. He would not say "I'd rather take the money," since without the operation he would not be around to enjoy the money. It suffices to impose a thin veil of ignorance, however, to make it plausible that he might prefer the money. For "the poor lack a great many goods. Perhaps they would prefer to have some of their other needs met with the money that could be set aside for organ transplants."11 This looks like a knock-down argument, and perhaps it is. Let me try to show, nevertheless, how it could be met, somewhat (if I have understood him right) in the spirit of Williams' own argument. One can imagine two sorts of replies to the objection. First, one might say that paternalism ^justified under circumstances in which some people are so poor that they are tempted, against their real interest, to trade off their long-term health against immediate IO JON ELSTER betterment. Their capacity for making autonomous decisions is so impaired by poverty that they have to be protected against themselves. Secondly, one might argue that in capitalist societies the only redistributive policies that work are in-kind guarantees like free access to education or medical goods, because cash transfers, even when intended to benefit the poor, are largely captured by the middle class.12 In short, the poor would not get their share of the money set aside for transplants; and if they did, they would use it unwisely. These rebuttals are powerful, but insufficient. Empirical claims like this cannot be part of the argument for a necessary truth. Also, the claims themselves are far from being obviously true. A decision to forego costly insurance against a highly improbable event need not be irrationally myopic. A system of largely compulsory health insurance might include some optional features that could be traded in for cash.13 Transplantation could probably not be one of these features. Anticipation of the public outcry when non-insurers are turned away from transplantation centers would prevent any such scheme from getting off the ground.14 But perhaps I could be allowed to forego my right to be operated for varicose veins and take the cash equivalent instead? The third objection is that Williams' principle might partly be based on envy. In the provision of expensive life-saving health care, there is an "all or none" tendency, and a corresponding aversion to selective provision. If it is technically feasible to give the treatment to all who need it, one should do so; if not, no one should get it.15 In both cases, equality of medical care is realized. The American end-stage renal disease program is an example of the former. Eventually, dialysis was made freely available to virtually everybody who was medically indicted for it. The recent Oregon morato- rium on heart and liver transplantations is an example of the latter.16 A partially similar policy was followed in Massachusetts,17 where transplantations were allowed only within very tight budgetary constraints. The following comment on the Report of the Massachusetts Task Force on Organ Transplantation should provide food for thought for those who still believe in the "necessary truth": Suppose we then assume that there is no intrinsic merit to a fixed medical or hospital care budget. Suppose we also assume, as the Report itself suggests, that with feasible arrangements the supply of organs for trans- The idea of equality revisited 11 plantation will be adequate for all who desire transplants. Suppose that heart and liver transplant have passed the research stage, and are known to be effective, if expensive, ways of extending life, and that accurate information concerning the transplants is transmitted to patients and insurers. Finally, suppose that a financing method is developed in which someone who receives a transplant is charged the full resource costs of "producing" that transplant. The purchaser, in effect, causes no additional costs to be imposed on anyone except his household if he obtains a transplant. There could be insurance coverage of such expenses, but the premiums would apply only to those who had specifically elected transplant coverage; there would be no general spillover onto other insureds, either for transplant surgery, or for the follow-up care. My understanding of the Task Force's recommendations is that they would prohibit a person who lives in Massa- chusetts from buying a transplant under such circumstances. In effect, the Task Force finds objectionable a family's decision that it is willing to sacrifice other things it might consume in order to prolong the life of one of its members . . . What can be said in favor of this sort of distribution? In a society shot through with envy, such a view might make sense, but the Task Force offered no empirical evidence for such envy (or, for that matter, for its assertions about citizens' belief about fairness). In the absence of such evidence, I have serious difficulties about raising envy as a moral principle equal to altruism. In any case, envy would call for at most an excise (sumptuary) tax on purchased transplants, not a total prohibition.18

The second and third objections to Williams' argument may be combined as follows. Suppose that we are dealing with a scarce indivisible medical good in totally inelastic supply. The poor ill and the rich ill have the same chance of being drawn from the waiting list. Why should not a poor person be allowed - prior to the development of any illness — to sell his right to be placed on the waiting-list to a rich person, so that the latter would, in effect, have two tickets in the lottery? One might even impose the condition that the rich purchase the extra ticket or tickets prior to his development of the illness, so as to reduce the risk that he might use his wealth to coerce others to give up their rights. In Nozick's phrase, this is a capitalistic act between consenting adults that imposes no harm on third parties. In forbidding it, we express paternalism towards the poor or envy towards the rich, or both. We may deplore inequalities of income. We may wish for a society in which there were no millionaires who could buy transplantations when the number of publicly financed operations was exhausted, and no poor who might prefer, however autonomously, the cash 12 JON ELSTER equivalent of some of their medical rights. Note, however, that the purchase and sale of medical options might also occur, if allowed, in a society of complete equality of income, as long as preferences differ. It is not just the poor who might prefer the cash equivalent of the right to be put on the waiting list for transplantation; a less than average degree of risk aversion might have the same effect. Others might be so risk-averse as to buy expensive assurance for very improbable events. In that case, would not "equality of persons" enjoin us to respect the desire to opt out of or buy into the medical system? Totally reckless behaviour, like extreme myopia, is a sign of irrationality. When we find it, as we often do in young children, it provides sufficient grounds for paternalism. Extreme risk-aversion can be assimilated to a phobia, which is also a sign of irrationality. Less extreme attitudes towards risk may, however, be part of the quiddity of individual character that, in other writings, Williams has urged us to respect. How can one respect a person if one treats his central character traits as grounds for protecting him against himself? I am not quite sure where this medley of arguments leads me. One conclusion — or is it a premise? — is the following. Unlike Williams and Walzer, I do not believe in the inherent autonomy of spheres of justice. There are no necessary truths that regulate the distribution of sphere-specific goods, like medical care or exemption from military service.19 Unlike Walzer (and Williams?), I do believe that the distribution of income should be much more equal than it is in most Western societies today. For one thing, equality of income is a good thing in itself, as long as it does not interfere too much with efficiency. For another, income inequalities detract from the autonomy of choices, in two ways. If I am poor, I may not be able to recognize where my interest lies. And, even if I do, the rich may use their wealth to coerce me to act against it. Under these imperfect circumstances, the insistence on the autonomy of spheres may be justified, not as a first-best principle, but as a way of coping with weakness of will or understanding and with coercive power. Under a more equal distribution of income, the autonomy of spheres would be less important. It might, in fact, become a pointless obstacle to the free choice of life style and priorities. I now turn to a different set of issues. I said earlier that, in Williams' view, the principle of the equality of persons might conflict doubly with the meritocratic allocation of education. The first The idea of equality revisited 13 conflict arises because "one is not really offering equality of opportunity to Smith and Jones if one contents oneself with applying the same criteria to Smith and Jones at, say, the age of 11; what one is doing there is to apply the same criteria to Smith as affected by favourable conditions and to Jones as affected by unfavourable but curable conditions."20 Equality of persons - as they are in themselves, in abstraction from the environment - then demands that the curable environment be cured. Assume that the cure has been effected, so that talents are allowed to develop unfettered by unfavorable environments. Williams argu- es21 that in this truly meritocratic world, people would be "over- concerned with success" and place far too much emphasis on abili- ties. The values of the community and mutual respect would wither. The ideal of equality of persons would be undermined, as a result of reforms motivated by that very principle. I agree with Williams that this is a possible and undesirable side effect of otherwise desirable reforms. Let me also assume, for the sake of argument, that the side effect could outweigh the main intended effects. What, then, are the alternatives? One answer might be that the present system is, all things considered, superior to the meritocratic nightmare. Because everybody knows that talent is often fettered by circumstances, less blame is attached to low achievements than if it was known that achievement and ability were perfectly correlated. Losers in the rat race can retain their self-respect and the respect of others as long as there is sufficient uncertainty about the relative importance of social and genetic causes of success and failure. If social causes were to be eliminated, so would the salutary uncertainty. Better let things be as they are. I cannot believe that Williams would accept this proposal. It smacks too much of Evelyn Waugh. But it is not at all clear what he would say. In his concluding paragraph, he essentially throws up his hands to confess ignorance and advocate ad-hoc pragmatism. I shall attempt to carry the discussion somewhat further by discussing a proposal due to John Broome22 and already implemented, un- beknownst to him, in the Dutch educational system. The issue before us is this. On the one hand, many of us share with Williams the belief that the equality of persons is an important, if vague, principle. On grounds of common humanity, people have a right to equal concern and respect. We would like, moreover, this 14 JON ELSTER principle to have implications for actual policy-making in allocative arenas. On the other hand, most of us also share his belief that there are differential grounds for giving people access to scarce goods. Williams focuses on the grounds for admitting people to higher education, in terms of "the economic needs of society for certain skills." The more talented, presumably, are to be given priority by virtue of these needs. A similar point could be made with respect to the allocation of scarce medical resources. There are prima-facie grounds for preferential treatment of those who can benefit most from the scarce good.23 The Dutch solution is to admit students to medical school by a scheme of weighted lotteries, high school grades being used as the weights.24 Broome would generalize the principle. He stipulates that one can usually ascertain the strength of a person's claim to the scarce good in question. A person with high grades has a stronger claim to being admitted to medical school than a person with lower grades. The claim of the latter, however, is weaker rather than non-existent. Similarly, the claim of an old person to receive an organ for transplantation is weak, but not non-existent. Broome argues, therefore, that the appropriate compromise between the general equality of persons and specific grounds for preferential treatment is to have a weighted lottery, with the strength of the claims being used as the weights. The proposal is attractive, if controversial.25 Let us try to see how it might work in the case of allocating organs for transplantation.26 It seems clear to me that the grounds for preferential treatment are expected probability of success of the transplantation multiplied by expected lifetime after a successful transplantation.27 Against the category of patients who best satisfy this criterion, there are two groups of patients who might press their claims. On the one hand there are those — notably the old — who can be expected (let us assume) to survive the operation as well as any other patient, but who are likely to die soon of other causes. On the other hand, there are the urgent cases who (let us assume) would live as long as any other patient if the operation succeeds, but for whom the very urgency of the illness makes success less likely. How do we measure the strength of the claims of these two patient categories? For the first category, we would probably measure the strength of the claim by expected life extension. For the second category, we might measure it by expected increment in likelihood of survival. The idea of equality revisited 15 This, however, makes for a problem. If on these grounds we allow transplantation of some patients at A in Figure 1.1, we would also have to allow some patients at C to get a ticket in the lottery. But this seems counter-intuitive. Fairness and compassion both suggest that the very ill get a chance, although it involves a near-certain waste of resources. But, in a situation of extreme scarcity, it is hard to see why people who already have a good prognosis should get a chance to improve it. Nor do I see any non-ad-hoc way out of the dilemma that would allow us to acknowledge the claims of patients at A without also recognizing those of patients at C. Regardless of this and other problems of implementation,28 I suspect that Williams might not want to go this way either. In a scheme of weighted lotteries he would hear "the rattle of machinery." It would represent a concern for the abstract human being, not for specific individuals. But I might be wrong here, mistakenly reading Williams through the lenses of the communitarian writers whom he has influenced and with whom he shares at least some enemies. Be this as it may, I do not think viable compromises between the equality of persons and grounds-based differential treatment can be reached otherwise than by some scheme of this general sort. For some purposes, weighted lotteries might serve; for other purposes different compromises might have to be found. This might seem to be fairly close to what I called Williams' "ad hoc pragmatism," and I suppose it is. The difference, if there is any, might lie in my greater willingness to accept mechanical schemes, and my emphasis on the virtues of publicity and predictability embodied in such schemes.

NOTES 1 In the following I quote from the reprint in Problems of the Self, Cambridge University Press, 1973. 2 I also believe we must go beyond laboratory studies of perceptions of justice. The otherwise invaluable work by M. Yaari and M. Bar-Hillel (see notably their "On dividing justly," Social Choice and Welfare, 1 [1984], 1—25) cannot provide a full empirical foundation for a theory of justice. 3 This is not the place to make this argument in greater detail; nor am I certain that, if challenged, I could make a strong and coherent case. My views on the matter are linked to work in progress on "local justice." For a rough overview, see my "Local justice and interpersonal com- l6 JON ELSTER parisons," in J. Elster and J. Roemer (eds.), Interpersonal Comparisons of Well-Being, Cambridge University Press, 1991, pp. 98-126. 4 Bernard Williams, "The idea of equality," p. 240. Among those who have taken their lead from this statement one may notably cite M. Walzer, Spheres of Justice, New York: Basic Books, 1983. 5 This resembles the familiar distinction between maximin justice and utilitarianism, but the two distinctions also differ in many ways, as is easily seen. 6 In that argument the two principles do not coincide fully, even under the stated assumptions, since one must take account of the possibility of a negative impact of distribution on the distribuendum. Throughout this chapter, however, I assume that the total to be distributed is unaffected by the way it is distributed. Even in the case of organ transplantation, this need not be true. The argument has been made, for instance, that if organs are allocated to non-resident aliens, citizens might be less willing to donate organs for transplantation. These are, I believe, at most second-decimal phenomena. 7 "Si la pluralite des personnes a aider fait obstacle, il faut choisir celui dont resultera au total le plus grand bien; ainsi done, en cas de competition, toutes choses egales d'ailleurs, le meilleur, e'est-a-dire le mieux dispose pour le bien public. En effet ce qu'on lui apportera sera multiplie en etant repercute sur beaucoup et de la sorte en l'aidant, on aidera plusieurs. Et meme en general, toutes choses egales d'ailleurs, il faut choisir celui qui a dtja la situation la meilleure. On montrera en efFet que l'aide procede non pas comme une addition, mais comme une multipli- cation ... Aider, e'est multiplier et nuire diviser; la raison en est que celui qui est aide est un esprit; or un esprit peut par sa demarche appliquer tout a tout, ce qui revient a faire un produit ou a multiplier. Prends quelqu'un dont la sagesse soit egale a 3, le pouvoir a 4, la valeur totale de cet homme-la sera egale a 12, et non a 7; en effet il peut user de la sagesse a n'importe quel degre de puissance. Et bien plus, dans le cas d'objets de meme nature, celui qui possede cent mille pieces d'or est plus riche que ne le sont cent personnes dont chacune possede mille pieces. En effet, l'union fait l'utilite: lui, il gagnera meme en ne faisant rien, les autres perdront meme en travaillant." (Leibniz, Philosophische Schriften, ed. C. I. Gerhardt, 1, p. 74. Italics added.) 8 Support for the following statements is found in J. A. Robertson, "Supply and distribution of hearts for transplantation: legal, ethical, and policy issues," Circulation, 75 (1987), 77-87 and in a number of places in D. Mathieu (ed.), Organ Substitution Technology, Boulder and London: Westview Press, 1988, for example, pp. 44-5, 91-2, 98-9, 278-9. Many of these statements compare patients who have had an earlier graft rejected or are on an artificial heart with other categories of patients. The former have a more urgent need for transplantation, but also higher rates of failure. 9 There are other difficulties as well. Williams would have to exclude the The idea of equality revisited 17 number of family dependents as grounds for giving a scarce organ to one candidate rather than to another. I cannot see why this could not sometimes be a valid consideration. 10 As mentioned in note 8, patients often get several transplants. In itself, this may not be objectionable, even though the organs used in the later grafts could have been more effective if given to others. But what shall we say if the need for the later transplants is created by the patient's resumption of the eating, drinking, or smoking habits that brought about the need for the first one? 11 Editorial comment in D. Mathieu (ed.), Organ Substitution, p. 146. 12 B. Page, Who gets What from Government? Berkeley and Los Angeles: University of California Press, 1983. 13 Two issues are involved here: levels of income and equality of income. We must imagine a society in which the minimal income is high enough to ensure that no one needs to be protected against their own irrational myopia, while the income distribution is sufficiently unequal to ensure that some people might take the cash option because, at their level of income, they would rather spend the money on other things. The Scandinavian societies, in my opinion, approach this state of affairs. 14 Note, however, that this outcry could work against the ex ante interests of the group on behalf of which it is made. 15 This is apparently a central principle in Jewish ethics. See F. Rosner, Modern Medicine and Jewish Ethics, New York: Yeshiva University Press, 1986, pp. 347-8. 16 H. Gilbert and E. B. Larson, "Dealing with limited resources: The Oregon decision to curtail funding for organ transplantation," New England Journal of Medicine (July 21, 1988), 171-3. 17 CM. Havigurst and N. M. King, "Liver transplantation in Massa- chusetts: Public policymaking as a morality play," Indiana Law Review, !9 (1986), 955~87- 18 M. V. Pauly, "Equity and costs," in D. Mathieu (ed.), Organ Substitu- tion, pp. 172-3. 19 I enjoin the reader to go through the following thought experiment. Suppose that the good society has been reached. There is full equality of income. There are no wars, and no need for military service. There is, however, a need for "national service" to cope with frequently recur- ring natural disasters. A small number of young men are chosen each year by lot for this vital but risky task. Should one allow substitution, so that a man selected by lot could pay another to take his place? Should one allow commutation, so that he could go free by paying a fee to the state, set at a level sufficient to induce volunteers? If not, why not? 20 Williams, "The idea of equality," pp. 245-6. 21 At least, I believe the following represents his views on the subject. What he actually says refers to a science-fiction example so removed from the real world that it is hard to know what to infer from it. 22 John Broome, "Fairness and the random distribution of goods" l8 JON ELSTER

(unpublished manuscript 1987). 23 Unlike the reasons given for giving the talented preferential access to higher education, this argument does not rely on the economic benefits which the recipients of medical treatment will be able to provide for other people. People who have received transplants are, by and large, unable to do much productive work. 24 For a description of the system, see W. K. B. Hofstee, "The case for compromise in educational selection and grading," in S. B. Anderson andj. S. Helmick (eds.), On Educational Testing, San Francisco: Jossey- Bass, pp. 109-27. 25 For some inconclusive comments, see my Solomonic Judgements, Cam- bridge University Press, 1989, pp. 114—15. 26 For one version of this idea, see D. Brock, "Ethical issues in recipient selection for organ transplantation," in D. Mathieu (ed.), Organ Substi- tution, pp. 86—99, at P- 97- His proposal is rather coarse-grained, as it does not distinguish between the two reasons why people might not benefit from an organ transplantation: low probability of success of the operation and low expected lifetime following a successful operation. 27 For reasons briefly stated in my "Local justice and interpersonal comparisons," I do not think one should measure the life extension in "quality-adjusted life years." 28 Among the other problems, consider claims based on family dependents (note 8 above). If these are admitted, how should they be integrated with other claims into one overall claim? CHAPTER 2 Is the idea of purely internal consistency of choice bizarre?1 Amartya Sen

I. INTRODUCTION I intend to argue for an affirmative answer to the question in the title. But why ask it in the first place? That last query is easy to answer. The axiomatics of internal consistency of choice have been given a central place in many approaches in decision theory, in economics, in game theory, in social choice, in axiomatic choice theory, and in many related disciplines. Many such conditions of internal consistency are standardly used to establish theorems and derive results, and they include axioms of various different types, for example, "the weak axiom of revealed preference," "the strong axiom of revealed preference," "basic contraction consistency," "basic expansion consistency," "binariness of choice," "strong independence axiom," "sure thing principle" (to mention just a few of a very large set of conditions).2 The problem that arises immediately is whether a set of choices can be interpreted and seen as consistent or inconsistent without bringing in something external to choice, such as the underlying motivations or objectives or values that are pursued or acknowledged by choice. Hence the question, which - despite a slight air of mud-slinging — is meant as a serious proposal for investigation.3 I would also argue that the pragmatic reasons for invoking conditions of internal consistency of choice continue to exist, even though any attempt to defend, or motivate, or indeed properly interpret, these conditions without an external reference is bound to be flawed. There is a reason for choosing this topic for a Festschrift for Bernard Williams. Williams has - perhaps more than anyone else - provided far-reaching critiques of the idea of consistency in other

19 20 AMARTYA SEN areas, for example, consistency of beliefs, desires, judgments, imperatives, and also of ethical consistency and moral conflicts in general.4 He has not, as far as I am aware, specifically scrutinized the idea of internal consistency of choice, but that subject does have inter alia a clear bearing on things that he has actually investigated.

2. INTERNAL CONSISTENCY AND EXTERNAL CORRESPONDENCE What is internal consistency of choice? The idea refers to relations of apparent "agreement" or "coherence" that are required to hold between what are chosen from different sets, i.e., different parts of the choice function.3 An internal consistency condition takes the form of imposing constraints on what can or cannot be chosen from a set S given what are being chosen from other sets T. To illustrate, Samuelson's well-known "weak axiom of revealed preference" is a condition of internal consistency that demands that if some element x is chosen from a set that includes y (i.e., y could have been chosen instead), then y must not be chosen from any set that includes x.6 To give another example, the so-called Property a (sometimes also called the "Chernoff condition," or "basic contraction consistency") demands that if x is chosen from a set T and belongs to a subset S of T, then x must be chosen also from that subset S.7 Similarly, other constraints on what can or cannot be chosen from a subset, given the choices over other subsets, are variously imposed by other conditions of internal consistency of choice (such as "expansion consistency," "strong axiom of revealed preference," "weak congruence axiom," "binariness" or "norma- lity" of the choice function, acyclicity or transitivity of the base relation of choice, and so on).8 Internal consistency can be contrasted with what may be called "external correspondence." A condition of external correspondence relates choice to something other than choice (for example, an objective function to be maximized).9 "Optimization" is a requirement of external consistency, to wit, that the choices must correspond to picking the alternatives judged to be at least as good as all the others in the available set. So is "maximization" in the broad sense, i.e., picking an alternative that is not worse than any other available alternative.10 External correspondence takes us beyond the choices themselves.11 Purely internal consistency of choice 21 It should be obvious that any condition of external correspondence will entail some internal characteristics to be satisfied by the choice function, and they can be, then, seen as derived conditions of internal consistency. For example, consider optimization in the form of maximizing a numerical objective function with a specific numerical value attached to each alternative (independently of the rest of the "menu"). With such optimization, Property a ("basic contraction consistency" or "the Chernoff condition") must be automatically satisfied (as an implication). Suppose x is chosen from the larger set T and belongs to a subset S of T. Clearly, x has at least as high a value of that numerically measured objective as any other member of T. This entails that x has at least as high a value as any other member qfS (a subset of T). Therefore, x is optimally choosable from S, in line with the maximization of that numerical objective function. Thus the internal consistency condition of Property a will be satisfied, as an entailment of optimization. This internal characteristic is merely a consequence of the external correspondence (in this case, the maximization of the objective function), and its justification involves external reference. Such "exogenous" internal consistency can be contrasted with "pure" or "endogenous" internal consistency, demanding that if x is chosen {no matter why) from T and belongs to a subset S of T, then x must be chosen from S.

3. VIOLATION AND THE NATURE OF OBJECTIVES Before taking up the thesis of the "bizarreness" of purely internal consistency, let us evaluate the claim that a violation of some internal consistency indicates some failure of rationality in the sense of making reasonable and intelligent choice. The oddity in the claim lies in the expectation that some condition of internal consistency of choice can be specified without knowing what the motivation behind the choice is, and in the insistence that this condition must apply no matter what the motivation is. Is that belief sustainable? It is not obvious how it can be sustainable. Consider, first, a deliberately contrived counter-example (I will presently move to more "natural" ones). Suppose my object is precisely to establish that I am "inconsistent," as that idea is understood in this context. I may have reasons to try to establish that view of me. I may want to be taken as unusual, or eccentric, or unfathomable. I may believe 22 AMARTYA SEN that, as an oriental, it is right that I should be immensely inscruta- ble. I might, like the Maid of Orleans, be entirely convinced that St. Michael, St. Margaret and St. Catherine have been speaking to me - in this case exhorting me to rise against the invasion of consistency- mongers. Or — more simply — I could just find inconsistency "great fun" ("what joy to see the puzzled faces of the wise and the learned - especially decision theorists"). No matter what the specified conditions of consistency are, I can have fair reasons to violate them. How could it be the case that the violation of some specified consistency condition must indicate a failure of reasoning and a violation of rationality, since my reasoned objective can be precisely to act against that condition of consistency? But enough of that line of argument, with its devised feature of inconsistency-seeking. Suppose the person in question does not "give a damn" one way or the other as to whether she is consistent or not. Even then, no matter what specific condition of internal consistency is prespecified, she can possibly have good reasons for violating it. We must distinguish between several layers of problems. There is a practical issue of using internal conditions of consistency of choice when the chooser is, in fact, pursuing maximizing behaviour, but when there are more than one "optimal" or "best" alternatives between which the chooser is "indifferent". If a person finds x and y both equally good and optimal in a given set S, then he may choose the alternative x - not y - in the presence of y, and then again choose y - not x - in the presence of x, without in any way contradicting himself, even from the rigid perspective of optimization. He has to choose one or the other, since on each occasion the exercise takes the form of picking one alternative, and he is sure that by choosing either x or y he has nothing to lose, since either is at least as good as every other alternative (including y and x respectively). However, as the so-called Weak Axiom of Revealed Preference is formulated,12 choosing x (not y) in the presence of y and choosing y (not x) in the presence of x is a violation of this widely used condition of internal consistency. This problem can be tackled within the framework of optimization by ruling out non-singleton optimality, i.e., demanding that in each set there must be only one best element. This can be done in several analytically different ways. The simplest — and the crudest — is to assume that the person is never indifferent between any two alternatives. A more sophisticated route is to assume that the person Purely internal consistency of choice 23 has a strictly concave objective function (so that the "indifference curves" do not have flat bits), which would entail for given prices that there will be a unique best element for any total budget.13 But, whichever way the choice sets are "made" element-valued (with singleton best elements respectively), any internal-consistency condition that relies on that singleton property for its rationale would have to depend thoroughly on some external assumptions that limit its applicability and scope. The problem becomes even sharper when the person is maximizing in the broad sense (discussed earlier) of choosing an element that he finds no worse than any other (but not necessarily decidedly at least as good as all the other alternatives). Such maximization can be based on objectives reflected by an incomplete ordering. In this case, even "ruling out" indifference will not solve the problem. Consider the much-discussed case of "Buridan's ass." There are many versions of the dilemma - and the tragedy - of that thoughtful donkey, but consider the one that suggests that the ass in question could not decide which of the two haystacks was better (for example, larger). It died of starvation - dithering between the two haystacks which it could not rank. That was, it is not hard to argue, a silly thing to do, since no matter which haystack was the better of the two, either would have been - in view of what we are told about the ass - better for (and preferred by) it than dying of starvation. But suppose that Buridan's ass were to behave intelligently. Since either haystack is better than starvation, and since neither haystack is known to be worse than the other, Buridan's ass has a good reason to choose either. Suppose it chooses haystack A, rejecting haystack B as well as starvation. The Weak Axiom of Revealed Preference will now demand that Buridan's ass must not choose haystack B from any set containing A. But why is that reasonable? If Buridan's ass is subjected to a fresh dilemma, having to choose among haystacks A and B - still unrankable - and a new tiny haystack C (clearly worse than both A and B), and, of course, starvation, then why would it be irrational for it to pick haystack B now? Haystack B is clearly better than both starvation and the tiny haystack C, and it is not known to be worse than haystack A. The dithering can be broken as reasonably with choosing B now as it was with choosing A earlier. This example indicates that, depending on the circumstances, there may be good reasons for not worrying about the violation of even such elementary requirements of internal consistency as the 24 AMARTYA SEN Weak Axiom.14 The problems considered in this section have arisen from the nature of the objectives or maximizing targets pursued by the person. Being inconsistent or creating an image of inconsistency may be among the objectives pursued, and, even when that is not the case, the possibility that indifference or incompleteness may influence the targeting can lead to the violation of formal conditions of internal consistency on entirely reasonable grounds. In the next section I consider a different type of problem which, I believe, is deeper in significance and harder to deal with. This concerns the epistemic relevance of the menu from which we find ourselves choosing. This consideration provides very different grounds for violating the purely internal consistency conditions that relate the choice over one menu to choices over other menus.

4. VIOLATION AND THE EPISTEMIC RELEVANCE OF THE MENU In the social sciences we are much inclined to observe what people choose, and, on that basis, form a view of their aims, objectives, values, and so forth. But this observational feature need not work only in that one direction. The person making the choice also takes a view of the nature of the world - or the characteristics of other parties — on the basis of examining what choices are being offered to her. Thus, the "menu" that she faces gives her information about things that may well be directly relevant to what she has reasons to choose. I will begin with a simple example, which is not elegant, but has the virtue of being a true account of a personal experience. When I first came to the United States on a visiting appointment (in i960, as it happens), we were invited to an American colleague's home "at 8 p.m." With Indian and British experiences behind me, I antici- pated supper, and, on arrival, refused impediments in the form of nuts and crackers which were being offered, so as to retain my appetite. Only when coffee and ice-cream were also offered, I knew where I had erred and immediately fell on the nuts and crackers. A rejected alternative (viz., nuts and crackers) in the smaller set got firmly - and ravenously - chosen in the enlarged set. That violation of Property a might not be seen as a "real" violation, since my knowledge was greater (to wit, the awareness of the non-existence of supper) when I chose the neglected nuts and crackers. But the important point to note is that this expansion of knowledge was based entirely on the enhanced menu that was being Purely internal consistency of choice 25 offered. The value I had reason to attach to nuts and crackers shot up the moment coffee and ice-cream were also offered. Observation is not a one-way process. Just as the decision theorist "reads" what people choose, people also "read" what is being offered. A variation of the "menu" can give us good reasons to re-evaluate the alternatives. This plays havoc with almost all the conditions of internal consistency which have been proposed and used in the literature, since they look for "internal consistency" of choices over different menus as if menu variations cannot make any differences to the relative placing of different alternatives. It is easy to consider many other types of cases in which variation of the menu give reasonable grounds for re-evaluating the alternatives. As the library closes, Jack asks Jill whether she would like to join him for a cup of coffee in the neighbouring cafe. Jill chooses to do that rather than going immediately home. In an alternative scenario, Jack asks Jill whether she would like to join him for a cup of coffee in the neighbouring cafe, or come to his home and share some cocaine. Jill has no interest in cocaine or in visiting Jack's home. But what is relevant here is the possibility that with this enhanced menu that Jill faces she may now reject even the alternative of having coffee with Jack in the neighbouring cafe. The appearance of the third alternative has altered Jill's evaluation of the relative merits of the two alternatives in the former scenario. To be sure, this is because, in the latter case, Jill knows some things about Jack that she did not in the former case, but that additional knowledge or understanding has been obtained through the nature of the "menu" offered — the set from which choices are being made. To take quite another type of case, consider the proposed — and happily rejected - legislation in Britain that British residents origi- nating from Asia, Africa, and the West Indies be paid substantial sums of money to induce them to "go home." The idea (outlined by the Right Honourable Enoch Powell, if my memory serves me right) was not to force anyone to go, but to give the dark-skinned British residents of outside origin the option of being paid to go away. Could this option, if passed, give dark-skinned British residents reasons to behave differently in any way, even if they were not to take up that specific option? It is hard to see why not. The option offered might have made clear how keen the natives were in Britain to get rid of the dark-skinned immigrants, and this could have easily 26 AMARTYA SEN influenced many choices (including ones relating to political organization and activism) on the part of the disfavored residents. Once again the expanded "menu" offered has a material impact on the evaluation of pre-existing alternatives. In general, choosing an alternative x from a set S is not quite the same choice as choosing the same alternative x from a set T. Often, even our language of choice - describing what we are doing - brings in the counter-factual situation. A person who fasts is not just starving, but choosing to starve when he does have the option of eating. A person who chooses firmly and happily to fast for a cause, when he has the opportunity of eating what he likes, may not choose starvation that easily if he is too poor to be quite sure of the next meal. For one thing, one cannot adequately "deny" an opportunity when it is not fully there. Of course, it is possible to "redefine" the choice of alternatives in such a way that this violation of Property a will be avoided. For example, choosing x from set S (call it x/S) may be taken to be a different alternative from choosing the same x from another set T (call it x/T). Then the choice of x/S and the rejection of x/T cannot possibly be in any tension with one another. That will save Property a all right, but only by making it silent. If every time the set from which the choice is being made changes, the choice of any given alternative (for example, x) is taken to be a different choice (for example, x/S), then no condition of internal consistency of choices from different subsets can make any demand whatsoever. Property a could not be violated since it demands nothing. Nor will any other condition of internal consistency of choice from different subsets. To see these internal consistency conditions as invariable requirements of rational choice, and to diagnose any violation of these conditions as failures of reasoning, are clear mistakes. On their own the consistency conditions are not demands of reason. They may well be entailed as demands of reason in particular circumstances for particular types of choices, but that would be so derivatively and contingently.

5. STATEMENTS 'VIS-A-VIS' CHOICES If the foregoing argument is correct, then we may conclude that conditions of internal consistency of choice cannot be, in themselves, demands of reason, even though in specific circumstances they may Purely internal consistency of choice 27 become - derivatively and contingently - implications of reasoned pursuit of objectives and values (external to the choice function as such). But this does not yet show that to demand internal consistency conditions as autonomous or invariable requirements of reasoned choice is actually bizarre. The rejected claim may be mistaken, without its being odd enough to qualify as bizarre. Not only is the claim mistaken, it is also extraordinarily hard to understand how it could have been seen as plausible. It is, of course, not difficult to think of statements about alternatives that would in fact be internally contradictory, for example, that x is better than y and that x is not better than y.15 Or that x should be chosen over y and that x should not be chosen over y. But an act of choice (for example, choosing x from a set containing y) is not in itself a statement of this kind. An act of choice is not a statement of any kind. Nor does it entail a statement, on its own.16 It is only with some idea of the motivation underlying the choice - an external reference - that one can begin to "read" the choice as having made a statement of some kind. With such an external reference, we can, of course, examine the coherence and contradictions of the interpreted statements, but that analysis would not be, in any sense, a purely internal one, but one that is contingent on the external reference. The choice of x may reflect one of a great many different motivations. From the fact of the choice itself (i.e., without bringing some reference that is external to the fact of the choice), we have no way of knowing what characteristic of x (if any) led to the choice of x from the set S.17 Thus, we cannot decide whether two choices, for example, x from S and y from T, are in any tension with each other, without knowing the motivation underlying the choice and the contexts of the choice (i.e., without knowing some things that are different from the facts of the choices themselves). The bizarreness lies not in the simple fact that the claim under dispute is false, but that it could not possibly make sense unless the choices are taken to be statements in a way they cannot possibly be. Underlying the mistaken claim, there is a basic disorientation about the nature of choice acts as such. Despite the bizarreness of the idea of purely internal consistency of choice, the testing of internal consistency can be quite useful as a derivative exercise, given an appropriate external reference. When the nature and the demands of the motivations underlying the choices are understood, it may well emerge that particular corres- 28 AMARTYA SEN pondences between choices from different sets could be expected to hold, and that the absence of such correspondences may indicate some failure of the use of reasoning in choice. For example, when a person is optimizing with a completely ordered objective function (an ordering that is invariant with respect to variations of the menu), many internal consistency conditions will be automatically satisfied (including Property a), as was discussed earlier. When a person is attempting to optimize in this way, checking whether those derived internal consistency conditions are being met may well be a good practical way of assessing how that attempt is generally proceeding.18 But, in all these cases, the reason for satisfying these internal consistency conditions arise from something external to the choice function itself (for example, the demands of optimization on the basis of a given and complete objective function). The bizarreness of seeing internal consistency conditions as autonomous or invariant demands of rationality does not in any way dispute the possible usefulness of these conditions employed in a derivative and contingent way with proper external reference. The contingent usefulness of exogenous conditions of internal consistency has to be distinguished from the alleged viability and virtue of purely internal consistency. The first step is to sort out what is being done and why.

NOTES 1 For helpful discussions on these and related matters, I am most grateful to Jerry Green, Frank Hahn, Isaac Levi, Tapas Majumdar, Robert Nozick, Emma Rothschild, and Richard Zeckhauser. Some of the technical issues referred to in this chapter are further discussed in my Presidential Address to the Econometric Society, "Internal consistency of choice," Econometrica, 61 (May 1993). For research support I am indebted to the National Science Foundation. 2 The literature on internal consistency conditions is vast. A selected list (with an eye to pioneering contributions as well as variety) must include inter alia: P. A. Samuelson, "A note on the pure theory of consumer's behaviour," Economica, 5 (1938); I. M. D. Little, "A reformulation of the theory of consumer's behaviour," Oxford Economic Papers, 1 (1949); H. S. Houthakker, "Revealed preference and utility function," Economica, 17 (1950); J. F. Nash, "The bargaining problem," Econometrica, 18 (1950); K. J. Arrow, Social Choice and Indi- vidual Values, New York: Wiley, 1951; H. ChernofT, "Rational selection of decision functions," Economica, 22 (1954); R- M. Thrall, D. H. Coombs, and R. L. Davis (eds.), Decision Processes, New York: Wiley, Purely internal consistency of choice 29 1954; R. D. Luce and H. Raiffa, Games and Decisions, New York: Wiley, 1957; KJ. Arrow, "Rational choice functions and orderings," Economica, 26 (1959); T. Majumdar, The Measurement of Utility, London: Macmillan, 1962; M. K. Richter, "Revealed preference theory," Econo- metrica, 34 (1966); B. Hansson, "Choice structures and preference relations," Synthese, 18 (1968); P. C. Fishburn, Utility Theory for Decision Making, New York: Wiley, 1970; J. S. Chipman, L. Hurwicz, M. K. Richter, and H. F. Sonnenschein, Preference, Utility and Demand, New York: Harcourt, 1971; H. Herzberger, "Ordinal choice structures," Econometrica, 41 (i973);J- Kelly, Arrow Impossibility Theorems, New York: Wiley, 1978; K. Suzumura, Rational Choice, Collective Decisions and Social Welfare (1983); I. Levi, Hard Choices, Cambridge University Press, 1986. 3 The mud, if slung successfully, will certainly find its way on to my own shirt. I mention this only to make clear that the paper is not really meant as a "wiser than thou" exercise. 4 See Bernard Williams, Problems of the Self, Cambridge University Press, 1973; Utilitarianism: For and Against, jointly withj. J. C. Smart, Cam- bridge University Press, 1973; Moral Luck Cambridge University Press, 1981; Ethics and the Limits of Philosophy, Cambridge, MA: Harvard University Press, and London: Fontana, 1985. 5 Formally, a choice function C(.) denned over the universal set X maps the set of nonempty subsets (2X - o) of X into itself. It specifies for each nonempty subset S of X, a non-empty subset C(S) of S. Internal consistency conditions are restrictions on C(S) given C(T) for T different from S, i.e., they constrain what can or cannot be chosen from subset S given what are being chosen from other subsets T. 6 See Paul A. Samuelson, "A note on the pure theory of consumers' behaviour." 7 This condition has been used in a number of well-known contributions to individual and social choice theory, including J. F. Nash, "The bargaining problem"; KJ. Arrow, Social Choice and Individual Values; H. Chernoff, "Rational selection of decision functions"; R. Radner and J. Marschak, "Note on some proposed decision criteria," in Thrall, Coombs, and Davis, Decision Processes. I have investigated the content and implications of Property a in my "Choice functions and revealed preference," Review of Economic Studies, 38 (1971), reprinted in Choice, Welfare and Measurement Oxford: Blackwell 1982. See also H. Herzber- ger, "Ordinal Choice Structures"; K. Suzumura, Rational Choice; I. Levi, Hard Choices. 8 Many of the apparently different conditions of internal consistency entail each other, and some entail others without being entailed by them; on these relations, see K. J. Arrow, "Rational choice functions and orderings"; A. Sen, "Choice functions and revealed preference"; H. Herzberger, "Ordinal Choice Structures"; K. Suzumura, Rational Choice. 9 That is, whereas internal consistency is a characteristic of the choice 30 AMARTYA SEN function itself (and involves no reference to anything other than that function), external correspondence is a characteristic of the relation of the choice function to something else, e.g., a demand that choices must correspond to the maximization of an objective function (not in itself a part of the choice function). 10 When the ranking relation "at least as good as" is a complete ordering, the "optimal set" (the set of alternatives that are "at least as good as all the others) and the "maximal set" (the set of alternatives that are "no worse than any other") will coincide. But they are not in general equivalent. 11 A terminological point may be worth commenting on, at the risk of laboring the obvious. In his paper "Internal and external reasons" (in R. Harrison [ed.], Rational Action, Cambridge University Press, 1980; reprinted inB. Williams, Moral Luck, 1981), Bernard Williams discusses the distinction between a reason being "internal" in the sense of relating directly or indirectly to "the agent's subjective motivational set" and its being "external" when it does not have that connection. (A possible example of external reason, considered by Williams, is that of Owen Wingrave's having "a reason," according to his father, to join the army "since all his male ancestors were soldiers," even though Wingrave himself has "no motivation to join the army at all.") An "internal reason" (invoking the person's motivations) would yield an "external correspondence," as defined in this paper (relating the person's choices to something other than those choices - to wit, the person's motivations). The internality of "internal reason" (in Williams' sense) refers to being internal to the person, whereas the externality of "external correspondence" refers to the choices being linked to something external to the choices. A person's motivations, objectives, etc. are internal to the person, but they are external to the observed choices. 12 Samuelson, "A note on the pure theory of consumer's behaviour." 13 This result will not generally hold even for consumer's choice in non-competitive markets; on these and related matters, see Tapas Majumdar, "Revealed preference and the demand theorem in not- necessarily competitive markets," Quarterly Journal of Economics, 83 (1969)- 14 The example considered will violate even the most basic requirement of Property a if the element that is actually chosen from a set is taken to be the only "chosen" element from that set. However, there is a way of characterizing maximization, even with an incomplete ordering, that would satisfy Property a. Rather than taking the demand of maximization to be the choosing of some alternative that is not established as worse than any other, it could be demanded that all alternatives that are no worse than any other are recognized as "choosable." In this case, for the "choosable" sets, Property a will be satisfied. (In the example considered in the text, both A and B will be "choosable" in both the Purely internal consistency of choice 31 cases.) Note, however, that with this reformulation, it would never be decidable by simply observing two choices over different subsets ("menus") whether or not Property a has been violated. It should be noted that this general problem is present even with full optimization (i.e., not just maximization in the broad sense) since the "optimal set" may contain more than one alternative (given the possibility of "indifference"). 15 Note that the statements "x is better than y" and "y is better than x" are also contradictory given the asymmetric nature of the binary relation "better than." 16 In particular, to choose x from a set S containing y does not, on its own, entail the statement that x is better than y, nor the statement that x should be chosen from S. An analysis of the person's views on these things, if it is to be related to choice information, must take note not only of the acts of choice, but also of the motivations leading to those choices. 17 As Donald Davidson notes, the "pro attitude" towards an action that may give a person reason to act that way may include "desires, wantings, urges, promptings, and a great variety of moral views, aesthetic principles, economic prejudices, social conventions, and public and private goals and values" (D. Davidson, Essays on Actions and Events, Oxford: Clarendon Press, 1980, pp. 3-4). 18 It must, however, be recognized that the kind of motivational complexity that Bernard Williams has particularly studied (in Problems of the Self Moral Luck, Ethics and the Limits of Philosophy, and elsewhere) is ultimately relevant even to assess the derivative conditions of internal consistency of choice. The derived consistency conditions may, in many circumstances, be far from simple. CHAPTER 3 Science, ethics and objectivity Nicholas Jardine

In science, Bernard Williams maintains, we can hope that on each question our inquiries will yield 'a convergence on an answer where the best explanation of the convergence involves the idea that the answer represents how things are'.1 Two features of science are, on Williams' account, crucial for the coherence of this hope. First, we can coherently hope of scientific concepts, theoretical as well as observational, that they are 'world-guided' - that is, that when they are competently applied the reliability of the resultant judgments can be explained in terms of their causal relation to the relevant states of affairs 'out there in the world'. Secondly, our scientific beliefs can provide the basis for an 'absolute conception of the world', a conception that is 'to a maximum degree independent of our perspective and its peculiarities' and in terms of which 'we may hope to explain the possibility of our attaining the conception itself, and also the possibility of other, perspectival, representations'.2 In ethics, Williams argues, we cannot coherently hope that our inquiries will yield any such convergence of belief guided by the way things in reality are. For the two features of science that sustain this hope are lacking in ethics. The most general concepts of ethics are definitely not world-guided: 'discussions at the reflective level, if they have the ambition of considering all ethical experience and arriving at the truth about the ethical, will necessarily use the most general and abstract ethical concepts such as "right" and the concepts do not display world-guidedness'.3 Moreover, nothing in ethics corresponds to scientific explanations of our perception of secondary qualities, accounts which relate the way the world is to the ways it seems to us. Specific 'thick' ethical concepts — promise, courage, treachery - are analogous to concepts of secondary qualities in

32 Science, ethics, and objectivity 33 being both perspectival and world-guided.4 However, at the explanatory level the analogy breaks down. In the case of scientific accounts of perception of secondary qualities 'the explanation also justifies, because it can show how perceptions are related to physical reality, and how they can give knowledge of that reality, which is what they purport to do'.5 This is not the sort of question that is posed when we reflect on the thick ethical concepts in use in particular societies. Instead we raise ethical questions: about the quality of life in such societies, the justice of the societies, etc. Far from justifying the use of the perspectival ethical concepts, such reflections will frequently discredit them through critique of the social worlds of which they are constituents.6 Taken together these discrepancies imply a dramatic contrast between science and ethics. In science inquiry can, and often does, yield an accumulation of objective knowledge. In ethics inquiry can only reduce our knowledge - it can yield no new knowledge and its critique of pre-reflective concepts is apt to render existing ethical knowledge inaccessible to us. In this chapter I consider Williams' contrasts between science and ethics on the score of objectivity. These contrasts are, I suggest, exaggerated. Scientific as well as ethical inquiries are, I argue, liable to alienate us from pre-reflective knowledge; and in both science and ethics there is reason to doubt whether we can coherently hope that our inquiries will yield a world-guided convergence. I suggest that, though world-guidedness is not a tenable standard of objectivity, we should hold on to Williams' suggestion that the standard of objectivity be set by a type of convergence in belief whose explanation shows it to be the product of reliable processes. I outline an alternative type of convergence, precedent-guided convergence, which satisfies Williams' condition for a standard of objectivity. I argue that both in science and ethics we can coherently hope that our inquiries will yield knowledge that is objective by the standard of precedent-guidelines. I conclude with a suggestion about the ethical implications of acceptance of precedence-guidance as a standard of objectivity.

11 In contrasting the impact of scientific inquiry with the impact of ethical reflection on pre-reflective knowledge, Williams considers 34 NICHOLAS JARDINE three components of the conceptual repertoire of a traditional society: concepts of such familiar secondary qualities as colours, smells, and so on; concepts of magical agency and influence; specific 'thick' ethical concepts such as courage and treachery. Science, Williams suggests, deals cleanly with the concepts that fall within its domain. The scientific explanation of the use of secondary quality concepts justifies that use by showing how it yields knowledge of the physical world. The magical concepts, on the other hand, are discredited; for the beliefs in which they figure are, by the lights of science, straightforwardly false.7 By contrast, Williams claims, ethical reflection yields no such tidy partition of the ethical concepts of traditional societies. From our ethical standpoint, consideration of the uses of 'thick' ethical concepts will often yield a strong prima facie for holding both that their use does embody ethical knowledge, and that it is not an option for us because it is inextricably bound up with a social world that is for us both inaccessible and unacceptable. Consider a fourth component of traditional conceptual reper- toires, namely, 'folk taxonomies', the systems of classiflcatory concepts that they apply to plants and animals.8 Folk-taxonomies commonly resemble our scientific taxonomies in showing a measure of hierarchic structure; moreover, they generally involve some concepts roughly coextensive with classificatory concepts of ours — animal, elephant, tree, cat, etc. Taken as wholes, however, folk-taxonomies are often such as to call to mind Borges' Chinese encyclopedia which divided animals into '(a) belonging to the emperor, (b) embalmed, (c) tame, (d) suckling pigs, (e) sirens, (f) fabulous, (g) stray dogs, (h) included in the present classification, (i) frenzied, (j) innumerable, (k) drawn with a very fine camel-hair brush, (1) etcetera, (m) having just broken the water pitcher, (n) that from a long way off look like flies'.9 Like concepts of ordinary secondary qualities, the concepts which figure in folk-taxonomies are perspeeti- val and embody knowledge of the environment. Here, however, scientific explanation fails to provide justification. For the knowledge it attributes to the folk involves concepts that are not only perspectival, but also local - local in the sense that they apply only to a limited range of the earth's flora and fauna, and would break- down were the attempt made to apply them outside that range. Moreover, the concepts involved are generally local in the further sense that they are tied to the specific interests - culinary, medical, ceremonial, etc. that the folk have in the plants and animals in Science, ethics, and objectivity 35 question.10 Far from justifying the folk-taxonomic concepts, scientific reflection on them will often show their use not to be an option for us because it is inextricable from standpoints that we cannot occupy, and ways of life that we cannot pursue. The example of folk-taxonomies shows that ethical reflection is not, as Williams suggests, to be distinguished from scientific reflection by its propensity to cut us off from the local knowledge embodied in the conceptual schemes of traditional societies.

in According to Williams, the possibility of objective knowledge requires the possibility of a strong form of convergence, convergence on a body of truths 'which is brought about and explained by the fact that they are truths'. We can coherently hope for such world- guided convergence in science; indeed the occurrence of such convergence is attested by the history of science.11 In ethics, at least at the level of general theory, such convergence is inconceivable. General ethical theory is bound to make use of such abstract 'thin' concepts as good and right; and convergence in beliefs about what is good and right could not possibly be explained in terms of what is in reality good and right. Though I shall argue that in ethics as in science the notion of world-guidedness is conceptually problematic, I have doubts about Williams' case for the impossibility of world-guided convergence in ethics. To start with, he appears to overlook the possibility that ethical reflection could add to a community's repertoire of 'thick' ethical concepts, concepts which would be world-guided and hence allow addition to their stock of ethical knowledge. Suppose, for example, that 'industrious' had started life as a purely descriptive predicate, but that its habitual use had become value-laden as a result of moralistic reflections along the lines of the adage that the devil finds work for idle hands. Indeed, unless it is possible for ethical reflection to lead to value-loading of the habitual uses of predicates, it is hard to see how moral teachers and preachers have been able to change people's day-to-day ethical attitudes. Further, the forms in which Williams states the condition for world-guided convergence make it look much more obvious than it really is that the 'thin' theoretical concepts of ethics cannot be world-guided. Though he does not assert this explicitly, his account 36 NICHOLAS JARDINE makes it appear that world-guided convergence on the belief that p requires that p be a premise in the causal explanation of the convergence. This certainly debars such ethical claims as 'self-restraint is a virtue' from constituting objective knowledge, simply on the grounds that this is not the sort of statement that could possibly figure as a premise in a causal explanation. (On the same ground it also debars a host of well-founded scientific claims.) However, the condition is unnecessarily strong. The core of the notion of world- guidedness is the thought of a convergence on the belief that/) whose explanation shows the consensus onjfr to be reliably indicative that/>. This, however, requires only that p be a consequence of the premisses of the explanation, not that it be itself a premise. Given this weaker condition, it ceases to be entirely obvious that there could not be world-guided convergence on the belief that self- restraint is a virtue. Consider a quasi-Aristotelian account according to which a virtue is a disposition whose prevalence is necessary for the operation of those societies in which human well-being is opti- mised. Let us suppose, moreover, that exercise of self-restraint is a necessary condition for the operation of those societies which opti- mise human well-being. On such a naturalistic account we can envisage premises that both entail that self-restraint is a virtue, and that might figure in a causal explanation of convergence on the belief that self-restraint is virtue. Of course, world-guided convergence in the deliverances of ethical reflection is on the cards only if a naturalistic approach to ethics is valid. However, even a naturalistic approach cannot salvage the possibility of world-guided convergence in ethics unless such convergence is a genuine possibility at the level of general theory in the sciences. Is it? As noted above, Williams holds that the history of Western science manifests world-guided convergence. One way in which the history of the sciences might be thought to imply the occurrence, and a fortiori the possibility, of world-guided convergence has to do with observation and experiment. The history of the science surely shows, at least for some disciplines in some periods, that observation and experiment have played substantial roles in the achievement of stable consensus on theoretical issues. On standard causal accounts of perception and of the functioning of measurement instruments, this certainly implies that the theoretical deliverances of those sciences are strongly constrained by worldly states of affairs. Science, ethics, and objectivity 37 However, far more than this is required for world-guided convergence in theoretical beliefs. Consider our present consensus about the mass and charge of the electron. Our current scientific theories can doubtless provide us with causal stories which connect states of electrons to some of the experimental findings which played important roles in the genesis of that consensus: causal stories which show those experimental findings to be reliably indicative of states of electrons. Further, current perceptual psychology provides at least a part of the materials needed for causal explanations of our capacity to read the relevant instruments reliably. But what is required is a causal account which stretches all the way from states of electrons via experimental outcomes, their perception, interpretation, and publication, and thence via a whole international complex of debates and negotiations about instrumental and theoretical issues to our present consensus on the body of high-level theory in which these beliefs about electrons figure — an account which taken as a whole shows why our consensus on the charge and mass of the electron is reliably indicative of the charge and mass of the electron. Even in the case of 'straightforward' experimental findings in the sciences, the processes that lead to consensus - processes of calibration of instruments, of replication of experiments, of debate over the reliability and credibility of experimenters, of selection and presentation of findings for publication, etc. - are much more complex and protracted than one might be led to believe by the standard textbook accounts of famous and crucial experiments.12 It is far from clear what form causal accounts of such processes would take, or how such accounts could show the processes to be reliable. However, in the face of these difficulties, the notion of world-guided convergence preserves at least the appearance of coherence at the level of consensus on experimental findings. The realization that we cannot begin to say how the social processes that eventuate in consensus on such matters might constitute causally reliable processes damages, but does not destroy, the metaphor of whole communities of scientists as 'detectors' producing 'readings' in the form of consensuses. When, however, we move from formation of consensus about experimental findings to formation of consensus at the level of high theory, we lose our hold on the notion of world-guidedness altogether. Theories are certainly not established piecemeal by first detecting theoretical entities and properties, and then fitting the 38 NICHOLAS JARDINE results together. Yet a more holistic view of theory formation kills the metaphor of communities of scientists as detectors of theoretical states of affairs.13

IV Both in science and in ethics world-guided convergence is, I suggest, a notion that is of doubtful coherence at the theoretical level, and deeply problematic even at the level of perceptual judgments. However, in distancing ourselves from this false ideal of objectivity we should, I think, hang on to many of the assumptions that lead Williams to propose it. It is surely right, a mark of due respect for the subject, to insist with Williams on measuring the deliverances of ethical reflection by the most austere standards available to us. It is surely right to regard the history of the sciences as the primary source for our most austere standards, and to see those standards as having as their primary application the resolution of conflict. It is an important insight that an objective convergence of belief is one that is justified by its explanation. Moreover, Williams is surely right to insist that the justification required for objectivity is in terms of reliability. (His trenchant criticisms of the various forms of ration- alism in ethics do much to discredit the salient alternative, justification in terms of reasons.) Is world-guidedness the only way in which these stringent demands on a standard of objectivity can be met? Let us distinguish two ways in which a belief can be justified or vitiated by appeal to considerations of reliability.14 In a causal justification the methods or techniques which have engendered and sustained the belief are spelled out in causal terms in such a way as to demonstrate their reliability; conversely, in a causal vitiation the causal account of the relevant methods or techniques is such as to demonstrate their unreliability. In a. justification by precedent the reliability of the methods and techniques that have engendered or sustained a belief is demonstrated by testing against independently warranted precedents and standards; conversely, in a vitiation by precedent the unreliability of the relevant methods and techniques is demonstrated by testing against precedents and standards. Williams' world-guided convergence requires explanations of consensus which provide causal justification of beliefs accepted and causal vitiation of those rejected. Suppose, instead, that we are able Science, ethics, and objectivity 39 to explain convergence in such a way as to provide justification by precedent of the beliefs accepted and vitiation by precedent of the beliefs rejected en route to consensus. Let us call this precedent-guided convergence. Precedent-guidedness of convergence satisfies all Williams' basic requirements for a standard of objectivity. It is tempting, but mistaken, to dismiss such explanations outright on the grounds that they are inevitably circular.15 Any attempt to provide such an explanation for convergence on an entire all- encompassing world-picture would fail on grounds of circularity. For there would be no independently warranted beliefs to provide the bench-marks against which to test the reliability of the methods and techniques that had engendered and sustained the world- picture. But the same objection would apply to any attempt to use causal stories of world-guidedness to justify a total world-picture. For the causal stories would inevitably be part of the world-picture itself. However, when we turn from fantastic projects of explanatory justification of total world-pictures to mundane projects for piecemeal explanation and justification of fragments of a world- picture, circularity is no longer inevitable. We must, of course, observe certain precautions. The case for the reliability of a method or technique M should not depend on an application of M itself. Moreover, it should not depend on premisses whose warrant is seriously called in question when the reliability of M is called in question. In trying to show convergence on particular beliefs to be precedent-guided, care in the choice of precedents against which methods and techniques are calibrated should enable us to avoid these pitfalls. Is precedent-guided convergence something that we can coherently hope for both in ethics and in the sciences? Part of what is required for precedent-guided convergence is relatively unproblematic in both cases. Calibration of methods and techniques against independently warranted standards is a routine aspect of both scientific inquiry and ethical reflection. It is in operation in science when instruments are calibrated and when research strategies are promoted or attacked by appeal to their track records. Likewise it operates in ethical reflection whenever arguments whose validity is at issue in one context are criticised in terms of their acceptable or unacceptable results in other contexts. Much more problematic in the ethical case is the destructive achievement required of a precedent-guided convergence - the 40 NICHOLAS JARDINE vitiation by precedent that 'explains away' divergent and rejected beliefs as the products of unreliable procedures and methods. Williams' treatment of ethical divergence raises two main difficulties for such explaining-away of divergent ethical beliefs. The first difficulty arises when the beliefs in question are those of societies that are culturally very distant from our own - societies whose outlooks are not 'real options' for us.16 Then, Williams argues, the language of appraisal - 'good', 'bad', 'right', 'wrong', 'true', 'false', etc. - may be inapplicable. He contrasts such ethical divergence with the case of scientific theories radically divergent from our own. They too may not be real options for us; but that is precisely because the vocabu- lary of appraisal is applicable: 'Phlogiston theory is not a real option because it cannot be squared with a lot we know to be true.'17 It must be conceded that appraisal of the ethical systems of remote societies is deeply problematic. Over-preparedness to deliver negative judgments on an alien system of ethical beliefs may be indicative of cultural insensitivity or of cultural arrogance, a failure to realise that it may be we who should learn from them. The recognition of conflict between the alien beliefs and our own may be critically dependent on intractable issues of interpretation. There may be acute difficulties in deciding at what level it is appropriate to venture appraisals - at the level of particular ethical beliefs, at the level of systems of ethical belief, or at the level of entire social worlds. But all these difficulties have their parallels in the interpretation of historically distant beliefs in the sciences. Moreover, an appreciation of them does not help us to answer the obvious question raised by Williams' claim. How is it possible simultaneously to hold a set of beliefs, to recognise that an alien belief system is in conflict with it, and to deny that appraisal of the alien beliefs is even on the cards?18 It is tempting to answer this question by appealing not to problems of interpretation of alien belief systems, but to the phenomenon that is sometimes called 'dissociation'. Even when we have a good grasp of alien beliefs, those beliefs may remain baffling to us because the questions to which they constitute answers are, for us, unreal questions, questions which could not possibly arise. This may happen when the questions have presuppositions that we hold to be false. At a more fundamental level it may happen when they are such that we can envisage no conceivable way of getting to grips with them. When beliefs are in this sense dissociated from our own, it generally seems inappropriate to judge them true or false; for they Science, ethics, and objectivity 41 appear, in a phrase of Ian Hacking's, not even to be 'candidates for truth'.19 The claim that it is inappropriate to attribute truth or falsity to beliefs dissociated from our own is contentious.20 But, even if we do concede it, the phenomenon of dissociation does not give genuine support to Williams' position. For, in treating alien beliefs as dissociated from our own, we do not view them as being both in conflict with our own beliefs and immune to appraisal by us. On the contrary, in such cases we appraise the alien beliefs as 'unreal' or 'not even candidates for truth'. The second obstacle to explaining away divergent ethical beliefs is posed by Williams' claim that ethical reflection cannot 'generate an adequate theory of error and account generally for the tendency for people to have what, according to its principles, are wrong beliefs'.21 We have already dealt with this objection to the extent that it rests on an insistence that ethical reflection is not world-guided and hence cannot yield objectively true premisses for use in a theory of error. However, a more specific objection has considerable prima facie cogency, namely, that the resources of ethical reflection are too impoverished to provide a systematic theory of error. This objection may be reinforced in the spirit of Williams' approach by pointing to resources of the sciences in this area for which there are no evident analogues in ethics: explanation of erroneous belief in terms of malfunctioning or misuse of instruments; explanation of error in terms of abuse of statistical methods; explanation of error in terms of ignorance of relevant experimental findings; etc. This is, I think, a telling point. However, its force should not be exaggerated. There are, after all, certain basic sources of error to which both scientific inquiry and ethical reflection are routinely liable: factual misinformation; fallacious argument; prepossession by received ideas; etc. Sociological studies of the sciences - both those conducted by sociologists and those essayed by scientists out to discredit rival factions - suggest a variety of further sources of error that undoubtedly afflict ethical inquiry as well: susceptibility to appeal to vested interest; sensitivity to rhetorical and aesthetic blandishments; 'blind' deference to authority. In addition, there are types of vitiation of ethical beliefs that are without obvious analogues in the sciences. Thus social explanations of the ways in which institutions function, and of the ways in which they legitimate themselves, can provide bases for ethical critique. For example, in 42 NICHOLAS JARDINE the manner of critical theorists of the Frankfurt School, we may attempt to vitiate ethical beliefs and dispositions by showing them to be, as Williams neatly puts it, 'merely effects of the power they are supposed to legitimate'.22 Even on the most optimistic estimates of the powers of critical theory, there can be little doubt that the resources for a theory of ethical error are substantially poorer than those customary in the sciences. But they are surely not so impoverished as to rule out the very possibility of explaining away alternative ethical beliefs as the products of unreliable methods and procedures. I have argued that we can coherently hope for precedent-guided convergence in ethics — that is convergence whose explanation provides justification by precedent of beliefs accepted and vitiation by precedent of beliefs rejected en route to consensus. That is enough to make objective ethical knowledge possible.

I have challenged the contrast that Williams draws between science as a domain in which inquiry can, and often does, yield objective knowledge, and ethics as a domain in which inquiry cannot, in principle, yield objective knowledge. In science as in ethics, I have claimed, it is doubtful whether we can coherently hope that our inquiries will yield substantial bodies of knowledge that are objective by Williams' world-guidedness standard. I have proposed an alternative standard, precedent-guidedness of convergence, which satisfies the eminently sensible demands Williams makes on a standard of objectivity. In ethics as in science we can, I have argued, coherently hope that the deliverances of our inquiries and reflections will measure up to this alternative standard of objectivity. Coherent hope is one thing, realistic hope another. In the sciences we have, I think, ample grounds for claiming that precedent-guided convergence often occurs. Not so in ethics. Here the historical record manifests little enough convergence, let alone of the precedent- guided sort; and the relative poverty of theories of ethical error provides a substantial further ground for pessimism about the prospects for precedent-guided convergence on a substantial body of ethical theory. It may well seem that the criticisms I have ventured, even if well founded, do remarkably little to disturb the overall arguments of Science, ethics, and objectivity 43 Ethics and the Limits of Philosophy. For the insights about objectivity that Williams uses to underwrite his world-guidedness standard are satisfied by my precedence-guidedness standard. And, hardly sur- prisingly given their similarities, the two standards yield similarly dismal conclusions about the prospects for attainment of a substantial and general body of objective ethical knowledge through ethical reflection: nil in principle by Williams' criterion, poor in practice by mine. Does it really matter whether the attainment of a substantial body of objective knowledge through ethical reflection is impossible in principle or merely unlikely in practice? On at least one important score I think it matters a great deal. In a brief but suggestive passage Williams introduces the notion of ethical confidence, the disposition to live and act confidently in accordance with certain of the values prevalent in one's society.23 The type of ethical confidence of which Williams approves is not an uncritical disposition, for it gives practical endorsement only to values that stand up to scrutiny of their psychological and sociological pedigrees. Ethical confidence is presented as an alternative to two other ethical stances, the ethics of conviction and the existential ethics of decision. Both the ethics of conviction and the ethics of decision are, Williams claims, undermined by the demonstration that ethical reflection cannot yield objective ethical knowledge.24 Not so ethical confidence. The confident stance is surely compatible with the belief in the impossibility in principle of reflective attainment of objective ethical knowledge. Indeed, it may even be fostered by the belief, at least in the small class of intellectuals who might otherwise have hoped, or feared, that the objective deliverances of moral philosophy could discredit old values and lend credence to new ones. Contrast this with the implications of the straightforwardly sceptical belief to which the precedent-guidedness condition of objectivity is conducive: that our values may embody objective ethical knowledge, but for the most part probably do not. Such scepticism I believe to be healthy, not least because it militates against the whole range of 'strong' and potentially dangerous ethical stances: ethical conviction, ethical decisionism, and ethical confidence. 44 NICHOLAS JARDINE

NOTES 1 B. Williams, Ethics and the Limits of Philosophy, London: Fontana/Collins, 1985. P- !36- 2 Ibid., pp. 138-9. 3 Ibid., p. 152. 4 Ibid., pp. 140-50. 5 Ibid., p. 150. 6 Ibid., pp. 150-1. 7 Ibid., p. 145. 8 On folk taxonomies see, for example, R. F. Ellen and D. Reason (eds.), Classifications in their Social Context, London: Academic Press, 1979. 9 Cited in Michel Foucault, The Order of Things, London: Tavistock, 1970, p. xv. 10 Following Durkheim and Mauss it has often been supposed that folk taxonomies are local in the further respect that they embody symbolic representations of the familial and social structures of the folk who employ them: E. Durkheim and M. Mauss, Primitive Classification (1903), translated and introduced by R. Needham, London: Cohen and West, 1963. This thesis remains the subject of heated controversy amongst anthropologists. There are good grounds for denying the coherence of strong versions of the thesis according to which folk taxonomies have social structures rather than plants and animals as their real objects: see J. Skorupski, Symbol and Theory, Cambridge University Press, 1976. 11 ELP, pp. 136-7. 12 On the complexity and protraction of the social negotiations that lead to closure of debate concerning experimental findings in the sciences see, for example, H. Collins, Changing Order. Replication and Induction in Scientific Practice, London: Sage, 1985; S. Shapin and S. Schaffer, Levia- than and the Air-Pump. Hobbes, Boyle, and the Experimental Life, Princeton University Press, 1985. 13 I raise some further difficulties for the ideal of world-guided convergence at the level of high theory in The Fortunes of Inquiry, Oxford, Clarendon Press, 1986, ch. 8. 14 'Reliabilist' epistemologists, who say that it is reliability, not evidence for reliability, that constitutes justification, are at liberty to substitute some other term here — 'vindication', perhaps. 15 R. Firth, for example, rejects on grounds of circularity all attempts to demonstrate reliability by appeal to precedents: 'Epistemic merit, intrinsic and instrumental', Proceedings and Addresses of the American Philosophical Association, 55 (1981), 5-23. He is effectively answered by J. Heil, 'Reliability and epistemic merit', Australasian Journal of Philosophy, 62 (1984), 327-38. 16 ELP, pp. 160-2; see also 'The truth in relativism', Proceedings of the Aristotelian Society, 75 (1974-5), 215-28. Science, ethics, and objectivity 45

17 ELP, p. 162. 18 This question is forcefully raised byj. McDowell in his critical notice of ELP in Mind, 95 (1986), 377-86, p. 384. 19 Ian Hacking, 'Language, truth and reason', in M. Hollis and S. Lukes (eds.), Rationality and Relativism, Oxford: Basil Blackwell, 1982, pp. 48-66.1 discuss dissociation in the history of the sciences in The Scenes of Inquiry: On the Reality of Questions in the Sciences, Oxford: Clarendon Press,

20 In particular, William Lycan has produced powerful arguments for the view that in what linguists adduce as the central cases of presupposition failure the assertions at issue do have truth values: Logical Form in Natural Language, Cambridge, MA: MIT, 1986, ch. 4. 21 ELP., p. 162. 22 Ibid., p. 166; cf., the remarks on p. 199 on 'transparency' of institutions and social practices. 23 Ibid., pp. 170-1. 24 Ibid., pp. 168-70. CHAPTER 4 Fallibilism and objectivity: science and ethics Christopher Hookway

I. INTRODUCTION Debates about whether evaluative claims (particularly ethical ones) meet standards of objectivity set by such paradigms as scientific theory and ordinary empirical statements used often to dwell upon the supposed fact that irresoluble disagreement was endemic in ethical discussion, but rare in scientific debate. In so doing, they exploited the observation that we take it to be one of the marks of a statement having an objective truth value that an unforced consensus can be secured concerning which truth value it has. It is now generally recognized that formulating the question in this way distorts the position in both science and ethics. It seems to be at best a hope that apparently irresoluble differences over scientific theory will be resolved by rational means. And consensus over ethical matters is more extensive than such debates admitted: although the plurality of values may produce irresoluble differences over what is to be done in particular circumstances, dissension is rarer over whether an act is cruel or kind, or a person vindictive or generous. In chapter 8 of Ethics and the Limits of Philosophy, 'Knowledge, Science, Convergence', Bernard Williams argues that it is still possible to employ the notion of convergence of opinion to mark a fundamental difference between ethical beliefs and the products of scientific inquiry. The difference he discusses does not concern whether such convergence will be achieved; scientific research and ethical reflection both hope to arrive at opinions which will be endorsed by the independent findings of other people. Williams' distinction turns instead upon how the hoped for convergence of opinion is to be explained. Explaining the claim that a belief is descriptive of reality by reference to whether it could become the object of an unforced

46 Fallibilism and objectivity: science and ethics 47 consensus is associated with Charles Peirce. He was aware that a consensus (albeit a potentially unstable or temporary one) could be secured through unsatisfactory means: agreement resulting from the arbitrary whim of an intellectual dictator or from the possibly contingent fact that we all find the same propositions 'agreeable to reason' does not describe the character of'reality'. Scientific investigation is guided by the aim of arriving at a consensus which is explained by the facts that 'realities affect our sense according to regular laws', and that 'by taking account of the laws of perception, we can ascertain, by reasoning, how things really are'. In other words, we seek a consensus on a proposition p where the fact that p enters into the explanation of the consensus.1 Williams' claim, very roughly, is that such an aim is appropriate in scientific investigation, but not in ethical inquiry:

In a scientific investigation there should ideally be convergence on an answer, where the best explanation of the convergence involves the idea that that answer represents how things are; in the area of the ethical, at least at a high level of generality, there is no such coherent hope. The distinction does not turn on any difference in whether convergence will actually occur, and it is important that that is not what the argument is about. It might well turn out that there will be convergence in ethical outlook, at least among human beings. The point of the contrast is that, even if this happens, it will not be correct to think that it has come about because convergence has been guided by how things actually are, whereas convergence in the sciences might be explained in that way if it happens. (Williams 1985, p. 136)

At first sight, this seems a promising way of articulating the attractive thought that science seeks to attune itself to the structure of reality in ways in which our reasoning about ethical matters does not. Williams goes further, drawing the seemingly sceptical conclusion that, whereas 'science has some chance of being more or less what it seems, a systematised theoretical account of how the world really is', 'ethical thought has no chance of being everything it seems' (p. 135). Williams' contention seems to be that, while theoretical and methodological reflection can contribute to the growth of scientific knowledge, in ethics, ''reflection can destroy knowledge1 (p. 143). Interestingly, Peirce himself distinguishes between inquiry in science and ethics in a similar way. He allows that, for certain purposes, the aim of (for example) arriving at an opinion which is 'agreeable to reason' may be more appropriate than that of finding 48 CHRISTOPHER HOOKWAY an opinion which can be the object of a final settled consensus meeting the condition described. That the latter goal is appropriate primarily for science is clear from a later series of lectures, 'Detached ideas on vitally important topics', where he scorns the suggestion that logical reflection or systematic philosophy can yield satisfying response to live or controversial ethical questions. The view that a range of beliefs purports to represent how things are, and the related suggestion that their subject matter is objective, makes intelligible the possibility that the beliefs are fallible. If our methods for acquiring beliefs are unreliable, or if they are imperfec- tly implemented, or if relevant evidence is not available to us, we can arrive at beliefs that are false. One corollary of the distinction Williams defends should be a difference in the role of thought about our fallibility in connection with scientific beliefs and ethical ones: we see ways in which scientific beliefs are fallible to which nothing corresponds in the ethical case; or we see no grounds for hoping that our fallibility can be overcome in connection with ethical beliefs, but experience no such pessimism about science. Since reflection directed towards estimating our fallibility and deciding how to respond to it in particular cases seems an inevitable consequence of the recognition that we are fallible, the source of the connection Williams appeals to between views about objectivity, views about the role of reflection, and the possibility of various kinds of scepticism is clear. We should start from a striking empirical difference between inquiry in science and ethics: the latter has nothing corresponding to the institutional framework for relatively disinterested research which characterises theoretical science. Organized international communities of physicists, for example, are absorbed in urgently seeking solutions to problems whose origins lie within the tradition of theoretical inquiry itself. Moreover, there is substantial agreement about what would count as an acceptable answer to questions under consideration, and about what represents a challenge to accepted views. Reflection on ethical matters becomes urgent only in response to ethical problems that have become pressing in practice: there is no established tradition of extending our moral responses by constructing situations which display the strains in our current ethical views; little value is seen in exposing those strains and tensions. And there is little consensus about when a response to a problem of ethics is acceptable. While this is to be expected if Fallibilism and objectivity: science and ethics 49 Williams' claim is correct, it offers only weak confirmation of his position. Alternative explanations of the difference are available: ethical thought is at a primitive stage in its development; various philosophical projects propose to initiate such research traditions; or this is not a necessary feature of a discipline guided by the scientific paradigm of objectivity. Peirce's characterisation of the aims and methods employed in science focuses on one important feature of such a tradition of theoretical inquiry: investigators are 'distanced' or 'detached' from their current scientific views. Participating in science involves a willingness to postpone current, short-run, settlement of opinion for an opinion which will be genuinely and permanently settled: the method of science makes no promises about the speed of convergence. We view our contributions as located within a co- operative search for the truth, and as of value for the sake of the contribution they make to this wider inquiry. For Peirce, the value we see in our scientific activities does not rest upon believing they will lead either to our grasping the truth or to the truth being reached in our lifetimes. Although there are hypotheses or background beliefs which we assert during our scientific activities, and although we may claim to 'believe' these things, such 'belief and 'assertion' may not involve a commitment to truth, but merely the sense that holding to those propositions at this stage contributes to our steady progress towards the truth.2 Although the search for objective truth in natural science may call for an inquiry to which we take the detached attitude, perhaps other disciplines achieve similar objectivity without deferring to the future in this way. If such deference is likely wherever a scientific conception of objectivity is in play, then we might find a basis for locating a deep difference between science and ethics. Peirce certainly holds that recognition of our fallibility in exploring an objective subject matter will prevent our current opinions serving as live certainties; it encourages a picture which calls for a more tentative or 'distanced' form of acceptance coupled with a deference to future inquirers who may correct our views. And this is possible only within a community of inquirers whose understanding of the sources of error is sufficiently sophisticated that we can reasonably hope that the progress of inquiry will overcome our fallibility. Peirce's scepticism about the role of reflection in connection with 'Vital matters' rests upon the belief that tentative acceptance which defers to the future has no 50 CHRISTOPHER HOOKWAY authority over spontaneous or common-sense ethical reactions whose value is vindicated by their ability to enable us to enter into a harmonious relation with our surroundings. A deeper scepticism can question whether an 'ethical community' meeting these conditions is possible at all. This chapter examines how far we can go along with Williams' attempt to distinguish science and ethics. The following section distinguishes two theses about the convergence of opinion, and elaborates some difficulties in explaining the difference in the way Williams proposes. The remainder of the paper explores the distinctive claims about science and ethics that he makes in arguing for his position. Section 3 disputes some of his views about convergence in science and the role of the 'absolute conception of the world' in characterizing the goals of scientific inquiry; and section 4 challenges his understanding of how an ethical position could be developed which insisted upon a parallel between the two. Section 5 attempts to arrive at a conclusion similar to Williams' own, but by a rather different route. Ethics and the Limits of Philosophy discusses a wide range of interlocking themes and I am conscious of having failed to take account of all of Williams' contributions to the topics I am discussing: I hope that this does not destroy the interest of the following argument.

2. SCIENCE AND CONVERGENCE: TWO THESES We can distinguish two claims about convergence. /. Local convergence: If S expresses a definite proposition which is true, then, any inquirer who understands that proposition and who inquires into its truth for long enough, employing sound methods of investigation, and who is not handicapped by being spatially or temporally prevented from gathering evidence necessary for investigating the truth value of that proposition, would come to believe the proposition expressed by S. II. The aim of science: The aim of scientific activity is to progress towards a body of opinion which can be accepted by all rational agents who adopt the goal of seeking the truth in a disinterested spirit and which represents itself as accepted because it is true. It seems clear that / and // are independent. If the identity of a Fallibilism and objectivity: science and ethics 51 proposition (the question raised when we ask whether it is true) depends upon the background knowledge determining one's understanding of it, then (even in a scientific context) local convergence could occur for a proposition that will not feature in any ultimate consensus. //, the 'aim of science', is the source of detachment; although /may allow that such judgments are generally fallible, this only provides grounds for detachment if we always have an eye on future developments. Hence we might reasonably hope for /-type convergence in our ethical judgments although no ethical ideal corresponding to // drives our ethical thinking. So long as the common-sense presumptions which guide our moral responses contribute to fixing the identity of the questions asked in moral 'inquiries', it seems reasonable to suppose that this hope could be fulfilled. We could look on / as relating a kind of internal convergence to our understanding of truth, and // as a kind of unlimited external convergence to our idea of science. If'scientific objectivity' is to be available in ethics without ethical opinion having a detached character, we may be required to retain / in ethics, while denying the relevance to ethical reflection of the ideals reflected in //. These remarks oversimplify the situation in both scientific and ethical inquiry. For example, most scientific realists support accepting current theories in the hope that subsequent investigation will reveal that they were at least approximately correct. And there is no absurdity in the idea of an ethical outlook characterized by a vivid sense that our current moral views might be unmasked as confused, prejudiced, or otherwise mistaken by further developments. In spite of this, there is value in using this distinction to sharpen our sense of how inquiry in science and ethics differs, and to use this to question Williams' framework for explaining the difference. In the early writings in which his theory of truth is developed, Peirce gives the impression that the method of science has no role in connection with vital matters: for these, we happily rest with what is 'agreeable to reason'. David Wiggins has recently disputed this suggestion, offering a characterization of the goals of ethical investigation intermediate between such an unconstrained appeal to what seems right and the scientific demand that our beliefs be fixed 'by nothing human, but by some external permanency - by something on which our thinking has no effect'.3 In evaluating ethical beliefs, we are sensitive to the demand that a belief be accepted only if it is 'determined by causes that are not accidental relative to its content'. 52 CHRISTOPHER HOOKWAY Although ethical beliefs and methods can be criticized, our current ethical concepts, which fix the contents of these beliefs, provide the boundaries within which such investigations develop: we are not guided by a readiness to transcend our ethical perspective in pursuit of an ultimately stable set of opinions. Such a view would accord with the possibility just described: it insists on the role of critical reflection in the search for local convergence; but it denies the appropriateness of valuing our ethical reasonings for the contribution they make to an ongoing tradition of ethical inquiry. Although it is not to our current purpose to discuss how far Peirce could have endorsed this suggestion, it is relevant to point out that, if it becomes controversial how an ethical belief is caused, or it is contested which kinds of causation are 'not-accidental relative to its content', reflection could lead to our becoming distanced from our ethical beliefs even on this more modest view.4 More immediately relevant is the degree of illumination cast upon Williams' thesis if we distinguish science and ethics in this fashion. Does it entail that science aims at a convergence of opinion guided by how things actually are, while ethics does not? Much turns on the content of the claim that convergence is 'guided' or 'explained' by reference to how things 'actually' are. If the claim is just that a scientific explanation of our making of moral judgments cannot show that they track the truth because evaluative predicates have no place in pure science, it may seem that questions are being begged. We now exploit these distinctions drawn in order to show that the distinction Williams makes is less obvious, and less easy to draw, than we might suppose. This discussion is partly unfair to Williams since I am ignoring the terms he himself uses to draw the distinction; I hope that the benefits of doing so will emerge sub- sequently. We shall consider Williams' thesis as a claim about, first, / convergence, and, second, // convergence.

(a) I convergence We have little trouble comprehending the thought that, where p is some true definite proposition, then if it had not been true, careful inquirers who understand the proposition in question, would have been prevented from asserting it. Within the context which fixes the question raised by asking whether p, we can find room for the Fallibilism and objectivity: science and ethics 53 thought that we can track the truth in our methods for deciding whether p is assertable. Moreover, guided by uncontroversial standards of evaluation, we can allow our evaluations to be defeated by discoveries that are paradigmatically factual. So long as / convergence is what we are concerned with, it may seem hard to find a radical asymmetry between the position of the scientist investigating a proposition against a background of presumed theory and background knowledge and that of someone employing evaluative predicates against a background of common-sense standards of ascription and evaluation. It can enter into the explanation we offer of a proposition being asserted that that proposition is assertable: we find an act cruel or despicable because it is cruel or despicable. So long as our ethical concepts provide us with what are seen as reasons or criteria for applying these ethical terms, so that convergence results from the satisfaction of those criteria, it is not absurd to describe the resulting convergence in these terms. It may be another expression of this thought that, in David Wiggins' phrase, we find that 'there is nothing else to think on the matter'.5 If/convergence is all that is at issue, it is hard to see what basis there is for distinguishing science and ethics.

(b) II convergence Suppose that science, unlike ethics, is guided by an aspiration to // convergence as well: this reflects a goal governing any reflective participation in the activity. This, too, seems unable to help us to formulate Williams' point in the terms he favours. Compare: (i) My asserting that/) is best explained by reference to the fact that p is actually true. (ii) My asserting that/) is best explained by the fact that doing so at this time is proper, given the current state of scientific understanding, as a contribution towards our eventual understanding of the fundamental laws governing the universe. From the standpoint of//convergence, (ii) is what is relevant to my current understanding of my assertions. The phenomenon of 'detachment' or 'distancing' suggests that (i) is not required by // convergence at all. At best I aspire to contribute to reaching a point in which (i) can be asserted; but that says nothing about my current cognitive position. Moreover, as stressed above, unless 'actually true' is given a special significance, then, in many cases, / con- 54 CHRISTOPHER HOOKWAY vergence provides us with (i) anyway. Hence, although there may be this major difference between ethics and science, that only the latter is guided by an aspiration to // convergence, it is unclear as yet that this is usefully expressed as a claim about how convergence is to be explained. // convergence reminds us that our current scientific conceptions are likely to be superseded as inquiry progresses; hence (i) above has a limited role and our attitude towards our current 'beliefs' is detached. Detachment is mitigated: in some measure we are attached to current scientific views, and we often take them to be true. However, likely explanations of that do not help us to make Williams point at all clearly. We might point to successful applications, defending beliefs prag- matically because they yield accurate predictions which are relevant to our purposes. Although this involves something other than a detached attitude towards the beliefs in question, it fails to ground the sort of claim about explanation which interests Williams. In so far as the assertion is made by a scientist with the detached attitude, pragmatic success is, at best, a weak indication of truth. Due to our limited experience and to the incomplete state of the theoretical background to the assertion, we can expect that we currently possess only a partial grasp of generalizations with wider scope and more explanatory force. In so far as we simply treat the assertion as limited in its scope to the range of phenomena to which we apply it, then our attitude towards it is that of the engineer rather than that of the scientist, /convergence is all that is at issue; and the position is not different from the employer of ethical concepts whose acceptance of his system of evaluations is grounded in their success in enabling him to order and systematize his evaluative responses. Secondly, we can be committed only to the approximate truth of the assertion, confident that subsequent theory will explain its current appeal to us by reference to the truth of some broader generalization over a limited range of phenomena. 'Approximate truth' introduces a host of problems which must be side-stepped here. Two main points should be stressed. First, unless science is to discard its achievements as it progresses, we should expect it to endorse the approximate correctness of earlier theories, explaining the empirical regularities which have been uncovered, and describing the partial perspectives which have been a source of error. Hence we may expect the current theories, suitably qualified, to be Fallibilism and objectivity: science and ethics 55 preserved as something to be saved within new theories. But, if this sort of attachment to current scientific theories is what is at stake, it is unclear that, once again, we can discriminate science and ethics. For, in each case, we expect that any development in our concepts will endorse the approximate correctness (or intelligible erroneous character of) our current application of them. We tend to concep- tualize changes in our ethical thinking by discovering that, at an earlier stage, we were blind to a relevant defeating or supporting consideration which, in some sense, is now deemed relevant in virtue of kinds of considerations that were deemed relevant before.6 The second point forces us to recognize an element of idealization in our distinction between convergence / and convergence //. The local convergence thesis (I) applied to 'definite questions'. It is plausible that most scientific theories employ concepts that are vague: our present acceptance of such a theory amounts to no more than the assurance that some precisification of it will be successful as inquiry proceeds; current acceptance involves a nod towards the future. In that case, a current theory may be accepted under the understanding that some precisification of it will feature in a stable consensus. This will be important for subsequent discussion: in so far as the presence of vague concepts in formulations of our current views does introduce a 'nod towards the future', they suggest a limitation on the possibility of maintaining the objectivity of an area of discourse by insisting that /convergence is all that is involved. So far, our attempt to describe some assumptions about convergence which inform our thoughts concerning science and other forms of inquiry have not enabled us to arrive at a formulation of Williams' claim which is either plausible or defensible. This is unsurprising, because we have failed to take into account a conception which is central to his exposition, the 'absolute conception of the world'.

3. THE ABSOLUTE CONCEPTION OF THE WORLD The 'absolute conception of the world' involves a way of thinking about a world that is 'there anyway, independent of our experience': In reflecting on the world that is there anyway, independent of our beliefs, we must concentrate not in the first instance on what our beliefs are about, but on how they represent what our beliefs are about. We can select among our beliefs and features of our world picture some that we can reasonably 56 CHRISTOPHER HOOKWAY claim to represent the world in a way to the maximum degree independent of our perspective and its peculiarities. The resultant picture of things, if we can carry through this task, can be called the 'absolute' conception of the world. So when Williams claims that, unlike science, ethics does not hope to arrive at beliefs which are held because they describe 'how things actually are', there is in the background the assumption that 'how things actually are' is determined by reference to this absolute conception. For Williams, this is offered as an elucidation of an ideal which is implicit in scientific practice. As he says, it provides a distinction between 'the world as it is independent of our experience' and 'the world as it seems to us' by clarifying the latter as 'the world as it seems peculiarly to us' (p. 139). The absolute conception is an account of reality available to any investigator, even those whose perspective is radically different from our own; and Williams insists that the absolute conception will contain materials for explaining how different perspectives upon reality are possible. There are several routes from this characterization of science to the conclusion that ethical judgment does not employ a comparable or similar conception of objectivity. For example, scientific investigation may explain our practice of ethical valuation by showing that it inevitably rests upon a distinctive, 'peculiar' perspective; or the aspiration to transcend such peculiar perspectives through reflection and inquiry could be shown to be destructive of our claims to ethical knowledge. However, since the significance of such demonstrations depends upon accepting Williams' view about the role of the absolute conception in science, we shall begin with some critical remarks. To this end, we should distinguish two themes. Firstly, suppose we agree that scientific inquiry is characterized by a hope (and an ability) to transcend limitations of perspective and arrive at conclusions that reflect how things actually are: is this hope best characterized in terms of the absolute conception of reality? Does that conception provide the appropriate distinction between appearance and reality? Secondly, and relatedly, is it correct to assign to the absolute conception a role in characterizing the aims of science? I shall question whether Williams' response to these questions is as obvious as he supposes. The claim that a field of inquiry has the ability to discover how things actually are suggests that it is able to overcome possible sources of error. We might gloss this: if it appears to an inquirer that Fallibilism and objectivity: science and ethics 57 p, then, if it is not true that p, further inquiry will reveal that this appearance is in error. It is also natural to claim that error can result from peculiarities of individual perspective: if an inquirer is prevented from gathering relevant evidence, or if he is in the grip of a mistaken theory which distorts interpretation of evidence, then error will emerge. Use of a distinction between appearance and reality, like reference to the peculiarities of individual perspective, can have a role in articulating ways in which inquirers are fallible and in capturing the hope that our fallibility can be overcome. The contrast between appearance and reality which enters Williams' characterization of the absolute conception is distinct from this epistemological one. The objection to a peculiar perspective is not that it is a source of error or that it fails to provide knowledge. Rather, the knowledge it provides falls short of a kind of generality to which science, properly so-called, aspires. The observation that a field of knowledge is essentially perspectival in this sense has sceptical consequences only if inquiries in this area aspire to this kind of generality. Hence, issuing from a peculiar perspective is not inevitably a source of fallibility in inquiry - unless we take being a potential source of error as requisite for a perspective being peculiar. If the latter proviso is accepted, then room is allowed for knowledge which is tied to perspectives which are particular, but not peculiar. This leads to our question about the role of the absolute conception in fixing the aims of science. Williams acknowledges the absurdity of the picture of scientists adopting the goal of discovering the way the world actually is, unless that goal is given a more substantive content. The 'absolute conception' is intended to provide this fuller specification of the goal. I am unconvinced that it does so. Distinguish three elements in Williams' claims about the role of the absolute conception in science: (i) It offers a description of reality whose acceptance does not depend upon occupying any 'peculiar' perspective: it can be endorsed by any rational investigator. (ii) Scientific knowledge contains materials for explaining how we acquired scientific knowledge. The explanation provided shows that we believe these scientific propositions because they are true. (iii) Science explains how the knowledge which reflects our 'peculiar perspectives' results from the interplay of perspective and the character of reality. 58 CHRISTOPHER HOOKWAY Our task is to understand the role of these ideals in the activities of the community of scientific investigators. Compare (i) and (ii) with two weaker claims. If we are not to lose predictive control, or understanding of general principles, any theory which has seemed plausible should not be abandoned unless we either can, or hope to be able to, explain why it was as successful as it was. Since we think of science as a tradition, and as the product of a community, the same observation holds interpersonally: theories held by others should not simply be discarded if we see no hope of explaining the degree of their success. It is plausible to see such a regulative principle recommending pursuit of theories which can accommodate all perspectives which have a place within the scientific tradition. Secondly, although opposing schools within the scientific tradition can be viewed as in competition, this is a superficial phenomenon: it is a criterion of membership in such a tradition that a value is attached to arriving at a body of explanatory theory whose adequacy can be acknowledged by all; once the overall adequacy of a body of theory is established, rivals tend to wither away. If this weaker characterization supports the contention that science is likely to advance towards an 'absolute conception', it differs from Williams' bold claim in two ways. First, the regulative ideals employed are characterized in terms of one's relations to other members of a community of inquirers, rather than by references to what might be acknowledged by any rational investigator. And second, convergence upon an analogue of the 'absolute conception' need not be introduced as the overriding 'aim of science'. It could be the consequence of a sequence of actions and inquiries directed towards much more short-term goals: retaining the insights of one's own earlier views; and recognizing the insights of others in order to maintain the links required for membership of the same cognitive community. I acknowledge that this is hardly less of an idealization than Williams' description of scientific practice. I hope that its merits will emerge when I turn to the comparison with ethics. A second dimension of Williams's characterization of the absolute conception was that it provided materials for explaining our own successes in pursuing it. This partly reflects the regulative principle described above, that successor theories should be expected to retain, and hence explain, the successes of theories which are rejected. It also, presumably, suggests that developments in cognitive science or biology would enable us to assess, and explain, the Fallibilism and objectivity: science and ethics 59 reliability of our cognitive apparatus. This has the corollary that our understanding of the sources of scientific error and of the sources of our fallibility grows with the progress of science. It is a guiding hope that such progress will not prompt the sceptical worry that such is the extent of our fallibility that we cannot trust the products of our own inquiries. Since a major difference between science and ethics will turn out to be that this guiding hope is hard to sustain in the latter, the most promising route for resisting Williams' claims here is to emphasize the limits of this sort of explanation in science. Even here there are general conceptions of explanatory intelligibility, or broad assumptions about the structure of the world, or basic statements of mathematics, such that the question how we know these things has no determinate answer. They are so deeply embedded in all of our ways of finding things out, that a description of particular processes that underwrite our reliance on them, or particular evidence war- ranting our use of them, seems out of place. Rather: anything that could put them in question would be so much less certain than they are themselves that they are effectively indubitable; 'everything counts for them and nothing counts against them'; 'there is nothing else to believe'. Beyond saying that if they had not been so, we should not have believed them, little that is specific can be said to ground our assurance that relying upon them is not a manifestation of our cognitive fallibility. A consequence of these remarks will be important for subsequent discussion. It bears on the remarks about vagueness in section 2. Compare two schools within a discipline which are developing different precisifications of a single vague picture of reality. In one sense these are in competition: and 'vital' matters concerning the institutional trappings of science could make winning very important to members of each school. But this competition is mitigated: according to our idealized pictures of science, the final 'victory' of one of the groups should be regarded as a triumph for science rather than as a defeat for the opposing group. Such ideals may prescribe that the losing group should welcome the scientific progress that their activities have provoked; not least, because their successes should be preserved and explained within the newly dominant framework. 'Detachment' from one's beliefs, 'deference' to the future and to the scientific community, and the structure of the community of scientific investigation, combine to require partici- 60 CHRISTOPHER HOOKWAY pants to value generalizations which preserve the achievements of all groups, and which can be a matter of consensus above the endorsement of any particular theories.

4. OBJECTIVISM AND ETHICS Much of Williams' discussion of how ethics differs from science focuses on difficulties facing what he calls an 'objectivist' conception of ethics. Someone who views his ethical concepts in an objectivist manner sees himself, and other members of his community, as trying, 'in their local way, to find out the truth about values, an activity in which we and other human beings, and perhaps creatures who are not human beings, are all engaged' (p. 147). This would appear to involve viewing communities whose values differ from our own, or others within the community with opposed ethical standpoints, as, so to speak, rival research traditions. As should be clear, this perspective raises problems about an inquirer's undistanced endorsement of his values. He should adopt an agnostic attitude towards the question of how to live: and cannot prize his own ethical framework as having a serious claim to represent the truth about values. In general, an 'objectivist' can only investigate values in a responsible manner if he possesses some reflective understanding of what such inquiries aim at, and of how they should be conducted. The relations between this conception and the absolute conception are clear: just as it is a mark of the objectivity of science that it aims at a body of knowledge which any open-minded inquirer could accept, so it is a mark of the objectivity of values that we seek a body of knowledge which only the hopelessly prejudiced or deceived could deny. We might contrast this perspective with that of a largely unreflective community who employ a range of ethical concepts, which are largely effective in enabling members of the community to live in ways that they find satisfactory. Nothing in their use of these concepts forces them to reflect upon their inadequacies or to try to amend them. There seems to be no reason to suppose that such a community are implicitly in the grip of the 'objectivist' picture. However, if their values differ from those of an 'objectivist', they present an epistemological challenge to the latter which he does not provide for the unreflective community. For they prompt the question whether their unreflectively accepted values might not be closer Fallibilism and objectivity: science and ethics 61 to the 'truth about values' than the products of the objectivist's careful inquiries. Williams clearly doubts that the question thus prompted can be resolved in a principled fashion; and, indeed, awareness of the variety of competing systems of values can lead to alienation from one's own ethical outlook. I think that he has the picture that unless we deny that inquiry in science and ethics are analogous, we are committed to an 'objectivist' picture of values. In so far as he asserts that ethics has no chance of being what it seems, this 'seeming' must stand in some sort of relation to the 'objectivist' conception. The suggestion that ethical reflection might distance me from my own ethical views, shaking my current confidence, does not seem to represent a contrast between science and ethics. We might have focused on the lack of such detachment as a more central focus for drawing a distinction. Moreover such distancing may not arise: the analogy with science allows for our attachment to our own current values being understood as ethical engineering. So long as our ethical conceptions are 'successful', we are not shaken by the observation that others use different conceptions or that we lack the desired theoretical justification for them. What does shake our confidence is encountering moral challenges which we cannot successfully cope with. What 'success' amounts to here may not be easy to formulate, but it is enough that we can recognize that, by our own lights, we would be worse off without them. If they contribute to fulfilment and help avoid disruption, we find it hard to doubt that an ethical theory underwritten by 'scientific' investigation could fail to recognize that it was 'all right' to live by these standards in these circumstances. If ethical inquiry does mirror scientific inquiry, then we should expect this engineering conception of our ethical standards to co-exist with a body of ethical views to which we have a more detached commitment. We might hope that the scientific inquiry will lead gradually to a transformation of our live ethical views; but a degree of insulation between them ought to be characteristic of the scientific conception. It is not obvious that the analogy with science requires the 'objectivist' conception of values. Suppose we were engaged in evaluations involving concepts such as justice or autonomy, joining with others in testing such evaluations and developing the conceptual scheme in which it has a role. We might reasonably hope to aim at a consensus which can accommodate all those who share our 62 CHRISTOPHER HOOKWAY interest in this inquiry, and who share our interest in guiding our lives in accordance with this concept. Rather than viewing an alien ethical outlook as a rival research programme it could, developing the analogy, be pictured as a distinct discipline: we picture a plurality of ethical 'sciences' rather than a unified science of value. There are, doubtless, problems in developing the analogy in this way; but it has several advantages over the objectivist picture. First, it retains a firm focus on the idea of a community of inquirers unified by a common purpose: the 'objectivist' picture appears to include many people among the ethical scientists who are not part of any project of communal inquiry. Secondly, it holds on to the picture that consensus is the product of cumulative investigations with a more limited focus in responding to short-run problems or anomal- ies. Some such picture would be required by the idea that convergence / is all that is strictly relevant to ethical inquiry. The claim that the question of how to live provides the focus for ethical investigation does not obviously conflict with this picture. That does not require me to be able to endorse my way of life as the best. Analogy with science need require only that I find it satisfactory, and that I can feel confident that it has the potential to improvement or for responding to such challenges as may be provided. Experience of other forms of ethical life need be threatening only if it disturbs my evaluation of my own evaluations or raises questions about the appropriate attitude to take towards the future development of my framework of ethical concepts. So long as the question of how to live is not immediately formulated as 'what is the best form of life for a human being?' but is rather seen as a question that arises against a background of a tradition of ethical debate, argument is required to show that the questions raised by the 'objectivist picture' are forced upon anyone who takes the analogy with science seriously. I shall suggest, in section 5, that a sceptical role can be assigned to observations of differing ethical traditions which does not depend upon the objectivist picture. It is reasonable to hope that, if this view is taken, ethical inquiry would share many of the features of scientific inquiry. For example, there would be considerable local convergence which could be explained by saying of particular inquirers that they would form such and such a conclusion only if it were true; we would expect successor ethical positions to explain the approximate correctness of those which are discarded; we would acquire a developing under- Fallibilism and objectivity: science and ethics 63 standing of the factors that prompt people to make errors in arriving at ethical judgments; we would be able to appeal to a body of knowledge about the marks of (for example) just actions in order to provide quite naturalistic explanations of the ability of particular people to make particular judgments; and our acceptance of the broad background framework against which we judge might well be supported by pointing out that anything used to challenge it would be so much less certain than it is itself. So, if there is a fundamental contrast between science and ethics, we have not yet clarified what it is. It may be sufficient to count as a difference that science aims at a wider unification of perspectives than ethics, but it is unclear why that should have sceptical consequences for ethical reflection: there can still be scope for talk of discovery and fallibility in ethics; it is hard to see why ethics cannot be 'what it seems'; and there does not seem to be any bar to ethical reflection becoming self-conscious about our practice of making ethical judgments or about how to guard against error. In the closing section, I shall make a suggestion about how a sceptical challenge could be mounted. It will have the merit of deriving from reflection upon the prospects of progress in our dealings with our own ethical concepts within what is recognisably a community of ethical inquiry. Although it allows a role for observations of ethical variety in prompting such scepticism, this will not be through committing us to a search for an ethical view which can integrate the concerns of a variety of communities.

5. SCIENCE AND ETHICS: THE TRUE DIFFERENCE Division of labour and expertise has a role in science to which nothing corresponds in ethics. There is a role for ethical 'expertise' in helping people to become clear about judgments that they make unreflectively; it can also help people by guiding them to the range of factual considerations they should take into account or to its bearing upon their judgments. However, for something comparable to the division of labour found in science to arise in ethics, we should require to find (for example) agents with no concept of justice deferring to the judgments of the guardians of that concept in planning actions. It would be a confession of moral failing were I to act on others' judgments of justice while unable to form such judgments for myself. That a theoretician should use experimental 64 CHRISTOPHER HOOKWAY results which he could not have obtained himself, or that he should borrow principles from a theory he does not understand, is not a failing. This is enough to suggest that the community of inquirers has a different role in the two cases: in science, members of the community contribute to the developing knowledge of the community; in ethics, it is something which helps me to shape my ethical responses. Distinct 'schools' within ethical debate are in more direct competition than in science. In analogy with remarks about science in sections 2 and 3, we might think of justice as a vague concept: we agree on a range of uncontroversial cases and on a range of considerations deemed relevant to the applicability of the term to those cases; but there is uncertainty about how the term is to be employed in connection with further cases, and about how our standards should be formulated or refined. If there are distinct schools of thought within the community about how this refinement should proceed, then this is likely to be reflected in differences over how to live: over how institutions should develop; and over how people should be viewed. Since members of the opposing schools will have vital interests in prevailing, for one to triumph is for the other to be defeated. Whereas the official view of science requires practitioners to accept, or indeed to welcome, such 'defeats', it is hard to see how that could be accepted in ethics. The emergence of essentially contested concepts, like the limited role for expertise, is a reflection of the fact that securing consensus can be a lesser value than maintaining one's own substantive outlook among ethical inquirers. These sketchy remarks suggest a way of endorsing Williams' insistence that hope for consensus in ethics has a much less secure foundation than the similar hope in science. Suppose that we grant that opposing factions in a debate over the refinement of a vague ethical conception possess explanations of the substantial correctness and partial error of their opponents. In the scientific case, this would be sufficient to underwrite the generalizations that the opponents formulated, and the practical (engineering) applications to which they put their knowledge. The latter is not to be found in the ethical case. The practical hopes and aspirations which the losing school defend by reference to 'justice' are dismissed along with the losing conception of justice: the saliences whose moral relevance was defended through the proposed extension of the term 'just' are apt to vanish or lose their force as the concept develops in other ways. In Fallibilism and objectivity: science and ethics 65 the scientific case, a successful resolution of debate ideally preserves all that either party would deem a practically relevant achievement; there is little reason to expect this in the ethical case. The hope for a consensus which does not rest upon the power behind the prevailing view, and which does not threaten to disturb the unity of the inquiring community, must rest upon the expectation that all will formulate compatible desires or notice similar saliences in attempting to use existing concepts to cope with a changing world. There does not seem to be any component in the cognitive values which guide ethical investigation to ensure that maintaining the integrity of the ethical community should be required for, or take precedence over, preserving one's own integrity. If we think that ethical concepts lack such vagueness, or if we suppose that no competing 'schools' will emerge, then there may be no grounds for scepticism. Hence the hope for unforced consensus has little to recommend it. This argument supports the view that in ethics, but not in science, it is controversial what is to be saved as one view succeeds another. In so far as the application of evaluative concepts is analogous to the making of spontaneous quasi-perceptual judgments, debates about how to develop a vague conception have consequences for the saliences which should be deemed 'perceptually' relevant. It has the consequence, too, that there is no guarantee that an ethical community is bound together by a shared understanding of possible sources of error in making ethical evaluations. We can form no clear shared understanding of our fallibility to warrant the hope that error will steadily be eliminated as inquiry develops. The differences between my argument and Williams' can be narrowed a little further: room can be found for 'objectivist' themes and the impact of experience of contrasting moral positions. I suggested in the previous section that the analogy between sciences and ethics might be maintained without forcing us to confront issues of the best way to live. If competition of the sort I have described does emerge in the course of ethical inquiry, then such evaluations seem unavoidable if there is to be any rational basis for choosing between the competing factions: such a decision must appeal to the merits of the consequences of emphasising the sets of saliences each side exploits. Finally, any evidence which suggests that I cannot be confident how I would respond to situations which call for the refinement of our concepts, or that I cannot be confident that others will respond as I do, will shake my confidence that any but a forced 66 CHRISTOPHER HOOKWAY convergence can be achieved. Even if I feel no obligation to arrive at an ethical outlook which would convince those whose ethical views are radically different from my own, evidence of the variety of human evaluative response can leave me with little basis for hoping that the harmony of desires and interests required for unforced consensus obtains within my own ethical community: if no 'objectivist' theory of value can be defended, then there is unlikely to be sufficient local intersubjective concordance for us to be confident that our framework of moral concepts can be refined uncontrover- sially in a changing social world. This argument has points of contact with Wiggins' invocation of the essentially contested character of some ethical concepts in order to characterise the kind of truth which can be predicated of ethical judgments. The following comments are also close to his view. It is a common observation that scientific beliefs depend upon the use of identifiable sense-organs whose reliability can be explained: nothing comparable can be found for ethics. In earlier discussion, I attempted to disarm this disanalogy, observing that little that is definite can be said about the foundation of many general scientific beliefs, and noting that we often do have specific views about sources of evaluative error. Once disagreement arises, then the scientific picture appears to require us to describe the methods we relied upon for forming our beliefs and instigate inquiries designed to calibrate them, or to assess their reliability. We take it for granted that they have a certain degree of reliability and attempt to discover what that is. The lack of an identifiable 'moral sense-organ', and the potential for contested questions of the sort described above, make it difficult to believe that this could be achieved. It is unclear how our understanding of our fallibility could be developed in a way which had the kinds of connections with our hopes for overcoming our fallibility that is characteristic of science. The remarks of the final section take the form of suggestions rather than developed arguments. The underlying theme of this chapter has been that an adequate comparison of ethics and science should focus on: our ability to think of our ethical investigations as fallible; our ability to view our contributions to ethical thought as part of a progressing historical process of ethical inquiry. It need not rest immediately upon a need to embrace an ethical outlook which can affirm the supremacy of our own way of life in relation to those of other communities. Moreover the 'scientific' picture of ethics Fallibilism and objectivity: science and ethics 67 which I have formulated and criticized does not seem to be a version of relativism.

NOTES 1 See, for example, 'The fixation of belief, in Writings of Charles S. Peirce: a Chronological Edition, volume 3, Bloomington: Indiana University Press, 1986. 2 If'belief is the willingness to risk to great deal upon a proposition' then it is 'no concern of science' (5.598): 'The scientific man is not in the least wedded to his conclusions' (1.635). (References are by volume and paragraph to Collected Papers of Charles Sanders Peirce Cambridge: Harvard University Press, 1931-5, 1958). 3 David Wiggins, Meeds, Values, Truth, Oxford: Blackwell, 1987, pp. 344-5. 4 This point can be missed because the structure of the argument in 'The fixation of belief can give the impression that Peirce holds to the 'innocent until proved guilty' doctrine quite generally. Peirce's own position may be closest to the 'engineering conception' described in section (4): this would permit the parallel development of ethical reflection guided by reflective methodological sensitivities guided by the goals Wiggins describes. Some passages suggest that this is his view. 5 Wiggins, Needs, pp. 164-5. 6 It is significant that epistemological studies of science focus on how scientific ideas change, and on how our methodological conceptions guide us in coping with such changes. The processes involved in radical change in ethical ideas are a minor concern of philosophical ethics. CHAPTER 5 Might there be external reasons? John McDowell

This chapter is directed to a question that has been posed by Bernard Williams, in his interesting and insufficiently discussed article 'Internal and External Reasons'.1 Statements to the effect that someone has reason to act in a specified way (say to (f>), or that there is reason for someone to <[>, are apparently susceptible of two sorts of interpretation. The first is the internal interpretation, on which the statement is falsified by the agent's lack of any 'motive which will be served or furthered by his

II It is a strength of Williams' argument that he bases it on a subtle and flexible conception of the materials available to the internal interpretation. On the crudest understanding of that interpretation, an agent would have a reason to <}> only by virtue of having a desire such that -ing either is its satisfaction or is conducive to its satisfaction as means to end. I shall note two main ways in which Williams refines this.

68 Might there be external reasons? 69 First, about the role of desire. What is required if there is to be an internal reason is that the action in question is suitably related to an element in the agent's 'subjective motivational set' (p. 102); and, if the term 'desire' is used for all such elements, we must be clear that this is only a 'formal' (or, as one might say, philosopher's) usage. Subjective motivational sets are not restricted to what would ord- inarily be called 'desires': they 'can contain such things as dispositions of evaluation, patterns of emotional reaction, personal loyalties, and various projects, as they may be abstractly called, embodying commitments of the agent' (p. 105). Second, about what it is for an action to be suitably related to an element in the agent's subjective motivational set. The crude view already allows that an agent can have a reason to <{> without being actually motivated to 4>> if he does not realize that <|>"mg is a means to something he desires. The general idea is that one has reason to do what practical reasoning, starting from one's existing motivations, would reveal that one has reason to do - even if one has not realized that one has reason to do it. But the effect of the crude view is to limit the way in which practical reasoning can expand one's awareness of one's internal reasons: the expansion is restricted to cases where practical reasoning brings to light matters whose bearing on one's practical situation is, roughly speaking, technical.4 Williams' refinement drops this exclusive concentration on means and ends. The conception of practical reasoning or deliberation involved in his version of the general idea is much less restricted; he sketches it only by means of examples. (Such a procedure is unavoidable, he suggests, since 'there is an essential indeterminacy in what can be counted a rational deliberative process' (p. no).) He writes: A clear example of practical reasoning is that leading to the conclusion that one has reason to 4> because -mg would be the most convenient, economical, pleasant etc. way of satisfying some element of S [the agent's subjective motivational set], and this of course is controlled by other elements in S, if not necessarily in a very clear or determinate way. But there are much wider possibilities for deliberation, such as: thinking how the satisfaction of elements in S can be combined, e.g. by time-ordering; where there is some irresoluble conflict among the elements of S, considering which one attaches most weight to (which, importantly, does not imply that there is some one commodity of which they provide varying amounts); or, again, finding constitutive solutions, such as deciding what would make for an entertaining evening, granted that one wants entertainment (p. 104). 70 JOHN MCDOWELL One important ingredient in this freeing of the idea of practical rationality from restriction to satisfactions, and means to satisfactions, of given desires is that Williams is able to insist on the relevance of imagination to deliberation. By letting his imagination play on an outcome which he supposes he has reason to promote, an agent 'may come to have some more concrete sense of what would be involved, and lose his desire for it'; conversely, 'the imagination can create new possibilities and new desires' (p. 105). So the idea is this: what one has reason to do, on the internal interpretation, is whatever one can conclude that one has reason to do by an exercise of practical reasoning, conceived on these unrest- rictive lines. Practical reasoning is 'a heuristic process, and an imaginative one' (p. no). The significance of elements in one's subjective motivational set is not that one has reason to do only what is conducive to, or constitutes, their satisfaction, but that they 'control' the thinking by which one determines what one has reason to do, in ways like those exemplified in the quotation above: ways that it is impossible to codify in some simple theory.

in Does this leave any room for the external interpretation? Williams argues, on the following lines, that it does not. Any reason for action must be something that could explain someone's acting in the way for which it is a reason. If a reason did explain an action, the agent would have a motivation towards acting in the way in question — a motivation that the reason-giving explanation would spell out. But ex hypothesi an external reason statement can be true of someone without his actually having any motive that would be 'served or furthered' by his doing what he is said to have a reason to do — not even one whose relevance to his doing that would need to be uncovered by deliberation. (This simply spells out the idea of the external interpretation.) We can reconcile this with the point that any reason, even an external one, must be potentially explanatory of action, on these lines. Consider an agent who is not motivated by an external reason. It must nevertheless be true that some consideration constitutes a reason for him to act in a certain way; his not being motivated by it is a matter of his not believing, of the consideration, that it is a reason to act in that way.5 If he came to believe that, he would come to be motivated. That is, certainly, he Might there be external reasons? 71 would come to have an internal reason statement true of him. But this hypothetical internal reason need not pre-empt all the space an external reason might occupy: we can preserve the external reason if we can make sense of something which would be true throughout such a transition, and which is such that we can see how coming to believe it would be coming to have the motivation that makes the internal reason statement true. So we need to consider the transition from not being motivated by a putatively external reason to being motivated by it. The question is whether we can make sense of this as coming to believe something that was true already in advance of the agent's being motivated. What 'the external reasons theorist' needs, Williams says, is 'that the agent should acquire the motivation because he comes to believe the reason statement, and that he should do the latter, moreover, because, in some way, he is considering the matter aright' (pp. 108-9). At this point Williams makes a claim that is crucial for his argument (p. 109): If the theorist is to hold on to these conditions, he will, I think, have to make the condition under which the agent appropriately comes to have the motivation something like this, that he should deliberate correctly; and the external reasons statement itself will have to be taken as roughly equivalent to, or at least as entailing, the claim that if the agent rationally deliberated, then, whatever motivations he originally had, he would come to be motivated to <|>. The effect of this is to represent the external reasons theorist as committed to returning an affirmative answer to Hume's question, whether reason alone can give rise to a motivation. The predicament itself determines an interpretation for 'alone' here. As Williams describes his position, the external reasons theorist must envisage a procedure of correct deliberation or reasoning which gives rise to a motivation, but which is not 'controlled' by existing motivations, in the way that figures in the account of internal reasons; for, if the deliberation were thus 'controlled' by existing motivations, the reason it brought to light would simply be an internal reason. So the external reasons theorist has to envisage the generation of a new motivation by reason in an exercise in which the directions it can take are not determined by the shape of the agent's prior motivations — an exercise that would be rationally compelling whatever motivations one started from. As Williams says (p. 109), it 72 JOHN MCDOWELL is very hard to believe that there could be a kind of reasoning that was pure in this sense - owing none of its cogency to the specific shape of pre-existing motivations - but nevertheless motivationally efficacious. If the rational cogency of a piece of deliberation is in no way dependent on prior motivations, how can we comprehend its giving rise to a new motivation?

IV But need the external reasons theorist fight on this ground? Let us retrace our steps. We have to consider the transition from not being motivated by a supposedly external reason to being motivated by it. The external reasons theorist must suppose that the agent acquires the new motivation by coming to believe the external reason statement. To be an external reason statement, that statement must have been true all along; in coming to believe it, the agent must be coming to consider the matter aright. The crucial question is this: why must the external reasons theorist envisage this transition to considering the matter aright as being effected by correct deliberation? He cannot make sense of the motivational effect of the transition by crediting it to deliberation 'controlled' by prior motivations, since that would merely reveal the reason to be internal. So, if there must be deliberation — reasoning — that could bring about the transition, he needs to invent an application of reason in which it can impel people to action without owing its cogency to the specific shape of their prior motivations; and this is what Williams rightly says it is hard to believe in. The argument debars the external reasons theorist from supposing that there is no way to effect the transition except one that would not count as being swayed by reasons: for instance (p. 108), being persuaded by moving rhetoric, and, by implication (p. no), inspiration and conversion. But what is the ground for this exclusion? Williams' wording may seem to answer this question. To repeat a passage I have already quoted, he says that what the external reasons theorist needs is 'that the agent should acquire the motivation because he comes to believe the reason statement, and that he should do the latter, moreover, because, in some way, he is considering the matter aright' (pp. 108-9). If'considering the matter aright' figures, as this suggests, in an explanation of the agent's coming to believe the reason statement, suited to reveal the transition as one to Might there be external reasons? 73 a true belief, it may seem that the phrase must single out and endorse something like a procedure of argument or reasoning. But all that the external reasons theorist needs at that point in the argument — this is quite clear from the rephrasing of it in my last paragraph - is that in coming to believe the reason statement, the agent is coming to consider the matter aright. This leaves it quite open how the transition is effected. It is worth emphasizing that there need be nothing philosophically mysterious about the notion of considering matters aright in this kind of context: no implication of a weird metaphysic, for instance, in which values or obligations are set over against our subjectivity, as independent of it as the shapes and sizes of things.6 If we think of ethical upbringing in a roughly Aristotelian way, as a process of habituation into suitable modes of behaviour, inextricably bound up with the inculcation of suitably related modes of thought, there is no mystery about how the process can be the acquisition, simultaneously, of a way of seeing things and of a collection of motivational directions or practical concerns, focused and activated in particular cases by exercises of the way of seeing things.7 And if the upbringing has gone as it should, we shall want to say that the way of seeing things - the upshot, if you like, of a moulding of the agent's subjectivity — involves considering them aright, that is, having a correct conception of their actual layout. Here talking of having been properly brought up and talking of considering things aright are two ways of giving expression to the same assessment: one that would be up for justification by ethical argument. Let me emphasize what is implicit in this last sentence: I am using the notion of proper upbringing only to defuse the threat of metaphysical peculiarity, not as a foundational element in some sort of ethical theory — as if we had independent access to what counts as a good ethical upbringing, and could use that to explain ethical truth as a property enjoyed by the judgments that a properly brought up person would make. What if someone has not been properly brought up? In order to take seriously the idea that someone who has been properly brought up tends to consider matters aright in the relevant area, we surely do not need to embrace the massively implausible implication that someone who has not been properly brought up — someone who has slipped through the net, so to speak - can be induced into seeing things straight by directing some piece of reasoning at him. On the 74 JOHN MCDOWELL contrary, reasoning aimed at generating new motivations will surely stand a chance of working only if it appeals to something in the audience's existing motivational makeup, in something like the way exploited in Williams' account of the internal interpretation; and the trouble with someone who has in some radical way slipped through the net is that there may be no such point of leverage for reasoning aimed at generating the motivations that are characteristic of someone who has been properly brought up. What it would take to get such a person to consider the relevant matters aright, we might plausibly suppose, is exactly the sort of thing that, according to Williams' argument, the external reasons theorist may not appeal to: something like conversion. Admittedly, it is not straightforwardly obvious how we should think of this, or some better substitute, as operating; the bare idea of conversion points at best to a schema for explanations of shifts of character, and the weight of the explanation in any real case would rest on our comprehension of the psychic efficacy of the specific converting factor (a religious experience, say). But it does not seem hopeless to suppose that at least sometimes we really might be able to understand on these lines how someone who had slipped through the net might suddenly or gradually become as if he had been properly brought up, with the interlocking collection of concerns and way of seeing things that he failed to acquire earlier. The idea of conversion would function here as the idea of an intelligible shift in motivational orientation that is exactly not effected by inducing a person to discover, by practical reasoning controlled by existing motivations, some internal reasons that he did not previously realize he had. But if its upshot is a case of considering matters aright, why should such a process not count as someone's being made aware of some external reasons, reasons that he had all along, for acting in the relevant ways? It is plausible, then, that from certain starting-points there is no rational route — no process of being swayed by reasons — that would take someone to being as if he had been properly brought up. (Being properly brought up is not itself a rational route into being that way.) But this has no evident tendency to disrupt the natural connection between being that way and considering matters aright. So why not suppose that the kind of conversion I have envisaged might be a case of what the external reasons theorist needed: acquisition of a new motivation by way of acquiring correct beliefs? Might there be external reasons? 75

v Williams simply assumes what rules this out, that the external reasons theorist must envisage a transition to considering matters aright that would be effected by reasoning. This assumption is held in place by what Williams takes to be the only point of believing in external reasons. According to Williams, the external reasons theorist wants to be able to bring a charge of irrationality against anyone who is not motivated in some direction that the theorist thinks he should be motivated in; 'he wants any rational agent, as such, to acknowledge the requirement to do the thing in question' (p. no). There is certainly a recognizable temptation hereabouts. Moral- ists in particular are prone to suppose that there must be a knock- down argument, an appeal to unaided reason, which, if one could only find it and get people to listen, would force anyone capable of being influenced by reasons at all into caring about the sorts of things one ought to care about. In itself this might be no more than a harmless fantasy (although when the expectation of finding such an argument is disappointed, this can lead to morality's seeming problematic in ways that it should not).8 What is in question at present, however, is a related temptation, which one can see that one should avoid however optimistic one is about the prospects for the knock-down argument: the temptation, lacking the knock-down argument as we all do, to talk as if the argument is out in the open and people who do not care about the sorts of things they ought to care about are flying in the face of it. Williams' excellent point is that an accusation of irrationality that is supposed to convey something on those lines is nothing but 'bluff' (p. in).9 As Williams notes, there are plenty of things that the internal reasons approach allows one to say against someone who is not motivated by the considerations one thinks he should be motivated by: for instance, 'that he is inconsiderate, or cruel, or selfish, or imprudent; or that things, and he, would be a lot nicer if he were so motivated' (p. no). (We can add: that the shape of his motivations reveals that he has not been properly brought up.) So what is the point of holding out for the right to make an accusation of irrationality as well, if it is not to bluff the person into mending his ways by means of a fraudulent suggestion that he is flouting considerations that anyone susceptible to reasons at all would be moved by? Perhaps the answer is 'None'; it may be that calling a person 76 JOHN MCDOWELL 'irrational' carries, beyond the possibility of cancellation, the illicit implication that he is unmoved by something that would move anyone capable of being moved by reasons at all. (I shall come back to this later.) But even if we renounce any right to direct an accusation of irrationality against a person when he is unmoved by considerations we think he should be moved by, that does not quite settle the question whether, as Williams puts it (p. m), 'the only rationality of action is the rationality of internal reasons'. What is at issue here is the relation between the explanatory role of the concept of reason and a critical or normative dimension that it must have. Williams is considering this issue when he raises the question (pp. 102—3) whether someone who believes of some petrol that it is gin, and wants a gin and tonic, should be said to have a reason to mix the petrol with tonic and drink it. If he does that, we shall have an explanation, of the reason-giving kind, for his doing what he does. But Williams suggests that an internal reasons theorist should not say, on those grounds, that the person does have a reason to drink the stuff which is in fact petrol: It looks in the wrong direction, by implying in effect that the internal reason conception is only concerned with explanation, and not at all with the agent's rationality, and this may help to motivate a search for other sorts of reason which are connected with his rationality. But the internal reason conception is concerned with the agent's rationality. What we can correctly ascribe to him in a third-personal internal reason statement is also what he can ascribe to himself as a result of deliberation. Here Williams is making room for a thought on these lines: the explanatory power of reason-giving explanations depends on there being a critical dimension to the concept of rationality. These explanations do not merely reveal actions as the outcome of some way — to whose detail we can be indifferent — in which the agent happens to be internally organized; that the operations of an internal organization from which an action is represented as flowing are recognizably the right kind of operations for the explanation to be of the reason-giving sort requires that they approximate sufficiently closely to something in the nature of an ideal.10 Reason-giving explanations require a conception of how things ideally would be, sufficiently independent of how any actual individual's psychological economy operates to serve as the basis for critical assessment of it. In particular, there must be a potential gap between the ideal and the specific directions in which a given agent's motivations push him. Might there be external reasons? 77 Williams secures an independence that conforms to this abstract description, in the passage I have quoted, by his appeal to deliberation: what practical rationality requires of an agent is not simply read off from his specific motivations just as they stand (including, in Williams' example, a wish to drink some stuff which is in fact petrol), but is determined, from these motivations, by deliberation, whose capacity to correct and enrich the specific motivations one starts with is supposed to open up the necessary gap between actual and ideal. But it is open to question whether this puts the right distance between the basis for criticism of the way an agent's actions flow from his psychological states, on the one hand, and the way his psychology happens to be, on the other.11 Certainly the appeal to deliberation interposes some space between these things, but the standard is still, albeit with the indirection which that imposes, fixed by the agent's motivations as they stand. There is thus a sense in which it can be claimed that the resulting picture of the critical dimension of the concept of practical rationality is psychologistic. That term has become attached12 to Frege's reproach against positions that treated the principles of logic, which we can think of as delineating an important part of the structure of theoretical rationality, as 'laws of thought' rather than 'laws of truth':13 Frege's point was, in effect, that, if logic is to be able to stand in judgment over the workings of minds, it cannot be constructed out of mere facts about how psychological transitions take place. One reason, then, for doubting that 'the only rationality of action is the rationality of internal reasons', quite distinct from the wish to browbeat people into (for instance, and especially) morality by a fraudulent accusation of irrationality, might be the idea that the critical dimension of the notion of practical rationality requires an analogous transcendence of the mere facts of individual psychology - even as corrected by the sort of deliberation that the internal reasons conception allows. Beliefs are not rationally self-contained psychic phenomena, which could be aggregated just anyhow, with no restrictions on their interrelations and their relations to the subject's world, and still add up to a state of mind. Theoretical reason, both formal and substantive, puts limits on the attributions of belief that make sense. It would get things back to front to think that an adequate conception of theoretical reason could be derivative from a set of supposedly independent data about the workings of minds. Now a point of this 78 JOHN MCDOWELL sort should be applicable to practical, no less than to theoretical, reason. Desires, in the broad 'formal' sense of ingredients in subjective motivational sets, are similarly not rationally self-contained psychic phenomena, which can be unproblematically conceived as determinants, from the outside, of the shape that practical reason takes, for the individual agent whose desires they are.14 But the idea that 'the only rationality of action is the rationality of internal reasons' seems to involve thinking of the desires (in that sense) from which an individual agent starts in just this way. In the context of this kind of refusal to find 'the rationality of internal reasons' sufficient, the idea of not reasoning correctly might be glossed in terms of not giving a consideration the right weight in deliberation. On these lines, deliberating correctly would be giving all relevant considerations the force they are credited with in a correct picture of one's practical predicament. This yields a sense in which to believe an external reason statement is, as Williams indeed suggests it must be (p. 109), to believe that if the agent deliberated correctly he would be motivated (of course not necessarily conclusively) in the direction in which the reason points.15 But there is no implication, as in Williams' argument, that there must be a deliberative or rational procedure that would lead anyone from not being so motivated to being so motivated. On the contrary, the transition to being so motivated is a transition to deliberating correctly, not one effected by deliberating correctly; effecting the transition may need some non-rational alteration like conversion. Perhaps a sense can even be given to the accusation of irrationality within this framework. There would now be no question of a bluff, any more than one need be bluffing if one says, to someone who cannot find anything to appreciate in, say, twelve-tone music, 'You are missing the reasons that there are for seeking out opportunities to hear this music' (It might take something like a conversion to bring the reasons within the person's notice; there is no suggestion that he is failing to be swayed by something that would sway anyone capable of being influenced by reasons at all.) However, it is (at least) difficult to separate calling someone 'irrational' from the suggestion that he is missing the force of something in the nature of an argument. (It would be odd to say that a person who finds no reasons to listen to twelve-tone music is irrational, even though one thinks that the reasons are there.) So far as I can see, scepticism about the adequacy of the internal reason conception has no need to Might there be external reasons? 79 insist on classifying people in these terms, and it is probably safest for it not to do so.

VI There is no question of my trying to establish the tenability of a position on these lines in this chapter: my main point is only that, though Williams is surely right about the bluff he attacks, that does not show that the internal conception of practical reason gives us everything we want. In this section, I shall mention very briefly some possible different lines on which a scepticism about external reasons might be defended, and suggest responses to them. On the internal conception, practical reason can be, in a sense, content-neutral: deliberation is indifferent to the nature of the motivations that constitute input to it, and might be conceived as a procedure for imposing coherence and practical determinacy on whatever collection of prior motivations one presents it with. On an external conception, by contrast, practical reason is not something that can be equally well exemplified whatever the content of the motivations one begins with. If one is tempted to think that the content-neutrality of deductive rationality ought to generalize over all applications of reason, one will be inclined to see an argument for the internal conception here. But the argument would be no stronger than the case for generalizing that feature of deductive logic. And surely it is no more obvious that we should expect practical reason to be content-neutral than it is that we should expect, for instance, scientific reason to be so.16 Reasons must be capable of explaining actions; and it can seem that, if we are to find that intelligible, we must conceive practical reason as directing the action-generating efficacy of a collection of motivations which are prior to it, and, as far as rational explanation goes, simply given. This idea is clearly congenial to the internal conception. But it rests on what looks like a misconception of the way in which actions are explained by reasons, one which assimilates that to the way in which events are explained by mechanical forces like the tension in a tightly wound spring. One can be suspicious of this assimilation without threat to the thesis that reason-giving explanations are causal.17 To respect the truism that reason-giving explanations work by revealing how actions are motivated, it is not necessary to picture motivations as antecedent 80 JOHN MCDOWELL quasi-mechanical sources of energy: in explaining an action, a reason-giving explanation can equally make the motivation of the action rationally intelligible, and there is no basis for insisting that this must be by way of representing that motivation as the upshot of reason's channelling a pre-rational motivational force in a certain direction.18 The opposition to psychologism that I have described pictures practical predicaments as structured by collections of values that are independent of any individual's motivational makeup, and this may seem to reintroduce the threat of a weird metaphysic which I discounted earlier. But this is a mistake. One way to avoid such a metaphysic is to regard values as reflections or projections of psychological facts involving affect or sentiment, and such a position might indeed have difficulties in accepting the kind of transcendence I have envisaged. But in order to acknowledge the constitutive connection of values to human subjectivity, it is not obligatory to suppose that the genealogy of value can be unravelled, retro- spectively, in such a way as to permit factoring out a contribution made by isolable facts about our individual psychology to the evaluative contours of our world. A sane subjectivism can allow that value transcends independently describable psychological fact.19 There may seem to be something suspect about the way I have exploited Williams' phrase 'considering the matter aright'. I suggested that, when the supposed external reasons in question are ethical ones, the assessment expressed in 'aright' would be grounded in ethical argument; and I meant argument internal to some specific ethical outlook, not argument that would somehow win over someone unmoved by what one wants to represent as external reasons. There is an intelligible temptation to suppose that such a use of 'aright' cannot be more than bluster. That is indeed what it would be if it were meant somehow to impress outsiders, even in the absence of any ability to persuade them; that is Williams' point about bluff. But I have disavowed any aim of manipulating the notion like this. Nothing more would be in question, in any particular appeal to a determinate conception of how the relevant matters are rightly considered, than confidence in some part of an ethical outlook. (Ethical external reasons are not external to ethics.) This can seem a second-grade application of the concept of correctness, but it is open to question whether we can do better in Might there be external reasons? 81 any region of our thinking.20 We do not conceive our values as owing their authenticity, and their relevance to what we do, to our motivational makeup or to anything in the psychological genesis of our coming to have them; and that, together with our managing to sustain confidence in them through reflection, is all that it takes for us to suppose that they yield reasons which are not internal in the sense Williams explains. This sustaining of confidence is fraught with difficulties; but it is a mistake to let a supposed metaphysical insight, expressed in the depreciating of the sense of 'aright' available in appraisal of ethical thinking, seem to add to them. I began this section as if it were going to be something of a digression, but in fact I believe that the last couple of paragraphs have touched on what is, in one sense, the heart of the matter. Williams' explicit argument has no deeper foundation than the assumption that the external reasons theorist wants to be entitled to find irrationality when someone is insensitive to the force of a supposed external reason; and, in its naked form, the assumption seems too transparently flimsy to be the real basis for his conclusion. It is too easy to drive a wedge between irrationality and insensitivity to reasons which are nevertheless there: recall the case of not appreciating twelve-tone music. But perhaps we can begin to find it intelligible that this simple point goes missing if we locate the real foundation of the argument deeper down, in an idea on these lines: the notion of truth or objectivity, implicit in the appeal to 'considering the matter aright' which the external reasons theorist needs, requires beliefs to be capable of being formed either under the causal control of the circumstances that render them true or as a result of exercising rationality, conceived in purely procedural terms, as something that can be compelling without need of substantive presuppositions. Beliefs about values or obligations had better not fall under the first of these disjuncts, on pain of a weird metaphysic; so, according to this line of thought, they would have to fall under the second. I do not mean to suggest that starting here would make the argument against external reasons any stronger: on the contrary, we might just as well argue by modus tollens from the continued ease of driving the wedge to the conclusion that there must be something wrong with the disjunction. But there are familiar philosophical concerns about truth and objectivity whose operation, in some such form as this, might make it seem that Williams' starting assumption 82 JOHN MCDOWELL would have to be common ground.21 All this, however, is off-stage so far as Williams' article is concerned, and in another sense this section remains simply a digression.

VII Williams' argument depends crucially on the basic premiss that a transition to a correct view of the reasons for acting that apply to an agent - to considering these matters aright - must be capable of being effected, for the agent, by reasoning. The internal reasons approach gives an account of such transitions which represents the possibilities for making them, and therefore the content of a correct view of reasons, as determined by the motivational orientation from which the agent in question begins. If one is suspicious of this account of how a correct view of reasons is determined, one's only recourse, given the basic premiss, is to postulate a kind of reasoning that is practical, but not shaped by the motivations of those who engage in it: this then looks like a supposed exercise of that bloodless or dispassionate Reason that stands opposed to Passion in a familiar and unprepossessing genre of moral psychology, one which Hume made it difficult to take seriously. What I have been suggesting is that the basic premiss distorts the issue. In order to urge that there is more substance to practical reason than the internal reasons conception allows, one need not seek to supplement the internal reasons picture of practical reasoning. This distinction enables us to decline the choice between, on the one hand, taking a correct view of an agent's reasons for acting to be determined (indirectly, via deliberation) by his 'passions' as they stand — Williams' internal approach — and, on the other, taking it to be determined by dispassionate Reason — in effect, the only alternative Williams allows. We do not need to choose between conceiving practical reason psychologistically and conceiving it as an autonomous source of motivational energy over and above the 'passions'. This should begin to suggest a proper location for an investigation of practical reason which can accommodate the surely indubitable relevance of human psychology to what human beings have reason to do: the right way to think about this topic belongs between the individualistic psychologism of the internal reasons approach and the apsychologism, so to speak, of the only alternative that Williams' argumentative structure allows.22 Might there be external reasons? 83

NOTES 1 In Ross Harrison (ed.), Rational Action, Cambridge University Press, 1980; reprinted in Williams, Moral Luck, Cambridge University Press, 1981; I shall give page references to the latter. 2 As will become clear immediately, this can be understood so as not to require that the agent is actually motivated towards <|>-ing. 3 One might say: about whether there are any external reasons. Williams allows this sort of formulation (for instance in his title; I have followed suit in mine). But I shall follow him in treating this only as a con- venience; nothing turns on whether we are looking for a classification of sorts of reasons for action or merely a classification of sorts of things that can be said. 4 Williams even suggests, on the basis of the fact that 'the mere discovery that some course of action is the causal means to an end is not in itself a piece of practical reasoning' (p. 104), that the effect is to obliterate practical reasoning from the picture altogether. 5 There are two possible cases here: one in which the consideration is in view, but not believed to constitute a reason, and one in which the consideration is not even in view, or, if in view, not in clear focus. 6 Here I am in agreement with Simon Blackburn (see, for example, chs. 5 and 6 of Spreading the Word, Oxford: Clarendon Press, 1984; although I do not think it is felicitous to represent the metaphysically undemand- ing notion of correctness or truth that is in question as a construction on an a«/«-realistic base, as in Blackburn's projectivism. 7 See Nicomachean Ethics, Book 11; but no specific Aristotelian detail matters to the point I am making — what is in question is barely more than common sense. The terminology I have used to describe what I allege to be unmysterious in the light of Aristotelian common sense comes from David Wiggins: see 'Deliberation and practical reason', in Amelie O. Rorty (ed), Essays on Aristotle's Ethics, Berkeley, Los Angeles, and London: University of California Press, 1980. 8 I use 'morality' here as a mere variant on 'ethics'; the point is not one about the special topic that Williams discusses in chapter 10 of Ethics and the Limits of Philosophy, Cambridge, MA: Harvard University Press, 1985- 9 A similar point is made by Philippa Foot, in 'Morality as a system of hypothetical imperatives', Philosophical Review, 81 (1972); reprinted in her Virtues and Vices, Oxford: Basil Blackwell, 1978. In 'Are moral considerations hypothetical imperatives?', Aristotelian Society Supplemen- tary Volume 52 (1978), I tried to defend a version of the idea of the categorical imperative whose point would not lie in this kind of intellectual dishonesty. 10 Donald Davidson has written in this connection of the constitutive role of the concept of rationality (in its normative or critical application) in organizing our understanding of the concepts of common-sense 84 JOHN MCDOWELL psychology. See Essays on Actions and Events, Oxford: Clarendon Press, 1980, especially essays 11 and 12. (The point conies out in the plausibility of saying that it would make no sense to suppose that an explanation was of the reason-giving sort while disallowing the question how good the reason is for which the agent is said to have acted. This suggests that Williams did not need to deny that the agent has a reason, in the case he describes: he could have made the essential point about the importance of the critical dimension of the concept of rationality by allowing that the agent has a reason, but insisting that the reason is open to objection.) 11 It is not only (or even, perhaps, primarily) actions that can be criticized on the basis of a conception of rationality; but the topic of this paper is reasons for action in particular. 12 See Michael Dummett, Frege: Philosophy of Language, London: Duck- worth, 1973. 13 See the opening of'The thought: a logical inquiry', in P. F. Strawson (ed.), Philosophical Logic, Oxford University Press, 1970. 14 See, again, Davidson, Essays in Actions and Events. Against the idea of desire as a self-standing phenomenon, so that particular desires stand in no need of a conception of rationality to underwrite their intelligibility, see G. E. M. Anscombe's well-known remarks about wanting a saucer of mud, in Intention, Oxford: Basil Blackwell, 1957, pp. 70-1. 15 This is essentially the point made by Brad Hooker, 'Williams' argument against external reasons', Analysis, 44 (1987); he credits it to Robert Gay. Hooker does not question Williams' claim that the transition to being moved by an allegedly external reason would have to be effected by reasoning. 16 Williams himself would clearly have no sympathy with the idea of a purely formal account of practical reason. 17 For which see Davidson, Essays on Actions and Events, essay 1. 18 See ch. 5 of Thomas Nagel, The Possibility of Altruism, Oxford: Claren- don Press, 1970. 19 See David Wiggins, 'A sensible subjectivism?', in his Needs, Values, Truth Oxford: Basil Blackwell, 1987. It was the possibility of a subjectivism of this kind that I had in mind in my remark about projectivism in note 5 above. 20 Against Williams' suggestion, in Ethics and the Limits of Philosophy, that natural science is in a relevantly different position, see my 'Critical notice' in Mind, 95 (1986). 21 The assumption seems to be conceded without question in Christine M. Korsgaard's acute discussion of Williams, in 'Skepticism about practical reason', The Journal of Philosophy, 83 (1986), 5-25: see in particular, p. 20. 22 I am grateful to Annette Baier and Paul Hurley for helpful comments on an earlier draft. T. M. Scanlon responded to a version of this paper Might there be external reasons? 85 which I delivered to the Eastern Division of the American Philosophi- cal Association in December 1987, and his remarks have been very useful (though I have by no means met all his points). More recently, I have been greatly helped by comments from Jonathan Dancy. CHAPTER 6 Aristotle on human nature and the foundations of ethics Martha C. Nussbaum

The whole race is a poet that writes down The eccentric propositions of its fate. Wallace Stevens, 'Men Made out of Words' And language he has taught himself And thought swift as the wind And city-dwelling inclinations. Sophocles, Antigone Aristotle repeatedly mentions human nature in connection with ethical questions. Appeals to facts about our nature as human beings enter at crucial junctures into ethical arguments of various types, as if they did some sort of ethical work. But it is difficult to determine what sort of work they are intended to do, and how they do it. Take one example. Towards the beginning of Nicomachean Ethics i, Aris- totle tells us that the best life would seem to be one that is self- sufficient, autarkes. This might naturally call to mind the solitary contemplative self-sufficiency of the Platonic philosopher. But now Aristotle adds a qualification. "By self-sufficient I mean not for the person himself, leading a solitary life, but for the person along with parents and children and a wife and in general his friends and fellow citizens, since the human being is by nature a political being" (ioo,7b7-n). A fact about human nature is used to defend an ethical conclusion, the conclusion that we should prefer a communal conception of the good life to a more solitary conception. How does this work? For it seems that a serious opponent could object to Aristotle's proposal in at least two different ways. First, he could concede that Aristotle is correct about the facts of human nature, but deny that they have the alleged normative implications. So it is a fact about us that we need to live with others. This gives us no reason not to try to climb above our nature and repudiate these parts of it. Certainly it gives us no reason to ascribe to social pursuits any

86 Aristotle, nature, and ethics 87 positive value. Or, second, he could deny that Aristotle has correctly described the human, saying, for example, that our true nature as human beings is to be pure intellectual souls, only contingently bound up with the elements (viz. the emotions and appetites) that give rise to our need for community. What can Aristotle now say in defense of his argument? My purpose here is to ask how the idea of human nature works inside Aristotle's ethical arguments. Where does this idea come from, and what sort of idea is it? What is its relationship to ordinary beliefs and to science? And how, if at all, does it yield normative conclusions? How, if at all, can it escape the charge that such accounts substitute for experience a useless abstract theory, which cannot really resolve any practical dilemma? Answering this question in some detail will contribute, I hope, to the larger inquiry pursued in Williams' Ethics and the Limits of Philosophy.' For Williams suggests that, among the attempts to give a foundation to ethics, the Aristotelian approach is the most promising. And yet he holds that it fails, for several reasons. So a clear look at his charges and at Aristotle's arguments may shed light on the larger question: is there any coherent way of establishing a foundation for our attempts to answer the question, "How should one live?"

I. WILLIAMS ARISTOTLE I shall be arguing that some of Williams' criticisms of Aristotle fail because the account he gives of Aristotle's human nature project (an account that he shares with numerous other influential writers, past and present)2 is, in some important respects, inaccurate. Aristotle's actual position is stronger than the position Williams describes and then attacks, and is not vulnerable to all of his criticisms. Williams does not give a detailed textual exposition of Aristotle's arguments on this issue; so I shall be reconstructing his interpretation both from the general account of Aristotle in Ethics and the Limits and from some revealing statements in the concluding section of his article published in 1986, "Hylomorphism," an article in which he appears more sympathetic to Aristotle's ethical use of the concept "human being" than he does in the book, but in which he offers approximately the same account of what Aristotle is attempting to do.3 OO MARTHA C. NUSSBAUM Williams' account of Aristotle makes, as I understand it, three claims: (1) Questions about essential human nature are, for Aristotle, matters of natural scientific fact, not of ethical value. They can be grasped from outside practices and the point of view that guides them, from the totally external viewpoint of a neutral observer - from, as ELP puts it, "an absolute understanding of nature" (52); and they are entirely independent of ethical conceptions and judgments. Nothing human beings do, believe, or choose can alter the status of some putative element in our human nature.4 (ia) It seems to be a corollary of this claim that the Aristotelian account of human nature is not based (or not primarily or solely based) upon the endoxa, upon ordinary beliefs and conceptions. Such facts about nature are what they are independently of what ordinary people believe, and ordinary belief will probably not even be a very reliable guide to such facts.5 (2) Questions of essential nature are (for Aristotle) uncontroversial in a way that ethical questions usually are not. Although the answers to such questions may be difficult to discover, the fact that they are matters for scientific inquiry means that they can, in principle, be settled in a more satisfyingly decisive way than questions of ethical value judgment.6 (3) The discovery of the essential nature of the human being does important work in reaching normative ethical conclusions. Its source in the external viewpoint provides us with fixed points that effectively rule out certain ethical alternatives, and perhaps also give positive support for certain others. In this way, a human nature inquiry contributes to the resolution of our most troublesome normative disputes.7 This seems to be the core of Williams' account of Aristotle, and the common ground between the article and the book. There are differences, of emphasis at least. The article thinks of the concept "human being" as providing hard value-neutral constraints against "atrocity" - so that, for example, a theory of human nature will tell us what we may not do to the helpless handicapped infant, or to the senile elderly.8 The book seems to envisage a richer and more ubiquitous role for the account of nature, thinking of it as giving us a full story of human natural perfection, from which a rather full account of ethical conduct would straightforwardly flow.9 There are also differences of evaluation. The article (written earlier, though Aristotle, nature, and ethics 89 published later) seems favorable to the Aristotelian project, and, in particular, to the idea that we might find, in an external account of humanness, the foundation for some absolute prohibitions that could not be altered by our choices. The book is more pessimistic, concluding that the strong assumptions on which the Aristotelian project is based "have collapsed," and that the project of grounding ethics in this sort of "outside view" has collapsed with them (p. 53). But these three claims are clear enough, in both versions; and it is clear, too, that for Williams it is precisely the alleged externality of the account of human nature that makes it (at least initially) promising as a starting-point for ethics, and as a source of solutions to difficult ethical problems. Williams' objections to Aristotle's account, as set out in the book, seem to be of two kinds. First of all, he suggests that Aristotle's external account of human capabilities and perfections is not available to us - is not one that we can plausibly believe. Furthermore, he suggests that it is very likely that no such external account of human nature will fill the gap left by Aristotle's: there seems to be no way of plausibly linking facts of nature to ethical considerations, at least in a way that "will adequately determine one kind of ethical life as against others" (p. 47). Second, he suggests that such external accounts may, in any case, be of dubious relevance to human ethical choice. If they do not already contain ethical thought, it is not at all clear what relevance they have to ethical thought, or why we should care about them.10 There is, I believe, a considerable tension in Williams' thought on just this issue - a tension revealed, as well, as by the juxtaposition of article and book, published at almost the same time. On the one hand, Williams appears to be deeply drawn to the idea that an external account of what we are might tell us what we should and should not do. The article approves of this idea, appealing to it (and to Aristotle) to criticize the grounding of ethics on an explicitly internal and ethical concept, the concept "person," which can, on account of its internality as part of a structure of human evaluative beliefs, be applied or withheld according to our ethical judgments. The book, again, shows considerable sympathy for the project of moving to an external view in order to ground ethical choices; it offers an attractive and detailed account of what such an external theory of human nature would have to look like if it were to offer us the foundation we want.11 And it suggests that when this project 9O MARTHA C. NUSSBAUM fails it is a considerable loss, the loss of a significant hope. On the other hand, the book's objection about relevance, repeated in more than one section, seems, to me at least, to cast doubt not only on some particular projects of giving ethics a scientific foundation (such as sociobiology, in connection with which Williams makes the objection), but on the whole idea that ethics could be meaningfully and convincingly grounded on an account given from the external perspective, or from what Williams calls the "absolute view of nature."12 I shall argue here that such an external account is not the account that Aristotle offers us; but that this is no great loss, because the external account, as Williams characterizes it, is neither necessary nor sufficient for providing ethics with a foundation. A foundation of a different and more promising kind is given in the account that Williams actually develops.

II. THE GENERAL STRATEGY A series of connected passages in which Aristotle appeals to questions of kind membership and essential nature will provide a starting-point, giving us a sense of his general approach. During the discussion ofphilia in Nicomachean Ethics VIII, Aristotle observes that, in general, friends wish for one another very great goods. But, he adds, perhaps we should not go so far as to wish our friends transformed into gods: for then they would be too far away from us to be our friends. He then makes a further, most significant remark about such wishing: If it is well said that a friend wishes goods for his friend for that person's own sake, then that person will have to remain the type of being he is (hoios pot' estin ekeinos). Then it is to him as a human being [or: on condition of his remaining a human being] (anthropoi onti) that the friend will wish the greatest goods. (1159^8-12) A later passage makes the same point in connection with well- wishing for oneself: Being is a good thing for the good person, and each person wishes good things for himself. But nobody would choose to have everything on condition of becoming other {genomenos allos) - as, for example, the god right now has the good. He will wish to have the good while continuing to be whatever he is (ho tipot' estin). (1166a 19) Aristotle, nature, and ethics 91 These passages restrict ethical aspiration by bringing in considerations of personal identity and kind membership. How do they do this? When I imagine a picture of the good or valuable life, and think of wishing it for myself or for another, I ought to get clear about the relationship between that valuable life and the conditions of my (my friend's) continued existence. I ought, that is, to ask closely whether this imagined life is a life that could be lived by a being such as I am - by a being, that is, who shares with me all those characteristics that I consider to be truly constitutive of my (my friend's) identity. For it might emerge that the life was so remote from mine that I could not imagine in it a person whom I could accept as identical with myself. Certain changes in a person and a way of life are compatible with continued personal identity; others are not. Certain changes make me better; certain other changes, however apparently splendid, bring into existence a new being who is not me. Aristotle assumes that ethical well wishing is aimed at realizing a good life for me and my friends, not at the realization of some ideal situation in which none of us exists. So our views about who we essentially are and what changes we can endure while remaining ourselves set limits of a kind upon what we can wish, on what our ethical theories can commend. As Aristotle remarks elsewhere, "It would be peculiar (atopon) if someone chose for himself not his own life, but that of someone else" (EN 117832-3). Strictly speaking, there might seem to be two separate questions here: (1) What are necessary and sufficient conditions of my continuing to exist as a human being? (What lives are and are not human lives?) and (2) What are necessary and sufficient conditions of my continuing to exist as myself? For there might be changes that would be sufficient to make me other than the individual I am, without making a creature of a different species come into being. Our first passage clearly focuses on the first question; and our second, though less explicit, appears (from its example) to have the same focus: the imagined transformation is to a different species. Now for Aristotle — and this seems plainly true — continued existence as a member of the species one is in is at least a necessary condition of continued personal identity.13 For Aristotle it is probably (if we add a fuller account of "continued") a sufficient condition as well14 - though here he might encounter more opposition. In any case, it is with the fulfillment of this necessary and central species condition that Aris- totle is concerned in these passages; and it is this condition that we 92 MARTHA G. NUSSBAUM shall be examining in the balance of this chapter. He is asking what transitions are such that they create a form of life that is not human at all, and so, a fortiori, not livable by a person whose identity includes humanness as (at least) a necessary element. Certain lives that gods cherish are unavailable as human lives. We can link these passages with others in which Aristotle crypti- cally contrasts the human form of life with the lives of both gods and beasts: for example, with Politics 1253a! ff. (to be analyzed in detail later), in which he states that it is part of the nature of a human being to have social or political concerns, and that a creature who totally lacks such concerns must be "either a beast or a god"; and with the similar passage in EM vn, in which Aristotle insists that ethical assessment itself is a part of the human form of life, poised as it is between the beast and the god:15 "And just as there is no ethical excellence or deficiency for beasts, so not for gods either: the one is above the excellences, and (the deficiency of) the other is not the same thing as ethical deficiency" (EN vn.i, ii45a25~7). Here Aristotle is, once again, pursuing our first question, but in a slightly different way, asking what forms of life, and what creatures, we count as human, and on what basis; what distinctions between the human and other kinds we employ when locating a creature or a life. The question about essential human nature, having been asked from the point of view of the subject who wishes to continue in existence, is now asked from the point of view of the classifier of lives, who wishes to know what beings or putative continuants of beings to count as beings of the same kind. It would be getting ahead of our argument to try to determine where the account of human nature comes from in these passages, and whether it is or is not independent of ethical beliefs. For the arguments that in the end explain these judgments are arguments that we shall examine later, concerning the political nature of the human being. But we must at least try to give these statements some philosophical motivation, asking what genuine human problem or problems they address, and how. Is there anyone who might have wished for a transformation into a godlike condition, not fully realizing that this condition is incompatible with his own continued existence as just the sort of being he is? Who might have made theoretical statements that clash with some of his deeper beliefs about identity and continuity? The answer is not simple: for many human beings, in many ways, Aristotle, nature, and ethics 93 at many times, have been guilty of some such incoherence. But one prominent opponent of Aristotle's, in the passage from -EJVvm—ix, is certainly Plato. And, if we try to see how considerations of our human species nature might be brought to bear against some of Plato's arguments, we will be in a better position to begin to see that Aristotle has identified a serious problem here, and what sort of problem it is. Let us, then, consider the following passage from the Phaedo: My friends, I have not convinced Crito that I am this Socrates here, the one who is now talking philosophy with you and setting in order each item of our discussion. He thinks that I am that being that he will shortly see as a corpse; he even asks how he should bury me. But I have already given you a substantial argument to the effect that, when I drink the drug, I shall not remain with you, but depart to the happy life of the blessed . . . When you see my body being either burned or buried, you must not fret as if I am undergoing something fearful. And at the burial you must not say that Socrates is being laid out, or carried out, or buried . . . You must be confident and insist that you are burying my body, and bury it in whatever way you like and consider most appropriate. (1150-E) Socrates here expresses the view that death, which he has defined as the separation of soul from body, is something that he can perfectly happily wish for himself; he urges his friends to wish it along with him. Aristotle's remarks about continuing press us to ask, however, whether this future post-mortem state is one in which the human being Socrates who is wishing, and whom his friends love, will continue to exist. We might ask, for example: how could Socrates talk philosophy and set the discussion in order without a body? But if he could not, and if the interlocutors believe that the ability to participate in the give and take of dialectic is essential, as part of being human, to the Socrates they know (as even Socrates strongly suggests), then we must concede that there is some incoherence in the wish that Socrates go ofTfrrjm his body to the happy life of the blessed. The wish for this event might seem to have just the peculiarity mentioned by Aristotle: Socrates is wishing not for his own life, but for someone else's, the life of an intellectual substance that is non- identical with him.16 In order to resolve this question we need an account of what it is to be a human being. But what we require from such an account, if it is really to speak to what troubles us, is an evaluation of elements of lives, one that will tell us what among the properties and doings of 94 MARTHA C. NUSSBAUM Socrates are the most important, the truly essential ones, the ones without which he could not continue. But it would seem that there is no answer to this question from a totally external perspective. A certain entity, let us grant to Plato, continues to exist - namely, an intellectual soul. The question is, whether this soul and its life contain enough of what makes Socrates Socrates to count, for us, as him. This is an evaluative question, and in the broad (Aristotelian) sense of the word, an ethical one; it is the question of what are the most important things in human living. And, if there is a resolution of the difference between Plato and Aristotle here, it will not come from the decree of an external judge, who will say, "Yes, that's right, this thing is (or is not) still human." It can only come by one party's showing the other (for example) that the internal view he actually has, on which he relies in life and action, is, taken as a whole, not consistent with his theoretical claims about the sufficiency of intellect for human survival. In any case, the inquiry, to be productive, will have to be internal (that is, conducted in and through human conceptions and beliefs) and evaluative, if it is to be at all relevant to the concerns Simmias and Cebes have when they consider whether they ought to wish their friend well on these terms. This idea that an account of what it is to survive as a human being (or: as a person) is an evaluative matter, requiring an evaluation of the elements of lives as to which are the most important, may appear alien to a dominant tradition in modern discussion of these problems — according to which questions of personal continuity are metaphysical questions. I think that, even where this philosophical tradition is concerned, matters are complex. For the best and deepest of the metaphysical arguments brought forward seem to contain an evaluative component: that is, they ask us (implicitly or explicitly) to consider which functions of an alleged human being are so important, so central, that their absence will mean the absence of a human being or human life. One could, I believe, support this claim by the analysis of many such arguments - including, probably, those of Williams himself. Furthermore, whenever such accounts are called upon actually to play a role in human practices - as, for example, in current debates about the definition of death and about the lives of severely demented or damaged patients, the evaluative nature of the arguments involved becomes especially clear — as medical ethicists and doctors ask themselves whether the functions we see before us constitute enough of what we value to count for us Aristotle, nature, and ethics 95 as the life of a human being (or: a person).17 The ancient tradition, then, is not only closer to the modern tradition than it might seem, it is also the one that emerges when people actually start to talk about their lives.

III. THE TRADITION Aristotle's concern with species and their ways of life does not arise in a vacuum. His appeals to the reader's beliefs about beasts and gods suggest that he is arguing in a tradition of reflection about identity that is old and pervasive in Greek ethical thought. This tradition tells stories about transformations from and to the human, appealing to the imagination of the audience to judge which are the features whose presence or absence makes the difference between humanness and its absence.18 It tells stories, too, of our near- relatives: of non-human anthropomorphs whose form of life is sufficiently different from our own that we judge that these beings do not share a "nature" with us. Such reflective stories are a preoccupation of Greek myth: they take their bearings from the traditional idea of phusis, "nature," the idea, that is, of a being's essential nature or character (that of its kind in almost all cases), as revealed in its characteristic activities.19 They are stories of communal self- definition and self-clarification, told to humans (especially to young ones) in order to initiate them into, or to keep them within, or, for those securely in, to promote reflective understanding of, the way of life that is constituted by the boundaries that the stories display. By the beast on the one hand, the god on the other. On the side of the god we have the countless stories of the Homeric gods, anthropomorphic beings rather similar to humans in their desires and characteristic activities, but differing only in their power and, above all, in their invulnerability to death. These stories (and related stories of transformations of former humans into gods) ask us to imagine for ourselves how much of what is important for us is inseparable from our mortality and our finitude. They ask us to what extent love, friendship, and the virtues could have a place in a non-finite life that would, accordingly, lack certain opportunities for risk, effort, and sacrifice. (Aristotle attaches himself explicitly to this tradition when he speculates in EN x about whether gods could have the human virtues, concluding that characteristics such as justice and moderation would have no place in the context of such a 0,6 MARTHA C. NUSSBAUM life.)20 By these means the stories lead us to grasp the extent to which limits make us who we are, and to see the importance of the practices and values constituted by these limits in making a life that we can call our own. On the side of the god, again, we have a more radical and severe tradition of reflection about divinity, present in Xenophanes and continued in Plato's Republic, which criticizes the gods of the stories for being more like us in their ways of life, than a being among whose essential characteristics are immortality and self-sufficiency could possibly be. These story arguments ask us to imagine a being who is truly perfect and complete; they then argue that such a being would surely not perform the recognizable human actions and motions ascribed to Olympian gods, such as telling lies, changing shape, travelling from place to place.21 Such a being would have no reasons for motion from here to there at all, no motive for self-change. These stories ask us to see that even the idea that one might retain a recognizable (partial) individual identity (say, as Heracles) while passing from the realm of finitude to the realm of the god is suspect. The boundary is starker, more complete than that; our needs and limits explain far more about us, even, than popular legends acknowledge. The popular legends - which might themselves have helped many young people to avoid the incoherence of wishing for immortality while cherishing as essential the parts of human life that cannot survive immortalizing - is now itself charged with incoherence, in that it seems both to assert and to deny that mortality and related limits are essential in making us who we are. The side of the beast is equally revealing - and, in a sense, even more so, since downward movement is possible in a way that upward movement is not. Thus the description of this boundary affirms as essential what might possibly have been discarded had humans decided not to accord it much importance. That beasts lack reason and language is proverbial and, in a sense, too easily seen to serve as the stuff of myth (though we shall see its function later in a very significant kind of story). More dramatic and more central to moral education are thought experiments in which the bestial creature is hypothetically granted speech and a kind of reasoning, and in which the story explores the difference between even this sort of reasonableness and a fully human use of practical reason. These stories usually focus on the importance of ethical Aristotle, nature, and ethics 97 awareness itself, and a responsiveness to the claims of others, in making a being that is truly human. Odysseus encounters Cyclopes, anthropomorphic creatures who live in isolation from community. "They have no assemblies that make decisions, nor do they have binding conventions, but they inhabit the summits of lofty mountains . . . and they have no concern for one another" (Od. 9.112-15). Above all, they entirely lack sensitivity to the most solemn or binding conventions that, in the human world, link one being with another. For they have no awareness that if you have received a man as a guest-friend you ought not to eat him. As Euripides says, they have no paideia (Cycl. 493) — nor do they have, it appears, the responsiveness to others that would enable them to receive a human paideia. Although these creatures look like humans, neither the characters nor the audience of Homer's poem classify them as human. In Aeschylus' Oresteia, the Erinyes begin the final play as strange hybrids of the human and the bestial (or bestial/divine), both in appearance and in action. Although in form they resemble human women, they also resemble dogs, sniffing after their prey with dripping eyes, thrilled by the scent of blood. Their deeper beast- liness shows in their speech, rational yet solipsistic, impervious to community, obsessed with revenge. When they allow themselves to engage in social and ethical discourse, giving and receiving promises, yielding their revenge to the "holy honor of persuasion," taking their place in the city and wishing its people well, they are physically transformed in analogous ways. Now Eumenides (so named for their kindly intentions) they cease to growl, to crouch, to sniff after blood. Standing erect, they "depart according to nomos," (1033), robed in human clothing, a gift from the citizens whom they promise to protect. This process has its inversion. In Euripides' Hecuba, as I have argued elsewhere, the story of Hecuba's transformation from woman to dog serves as the occasion for an exploration of the dark side of the Aeschylean reflection about the connection between humanity and social/political concern.22 Stricken by betrayal, immured within her solitary scheme of vengeance, cut off from all love and trust for others, Hecuba becomes, morally, a dog. The forecast that she will become a bitch with fiery eyes has already, in its deeper meaning, been fulfilled. For to become a being who trusts nobody, who accepts no promise and hears no persuasion, is already to be some- 98 MARTHA C. NUSSBAUM thing other than human. The human being is by nature a political being. These stories call for ethical reflection, connecting the reflection with our related questions of species membership and personal continuity. They ask us to evaluate the role of promising, of other- related responsiveness, of trust and sociability, in our sense of our identity. We should imagine these tragedies as they were staged, in a theatre in which each spectator looks across the staged action to the faces of his or her fellow citizens, as the assembled group, imagining, thinking, and feeling together, seeks to teach its identity to the young and to affirm it in themselves.23 The communal nature of this reflective process, and the fact that it works through emotional responses to tragic events, suggests that the process of participating in the exercise is already an affirmation of its content. If you care about this civic festival and respond to the events staged before you, you are responding as a political/social creature, and thus affirming the importance of the political in your own life. It would be peculiar to engage in a reflective exercise of this sort and yet to conclude that Cyclopes can be humans after all, that the political is non-essential to one's sense of the being one is. The "argument" is self-validating in a deep way; you cannot withdraw your assent from its conclusion, without withdrawing from the entire form of life that, as a procedure, it embodies.

IV. ONE PHILOSOPHICAL PRECEDENT! THE 'PHILEBUS' One example of the philosophical development of this tradition of argument will prove illuminating. For, as our literary cases focused on the role of sociability and responsiveness to others in human nature, so this one focuses on practical reason, Aristotle's other primary focus. Its concern with a hedonistic opponent and its strategy against the opponent make it appear to be the direct ancestor of Aristotle's argument in EN 1.7. And its explicit argumentative structure will show us, even more clearly than our literary cases did, exactly what is at stake in arguments of this sort, what type of foundation for ethics they do - and do not - provide. The argument occurs towards the beginning of Plato's Philebus. Its interlocutors are Socrates and the young Protarchus ("first-beginner");24 Protarchus has initially been drawn to the extreme hedonist position of Philebus, who asserts that pleasure is the (the only, the complete) Aristotle, nature, and ethics 99 good for human beings. Socrates, defender of reason and intellect, now attempts to adjudicate between the claims of pleasure and intellect, respectively, to provide the good life for a human being. (The specifically human nature of the subject matter is stressed at 11D5 and 19c). The interlocutors have agreed that the good life for a human being must be complete (teleon), sufficient (hikanon), and choiceworthy (haireton). Socrates now asks Protarchus to join him in a thought experiment. Let us imagine, he says, you, Protarchus, living a life full of large pleasures, but altogether bereft of reason and intellect. (Socrates stresses that the test will be carried on "in you"; it is important that Protarchus should try to see himself living the life.) This looks appealing to Protarchus at first glance, sympathetic as he is to the claims of hedonism. He says he would gladly accept such a life, and he would think himself lacking in nothing if he had it (21A). Even if all reasoning and thinking were omitted, "I would have everything, having pleasure" (21B2). Socrates now reminds him that, by omitting reasoning, they are also committed to omitting from Protarchus' life such things as the belief that he is enjoying himself, the memory of pleasure, the ability to calculate for future pleasure (2IB6-C6). In fact, Socrates now suggests, what Protarchus has left on his hands is "not the life of a human being, but one belonging to some jellyfish or some one of the living creatures in the sea with bony bodies" (21E6-8). He now asks Protarchus "A life of this kind isn't choiceworthy for us, is it?" (21D3). Protarchus replies, "Socrates, this argument has left me altogether speechless." The claims of practical reason have been acknowledged. What has happened here? Clearly some significant ethical work has taken place. Protarchus has been rapidly moved from a com- placent acceptance of extreme Phileban hedonism to a state of uncertainty; after a parallel thought-experiment concerning intellect, he is ready to concede that any good life for a human being must contain both pleasure and intellect. Clearly, too, the work has been done without moving outside Protarchus' own system of beliefs about himself and his life: the question is always what he would say, choose, accept. Furthermore, the responses elicited from Protarchus look more like judgments and evaluations than like statements of fact: for he is asked "Would you receive" such a life; and "Is this life choiceworthy?" In all of this, the role played by considerations of personal continuity and species membership seems to be as follows. IOO MARTHA C. NUSSBAUM Protarchus has been asked whether he would choose for himself a certain sort of life; he is asked to imagine himself living it. It now develops that Protarchus really believes (though, not being sufficiently reflective, he did not realize it) that this life is really not one that could be lived by a human being at all. It can be lived by, it is the life of, anyone who would live it would have to be, a lower sort of animal. But if this life cannot be lived by a human being, it cannot be the object of Protarchus' own self-wishing and self-choosing. The good life for him (for us) must first of all be a life for him (for us). Protarchus has been brought to see that he really has believed all along that any life that could be a life for him must involve certain uses of practical reason. Even the life of pleasure, if it is to be properly human, cannot do without thought and memory. Protar- chus' Phileban position had not taken account of some beliefs that he, Protarchus, deeply holds. Socrates' appeal to identity and species nature gets him to realize both that he holds them, and what this means for his ethical choices. He is left speechless by the surprise of self-discovery. No attempt is made to correct his beliefs by appeal to some fixed point on the outside; the aim is simply to adjust the totality of his beliefs. We could go further. The success of Socrates' argument seems to depend not just upon the nature of Protarchus' beliefs, but, more specifically, upon the nature of his evaluative beliefs. It would, in principle, be open to Protarchus to choose the life described by Socrates: no external barrier and no matter of fact seems to prevent this. What prevents Protarchus from seeing himself in such a life is the value that he attaches to matters of the intellect: to argument, to calculation, to memory. (Philebus might have tried, at least, to answer differently.) The life described seems too impoverished to be human because of what he cares about and thinks to be good, because he cares about these things so much that anyone who lacks them could not, he thinks, be identical with him, and would have to rank as a lower form of life. The question, "Could you accept," is a way of asking him about his values; and the statement, "This is not a life for a human being," is an evaluative judgment. In one respect, this argument seems different from the dramatic cases we have considered. For here, consistently with the customs of Socratic elenchus, the argument is with a single individual and not a group, and it works with this individual's beliefs, rather than the group's endoxa.25 (Here, as so often, we will find Aristotle returning Aristotle, nature, and ethics i o i to tragic traditions, in reaction against some aspects of Plato's procedure.) This difference is significant — and would be even more so if the content of the argument concerned sociability rather than individual practical reason. (An elenctic argument on that issue would not be self-validating in the way the communal argument is, since the mere fact of participating in personal self-examination would not by itself imply acceptance of the fundamental importance of ties to others.) But we should not overpress it. For Protarchus defines himself as a human being (cf. 11D5, 19c), and thinks of the shellfish life as a life "not choicewor thy for us" (21D3). He seems to accept the (tacit) premise that the humanness of a form of life is a necessary and central condition of its being a life for him, a life in which he could see a being identifiable as him. Conversely, the members of a tragic audience reason and respond as individuals, though aware of the group's presence; they must decide, one by one, what the tragic events mean in relation to their own lives and beliefs. The responsive imagination of the spectator works within and alone, however helped by, and committed to, the presence of others.26 The Philebus argument has the same self-validating character as the tragic arguments. For its procedures embody a respect for practical reason; they acknowledge the importance of practical reason to living humanly well. So, although it is, in principle, open to Protarchus to reject its conclusion, he cannot consistently do so without opting out of the form of life that the Socratic procedure exemplifies. Philebus does opt out, so he is to that extent not refuted. (No doubt he could be by a more extensive scrutiny of the role he implicitly accords to practical reason in his discourse and activity.) This self-validating structure, in addition to the argument's concern with basic considerations of personal identity and continuity, com- mends the argument as a good "foundation" for ethical inquiry. If we now return to the questions with which we began, we can see that the tragic and Platonic arguments do not try to move outside of ordinary evaluative belief to an external perspective of scientific fact. In speaking of human nature, they deal with beliefs that are both evaluative and, in the broadest sense, ethical - beliefs about what is worthwhile and worthless, liveable and not liveable. They are matters for communal judgment and decision, not for independent investigation and discovery. They are thoroughly internal to the community, and they serve to explain it to itelf. They IO2 MARTHA C. NUSSBAUM are less controversial than most ethical propositions — in particular, one cannot opt out of them without opting out the procedures of inquiry that support them. And they are embodied in stories that have shown their ability to move the imaginations and hearts of people in many different times and places. Their age and historical depth is part of their authority. Yet they can be disregarded and denied, especially on account of theoretical prejudices or political interests. In this sense, important work is done by reaffirming them: and in this sense, affirmed, they provide a foundation for further work. Now let us see where Aristotle stands in this tradition, examining his two most famous appeals to human nature: his claim that the human being is by nature political, and the "human function" argument in Nicomachean Ethics i, which defends the role of practical reason.

V. THE POLITICAL NATURE OF HUMAN BEINGS Williams claims that Aristotle's view of human nature, appealing as it does to a natural teleology, attempts to take up an extra-ethical, and even an extra-human, perspective on the human situation. We can begin to respond to this claim by pointing out that Aristotle's metaphysics of nature, and his biology, are neither value-free nor external. There is nothing anywhere in Aristotle's work precisely corresponding to a modern distinction between fact and value; and, furthermore, science, as well as ethics, is "internal" for Aristotle in the sense that it is the attempt to give an intelligent account of human experience of the world. Aristotle explicitly announces that his method in ethics is just the method he uses in all other areas: to preserve the greatest number and the most basic of the "appearances" - human perceptions and beliefs - on the subject.27 I am far more sympathetic with this as an account of what science achieves than Williams will, no doubt, be. His own distinction between the methods and results of science and those of ethics seems to me far too strong; but this is not the place to argue that point.28 Even the unsympathetic should, it seems to me, grant that a philosopher who does advance such an internal account, even for science, is rather unlikely to have sought a completely external foundation for the ethical. But even if this point is accepted (and, since it is controversial, it Aristotle, nature, and ethics 103 may well not be), we still need to say much more about whether the view of human nature in the ethical works is built upon beliefs that are specifically ethical in character — beliefs, that is, that concern the general question of the worthwhileness of various forms of life. And we need to see precisely what work the account of nature does in Aristotle's ethical project. "The human being is by nature a political being." With this claim, in the first book of the EN, Aristotle defended his ethical choice to make the self-sufficiency involved in human eudaimonia a communal and not a solitary self-sufficiency. That choice was controversial: for example, it might be rejected by Plato. We want now to determine how and from where it is defended. The social or political nature29 of the human being is brought forward against a Platonic account of self-sufficiency in two further contexts: in the accounts of the value of philia in the ethical works, and in the account of the naturalness of the polis in Book 1 of the Politics. We will have to look at these texts closely, in order to grasp the nature and tone of Aristotle's procedure. In the passages in the friendship books, it is abundantly clear that Aristotle takes himself to be setting down pervasive ordinary beliefs (which some philosophical views ignore); and that these are taken by him to include, prominently, evaluative beliefs, beliefs concerning what is worthwhile in a human life. In .EJVVIII, for example, he defends the claim that philia is "most necessary" for life in the following way (emphases mine):

For without friends nobody would choose to live, even if he had all the other goods. For even rich people and people who hold offices and the power of rule, it seems (dokei), have a particularly great need for friends. For what good are all those goods, if benefitting is taken away? But this happens above all, and in the most praiseworthy way, towards friends. And how could such a person be protected and saved without friends? For the greater he is, the more danger he is in. And in poverty and other misfor- tunes they think that the only refuge is one's friends. (1155a6-i2) Shortly after this, Aristotle defends the claim that friendship is a part of human nature:

It seems likely (eoiken) that it belongs by nature {phusei enuparchein) to the parent towards the child and the child towards the parent; not only among human beings, but even among birds and most animals; and that it belongs also to members of the same nation towards one another, especially among 104 MARTHA C. NUSSBAUM human beings. That is why we praise those who love and benefit other human beings (tousphilanthropous). And one might also observe in one's travels to distant countries the sense of recognition and affiliation that links every human being to every other. It seems likely that philia hold cities together as well. (1155a 16-23) In these passages, which defend the claim of philia and, in general, of interpersonal responsiveness, to be a necessary and natural part of human life, Aristotle leaves us in no doubt that he is recording and examining ordinary beliefs. The italicized phrases attest to this clearly. And, furthermore, it is rather clear that the naturalness of friendship is put forward not as a matter of detached discovery, but as a matter involving evaluation and judgment. It is, of course, in some sense physically possible to live without others. Aristotle simply points out that we do not choose that way and we do not approve of people who move in that direction. Furthermore, whereas we see ourselves as sharing with other animals a certain part of philia — the tendency to love and care for one's children or partner — we think that the human as such is distinctively characterized by a broader concern - by the ability to recognize and respond to the humanness of every other human. This is not a knock-down argument against someone who would insist on denying the claims ofphilia. But it raises considerations that might make such a person hesitate and ask himself whether he had explored all of his relevant beliefs about what makes a life humanly liveable. The reference to travel, with its "one might observe," looks like an invitation to the interlocutor/student to engage in a Protar- chan thought-experiment. Before you say that you can do without others, remember what it is like to travel to a strange place, and how terrifically glad you are, in such circumstances, to see any human face at all. If this were put forward from the outside, as a neutral anthropological observation, it would have little force or relevance for an interlocutor whose behavior exhibits a different pattern. But as an invitation to each reader to consider, within him or herself, the depth of certain evaluative beliefs, and their implications for his project of continuing as a member of a certain kind, it has considerable force. The nature of Aristotle's argument becomes even clearer when, in Book ix, he argues against an opponent who has claimed that the best life is a solitary life, insisting that the eudaimon, the flourishing person, has no need of friends. This is how Aristotle replies: Aristotle, nature, and ethics 105

It seems rather peculiar (eoiken d'atopoi) to give all good things to the eudaimon and to leave out friends, which seem (dokei) to be the greatest of the external goods . . . And surely it is peculiar (atopon) to make the happy person a solitary; for nobody would choose to have all good things in the world all by himself: for the human being is a political creature (politikon gar ho anthropos) and naturally disposed to living-with (suzen pephukos). (1 i6gb8~io, 16-19) Here we find, quite explicitly, an appeal to the essential nature of the human being, called in, as in EMi, to do some work in an ethical argument. The context, however, once again stresses the dialectical and "internal" nature of the procedure. Eoiken and dokei indicate the nature of the support Aristotle is claiming for his argument. And the emphasis on the word "peculiar" — which we have already seen once before, in a similar context - suggests that we have here not so much a knock-down proof of something from some fixed area of external fact, but rather an appeal to the reader to consider whether the opponent's project does not deal in an odd sort of incoherence, promising the eudaimon life to us, while depriving the eudaimon person of something without which we (including the opponent) believe no human life would be possible or worthwhile.30 The connection between the possible and the worthwhile here is of considerable interest. I have said that Aristotle's general strategy is to show that certain ethical proposals go against our deep beliefs about identity, asking us to choose a life in which (we really believe) no person identical with us could be. I have also shown cases in which the ethical proposal is ruled out by pointing out that no human being would choose such a life. We might think that these are two quite different strategies: for surely to ask whether I survive in a life is not the same as to ask whether I consider that a life worth the living. Now these would be two very different strategies if questions of personal identity were matters of fact that had no connection with judgment and choice. But is this so for Aristotle? (Is it so in any account of personal identity that has philosophical power?) Surely it was not so in the Philebus passage: for what told Protarchus that he would not survive in that life was just the fact that he would not consider that life worth accepting; it would not contain activities that he values so highly that he cannot regard himself as himself without them. What was really at stake, earlier, in our questions about the Phaedo passage, was, once again, an evaluative matter. The answer to the question whether that disembodied intellect is or is not Socrates will depend upon how highly the interlocutors rank 106 MARTHA C. NUSSBAUM those philosophical and human functions that Socrates could not do without his body: on whether they rank them highly enough that they are going to dig in their heels and insist that any being who lacked them could not be Socrates. In these two cases we are invited to imagine that some living being is living the life in question, a being that has at least a prima facie claim to be identical with the being doing the thought-experiment. In a case like this, the answer to the question about survival seems to require, and to rest upon, evaluative judgments. And in these Aristotelian passages the same thing seems to be going on. The claims that friendship is necessary and natural are joined to, and supported by nothing beyond, claims about our preferences and choices. Could we discover tomorrow that the opponent is telling the truth, that the human being is really by nature solitary and that the good life is therefore a solitary life? It seems that the nature of Aristotle's procedure here effectively rules out this possibility, short of a total change of beliefs and ways of life. Could we turn out to be naturally fitted for a life that none of us would find worth the living, and in which none of us would feel that the concerns and values with which we identify ourselves survive? This seems impossible, since to find out what our nature is seems to be one and the same thing as to find out what we deeply believe to be most important and indispensable.31 If the opponent really did not care at all about other people and happily opted for a solitary life after considering Aristotle's arguments, Aristotle could only say to him that he is not like us and will not be included by us in our community. He is asked to assess, in himself, the depth of a concern that most of us share. We must now turn more briefly to the most famous appeal to the political nature of the human being, the one in Politics i. It seems to me to tell the same story, though the ethical work it does against rival positions is less clearly set out. In this passage Aristotle advances two supportive considerations for thinking that to be politikon is part of our nature. One is a straightforward appeal to deeply rooted ordinary beliefs: It is evident that. . . the human being is by nature a political animal, and that the person who is citiless through nature (apolis dia phusin) and not through luck is either an inferior creature or greater than a human being: just like the person denounced by Homer as "without fraternal ties, without customs, without a hearth" - for this person is in his very nature Aristotle, nature, and ethics 107 (hama phusei) of such a sort and a lover of war, being "unyoked" like a piece in the dice-game. (125331-7)

Shortly before this, Aristotle had made reference to Homer's Cyclopes, whose specific difference from us is constituted by their lack of social and interpersonal concern alone. He thus reminded his audience of the depth of a tradition of thought about the human being, according to which an anthropomorphic being who lacked social concerns would not be classified as human. Now he goes a step further, considering a Homeric line32 that refers to a being whom an external scientist would classify as a member of the species homo sapiens — not, then, a mythical hybrid like the Cyclops — and reminds his audience of the way in which their hero Achilles (or their great poet Homer) denounces this being and relegates him to a distant place of inferiority. If it is really his nature to be a solitary and to love war for its own sake, not just as a means, then, Aristotle says, he is either below or above our kind, but he is not of it. If we encountered such a being, we would not count him as one of us, accord him the treatment we accord to our fellow humans. This passage clearly works within traditional beliefs, and emphasizes this fact by appeal to the fount of tradition, Homer himself. It works with beliefs that are clearly "internal" and evaluative: Achilles judges that he hates and despises this creature; and this, not some scientific inquiry, is the source of the judgment that he is less than human. The obscure reference to the dice-game33 suggests a further evaluation. The suggestion is that a human life is so thoroughly a life among others that it gets its meaning from those practices. Just as a piece in a game is not even itself, has lost its point and identity, when severed from the rules of the game, so the solitary human will be pointless and meaningless, bereft of her identity. This, once again, can only make sense as an internal and evaluative judgment. It is only from the point of view of social practices, and human immersion in, concern for, them, that the "unyoked" human looks incomplete. The argument that follows this one might initially seem to give more hope to the searcher for an external perspective: for Aristotle now defends the claim that our nature is politikon by pointing to the fact of the presence among humans of a language equipped to express ethical conceptions. Surely, he suggests, that language IO8 MARTHA C. NUSSBAUM would not be there without some function; its presence testifies to the importance of the social in our lives. Now, first of all, this argument does not really appeal beyond ordinary belief to science or metaphysics. As Newman stresses in his commentary on the passage, Aristotle, in referring to the fact that the human being alone uses language, is referring to an extremely common topos in contemporary rhetoric;34 this, no doubt, reflects a widespread ordinary belief about the importance of language as essential trait and differentia. Does the argument refer, even so, to a natural fact that can be established independently of ethical beliefs? It seems doubtful. For the beliefs about language to which Aristotle points already contain, it seems clear, an evaluative (and, broadly speaking, ethical) component. The claim that language is idion to the human reflects a rhetorical tradition in which language is made the important thing, the valuable gift that lifts us above the beasts. To a person who says or believes this, Aristotle's argument says that it is implausible not to acknowledge, also, the deep importance of interpersonal and social concerns: the two go together, they are "made for" each other;35 and it is implausible to exalt the one and deny the other. But, to a person who does not give language any special importance or value, it is not clear that Aristotle's observations would have, or ought to have, any force at all. There are many natural items that are idia to the human species, without becoming elements in an account of our essential nature; for many of these are trivial. It is only because language is esteemed that the activities with which it is associated have a claim to a high ranking. So this argument, like the others, seems simply to show a connection between two traditional evaluative beliefs. Aristotle's arguments, then, ask us to recognize the depth and pervasiveness of certain human beliefs and practices, claiming that they are constitutive of humanness, as we conceive it. Once we recognize how thoroughly sociability permeates our lives we will, Aristotle thinks, acknowledge that any search for the good life must go on inside a context of relatedness. The defender of solitary self-sufficiency is given a challenge: she must either show that our conception of our identity is not what Aristotle says it is — that it does, after all, accommodate Cyclopes and totally non-social anthropomorphs; or she must persuade us that other aims and beliefs require us to revise these deep beliefs about identity; or she Aristotle, nature, and ethics 109 must grant that the proper sort of self-sufficiency for us to aim at is Aristotle's sort; or, finally, she must dissociate herself from the other parties to the argument, announcing that her conception of identity and nature is truly and deeply not the same as theirs, and acting, henceforth, accordingly. This means that, from the group's viewpoint, she too must be or become the sort of being that Achilles shuns, apolis dia phusin, unresponsive to human cares. The fact that she had embarked as student or reader on this course of ethical lectures, whose goal is the communal clarification of ideas about the human good for the purposes of advancing the aims of the political - whose job it is to give a good life, in community, to all who are equipped to receive it - makes it unlikely that she will take this most radical of the alternatives. Thus, as in the tradition, the procedures validate their conclusion. But does this argument do any real work? It seems clear that it does. For the claim that each of us has, and lives by, a conception of personal identity according to which each of us is not only a "this" and "separate" (claims that Aristotle has already in a similar way established), but also a being essentially related to others by ties of recognition and concern, is a claim that a great deal of philosophical theorizing in ethics has not really taken to heart. If it is accepted, it follows that many philosophical projects and a great part of economic writing about human motivation and behavior, will need to be rewritten. For even the opposition to mainstream economic thinking, and to its focus on rational self-interest, usually accepts the opposition's characterization of the boundaries of self and other; but it is Aristotle's claim that in real life selfhood is much more other- entwined and other-identified than these theories let on. The claim made to (let us say) the economic theorist is that, even while defending a theory that treats the self as something essentially isolated from others, he or she actually has, and lives by, a different conception, a conception that he could not give up without understanding what he himself is in a new way. The shape of the argument is similar to that used by John Rawls against utilitarian moral theory.36 (Rawls himself distinguishes his own use of an ethical "concept of the person" from Aristotle's use of a theory of human nature; but his Aristotle resembles Williams'; and if Aristotle is understood as he has been here, he is doing something very close to what Rawls is doing.)37 Rawls argues that utilitarianism IIO MARTHA C. NUSSBAUM requires denying the moral salience of the separateness of persons and treating all desires and satisfactions as fusable into a single system. This, he then argues, is at odds with deeper beliefs about the person that the utilitarian himself shares and lives by, beliefs according to which a person is a separate unit, as Aristotle would say, a "this" and "one in number." Once again, the argument is claiming that the acceptance of a theoretical position entails a cost that the proponent of the position will be unwilling to pay - for these are beliefs in terms of which he defines himself. (Whether the argument has the self-validating character of Aristotle's is less clear, though there would certainly be something to be said in favor of the conclusion that it has.)

VI. THE HUMAN FUNCTION ARGUMENT38 Now we arrive at the material that is most often used to support the picture of a move from natural fact to ethical value, the material on which most accounts of Aristotle's allegedly metaphysical view of human nature most prominently rely. This is the famous "human function" argument in JVicomachean Ethics 1.7. This argument bears a striking resemblance to the Philebus argument which we have examined - though it is cast in the communal form of Aristotelian dialectic, rather than in the one-to-one form of the elenchus. We notice, first, that the context of Aristotle's argument is similar to that of the Philebus argument. It occurs towards the very beginning of a work concerning the good life for a human being, a work that will concern itself with the critical evaluation of proposed constituents of such a life and the construction of a best life out of the available constituents. It occurs prior to any detailed discussion of the life's content. It is also closely associated with the rejection of an extreme form of hedonism that would attempt to construct the good life out of pleasures alone. Aristotle has been telling us that one very popular view about the good life is that it is a life of pleasure. This, he says, is tantamount to "choosing the life of dumb grazing animals" (iog5b2o). This suggests that there is, in his view, something less than fully human about a life devoted to pleasure alone, the life that many people claim, at least, to want. It seems to leave out something that we think a human life should have; this something is hinted at in the word "choosing," prohairoumenoi - the Aristotle, nature, and ethics 111 exercise of practical reason. There appears to be something paradoxical in the choice to live without choice, just as in the Philebus there was something odd about addressing Protarchus' practical intellect with the question, "Would you choose a life altogether without reason?" This oddness suggests that the opponent's picture of the good life may be not only deficient by our lights, but also internally inconsistent: in his way of considering and going about things, he implicitly ascribes to practical reason a value that his argument denies it. This would give us a way of appealing to such a person and persuading him to revise his position;39 as with Protar- chus, we might be able to bring him into some measure of agreement with those he opposes, simply by working within his own beliefs about who he is. I shall now argue that the "human function" argument does this same sort of work in much more detail, attempting to establish some such basis of agreement from which people of diverse opinions can proceed to further work on the difficult matter of specifying the good life. The passage is introduced by a revealing observation: we agree in saying that eudaimonia is the best (to ariston), "but there is a strong desire (potheitai) that something clearer should be said about what it is." This remark about the limits of our agreement refers back to 1095a 19 ff., where Aristotle told us that both the "many and the refined" agree on the name for "the top (akrotaton) of practical goods," namely that it is 'eudaimonia''; and they agree that living well and doing well are the same thing as eudaimonein. "But concerning eudaimonia, as to what it is, they are in disagreement" - both with one another and, Aristotle points out, each person with himself. This situation of extreme disagreement as to the very "what it is" of the end in view seems to pose grave problems for an ethical inquiry that is going to pursue this "what is it" question, basing itself upon people's beliefs. It is no wonder, then, that in the passage before us Aristotle speaks of apothos for a more substantial agreement. Apothos is a strong desire for an absent object.40 He now continues in a more optimistic vein: "This [sc. the clearer account] might in fact perhaps come into being, if the function of the human being should be grasped." Aristotle's optimism is, apparently, justified by his discoveries in the ensuing passage. For he concludes it with the claim that it has provided an "outline sketch" for the inquiry into the human good (1098320-21). We shall return to this conclusion later; for it is very helpful in getting an under- 112 MARTHA C. NUSSBAUM standing of how Aristotle sees his accomplishment. But first we must consider the passage itself. An assumption stated at the outset is that if anything has a function or activity, then its good acting will consist in doing that thing well (ioo,7b25—28). This assumption needs scrutiny, since it has sometimes been taken to be the locus of a troublesome sort of slide from fact to value. The examples given are from the crafts: the aulos-player, the statue-maker, etc. What is being said about these craftspeople is, then, that if E is the function of the E-craftsman, then to be a good E-craftsman would be to do E well. So, presumably, a search for the specification of what good activity is for that sort of craftsman would do well to begin with an account of E; and a clear account of E would help such a search along. Now what does Aristotle mean by this notion of function or activity (ergon kai praxis)? Presumably he does not mean a task assigned to the person from without. The craft examples support no such suggestion, and an examination of Aristotle's use of function talk elsewhere would strongly tell against it.41 What would most naturally be meant by "the function or activity" of a certain sort of craftsman would be that craftsman's characteristic activity qua that sort of craftsman - the activity or activities in virtue of which he is, and is counted as, a craftsman of that sort. The ergon or praxis of an aulos-player is aw/cy-playing; this means that aulos-playing is that without which he would not be, or be counted as, an aulos-player.42 To say that the good activity of an E-craftsman is to do E, that characteristic thing, well, is not to say anything very surprising. It is just to make our by now familiar point that the search for the good for anything must remain within the confines of an account of what that thing is. It could not, logically, turn out that the good activity of an aulos-player as such was to make beautiful shoes or to play the lyre. This looks like a trivial point; but our previous examples have prepared us to consider that in certain circumstances such a trivial point might do important ethical work. Here it tells us that the search for the good activity (eudaimonia has already been identified with eu zen) of a human being must begin with an account of characteristic human functioning, just in the way that a treatise on the art of aulos-playing had better begin with an account of what the aulos is, including the ways in which it differs from other related instruments. A good active life for a being must first of all be a life for that being i.e. it had better include those activities that are essential to, definitive of, that sort of being. Aristotle, nature, and ethics 113 Aristotle now asks, rhetorically, does the human being as such have a function or activity, or is this creature idle by nature {argon pephuken)? In other words, would someone say {theie tis an) that there is something that we, as humans, characteristically (by nature) do? He is asking (in a way that strongly suggests an interest in ordinary beliefs) for a specification of a characteristic way of being active that belongs to the human being by the nature of what it is (rather than, say, by command from without) - something lacking which, then, the human being would not have its own nature, would not be itself. He now considers and rejects two rival candidates. We must study this passage with considerable care, for it has been understood in two quite different ways: What in the world might this be, then? After all, to live appears (phainetai) to be common to plants as well, and what we are looking for is something peculiar to the human {to idion). Then we must exclude the life of nutrition and growth {ten te threptiken kai ten auxetiken zoen). Next after this would be a life of perception {aisthetike tis); but this appears likewise to be common to the horse and the cow and to every sort of animal. There is left, then, the practical life of that which has reason (praktike tis tou logon echontos) . . . And this being said in two ways, we must set down the one according to activity {ten kat'energeian). {109^33-1 oo,8a6) So the ergon or praxis of the human being as such turns out to be "activity of the soul in accordance with reason or not without reason" {psuches energeia kata logon erne aneu logon, 1098a 7-8). To understand Aristotle's argument and its conclusion, we must, above all, clarify the force of the expressions "the life of E-ing" (in Greek, the -ike zoe). For these have been taken, in the literature, in two importantly different ways, and the question does not yet seem to have received the kind of attention that would be necessary to resolve the issue. A. "The life of E-ing" means the part of a total life that consists in E-ing: the threptike zoe is the self-nutritive part of a total complex life. On this story, Aristotle would be arguing that the life-activities that we share with plants and animals — nourishing ourselves, reproducing, perceiving, are no part of the essential, characteristic activity of a human being. We could have a properly human life altogether without these, and the good way of living for a human being does not involve doing these things well.43 B. On this interpretation, "the life of E-ing" means a life organized around E-ing, a life in which E-ing is the distinctive and guiding feature that gives the life its characteristic overall shape. On this reading, an E-ike zoe might 114 MARTHA C. NUSSBAUM well include G and H as well. What makes the life E-ike, as opposed to G-ike, is that E is in some way its distinctive organizing principle; around E the other component activities are structured, or by E they are infused. G and H are done in an E-ish way, or according to E.44 There are some very strong reasons for preferring reading B: 1. It is what Aristotle's conclusion actually says: he says, not that the ergon of the human being is the activity of reason, but rather that it is the activity of the soul according to reason or not without reason. This more inclusive formulation is captured by B and not by A. 2. To opt for A would entail a revision of the understanding of ergon kai praxis earlier in the argument. We had, following the suggestion of the examples, understood this phrase to mean "characteristic, defining activity": i.e. everything essential and not inci- dental, everything without which a being of a certain sort would not be a being of that sort. This would include both genus and differentia, both the shared and the non-shared elements. Reading A tells us that only the differentia is to be considered. (For there is ample and unequivocal evidence that self-nutrition, growth, reproduction, and perception are generically essential activities of the species anthropos and parts of its logos.)*5 This is tantamount to saying that an attempt to understand good clarinet-playing (to take the nearest contemporary relative of the aulos) can best begin not with an understanding of the entirety of the clarinet-player's activity, but rather with a narrow focus on just the elements of it that are not shared by other instrumentalists. In other words, we would have to omit mention of blowing or fingering or reed-making, not to mention more general aspects of musicianship, and focus narrowly on the properties of certain types of mouthpieces, etc. This would be an odd way to understand Aristotle's craft examples; and, so understood, they would be saying something blatantly false and rather absurd about where the search for good craft activity should begin in each case. Reading B, on the other hand would allow us to consider all the generic musical activities of the clarinet player, in so far as they are done according to the particular requirements of the clarinet, or in a way that is informed and shaped by the particular characteristics of the clarinet. This gives a much more reasonable picture of what a treatise on good clarinet-playing would naturally take for its starting-point. 3. B is the solution that fits best with what Aristotle goes on to do in the rest of the EN. For he very clearly does concern himself with Aristotle, nature, and ethics 115 good ways of nourishing oneself and reproducing. The virtue of sophrosune, as Aristotle emphasizes, is concerned with the appropriate way of doing these shared things. "Sophrosune and akolasia are concerned with those pleasures in which the other animals also partake" (m8a23-5; cf. also 1104.b34.-5). And, in discussing that virtue, Aristotle even remarks that a creature who did not find pleasure in these activities or did them too little would be "far from being a human being" (mgag-io). His point here is that to be human at all, to be part of a human life, they would have to be done according to, or not without, some reasoned plan or other, however bad or haphazard. {Akolasia, we recall, is a reasoned disposition.) To be done well, they would have to be done in accordance with practical wisdom. Since, as I hope we can agree, Aristotle makes activity according to sophrosune an end of intrinsic value in the best human life, in EN I-IX at least, it is clear that he does not omit the nutritive and reproductive life — in so far as it is pursued in a reason-infused manner. 4. The philological evidence concerning the meaning of an expression such as "the IL-ike zoe" strongly supports B against A. It will be necessary to consider this evidence in two stages, arguing, first, that parallel expressions with bios conform to B, not A, and, second, that the use of zoe is not in this respect different from the use of bios. We consider bios first because the evidence is more ample and because we can contribute, in the process, to the solution of a debate between John Cooper and David Keyt.46 Cooper has argued that expressions like 'the Y^-ikos bios" or "the bios of E-ing" always indicate a total way of life in which E is in some way a dominant function. (He takes this to imply two further facts: (1) that an individual can have only one bios at a time, and (2) that the livers of different bioi must be different people.) Thus, to leave aside the apolaustikos bios would not be to leave enjoyment out of a life altogether; it would be only to leave out a hit focused on this, a life that has this as its dominant or organizing principle. In other words, his interpretation is our interpretation B. Keyt has argued against this, claiming that certain passages permit a single individual to conduct, at different times, or in one case at the same time, different bioi. But Cooper has made a convincing reply about these passages; and in general, if one examines the evidence, both in Aristotle and elsewhere, Cooper is correct about what these expressions normally Il6 MARTHA C. NUSSBAUM imply: an E-ikos bios is a total way of life organized around E. No evidence, even anomalous, indicates that bios can be used in a way that supports A.47 Is zoe different? Cooper conjectured that it might be:48 but since this was not pertinent to his argument he did not study the question extensively. The only evidence he put forward came from this very passage, and from a twelfth-century Byzantine author whose linguistic practices are not significant for our question.49 In fact, if one comprehensively surveys the evidence, one discovers that zoe and bios function in exactly the same way: when they are used of a type or manner of life, they always designate a total mode or way of life, organized around the item named. This is very clear in Aris- totle's usage elsewhere, and also in other fourth- and fifth-century prose authors.50 With all this in place, reading B seems strongly supported.51 If it stands, then the ethical work done by the human function argument is similar to that done by the Philebus argument. Since no life will count as a good life for us unless it is first of all a life for us, and since a life for us must be a life organized, in some fashion, by practical reason, in which all functions are informed and infused by reason's organizing activity, then eudaimonia must be sought within the group of such lives, not in a life totally given over to bodily pleasure without reason, not in the sleeper's life of non-guided digestive functioning. To say so much is not to rule out much. But the example of the Philebus and of the earlier EN passage on hedonism shows that it is to rule out a serious ethical position, and one which Aristotle believes to have serious backing. Of rather, it is not so much to rule it out as it is to show the defenders of it an incoherence in their thinking. They say that they like this life; but at the same time they really believe (if they are like most of us) that a life without the guidance of reason is not a human life. Aristotle is aware that the defender can reply by refusing to value reason the way Aristotle believes most of us do. A passage early in the Eudemian Ethics, which appears to be closely related both to this passage and to the Philebus argument, tells us that the life of bodily pleasure without reason would seem a worthwhile life, a life worth choosing, only to someone "completely slavish: for it is clear that to a person making such a choice there wouldn't be any difference between being born a beast and a human being" (i2i5b28-37). (And if "choosing" is really what is going on, the person's endorse- Aristotle, nature, and ethics 117 ment of mindlessness is not consistent; it would be consistent only if, perhaps after an initial decision, choosing were renounced and drifting adopted as a procedure of selection.) The mindless life is not ruled out by external facts of nature; anthropomorphs might lead it consistently, though most of Aristotle's actual opponents certainly do not lead it consistently with the rest of their values. For this argument, like the political argument, is self validating; by participating at all in Aristotle's inquiry about eudaimonia, an inquiry that explicitly announces that its results will be drawn from the active practical reasoning of each participant, the interlocutor grants implicitly the importance of practical reasoning in a human life. And, even for a person who does not participate in Aristotle's discussion, it will be very hard in daily behavior and speech not to validate practical reason's claims. But still, as we have said, the mindless life is not ruled out; what is established is the cost of the choice, and how deeply it is at odds with some of our firmest convictions concerning who we are. The argument establishes, furthermore, that an anthropomorph who does lead the mindless life will not be regarded by us as fully human. This does not exclude many types of irrational people, who live "with" practical reason, though not "according to" it. A bad plan is still a plan; bad reasoning, even haphazard, careless reasoning, is still reasoning. It excludes only people who live without planning and organizing their lives at all: the sort of creature, we might say, who would be the survivor of a frontal lobotomy. Such creatures are discussed by Aristotle: they are the so-called "natural slaves," anthropomorphic beings who "completely lack the deliberative faculty" {holds ouk echei to bouleutikon, Pol. I266ai2, cf. I28oa32— 3). In virtue of this natural lack, they are said to be different in nature from the rest of us (1254^20-21), and, importantly, to lack a share in eudaimonia (i28oa33), as well as in philia.52 They are said to be as far apart from normal humans as a human being is from a beast (i254ai7). These people are, presumably, people who suffer from some severe degree of mental retardation. (Aristotle may believe that many barbarians are like this. This is a somewhat appalling flaw in his perceptions, but it should not prevent us from grasping his argument.) Aristotle is saying that they are, therefore, not really human; that no human individual could survive the change to such a state; and that they are, therefore, owed an ethical treatment that is different from the treatment we owe to members of Il8 MARTHA G. NUSSBAUM our own species. Similar judgments are at work in Aristotle's treatment of "monstrous births" and in his policies concerning abortion and the exposure of severely handicapped children.53 And here we notice that Aristotle's actual judgments do not employ the concept "human being" in the way suggested in Williams' "Hylo- morphism," drawing in a non-evaluative way an absolute line around all offspring of human parents. According to Aristotle it is possible to cease to be human and still to be survived by a living something; or to be born of two human parents and not to be human at all. In what sense does this argument provide a foundation for ethical inquiry? We have already said that it rules out certain extreme ethical positions, for example, Phileban hedonism. But we can now say more. Aristotle's ethical writing aims to provide instructions for the future legislator. The job of such a legislator is to give to the people in his city the necessary conditions for choosing a flourishing human life. The city aims at making people capable of such choices. A life that is not even human at all is, a fortiori, not a good human life. And it is Aristotle's claim that many conceivable political and economic arrangements do, either wholly or in part, remove the humanity from political life by removing choice from people - either in a single sphere or across the board. Schemes in which a Platonic state holds all the property remove choices of citizens with respect to property. Schemes in which certain classes are debarred by poverty from sharing in central civic institutions remove from them the choice to exercise civic virtue. Schemes that devote too little attention to education deprive all citizens of choice in a more fundamental and general way, depriving them of a development of practical reasoning that is necessary in order for adult rational choice to be possible at all.54 In such cases the individuals involved may not end up leading a life so devoid of practical reason as to be truly animal: for in many cases other associations - the family, religious organizations, and so forth — will bring the lives of individuals over the threshold of mere (human) life, if not all the way to good life. Even in circumstances of great constraint, practical reason and virtue can frequently find an outlet, as Aristotle stresses. But we can still criticize the form of government for its failure to promote the humanity of its citizens (or its people who ought to be citizens, but are not) ,55 The focus on practical reason as an essential necessary condition Aristotle, nature, and ethics 119 of humanness, and, therefore, a basis for political assessment, provides political thought with a direction strikingly different from that provided by a focus on wealth and commodities as primary objects of the legislator's concern.56 For it tells the legislator that these commodities have their place in a human life as means to the activities of practical reason. No good is done by giving people food, money, and medical care, unless government also promotes the truly human use of these objects, in a life governed by personal choice of the good. It is no use giving poor people what they need, if one at the same time maintains repressive forms of government, or alienating structures of labor that pretty well guarantee that this food will be used for mere "grazing," the medical assistance for merely animal hearing and seeing. Feeding, hearing, and seeing are parts of a human life only in so far as they figure in the praktike zoe tou logon echontos. As Marx, reflecting on these Aristotelian arguments, wrote: It is obvious that the human eye gratifies itself in a way different from the crude, non-human eye; the human ear different from the crude ear, etc. . . . The sense caught up in crude practical need has only a restricted sense. For the starving man, it is not the human form of food that exists, but only its abstract being as food; it could just as well be there in its crudest form, and it would be impossible to say wherein this feeding activity differs from that of animals.51 Marx and Aristotle would differ in some ways concerning the concrete implications of this emphasis on practical reason; but in some especially important areas they agree. In particular, they agree in condemning societies that promote money and commodities to the status of ends in themselves, on the grounds that this confuses and inverts values, debasing humanity to the status of a mere means, and alienates individuals from the truly human understanding of goods. "Assume man to be man" wrote Marx, "and his relationship to the world to be a human one: then you can exchange love only for love, trust for trust, etc."58 Aristotle would agree. And it seems likely that one could press this sentiment about the non-commensurability of valuable things in human practical reason further, criticizing as hostile to humanness any ethical or political system based upon the idea that values are all commensurable according to a single standard. This would involve a complex interweaving of the appeal to practical reason with the appeal to social/political nature. I shall not pursue the argument further here, but this should suffice to show in what ways the human nature arguments, though neither external I2O MARTHA C. NUSSBAUM nor value-free, are both foundational in a perfectly respectable sense, and also capable of yielding substantive results.

VII. HUMAN NATURE AND HUMAN LIMITS It will by now be obvious that in my account of these two arguments I have not pursued every part of Aristotle's project regarding human nature. I have focused on his attempts to show how we divide ourselves from the beasts, rather than on his equally important reflections concerning our distance from divine and self-sufficient beings. In reality, a full account either of our social responsiveness or of our practical reason would include both elements, since the gods have no need either of social bonds or of the practical use of reason in planning a life. Aristotle is explicit on this point, insisting that the characteristic operations and excellences of practical reasoning, and of our sociability, are inseparable from the particular ways in which we find ourselves in the world, as both capable and limited.59 That we are mortal; that we have bodies of a certain sort; that we need things that we do not control; that we feel pain and pleasure; that we have a certain range of bodily desires and appetites — all this is not, so to speak, external fact, but rather a more or less permanent part of the internal perspective of human beings in many times and places. The traditional emphasis on "biped," "featherless," and so on has great point here, in reminding us that some forms of life are open to us and others not. In another related paper60 I have addressed this part of Aristotle's human nature project, arguing that Aristotle's attempt to found a theory of the virtues on a description of these "grounding experiences" of human life is not at odds with, and can incorporate whatever is true in, the cultural relativist's claim that even these basic experiences have, to some extent, been differently constructed in different societies. Valuable though this historical insight may be, we are none the less left with a great deal of overlap and convergence among cultures at the level of these experiences. We can see ourselves in their accounts of fear and grief and hunger and delight, and they themselves in ours. Aristotle, we recall, said, "One might also observe in one's travels to distant countries the sense of recognition and affiliation that links every human being to every other." This seems true, and permits the human nature project, here again, to have broad applicability. Nature comes into the ethical Aristotle, nature, and ethics 121 enterprise, here again, not as an external fixed point, but as a humanly experienced context for human lives, evolving in history, yet relatively constant, presenting certain possibilities and foreclos- ing others, our sphere of hope and finitude.

VIII. INTERNAL AND EXTERNAL Williams will have many questions to press about this account. Let me try to anticipate just three of them, and to sketch the outline of an Aristotelian reply. First and most obviously, Williams will point out that in making the conception of human nature internal and evaluative, we have lost the independence and externality that give such an account its value as a foundation for ethics. We have already tried to argue that this internal account, because of its broad acceptability across time and place, and its self-validating structure, does provide a foundation for further ethical inquiry. Now we can add that it is just because it is internal that it can provide a foundation that is really pertinent to the lives human beings are actually trying to lead. Williams vigorously insists that ethical reflection is as natural as anything else for humans - and, indeed, even a part of our selfhood (ELP, 47-8). And he eloquently and sympathetically describes a kind of ethical discourse whose "politics" are not authori- tarian, imposing something on people from above, and whose aim is to "help in continually creating a community held together by that same disposition" — viz., the disposition to listen to ethical reasoning.61 This is precisely the sort of "foundational" discourse that we have found in Aristotle's text. It seems to me strange, then, that the idea of the external viewpoint and the externally dictated basis has such appeal for him, so that he at times suggests that ethics would have a foundation only if it had a basis of this sort. I want to say that this sort is neither available nor of any interest. It matters a great deal what we ourselves think about our selfhood and our possibilities; what a being who stands apart from our experiences and ways of life thinks seems to matter little, if at all. Human nature cannot, and need not, be validated from the outside, because human nature just is an inside perspective, not a thing at all, but rather the most fundamental and broadly shared experiences of human beings living and reasoning together. But, even if we accept the relevance of the internal perspective and grant, as well, that Aristotle's results will be wisely and deeply 122 MARTHA C. NUSSBAUM shared in many times and places, still, we must ask whether there is anything in the account that prevents its being applied in morally objectionable and unequal ways to different classes of beings. Aris- totle, for example, excludes many beings whom he judges to be incapable of practical reasoning. For his so-called natural slaves there is no foundation for ethics in human nature; and, in a more limited way, this is so for all women, because these beings are held not to exhibit (full) human nature in the first place. They are held to lack a set of basic capabilities to function that must be present if a being is going to count at all as a reasoner and to be enrolled in our enquiry, which validates practical reason. Should not an ethical foundation guard against these abuses, preventing those in power from arbitrary exclusions by setting up some hard barrier of fact? This objection comes from the Williams of "Hylomorphism," who wanted to avoid a "slippery slope" along which, without such externally fixed points, we would be carried to atrocity. Biological fact must inform us that certain creatures just are human, and worthy of our attention. To this we can only say, once again, that points fixed externally by some matter of extra-human fact do not seem to be the action- guiding criteria we need; nor are they all we can find. If we are going to avoid a certain course of action, it ought to be because atrocity is what it is — that is, for ethical reasons — and not because some wall of scientific fact has blocked us. And it seems perfectly correct in Aristotle to hold that an ethical theory of human nature should force us to answer for ourselves, on the basis of our very own ethical judgment, the question, which beings are the fully human ones and the members of our community. It seems reasonable, furthermore, that people with very severe mental handicaps should not have exactly the same political and ethical rights as people whom we all regard as fully human. Aristotle is wrong because he makes silly and unfounded judgments about barbarians and women; but the question he raises is a perfectly legitimate one. Raising it will not license wholesale slaughter of the deformed and the elderly, as Williams seems to suppose. For there will be another account of our obligations to the varied levels of beings in the world around us; and this account, which in Aristotle will be based on the level of capability of each creature, is well equipped to generate an attractive story of the moral foundations for our treatment of all animate beings. Nor is it ruled out that we may be ethically entitled to give a Aristotle, nature, and ethics 123 special place in our account to our own parents and children, no matter what their level of capability. Our social nature might well support this conclusion. The important point, however, is that the conclusion will be reached, if at all, by an ethical argument and not because a biologist has discovered something. Finally, Williams will surely object that too little has been done by this account. It is so indeterminate in its results, so general in its conclusions, that it does not go far towards solving our most troublesome ethical perplexities. I have tried to show that, in fact, a surprising amount has been accomplished for moral, and even for political, theory. But we must now grant that, even so, the results are, in two ways, indeterminate: they do not talk about the whole of human life; and the key concepts (social responsiveness, practical reason) are susceptible of multiple specifications. Now the question is, is this a bad thing? Aristotle explicitly presents his results as a "sketch" that can be filled in later: So much for our outline sketch for the good. For it looks as if we have to draw an outline first, and fill it in later. It would seem to be open to anyone to take things further and to articulate the good parts of the sketch. And time is a good discoverer or ally in such things. That's how sciences have progressed: it is open to anyone to supply what is lacking. (iog8a2O-6) This verdict seems appropriate, since we have in mind here (inter- preting Aristotle) a historically evolving project of self- perpetuation/inquiry/creation that can only be done well if room is left always for the play of the imagination and for improvisatory responsiveness to the concrete flow of history. Heraclitus said, "You would not find out the boundaries of the soul, even if you should travel along every path: so deep is its account."62 This idea of an ongoing journeying and an ever deeper searching into ourselves is appropriate as an image of the Aris- totelian search for human nature. For what is proposed here is a scrutiny that seeks out, among our evaluative judgments, the ones that are the deepest and the most indispensable, the ones that lie at the heart of the ways in which human beings over time have defined themselves to themselves, through the play of the story-making imagination and in countless other activities of self-expression and self-perpetuation. Such a search is truly unending, since, as Heracli- tus elsewhere reminds us, the story of human life is "ever-flowing" in history, and the soul has a way of "increasing" its own logos.63 Such a logos may seem too elusive, too open-ended, to serve as a foun- 124 MARTHA C. NUSSBAUM dation — if what one wants from a foundation is a once-for-all hard-edged solution to matters that actual human communities find perplexing.64 The Aristotelians claim, however, that no other sort of foundation is truly deep or truly pertinent. Much though at certain times we may long to be told from outside what to do and to be, it is only from ourselves that we can find (or rather, go on searching for) the answer to the question that we have actually asked. It is only if it remains rooted in the human and the ethical that our search can be about, and towards, the human soul - that is, about what is deepest and most essential about human living.65

NOTES 1 B. Williams, Ethics and the Limits of Philosophy, Cambridge, MA: Harvard University Press, 1985. (Hereafter ELP.) 2 Elements of this account can be found in T. H. Irwin, "The metaphysical and psychological basis of Aristotle's ethics," in A. Rorty (ed.), Essays on Aristotle's Ethics, Berkeley: University of California Press, 1980, pp. 35-54; in Alasdair Maclntyre, After Virtue, Duckworth, London, 1981, pp. 5off., and Whose Justice? Which Rationality? London: Duckworth, 1988; in David Wiggins, Sameness and Substance, Basil Black- well, Oxford, 1980, ch. 6. There are numerous earlier accounts that present a similar picture; some of these were discussed in essay 5 of my Aristotle's De Motu Animalium, Princeton University Press, 1978.1 discuss Maclntyre's Aristotle in detail in a review article on Whose Justice? Which Rationality?, New York Review of Books, 36 (December 1989). 3 "Hylomorphism," Oxford Studies in Ancient Philosophy, 4 (1986), 189-99. This paper was originally presented, in very much the same form, at the Princeton Conference on Ancient Philosophy, December 1978; thus it actually predates the writing of ELP. 4 Cf. also ELP pp. 43, 45-6. Compare Wiggins' claim that the substance concept human being, is something that is "determined from outside of us" {Sameness, p. 181), something that gives "answers independent of ourselves to the question what we are or even who we are and what matters to us" (p. 186). What it is to be human cannot "be freely reconceptualized and reinterpreted at any time" (p. 181). 5 On this see especially Irwin, "Metaphysical," who argues that the role played by extra-ethical considerations in Aristotle's ethics shows that Aristotle's own explicit claim that philosophy must both begin from and return to the ordinary beliefs can be "at best a half-truth." The metaphysical doctrines brought to bear in ethical argument do not themselves derive from ordinary belief, and are said by Irwin to be "beyond" and "more than" the common beliefs. The pages of ELP to which we have already referred indicate that Williams shares this view. Aristotle, nature, and ethics 125

Contrast the general claim about method made by Aristotle in £JV 1145b2~7, and about ethical method in particular at EE I2i6b26ff. — and see the discussion of these issues in my The Fragility of Goodness, Cambridge University Press, 1986, ch. 8. 6 See especially Williams, "Hylomorphism," pp. 198-9 and Wiggins, Sameness, pp. 181-7. 7 Cf. especially "Hylomorphism," p. 198, where the clear boundaries of the concept "human being" are said to avoid the "slippery slope" that might lead to moral atrocity against the elderly, apparently because it is a biological and not a (disputable) ethical concept. In a similar way, Wiggins claims that the boundaries that come with the concept human being will settle some tough ethical problems and protect our otherwise vulnerable ethical dignity: If freedom and dignity are what we crave, we shall find more promise of these things in the Heraclitean prediction, "You would not findou t the bounds of the soul, though you traversed every path: so deep is its logos" than in the idea that it is for men to determine the limits of their own nature, or mould and remake themselves to the point where they can count as their very own creation, (p. 187) 8 "Hylomorphism," 198—9. There are actually two different contrasts made in Williams' brief discussion. One seems to be the one I have emphasized: that "person" is evaluative and "human being" is not. (Thus, on p. 198 "person" is connected with "capacity for higher mental functions" - where the determination that some are "higher" than others is surely an evaluative matter.) The other point is that "person" is a "scalar" concept, like "dwarf" or "giant": whether something is a person is a matter of degree, and the place where the boundary is drawn is to some extent arbitrary. "Human being," on the other hand, is held to pick out a genuine natural kind that is either all there or not there. I suspect that these two contrasts are to some extent connected: that it is in part because the question is an evaluative one, involving further questions such as, "how high are these faculties?" and "how many of the really high ones are in operation?"; that its answer seems likely to involve differences of degree; and it is because "human being" is understood to be a matter of scientific fact that it does not appear that a being can be more or less human. For Aristotle, as we shall see, "human being" is not implicitly comparative to an average, like "dwarf" or "giant"; but it does involve marking off a certain segment in the continuum of nature in a way that would admit of dispute. 9 ELP pp. 43ff. 10 Cf. for example, pp. 26-7, 45-51. 11 ELP pp. 45-6. 12 Cf. Ibid., pp. 26-7, 51. 13 See Wiggins, Sameness, passim; and also R. Wollheim, The Thread of Life, Cambridge, MA, Harvard University Press, 1984, ch. 1. 14 For an account of what needs to be added, see Wiggins, Sameness, ch. 3. 126 MARTHA G. NUSSBAUM

15 Compare Williams, ELP, p. 51. 16 See the commentary on this passage by David Gallop, Plato's Phaedo, Clarendon Plato Series, Oxford 1975. 17 See, for example, D. Brock, "Quality of life measures in health care and medical ethics," in M. Nussbaum and A. Sen (eds.), The Quality of Life, Oxford: Clarendon Press, 1993. Elsewhere, Brock sometimes uses "person" as a strong evaluative concept, "human being" as implying only a lower, animal level of functioning: a senile patient who lacks (permanently and irreversibly) all sensation of pleasure and pain and all consciousness, but who still has enough brain function to be legally alive, is a human being but not a person. Both concepts are, it seems, evaluative here, since even the legal definition of death requires the judgment that mere cardiovascular function is not sufficient to constitute a human life. I am not clear what the purpose of having two such concepts is; and Aristotle would, I think, be right to prefer a single concept. 18 See G. Kirk, Myth, Cambridge University Press, 1970, pp. 162 ff.; J.-P. Vernant, "A la table des hommes," in M. Detienne and J.-P. Vernant (eds.), La Cuisine du sacrifice en pays grecque, Paris: Gallimord, 1981. (This theme recurs often in Vernant's work, as in related work by P. Vidal- Naquet and M. Detienne.) See also W. Burkert, "Greek tragedy and sacrificial ritual," Greek, Roman, and Byzantine Studies, 7 (1966), 87-121. 19 Cf. Emile Benveniste, JVoms d'agent et noms d'action en indo-europeen, Paris, 1948; D. Holwerda, Commentatio de vocis quae est phusis vi atque usu, Groningen, 1955; Charles Kahn, Anaximander and the Origins of Greek Cosmology, New York: Columbia University Press, i960). 20 On the complexities of Aristotle's position on this issue, see Fragility, ch. 11 and Appendix to ch. 12. 21 See Plato, Republic, Books ii-m; the same point is made by Xenophanes, DK B25. Cf. Fragility, ch. 5. 22 On this, cf. Fragility, ch. 13, with references. 23 Cf. Burkert, "Greek Tragedy": the particularly close connection between tragic themes and the initiation of the young is brought out in John J. Winkler, "The Ephebes' Song," Representations, 11 (1985), 26-62. On the Athenian democracy as a self-interpretive community, in which ethical reflection influences, and is influenced by, immersion in social practices, see the excellent discussion in Cynthia Farrar, The Origins of Democratic Thinking, Cambridge University Press, 1988. 24 This name, being, as seems likely, fictional, underlines the dialogue's concern with the theme of moral initiation and education. 25 Cf. Gregory Vlastos, "The Socratic Elenchus," Oxford Studies in Ancient Philosophy, 1 (1983), 27-58. 26 Thus it will be seen that I think that Vlastos' account of Aristotle in "The Socratic Elenchus" is not sufficiently complex, and assimilates Aristotle's procedure too much to one that proceeds by majority vote; see also Fragility, ch. 8. One can further support these claims by pointing to the insistence on the inner activity of emotions and imagin- Aristotle, nature, and ethics 127 ation - both in Aristotle's ethical thought and in the Athenian moral tradition that he develops: see Nussbaum, "The Discernment of Per- ception," in Love's Knowledge, Oxford, Claredon Press, 1990, and Fragility, ch. 10. 27 On appearances, see Fragility, ch. 8. 28 See, for example, the arguments of Hilary Putnam in Reason, Truth, and History, Cambridge University Press, 1981. 29 Neither of these words is precisely appropriate for Aristotle's politikon, which, as the passages we are discussing make clear, refers to the full range of the human being's other-related concerns, familial and friendly as well as civic and social. 30 Cf. also Magna Moralia I2o8b5: "We see that friendship stretches through the entirety of life and is present on every occasion, and that it is a good thing"; and EE 1235a! ff.: "Our entire life and voluntary association is with [sc. people with whom we have a relation ofphilia]: for we lead our day-to-day lives with our family or relatives or friends, or children or parents or wife" (emphasis mine). 31 A related type of connection between ethical truth and judgments of worthwhileness is made by Aristotle in discussing views about the role of luck in the good life: see Fragility, ch. 11. 32 Homer, Iliad 9.63-4; see Newman's astute comments on Aristotle's reinterpretation of the passage: he compares EM ii77bg. (W. R. Newman, Aristotle's Politics, 4 volumes, Oxford, 1887-1902. 33 See Newman, ibid., for references and discussion. 34 Cf. for example Isocrates De Antid. 253-7, Nicocl. 5ff. 35 This also reflects a popular belief: see, for example, Xen. Mem. 4.3.12. 36 John Rawls, A Theory of Justice, Cambridge, MA: Harvard University Press 1971) 189-92. 37 On human nature and the concept of the person, see Rawls, A Theory, and Dewey Lectures, Journal of Philosophy, 77 (1980), 515-72. 38 I first discussed this argument in Appendix 1 to Essay 1 of Aristotle's De Motu, Princeton 1978; but I do not find what I said there satisfactory. Some aspects of the argument are discussed in a companion paper to this one, "Nature, Function, and Capability: Aristotle on Political Distribution," Oxford Studies in Ancient Philosophy, Supplementary Volume 1 (1988), 145-84. The literature on the argument is by now very large, and I shall be referring only to some representative items, including: John Cooper, Reason and Human Good in Aristotle, Cambridge, MA: Harvard University Press 1975; T. H. Irwin, "Metaphysical"; David Keyt, "Intellectualism in Aristotle," Paideia, special Aristotle issue (1978) 138-57; T. Nagel, "Aristotle on Eudaimonia" in Rorty (ed.), Essays, pp. 7-14; J. A. Stewart, Motes on the Micomachean Ethics of Aristotle, Oxford: 1892; K. V. Wilkes, "The good man and the good for man in Aristotle's Ethics," in Rorty, 341-58. 39 On persuading the interlocutor to revise his position as a central task of an Aristotelian ethical inquiry, see EE I2i6b27—30. 128 MARTHA C. NUSSBAUM 40 Aristotle uses this strongly affective word of his relation to the object of inquiry in only one other passage: in Parts of Animals 644.1327, he speaks of our pothos for unavailable knowledge of the divine heavenly bodies, comparing our situation to that of a lover yearning for an unavailable beloved. 41 On this, see my Aristotle's De Motu, essay 1. 42 Cf. Wilkes, "The good man," p. 343. 43 A is defended by: Cooper, Reason, p. 160 n. 16; Stewart, Notes, n.443-5; Nagel, "Aristotle," p. 8. 44 B is defended by: Irwin, "Metaphysical," p. 48; Wilkes, "The good man," 3431! 45 Cf., for example, DA 415323!!, 4i6bo,ff., 434a22, 434b23, 435b4, 435bi6. 46 Cooper, Reason, pp. 159-60; Keyt, "Intellectualism." The debate is carried further in Cooper, "Contemplation and happiness: a reconsideration," Synthese, 72 (1987), 187-216, and in Keyt, "The meaning of bios in Aristotle's Ethics and Politics," Ancient Philosophy 9 (1989) 15-21. It should be noted that Cooper's interpretation is our interpretation B only where the word bios is concerned. Because EJVi.y uses zoe instead, Cooper is inclined to prefer reading A for that passage: see notes 42 and 47. 47 Some of the most pertinent Aristotle passages are: EN 1095b 17, EE I2i5a4ff., 36fF., Top. i02bi8 Pol. i254b3O, 1256310, 30, 1265325, 1295325, 1324a 15, 27, 30, I328b4o; PA 69436, 682b7, 693311, HA 487all, 14, b33, 588317; GA 736^3, 753b28, 77932, 75035; there are many others. Keyt brings forward several passages as counter-evidence: Politics I256a4off., and Plato, Laws 733D7 are alleged to show that a single individual can live different bioi at different times, or even, in the case of the Laws passage, at the same time. But these passages do not support reading A against B. The Politics passage actually stresses that the shift from one bios to snother is a shift from one mode of life to another, with a different focus and organizing principle. The Laws passage is anomalous because it deals with the unity of the virtues. It is not paradoxical for Plato to say that one person can, at one and the same time, combine the bios of temperance with the bios of wisdom and the bios of courage, only because it is precisely Plato's point that the virtues all supply a life with the same guiding principle or principle of organization, so that a life according to one virtue is also, of necessity, a life according to, guided by, the others. For other observstions on these psssages, see Cooper, "Contemplation." 48 Cooper, Reason, pp. 160 n. 16. 49 Cooper 3lso mentions Metaph. iO72b27, he gar nou energeia zoe; but clearly this case does not tell either way, since for the god the activity of nous is both a single life-component and a total way of life. Besides, zoe here msy mean simply "life" or "lifetime." 50 I have scrutinized the use of zoe in the following authors: Homer, Aristotle, nature, and ethics 129 Aeschylus, Sophocles, Euripides, Aristophanes, Herodotus, Thucy- dides, Lysias, Isaeus, Isocrates, Plato, and Aristotle. (It is not, in fact, a common word.) The verdict is clear and unequivocal: A cannot be correct and B must be. To be somewhat more precise: %pe has a number of uses that do not concern us. It can mean "life" (as opposed to "death"); it can mean "quantity (length) of life" (as in the expression pollezoe); it can mean "lifetime" and be used interchangeably with aim; like bios, it can also mean "livelihood" or "income." But when it is used of a type of life it always designates a total mode or way of life, organized around the item named. The following represents a complete list of uses of zoe apart from the meanings I have set aside above; and I would submit that all support B: Herodotus 1, 157; 2, 77; 2, 105; 4, 112; 4, 114. Aeschylus, fr. 401. Plato, Republic 344E3, 521A4; Laws 712C3, 94406; Epist. vn, 330B5 Aristotle, HA 584a3 (though this may be in the sense "lifetime"); 612b 19; PA 68ob3O (though some editors find this text corrupt, I see no reason to emend); EE 1215a5 (where bios and zoe are used interchangeably; Pol. i26oa4O, I278bi6, I28ob34, i28iai, i283a24-5, I288a27, I323a23. One representative example must stand for the whole list: Herodotus 4, 114. The Scythian men have succeeded in seducing the Amazon women. They are living with them near the battlefield; the Amazons have not shifted from their previous warlike way of life. Herodotus tells us that, although the Scythians prove unable to learn the women's language, the women soon learn the Scythians. The Scythians now make the following proposal: "We have parents and property. Let us live this sort of life (zoen toiende) no longer. Let us return to live with our people." The women reject the offer, saying: We and the women of your nation could never live together; our ways are too much at variance. We are riders; our business is with the bow and the spear, and we know nothing of women's work; but in your country no woman has anything to do with such things; your women stay at home in their wagons occupied with feminine tasks, and never go out to the hunt or for any other purpose. The Amazons' understanding of their new language is good. They get the point that the proposed change in zoe is not the innocuous adding or subtracting of one component activity: it is a total change in mode and organization of life. 51 Two potential objections suggest themselves; both can be met. (1) At iog8a2, the aisthetike zoe is held to be common to the horse, cow, etc. as well as (apparently) to the human being: phainetai de kai aute koine kai hippoi kai boi kai panti zoioi. Surely only on reading A could this be true: for humans precisely do not share a total mode of life organized around aisthesis (as B would read it). We may reply: (a) It is not in the least clear that the kai before hippoi does refer to human beings (... "as I3O MARTHA C. NUSSBAUM well"). It could just be coordinate with the two other kai's that follow: so the life would just be common both to the horse and to the cow, and to every (non-human) animal. There would then be no problem, (b) Furthermore, it is true, according to Aristotle's biological views, that there is a time when a human being leads a threptike zoe, and also another later time when it leads an aisthetike zoe: both are clearly marked stages of fetal development. At first, all embryos are said to zen phutou bion (GA 736^3, 753b27, 779,b33ff.); it becomes an aisthetike zoe when the embryo has acquired aisthesis (GA 779^33^, Pol. I335bigff.). Finally, with the advent of reason, if development continues along normal lines, it becomes a fully human life. Since these distinctions are important in Aristotle's political thought, one may suspect that the shifting use of koinon here reflects these concerns. (Embryos are mentioned in a similar context at 1102b 1). These two replies cannot be used together; but either one would dispose of the objection. (2) The other objection comes from iJjVi.13, where Aristotle denies to the threptike dunamis tespsuches a role in human excellence (iiO2b2— 12), on the grounds that digestion goes on during sleep, but ethical distinctions are not applied to what goes on during sleep. This, I believe, does not tell against reading B for the earlier passage. Reading B does not require that the good human way of life includes the good functioning of the threptike dunamis, except in so far as this functioning is a part of a life organized and guided by practical reason. In other words, it leaves in the chosen self-nutritive and reproductive activities that form part of a reason-guided human life, but not the unguided functioning of the bodily system in sleep, which could hardly be said to be either kata logon or meta logou (cf. also MA ch. 11). Aristotle tells us elsewhere that the sleeper's life is the life of a plant, nothing more (EN 1176334—5, EE I2i6a5): we are not ourselves when we sleep, we do not live the characteristic life of a human being. 52 On the complicated passages dealing with slavery and friendship, see my "Shame, separateness, and political unity," in Rorty, Essays, p. 434 n. 54 (the first reference should read, MM 1194b). 53 Pol. 1335b 19 ff. suggests that it should be mandatory to expose severely handicapped children, and permits abortion for the sake of population control up to the time when the fetus has developed aisthesis. It is worthy of note that the merely aisthetike zoe of an older fetus gives it a claim not to be aborted (Aristotle even says that to abort it is not hosion), whereas a living child that was handicapped in such a way that it would not progress beyond this level of life would probably be a candidate for exposure. It would be beyond the scope of this chapter to try to reconstruct the views about potentiality and actuality, as well as the views about the late thurathen entry of nous, that support these distinctions. What is of most importance for us is that these policies are based upon functional capability, not upon human parentage alone. 54 On all of these issues, see my "Nature, function, and capability." Aristotle, nature, and ethics 131 55 On the extent of the lawgiver's responsibility, see the discussion in "Nature, function, and capability." 56 Cf. ibid. 57 From Karl Marx, Economic and Philosophical Manuscripts 0/1844, translated by Martin Milligan, in R. C. Tucker (ed.), The Marx-Engels Reader, New York, 1978, pp. 88-9. 58 Ibid., p. 105. 59 See Fragility chs. 11-12 and Appendix; and especially "Non-relative virtues: an Aristotelian approach," Midwest Studies in Philosophy, 1988. 60 "Non-relative virtues." 61 Williams, ELP, pp. 26-7. 62 Heraclitus DK B45, see my "Psuche in Heraclitus, I," Phronesis, 17 (1972), 1—17. Notice that Wiggins appeals to Heraclitus in defense of his rather different position (above n. 7). We can agree that the humanness sought is not a function of people's immediate desires and preferences, and that people, according to Aristotle, are not simply free to remake themselves in any way at all. But I would argue that the idea of a human and internal account of depth is more true to Heraclitus (as well as to Aristotle) than is Wiggins' external account. 63 Heraclitus, DK BI 15. 64 Included among these will be problems concerning conflicts among goods; and it is one of Williams' charges against Aristotle that he underestimates the amount of such conflict. I agree with this to some extent (see Fragility of Goodness, chs. 11-12), though I do think that Aristotle does allow for some possibilities of conflict. But I believe that the larger measure of conflict permitted in Greek tragedy (see Fragility, chs. 2—3) is a reasonable amount to live with, and inseparable from a certain sort of richness of life. 65 A much earlier version of this chapter was presented at a conference on Aristotle's metaphysics at Florida State University; I would like to thank those present on that occasion, and many others who have read versions of this chapter since, for illuminating criticism. I am particularly grateful to Julia Annas for helpful discussion of the issues, and to J. Altham for most helpful written comments. After finishing this chapter I have found that some closely related ideas about self-interpretation are lucidly worked out in recent work by Charles Taylor: see especially Philosophy and the Human Sciences: Philosophical Papers, 11, Cam- bridge University Press, 1985, and "Explanation and practical reason," in M. Nussbaum and A. Sen (eds.), The Quality of Life, Oxford: Clarendon Press, 1993. CHAPTER 7 A most peculiar institution Charles Taylor

Bernard Williams' Ethics has put the distinction between morality and ethics back on the agenda, Williams making a swingeing attack on 'morality' as a sub-system of ethics, which has been misconceived as though it were the whole. His target is the kind of moral thinking which has been dominant in contemporary Anglo-Saxon philosophy, but beyond this has at least been immensely influential in the English-speaking world in general for the last two centuries. Perhaps the paradigm target of Williams' attack is Kant, but utilitarians are recognized as at least marginal members of the class (p. 178).1 These two schools together cover much of the mainstream. 'Morality' is characterized by its making obligation central: that someone ought to do something, or that something is a good thing to do, finds expression if it has moral relevance at all in someone's having an obligation to do some act. Williams points out how distortive this is of our moral experience. For one thing, this system has no place for acts of supererogation. We admire people for great acts of courage, or self-dedication, and all the more so just because they were under no obligation to do these (p. 179). 'Morality' is in fact a misguided Procrustean operation. But it is not an innocent error, an honest, even-handed attempt to make sense of our moral intuitions which happens to fall short in certain respects. It is motivated by its own moral vision, in terms of which it has its own reasons to delegitimate much of what it excludes. Moral philosophers who have articulated the system do not usually acknowledge this. Hare, for instance, sees himself as impartially making sense of our linguistic intuitions about such terms as 'moral'. Williams is sketching a Nietzschean genealogy of this mode of

132 A most peculiar institution i33 thought, which uncovers motivations its defenders are unaware of and would not accept. I am tremendously in sympathy with this attempt, I share much of Williams' impatience at the constricting modes of thought that dominate academic moral philosophy. Moreover, I agree very much that one has not got to the bottom of these until one has articulated the moral motivation underlying them, even in the teeth of their originators' disavowals. Partly because of this, I feel impelled to add a few footnotes to his chapter on 'Morality, the Peculiar Institution'. But, inevitably, I shall be doing slightly more than this. Because my way of defining the target, of picking out what deserves attack in it, and of articulating the underlying motives, are all somewhat different from Williams', and I have a sneaking suspicion that this might make a difference somewhere down the line — concerning where you go, for instance, once you have abandoned the Procrustean bed of moral thinking. Williams and I have different destinations, and this may emerge even in the way we attack the moral philosophy that we both love to hate. 'Morality' for both of us is characterized by a terrible constriction of ethical thinking. One of the things that gets left out, or distorted, is what Williams calls the notion of'importance' (p. 182). For the morality system, our sense of what we ought to do or be properly takes the form, in its higher reaches, of obligations we acknowledge to certain actions. These higher reaches are the ones we identify as 'moral', and obligations of this range override all other considerations of value, for example, those relating to aesthetic goods or personal ideals. The only thing that can trump a moral obligation is another moral obligation. Against this, Williams suggests a picture of ethical life which crystallizes at certain definite points around what we acknowledge as obligations, but where these are felt to hold against a background sense of what things are important - either to us or important tout court: what Williams calls the 'simply important'. This background is immensely varied: many things are important, and not all equally so or in the same way. There is not even a fixed and unvarying relative weighting, because some things which are usually not very highly ranked, can become urgent in certain circumstances. Obligations which we acknowledge can be lifted, or overridden, if it becomes inappropriate or even absurd to carry them through in the context, granted what is really important. 134 CHARLES TAYLOR For my money, it suffices to articulate this bit of moral phenomenology for one to feel its force; and this raises the question why the systems of thought Williams attacks have gained the currency they have, and why their protagonists resist his insights. We are brought straightaway to the question of motivation. Now I think the motivation will emerge more clearly if we rephrase Williams' 'importance' in more old-fashioned terminology, and speak of'good' and 'goods'. In the way I want to use the term, we have a sense of a good when we hold a certain way of living, or a component of a way of living - a kind of action, a mode of sensibility, a quality of feeling - as an object of strong evaluation. 'Strong evaluation' is a term of art I want to introduce, such that some end is strongly valued when we acknowledge that its being an end for us is not just contingent on our happening to desire or need it, when, in other words, we allow that we would be lesser beings if we should cease to want or need it. The ice-cream cone I now desire is weakly valued, because should I lose my interest in it, it would no longer have any claim on me. But I do not think of my commitment to Amnesty International in these terms. Someone who refused to contribute to this cause because they were not 'into' torture victims this week would be thought unbearably frivolous. Their change of interest does not reduce the claim, but rather shows them in a poor light. It should be clear that strong evaluation extends beyond the bounds of the 'moral'; we strongly value responses to people, or to beautiful things, as well as 'moral' acts. One might, for instance, find it impossible to admire someone for whom Mozart's music sounded monotonous or shallow. I would like to reformulate Williams' phenomenology, and say that our sense of what we ought to do or be is shaped by the (strongly valued) goods we acknowledge, and that the obligations we recognize hold against this background. The advantage of putting it this way, I believe, is that it brings out more clearly what is involved in 'morality'. These theories can be conceived as attempts at an ethics without the good. They offer the prospect of a stripped-down theory of right conduct, cast entirely in propositions denning what we ought to do. In other words, morality is conceived purely as a guide to action. It is thought to be concerned exclusively with what it is right to do, rather than with what it is good to be. Relatedly the task of moral theory is identified as denning the content of obligation, rather than the nature of the good life. In other words, morals A most peculiar institution 135 concern what we ought to do; excluding both what it is good to do, even though we are not obliged (which is why supererogation is such a problem for some contemporary moral philosophy) and also what it may be good (or even obligatory) to be or love, as irrelevant to ethics. In this conception there is no place for the notion of the good in either of two common traditional senses: either the good life (the sense that I introduced above), or the Good as the object of our love or allegiance (which I shall distinguish by using upper case 'G'). Moral philosophies so understood are philosophies of obligatory action. The central task of moral philosophy is to account for what generates the obligations that hold for us. A satisfactory moral theory is generally thought to be one that defines some criterion, or procedure, which will allow us to derive all and only the things we are obliged to do. So the major contenders in these stakes are utilitarianism, and different derivations of Kant's theory, which are action-focused, and offer answers exactly of this kind. What should I do? Well, work out what would produce the greatest happiness of the greatest number. Or work out what I could choose when I have treated other people's prescriptions as if they were my own.2 Or think what norm would be agreed by all the people affected, if they could deliberate together in ideal conditions of unconstrained com- munication.3 An ethics of obligatory action is an ethics without the good. But what is the attraction in this? There are in fact many. The adoption of this kind of theory is overdetermined. Some motivations are epistemological-metaphysical, some are themselves ethical in nature. Before unravelling these, we should see that the very sidelining of the good is itself complex and many-stranded.

11 The move to an ethics of obligatory action is sometimes described as according a priority to the right over the good. In one historic use, this was the slogan for a semi-Kantian rejection of utilitarianism. That is not the sense which I want to invoke here, because the move I am describing is common to both utilitarianism and Kantianism. The 'good' over which the intuitionists gave priority to the right was not the good in my sense, but the thin, weak good which is the object of utilitarian calculation. The good in my sense, i.e., as strongly valued, was exactly what utilitarianism wanted to sideline, I36 CHARLES TAYLOR in order to have a non-question-begging way of calculating the right. One sense in which we can give priority to the right is to have an ethic that makes no use of the good at all, and refuses to recognize it as a meaningful ethical concept. A view of this kind can be attributed to Bentham, but not to Kant. In relation to the utilitarian Enlightenment, Kant wants to rehabilitate one crucial distinction, that between actions done from duty and those done from inclination. This is grounded on a distinction of motives: the desire for happiness versus respect for the moral law. Kant deliberately takes this stance in opposition to the 'Gluckseligkeitslehre' of which he speaks in scathing terms. Following Rousseau, he breaks with the utilitarian conception of our motives as homogeneous. He returns to the Augustinian insight, that there are radically different qualities of the will. It is just this issue of the quality of one's will that is of ultimate moral importance.4 So Kant in my sense does recognize a good. He posits a qualitative distinction between two modes of life, defined by the quality of the will animating each, of which one is incomparably higher than the other, and remains the proper object of our moral aspirations whether we seek it or not. But there is another sense in which Kant can be said to accord priority to the right. The good will is denned entirely by its recognition of the obligation to act in a certain way, viz., in conformity with the categorical imperative. Kant utterly rejects the way of proceeding we find with Aristotle, for instance, where we define good ways to be, for example, traits of character, and then from these can determine in given particular circumstances what we ought to do. The good is defined exclusively by our stance to obligatory action. This kind of priority has its appeal, even to those who do not want to reject the good altogether. One way it can be applied (though it was not in this form by Kant) is to give a clear superiority to one kind of ethical demand over others. One set of demands concerns our obligations to others, for example, to fair dealing and benevolence; while another rather different set concerns our aspirations to fulfilment, or a valuable and worthwhile life. There can, of course, be overlap: some people's idea of a worthwhile life is one led in service to others. If this does not appear at all in someone's notion of the good life we might have cause to worry. But they are distinct as grounds for action, and they also fail to overlap at crucial A most peculiar institution 13 7 moments in our lives. Placing the right over the good can be a way of marking the systematic priority of obligations over fulfilments, as for example, Habermas does by dignifying only the first with the title 'ethics'.5 These two ways of sidelining the good, one by rejecting it altogether, the other by giving it low priority, are complexly interwoven in contemporary theories of 'morality'. One might think that the first way would be mainly motivated by epistemological or metaphysical grounds, the second by moral reasons. But it is more complicated than this. Utilitarians also had moral motives for their metaphysical stand, and a certain residual doubt and suspicion about the status of goods haunts some thinkers who would not go to the lengths of repudiating them altogether. The metaphysical-epistemological distrust of the good is fed by the naturalist temper of much modern thought. By 'naturalism', I mean the belief that humans as part of nature are in the end best understood by sciences continuous in their methods and ontology with modern natural science. The naturalist temper is a much more diffuse thing, stopping short frequently of explicit espousal of full- blooded naturalism, but tending to be suspicious of the things that naturalism cannot accommodate. Obviously, goods as I have denned them fall in this class. 'Higher' ways of being, acting, feeling, not simply reducible to ways we desire (for then they would be only weakly valued), must appear very strange entities indeed. Wherever can they fit in the furniture of the universe as revealed, say, by physics?6 So the naturalist temper tends to push towards an ethic of obligatory action. The reason for this becomes clearer when we look closer at the role of goods in our ethical consciousness and discourse. To speak of goods is to articulate our reasons for doing the things, taking the stands, making the judgments that we do, or for accepting the obligations which we take on. But they provide reasons in a rather different sense than the proponents of'morality' recognize. People of naturalist temper quite understandably look for their models of argument and proof to mathematics or natural sciences. In the latter sphere, at least, we strive for modes of argument which could ideally demonstrate to an opponent the truth of a thesis, starting from considerations which both sides have to accept, regardless of their starting theoretical positions. Uncontrovertible 'facts' provide one such range, as also do certain principles. Of 138 CHARLES TAYLOR course, it is hotly contested in the philosophy of science, whether argument can proceed this way, or whether theory is not crucially 'under-determined' by fact, but this is thought simply to render scientific argument less powerful; the model of what it is to prove something remains untouched. Applied to ethical arguments, this model would involve offering someone a reason for adopting a certain ethical position which would have to be recognized as a reason whatever the interlocutor's ethical outlook. Giving a reason in this sense means finding some external consideration, not anchored in our moral intuitions, which can somehow show that certain moral practices and allegiances are correct. An external consideration in this sense is one which could convince someone who was quite unmoved by a certain vision of the good that he ought to adopt this vision, or at least act according to its prescriptions. This is the kind of reason which a naturalistic picture of human life might seem to offer utilitarianism, or some ethic of 'material' welfare; or the kind of support that theories like Hare's prescriptivism derive from considerations about the logic of moral language.7 But this is not at all the way in which a good gives a reason for action. A notion of the good can be thought of as a qualitative distinction between ways of being, distinguishing a 'higher' or strongly valued way from its 'lower' contrast class. This does not offer us a consideration we have to acknowledge quite independently of our ethical intuitions; rather it is an attempt to articulate these intuitions, to make them more perspicuous and palpable. If you do convince your interlocutor by your discourse about a good, it will be because the good so articulated moves him, i.e., he comes to recognize it as an expression of his ethical intuitions. This is equally the case, if someone appeals to a Good, that is, something worthy of love or reverence, in ethical debate. Suppose someone recurs to his belief in God to explain some of his ethical views. If one interprets this as the offering of an external reason, then one will probably be very seriously misled as to what is being said. It is true that the invocation of God was taken as an external reason in an earlier age, where thinkers strove to 'prove' the existence of God from a standing start, as it were, and then deduce our obligations to the Creator. But it has no more relevance to our age. Characteristic- ally today, the belief in God offers a reason not in this sense, but as an articulation of what is crucial to the shape of the moral world in A most peculiar institution 139 one's best account. It offers a reason rather as I do when I lay out my most basic concerns in order to make sense of my life to you. But nor do our qualitative distinctions offer reasons in another sense which is often evoked in the literature of moral philosophy. In this sense, we give a reason for a certain moral principle or injunc- tion, then we show that the act enjoined has some crucial property which confers this force on it. I say: 'you ought to do A', and when you ask why, I add: 'because A = B', where 'B' allegedly offers a description of an act-form which we are morally committed to. So typical fillings for 'B' would go: 'obeying the law', or 'conducing to the greatest happiness of the greatest number', or 'saving your integrity'. We say that B gives a reason because we hold that the act picked out by the A-description is only enjoined because it also bears the B-description. Were this identity to fail, A would no longer be a moral obligation. A utilitarian will argue that I ought generally to pay my taxes, because this conduces to the general happiness. But in exceptional circumstances, where public revenues are being terribly misused, this identity fails, and I do not have to pay. A = B is an obligation-conferring identity. It is in its nature to work asymmetri- cally. B makes A obligatory, not vice versa. We can call B in this case a 'basic' reason. This form of argument is very frequent in ethics. We often ask what makes a given action right, and are answered with a basic reason. It is a big issue in moral philosophy how systematically our moral ends or obligations can be related to a small list of basic reasons. I shall return to this below. Relevant to my argument here is the fact that articulating a vision of the good is not offering a basic reason. It is one thing to say that I ought to refrain from manipulating your emotions or threatening you, because that is what respecting your rights as a human being requires. It is quite another to set out just what makes human beings worthy of commanding our respect, and to describe the higher mode of life and feeling which is involved in recognizing this. It is true that clarification on the second head is closely related to the definition of the basic reasons we invoke in the first kind of claim. Our conceptions of what makes humans worthy of respect have shaped the actual schedule of rights we recognize, the latter has evolved over the centuries with changes in the former. But they are none the less distinct activities. They offer reasons in quite different senses. Our qualitative distinctions, as definitions of the good, offer I40 CHARLES TAYLOR reasons in the sense that articulating them is articulating what underlies our ethical choices, leanings, intuitions. It is setting out just what I have a dim grasp of when I see that A is right, or X is wrong, or Y is valuable and worth preserving, and the like. It is to articulate the moral point of our actions. That is why it is so different from offering an external reason. I can only convince you by my description of the good if I speak for you; either by articulating what underlies your existing moral intuitions; or perhaps by moving you to the point of making my description your own. And that is why it cannot be assimilated to giving a basic reason. Relative to the most basic action-description, we can still strive to make clear just what is important, valuable, or what commands our allegiance, as with the above example of respecting human rights. This is not a step to a more basic level, because there is no asymmetry. But we can see how articulating the good may help further definitions of what is basic. So we can see the place that qualitative discriminations have in our ethical life. Pre-articulately, they function as an orienting sense of what is important, valuable or commanding, which emerges in our particular intuitions about how we should act, feel, respond on different occasions, and on which we draw when we deliberate about ethical matters. Articulating these distinctions is setting out the moral point of the actions and feelings our intuitions enjoin on us, or invite us to, or present as admirable. They have this place as much in the broader domain of goods that we pursue across the whole range of our lives, as in the more special domain of higher goods, which claim a status of incomparably greater importance or urgency.

in Now we can see better why naturalism tends to nudge us towards an ethic of obligatory action. In one way, it makes goods appear superfluous. Naturalism recognizes the value of an external reason, if there were any such. The theories of obligatory action also make a great deal of basic reasons, as I will explain further below. But goods offer reasons in neither of these senses, and so seem to add nothing at all to our moral understanding, nothing that we do not already have when we know what we ought to do. Besides seeming superfluous, this kind of articulation is bound to appear suspect to the naturalist temper. Articulating our qualitative A most peculiar institution 141 distinctions is setting out the point of our moral actions. It explains in a fuller and richer way the meaning of this action for us, just what its goodness or badness, being obligatory or forbidden, consists in. It is possible to know, for instance, as a child sometimes does, that a certain act is forbidden, but not to understand yet what kind of badness it exhibits. Later one may learn that it is something disho- nourable, or perhaps mean-spirited, in distinction from other forbidden things, which are ruled out just because they are dangerous, or because we cannot now pull them off. Many of our virtue terms belong to these richer languages of what I have been calling qualitative discriminations. The child or the outsider can be told what not to do, given a description which they can understand of what to avoid, before they can understand just what is wrong. We can get a sufficient grasp of the commandment: 'thou shalt not kill'; or can obey the order: 'don't talk like that to Grandad!', before we can grasp articulations about the sanctity of human life, or what it means to respect age. These two stages cannot be completely prised apart, because outside of the background articulated at the second stage we generally would not be able to project properly to new cases: to see, for example, what exclusions there are to the ban on killing (if we accept a view which allows some), or grasp what other persons are worthy of respect. But it is sometimes possible to give people at least a good first approximation of how to behave in external action terms. For example, you tell the new arrival: 'just remember, when you see someone with a red headdress, bow three times'. To move from external action descriptions to the language of qualitative distinctions, is to move to a language of 'thick description', in the sense of this term that Clifford Geertz has made famous,8 that is, a language which is both richer and more culturally bound, because it articulates the significance and point that the actions or feelings have within a certain culture. But that is just what naturalism strives to avoid. One of the denning characteristics of naturalism, as I am using the term, is the belief that we ought to understand human beings in terms continuous with the sciences of extra-human nature. Just as these last have progressed by turning away from anthropocentric language, by excluding descriptions which bear on the significance of things for us, in favour of 'absolute' ones, so human affairs ought to be maximally described in external, non-culture-bound terms. 142 CHARLES TAYLOR Thinkers of a naturalist temper, when considering ethics, naturally tend to think in terms of action. This temper has helped contribute to the dominance of moral theories of obligatory action in our intellectual culture. But this is not all, of course. The dominance is overdetermined, as I suggested above, and there are also moral motives for the exclusion. The various offshoots of the modern affirmation of ordinary life have engendered a suspicion of the claims made on behalf of 'higher' modes of life against the 'ordinary' goals and activities that humans engage in. The rejection of the higher can be presented as a liberation, as a recovery of the true value of human life. Of course, the moral value attaching to this liberating move itself presupposes another context of strong good. But, with that curious blindness to the assumptions behind their own moral attitudes, utilitarians and modern naturalists in general can just focus on the negation of the older distinctions, and see themselves as freeing themselves altogether from distinctions as such. There is also a closely related way in which the rejection of qualitative distinctions can be seen as a liberation, which we can understand if we explore moral phenomenology a bit further. To recognize a good is generally to want to measure well against that good. In the case of certain centrally important goods which are bound up with our sense of identity, this desire can be more like a craving. But precisely this craving can be the source of much suffering, or alternatively of self-delusion, or smug self-satisfaction. That is, I can feel the demand to incorporate the good in my life as crushing; it is a demand that I feel utterly unable to live up to, which I constantly measure up badly on, and which leads to an over- whelming depreciation of myself. This, besides being uncomforta- ble, can be immensely restricting and even destructive. To break my allegiance to this good can therefore be experienced as a liberation, which is what it is often represented as in much of the human potential literature of our day.9 In particular, Christianity has been attacked ever since the Enlightenment for laying a crushing burden on those in whom it inculcates a sense of sin; or, in contemporary language, for 'laying a guilt trip' on its devotees. To break with a good to which one cannot really subscribe is of course a liberation in anyone's language. The issue is whether flight is the only answer to unhealthy guilt. But we can readily see how, in a confused way, the possibility of such a liberation could seem to accredit the rejection of A most peculiar institution 143 qualitative distinctions as such. (Though not generally, it must be said, in the popular writings concerned today with human potential; these usually espouse their own crucial distinctions, around such goods as fulfilment, or self-expression.) Alternatively, my craving to be well placed to the good can make me a prey to illusions, either lead me to espouse a standard on which I measure well, but which I could not ultimately really defend; or make me blind to how badly I actually do compare. And to illusion is often added a certain smug satisfaction in contrasting myself to others. How much of my ability to live with myself comes from the repetition of such consoling thoughts as: at least I am not like those (wastrels, weaklings, blackguards, philistines, Rightists, Leftists, etc.)? Reflecting on all this can strengthen the sense that turning away from these notions of the good is somehow a liberation.

IV So metaphysical, epistemological, and moral considerations run together here. But we have not exhausted the moral considerations. As well as the affirmation of ordinary life, there is the modern notion of freedom. The ancient notion of the good, either in the Platonic mode, as the key to cosmic order, or in the form of the good life a la Aristotle, sets a standard for us in nature, independent of our will. The modern notion of freedom which develops in the seventeenth century portrays this as the independence of the subject, his determining of his own purposes without interference from external authority. Freedom so conceived came to be considered as incompatible with the good. The conflict was originally conceived in theological terms. Late medieval nominalism defended the sovereignty of God as incompatible with there being an order in nature which by itself defined good and bad. For that would be to tie God's hands, to infringe on his sovereign right of decision about what was good.10 This line of thought even contributed in the end to the rise of mechanism: the ideal universe from this point of view is a mechanical one, without intrinsic purpose.11 But, with the modern era, something analogous begins to be transferred on to humans. Norma- tive orders must originate in the will. This is most evident in the seventeenth-century political theory of legitimacy through contract. As against earlier contract theories, the one we find with Grotius and Locke starts from the individual. Being in a political order to 144 CHARLES TAYLOR which one owes allegiance presupposes, on this view, that one has given it one's consent. Rightful submission could not arise just by nature, as classical theories assume. The most radical development in moral philosophy of this line of thought had to await Kant. In the meantime the new conception of freedom, which arose partly from an anthropological transfer of the prerogatives of God, already has some role in the formation of utilitarian thought, and its proto-doctrines. Hobbes' political atomism is plainly linked with his nominalism, and with his view that the good is determined for each person by what he desires,12 both doctrines plainly owing a lot to late medieval defenses of theological decisionism. Political right is made by fiat.13 And, in mature utilitarianism, the stress on modern freedom emerges in the rejection of paternalism. Each man is the best judge of his own happiness. The rejection of the idea that our good is founded in some natural order is seen by utilitarians as the repudiation of paternalism. It is not only justified on epistemological-metaphysical grounds; not only appropriate as an affirmation of the value of ordinary life, but also seen as establishing the individual's freedom to determine the goals of his own life, and his own definition of happiness. This powerful array of motives converges, however mistakenly, to discredit qualitative distinctions, to make them appear as intel- lectually suspect and morally sinister, and to establish a model of moral thinking which tries to do without them altogether. Thence arise some of the perplexities and fudgings of utilitarianism, such as the difficulty in understanding what the moral motivation is that it appeals to, the relation between hedonism as a motivational theory and the benevolence that utilitarian practice seems to suppose, and the like. But the force of these motives is great enough to outweigh the intellectual discomfort of these unsolved puzzles; somewhere, utilitarians feel, there must be answers, and meantime they plunge forward into their homogeneous universe of rational calculation. (And the sense of power and control which comes from this latter is another strong motive for accepting the theory.) Utilitarianism is not the only philosophy responsible for this climate of modern thought. Kant rejects the one-dimensional utilitarian view of human motivation, but he none the less shares the modern stress on freedom as self-determination. He insists on seeing the moral law as one which emanates from our will. Our awe before it reflects the status of rational agency, its author, whose being it A most peculiar institution 145 expresses. Rational agents have a status that nothing else enjoys in the universe. They soar above the rest of creation. Everything else may have a price, but only they have 'dignity' (Wiirde),14 And so Kant insists strongly that our moral obligations owe nothing to the order of nature. He rejects vigorously as irrelevant all those qualitative distinctions which pick out higher and lower in the order of the cosmos or in human nature. To take these as central to one's moral views is to fall into heteronomy. It has, therefore, been easy for the followers of Kant to take this rejection of qualitative distinctions in the order of being for a rejection of any distinction at all, forgetting or putting into the shade Kant's doctrine of the dignity of rational agents. This has been the easier given the connection between the affirmation of modern freedom and the rejection of such distinctions that the naturalist Enlightenment gives currency to; and it is aided and abetted by all the epistemological and metaphysical doubts that it has entrenched in our modern outlook. This mixture of Kantian and naturalist conceptions has yielded the picture of the human agent so familiar in much contemporary moral philosophy. Iris Murdoch captures it in a memorable description: How recognizable, how familiar to us is the man so beautifully portrayed in the Grundlegung, who confronted even with Christ turns away to consider the judgement of his own conscience and to hear the voice of his own reason. Stripped of the exiguous metaphysical background which Kant was prepared to allow him, this man is with us still, free, independent, lonely, powerful, rational, responsible, brave, the hero of so many novels and books of moral philosophy.15 No one can gainsay the power of this ideal among our contemporaries, and, thanks to the confusion mentioned above, it has greatly served to strengthen the modern moral philosophies of obligatory action, which I described earlier, and which tend to sideline these qualitative distinctions, where they do not deny them a place altogether. The focus is on the principles, or injunctions, or standards which guide action, while visions of the good are altogether neglected. Morality is narrowly concerned with what we ought to do, and not with what is valuable in itself, or what we should admire or love. Contemporary philosophers, even when they descend from Kant rather than Bentham (for example, John Rawls), share this focus. Moral philosophy should concern itself with determining the I46 CHARLES TAYLOR principles of our action. Or, where it sees itself in a strictly 'meta- ethicaP role, it should concern itself with the language in which we determine extra-philosophically the principles of our action. Its starting-point should be our intuitions about what actions are right (Rawls), or some general theory about what morality is, conceived in prescriptive, i.e., action-guiding terms (Hare). The idea that moral thought should concern itself with our different visions of the qualitative higher, with strong goods, is never even mooted. Awareness of their place in our moral lives has been so deeply suppressed that the thought never seems to occur to many of our contemporaries. Their conception of freedom and their epistemological suspicion of strong goods bind together, utilitarians and naturalists of all sorts, as well as Kantians, in this suppression. And in this another motive concurs. A central feature of Enlightenment morality, in which it shows its roots in Christianity, is the stress on practical benevolence. This was a crucial theme with Bacon, where it still was expressed in Christian terms, one which he passed on to his more secular spiritual successors. Our scientific effort should not serve simply to create objects of contemplation for us, but should serve to 'relieve the condition of mankind'.16 Practical charity is enjoined on us. The Enlightenment took this up in intensified form; and it has become one of the central beliefs of modern western culture: we all should work to improve the condition of human life, to relieve suffering, overcome poverty, increase prosperity, augment human welfare. We should strive to leave the world a more prosperous place than we found it. This can seem to give independent justification to the exclusive focus on action in much of contemporary moral theory. The focus can be represented as being a sign of moral earnestness, of bene- volent determination. Those who are concerned about what is valuable, what one should love or admire, are worried about the state of their own souls. They are self-absorbed, prone to narcissism, not committed to altruistic action, to the improvement of the lot of mankind, or the defense of justice. Utilitarians seem particularly prone to have recourse to this argument.17 We can begin to appreciate how heavily overdetermined is this vogue of theories of obligatory action. Not only the epistemological and metaphysical predilections of naturalism are at work here, but also strong moral motives. I have spoken about three such: the A most peculiar institution 147 defense of ordinary life and desire against the (supposedly specious) demands of'higher' goods, the modern conception of freedom, and one reading of the demands of benevolence and altruism. But the desire for a fully universal ethic can also play a role. The goods that we articulate in qualitative distinctions are frequently those of a particular cultural group, and embedded in their way of life. If the aim is to avoid, above all, parochial ethical principles, then one has another reason to sideline these distinctions. This is undoubtedly an important factor in Habermas' position,18 and is, in addition to the motive that Williams identifies in his discussion, the attachment to a kind of purity, the aspiration to a moral existence which could be cleansed from 'other kinds of emotional reaction or social influence'. This in turn incorporates 'an ideal, presented by Kant... in a form that is the most unqualified and also one of the most moving: the ideal that human existence can be ultimately just' (p. 195).

Various combinations of these motives tend to bring Kantians and utilitarians together around theories of obligatory action, and, relatedly, lead them to share a procedural conception of ethics. I am using the word 'procedural' here in opposition to 'substantive'. These terms can be applied to forms of ethical theory by derivation from their use to describe conceptions of reason. I call a notion of reason substantive where we judge the rationality of agents or their thoughts and feelings in substantive terms. That means that the criterion for rationality is that one get it right. Plato has a conception of this kind. You could not be fully rational in his book, and believe, for instance, that Democritus was right about the natural world, or that the best life was the one where you fulfilled the most sensual desires. By contrast, a procedural notion of reason breaks this connection. The rationality of an agent or his thought is judged by how he thinks, not in the first instance by whether the outcome is substantively correct. Good thinking is defined procedurally. Des- cartes offers a paradigm example of this, with his model of clear and distinct thought. We end up with the assurance that this will give us substantive truth, but only after we have gone through the argument proving the existence of a veracious God. Correct thinking is not defined by substantial truth, because defining it is the prelude to raising the question as to whether its results are trustworthy. I48 CHARLES TAYLOR Practical reason was understood by the ancients substantively. To be rational was to have the correct vision, or, in the case of Aris- totle's phronesis, an accurate power of moral discrimination. But once we sideline a sense or vision of the good, and consider it irrelevant to moral thinking, then our notion of practical reasoning has to be procedural. The excellence of practical reasoning is defined in terms of a certain style, method, or procedure of thought. For the utilitarians, rationality is maximizing calculation, ^weckrationalitat is the crucial form. For the Kantians, the definitive procedure of practical reason is that of universalization. Beyond the common weight on them of modern epistemology, it is clear how for both the stress on the procedural is bound up with their allegiance to modern freedom. To make practical reason substantive implies that practical wisdom is a matter of seeing an order which in some sense is in nature and which determines what ought to be done. To reverse this, and give primacy to the agent's own desires or his will, while still wanting to give value to practical reason, you have to redefine it in procedural terms. If the right thing to do has still to be understood as what is rationally justifiable, then the justification has to be procedural. It cannot be defined by the particular outcome, but by the way in which the outcome is arrived at. This modern idea of freedom is the strongest motive for the massive shift from substantive to procedural justifications in the modern world. We can see the rise of social contract theory in the same light. Instead of defining legitimacy substantively, in terms of the kind of regime, or some conception of the good society, we define it by the procedure of its inauguration. It is all right, thought Grotius, no matter what its form, as long as it comes about through consent. And, if we leap from the earliest to the most recent such theory, Habermas' conception of a discourse ethic is founded, in part, on the same consideration. The idea that a norm is justified only to the extent that all could uncoercedly accept it, is a new and interesting variant of the procedural idea. It owes something to Kant, but offers a 'dialogicaP procedure in place of Kant's, which each agent could carry out on his own. But this change seems a step forward precisely because it involves a fuller acceptance of the free self-determination of diverse people. In a way, it unites both Kantian universality and the Benthamite refusal to decide for other people what is right for them.19 A most peculiar institution 149 Procedural ethics give an important place to what I called above basic reasons. In a sense, the procedure itself allows us to define some property which all obligatory acts must have, like having universali- zable maxims, or conducing to the greatest happiness of the greatest number. Moreover, they tend towards a unification of morality around a single, or at worst a small, list of basic reasons. There has been a tendency to breathtaking systematization in modern moral philosophy. Utilitarianism and Kantianism organize everything around one basic reason. And, as so often happens in such cases, the notion becomes accredited among the proponents of these theories that the nature of moral reasoning is such that we ought to be able to unify our moral views round a single base. John Rawls, following J. S. Mill, rejects what he describes as the 'intuitionist' view, which is precisely a view which allows for a plurality of such basic criteria.20 To see how far this is from being an essential feature of ethical thinking we have just to look at Aristotle's theory. Aristotle sees us pursuing a number of goods, and our conduct as exhibiting a number of different virtues. We can speak of a single 'complete good' (teleion agathon), because our condition is such that the disparate goods we seek have to be coherently combined in a single life, and in their right proportions. But the good life as a whole does not stand to the partial goods as a basic reason. There is no asymmetri- cal conferral of their status as goods. A good life should include, inter alia, some contemplation, some participation in politics, a well-run household and family. These should figure in their right proportion. But we cannot say informatively that contemplation is a good because it figures in the good life. It is much more that this life is good because, in part, it includes contemplation. This drive towards unification, far from being an essential feature of morality, is rather a peculiar feature of modern moral philosophy.21

VI I have been sketching the contours of a wide trend in modern moral philosophy. Having excluded qualitative distinctions for epistemological and moral reasons, so effectively that it has almost suppressed all awareness of their place in our lives, it proposes a view of moral thought focused simply on determining the principles of action. To the extent that it accords practical reason an important role in determining these, that it considers that the right action should 150 CHARLES TAYLOR always be in some strong sense rationally justifiable, it adopts a procedural conception of reason. One of its central issues concerns how we should understand practical reason, how we should rationally determine what we ought to do. But this suppression leaves perplexing gaps in the theory. It has no way of capturing the background understanding surrounding any conviction that we ought to act in this or that way, the understanding of the strong good involved. And, in particular, it cannot capture the peculiar background sense central to much of our moral life, that something incomparably important is involved. So this reappears in an odd form, for instance in a definition of moral reasoning, as a reasoning of a particular form whose outcomes have a special priority over others. What this priority consists in is left unexplained. For Hare, it belongs logically to the moral. This is just what we mean by 'moral'.22 For Habermas, the priority to a discourse ethic is a product of maturation both ontogenetically and in the history of culture.23 But clearly there is a gaping hole here. We might be tempted to put it by saying that they leave us with nothing to say to someone who asks why he should be moral, or should strive to the 'maturity' of a 'post-conventional' ethic. But this could be misleading, if we thought it to be asking how we could convince someone who saw none of the point of our moral beliefs. There is nothing we could do to 'prove' we are right to such a person. But imagine him to be asking another question: he could be asking you to make plain the point of your moral code, to articulate what is uniquely valuable in cleaving to these injunctions. Then the implication of these theories is that you would have nothing to say which could impart insight. You could wax rhetorical and propagandize, but you could not say what is good or valuable about them, or why they command assent. Or else you can take up the question, 'why should I be moral?' in the way Hare does in his recent book.24 The only kind of answer that Hare can conceive of offering here is one in prudential terms. He tries to show that it is in someone's interest to be brought up with the right moral principles. Within his philosophy, the question I articulated cannot be addressed at all. This conception of the moral is strangely skewed. It tries to account for the incomparable weight of certain considerations, which we should see in terms of the incomparable status of certain goods, by segregating off a domain of the 'moral', which is then A most peculiar institution 151 hermetically sealed off from other considerations. 'Moral' defines a certain kind of reasoning, which in some unexplained way has, in principle, priority. It is not clear how moral consideration can function with others in a single deliberative activity; we cannot see why these higher considerations should usually be given priority, but also why they might be denied this in certain circumstances. For this kind of deliberation would presuppose that we see them all as goods, with different levels of importance. At the same time this segregating boundary cannot just be dropped. For it is the only way a procedural ethic has of marking some considerations as incomparably higher. The more we are really (if inarticulately) moved by a good of overarching importance, the more fiercely we have to defend this boundary, however inadequate our explanation of it may be. For instance, in Habermas' case, the boundary between questions of ethics, which have to do with interpersonal justice, and those of the good life, is supremely important; because it is the boundary between demands of truly universal validity, and goods which will differ from culture to culture. This distinction is the only bulwark in Habermas' eyes against chauvinistic and ethnocentric aggression in the name of one's way of life, or tradition, or culture. It is thus supremely important to maintain it. The more one examines the motives - what Nietzsche would call the 'genealogy' - of these theories of obligatory action, the stranger they appear. Because it seems that they are motivated by the strongest moral ideals, like freedom, altruism, universalism. These are among the central moral aspirations of modern culture, the overarching goods which are distinctive to it. And yet these ideals drive the theorists towards a denial of all such goods. They are caught in a strange pragmatic contradiction, whereby the very goods which move them push them to deny or de-nature all such goods. They are constitutionally incapable of coming clean about the deeper sources of their own thinking. Their thought is inescapably cramped. It is striking how much has been suppressed by these strange theories of modern moral philosophy, which have the paradoxical effect of making us inarticulate on some of the most important issues of morality. Impelled by the strongest metaphysical, epistemological, and moral ideas of the modern age, these theories narrow our focus onto the determinants of action, and then restrict our I52 CHARLES TAYLOR understanding of these determinants still further by defining practical reason as exclusively procedural. They utterly mystify the priority of the moral by identifying it not with substance, but with a form of reasoning, around which they draw a firm boundary. They then are led to defend this boundary all the more fiercely because it is their only way of doing justice to the overarching goods which move them although they cannot acknowledge them. From this frequently follows another of the strange cramps they put in moral thinking, the tendency to unify the moral domain around a single consideration or basic reason, for example, happiness or the categorical imperative, thus cramming the tremendous variety of moral considerations into a Procrustes' bed. And there are other cramps as well. The notion that morality is exclusively concerned with obligations has had a restricting and distorting effect on our moral thinking and sensibility. Williams shows eloquently how badly distorting this is, and how it fails to cope with all that aspect of our moral thinking which concerns aspirations to perfection, heroism, supererogation, and the like. Once more, in Procrustean fashion, this is either assimilated to a foreign mould or rejected.

VII In drawing this portrait of'morality', I have in a sense been writing an elaborate footnote to Williams' chapter 10. I have been adding some details to his thesis, tracing some neglected strands in the complex genealogy he sketched. The spirit is one of agreement, one might say a fortiori agreement. But, at the same time, there is a displacement of emphasis. Williams shows how the morality system exalts one set of ideals over all others, illicitly and without taking responsibility for its preference, under cover of specious doctrines about the nature of moral theory as such. Williams is less than enamoured of these ideals, and the unmasking of morality would, ideally in his perspective, help to loosen their hold on us. To the extent that he sees the ideals as of Christian provenance, this loosening would further a development which he thinks he descries in modern culture, and which he approves, the 'departure of Christianity' (p. 198) and the corresponding renewed sense of the relevance of the philosophies of the ancient world. I heartily concur in denouncing the illicit privileging of some A most peculiar institution 153 goods on these specious grounds, but I do not think I judge these goods themselves as negatively as Williams does. I see my self, and also I must say modern culture, as deeply committed both to these ideals, for example, universal justice and benevolence, and to the goals which contend with them for space in our lives, the various notions of personal fulfilment, or those definitions of a worthwhile life which include more than heroic dedication to the welfare of others. In this sense, I see no sign of a serious 'departure of Chris- tianity' (or in any other sense, either). My main grievance against the morality system concerns the dead weight of enforced inarticulacy that it lays on modern culture. Williams could have recourse in the fight against morality to Nietzsche. I very much agree, provided one invokes the master himself and his ideal of'ja-sagen'. But, if it is a question of fashion- able neo-Nietzscheans, like Derrida or Foucault, they are no help whatever. They contribute to the climate of inarticulacy; they have their own ways of insinuating their ideals without directly espousing them. They have much in common with the moralists they attack. Just because we are attached to disparate goods, some of them in conflict, we need articulacy about the good.25 One of the laughable, if lamentable, consequences of modern moral philosophy is that you have to fight uphill to rediscover the obvious, to counteract the layers of suppression of modern ethical consciousness. Being able to talk again about goods is a condition of addressing the serious issues Williams raises when he impugns the ideals that 'morality' enshrines. Showing its genealogy is only the preliminary step. The serious contest can only begin when everyone is out from behind their pseudo-epistemological screens. We are still a long way from there, but we have been brought a little closer by Williams' excellent work. May it continue.

NOTES 1 Page references, unless otherwise attributed, are to Ethics and the Limits of Philosophy, London: Fontana, 1985. 2 R. M. Hare, Moral Thinking, Oxford University Press, 1981, 1.9, pp. 16-17. 3 J. Habermas, Moralbewusstsein und kommunikatives Handeln, Frankfurt: Suhrkamp, 1983, pp. 76, 103. 4 'Nothing in the world — indeed nothing even beyond the world — can possibly be conceived which could be called good without qualification 154 CHARLES TAYLOR

except a good will.'' This is the opening sentence of Kant's Foundations of the Metaphysics of Morals. 5 See Habermas, Theorie des kommunikativen Handelns, Frankfurt: Suhr- kamp, 1981. 6 See John Mackie, Ethics, Harmondsworth: Penguin Books, 1977. 7 See Hare, Freedom and Reason, Oxford University Press, 1963. 8 C. Geertz, The Interpretation of Cultures, New York: Basic Books, 1973, ch. 1. Geertz acknowledges the earlier coining of the term by Gilbert Ryle in his Concept of Mind, London: Hutchinson, 1949, but it is Geertz himself whose illuminating use of it has given the term its resonance. 9 See Gail Sheehy, Passages: Predictable Crises of Adult Life, New York: Dutton, 1976. 10 See Francis Oakley, 'Christian theology and the Newtonian science', Church History, 30 (1961), 433-57. 11 See Robert Lenoble, Mersenne et la naissance du Mecanisme, Paris: J. Vrin, 1943, 12 'Whatsoever is the object of a man's appetite or desire, that is it which he for his part calleth good,' Leviathan, ed. Michael Oakeshott, Oxford: Blackwell, n.d., ch. 6, p. 32. 13 'the pacts and covenants, by which the parts of the body politic were at first made, set together, and united, resemble that fiat, or the let us make man, pronounced by God in the creation'; ibid., Introduction, p. 5. 14 I. Kant, Grundlegung zur Metaphysik der Sitten, Berlin Academy edition, reprinted Berlin: Walter de Gruyter, 1968, iv, p. 435. 15 Iris Murdoch, The Sovereignty of Good, London: Routledge, 1970, p. 80. 16 F. Bacon, Novum Organum, 1.73. 17 Bentham in On the Principles of Morals and Legislation, ch. 2, paras. 5-6, takes the moral high ground with his opponents, and explains the adoption of the 'principle of asceticism' from the motives of pride and 'fear, the offspring of superstitious fancy.' 18 J. Habermas, Moralbewusstsein, pp. 113-14, 117-19. 19 See Habermas, Moralbewusstsein, ch. 3, and Theorie des kommunikativen Handelns, 1, ch. 3. An influential trend of contemporary thought has been applying this proceduralist move to political theory, as distinct from ethics, trying to develop norms of social justice or fairness, norms governing the coercive actions of political authorities. Thus Ronald Dworkin, who explicitly recognizes some notion of the good life (or at least of a life worth living) as crucial to moral choice, wants to exclude any considerations about the good life from political deliberations in order to be true to what he understands as the (defining liberal) principle of treating citizens as equal ('Liberalism', in Stuart Hamp- shire (ed.), Public and Private Morality, Cambridge University Press, 1978. See also 'What Liberalism Isn't', in The New York Review of Books, 20 January 1983, pp. 47—50). Rawls' celebrated theory again concerned social justice. Views in this domain about the neutralizing of some of our ends for the purposes of political deliberation have been put A most peculiar institution 155 forward by Scanlon and Nagel ('Moral conflict and political legitimacy', Philosophy and Public Affairs, 16: 3 (1987)). This is obviously a distinct issue from the one I am discussing in this section, although obviously these political doctrines are influenced by procedural views in morals; and the two have developed together since Hobbes and Grotius. But it is quite possible to be strongly in favour of a morality based on a notion of the good, yet lean to some procedural formula when it comes to the principles of politics. There is a lot to be said for this, precisely for the sake of certain substantive goods, for example, liberty, and that of respecting the dignity of all participants. Procedural norms have certainly been one of the crucial arms of liberal democracy. In the work of Rousseau himself, often thought the proge- nitor of 'totalitarian democracy' (J. L. Talmon, The Origins of Totali- tarian Democracy, London: Seeker & Warburg, 1952), the procedural requirements — common deliberation, the participation of all — have offered the strongest bulwark against the dangerous unanimist or 'Jacobin' thrust of his thought. If in the end I cannot quite agree with some such procedural view as the sufficient definition of the principles of liberal democracy, this is not because I do not see its force. The political issue is, quite distinct from that of the nature of moral theory. 20 A Theory of Justice, Oxford University Press, 1972, ch. 1, section 7; Mill uses the same expression for something similar in Utilitarianism, ch. 1. Rawls is here talking not of the whole of morality, but only of justice. 21 I have argued against this tendency in my 'The diversity of goods', in Philosophy and the Human Sciences, Cambridge University Press, 1985, ch. 9- 22 See Hare, Freedom and Reason, 9.3, and Moral Thinking, pp. 24, 55. 23 SeeJ. Habermas, Moralbewusstsein, ch. 4; also Theorie des kommunikativen Handelns. 24 Hare, Moral Thinking, ch. 11. 25 I have tried to make this argument at greater length in Sources of the Self, Harvard University Press, 1989. CHAPTER 8 Reflection and confidence J. E. J. Altham

Reflection is a pervasive theme of Ethics and the Limits of Philosophy (ELP), and it plays a predominantly discouraging role. The most startling claim made about it is that reflection can destroy knowledge, and in particular ethical knowledge. Indeed, reflection has destroyed at least some ethical knowledge. It does not destroy in order to rebuild: we cannot, by further reflection, acquire new ethical knowledge to replace what we have lost. For when reflection becomes philosophical, it reveals that it cannot provide a secure foundation for ethics. Ethical knowledge can be built neither on the requirements of practical reason, nor on any account of human nature, nor on any other basis attainable through reflective thought. We cannot avoid these ravages by avoiding reflection itself. For one thing it is too late: much ethical knowledge has already gone, and ceasing to reflect will not bring it back. For another, there is no route back from reflection: there are strategies to reduce it, but the costs of any substantial reduction are high, not least in honesty to ourselves. Williams thinks, however, that some ethical knowledge may survive reflection, and also that we do not have to support ethical life through thinking that we have knowledge of our values. We can in some circumstances have confidence without knowledge. In favourable circumstances, such confidence can survive reflection, but is not attained by it. People have it when social conditions are propitious, and they lead ethical lives that are sustained by those conditions. In this chapter I try to explore further the prospects for ethical confidence. I do not attempt to say anything about the lives and societies that may favour it. The task is rather to see what place there is for ethical confidence, given the erosive power of reflection. For 156 Reflection and confidence 157 there are considerations that may lead one to believe that, if we have some ethical knowledge, confidence can extend no further than the recognized possibilities for such knowledge, or, at least, can do so only at the cost of self-deception. If I have some ethical knowledge, how can I be confident about ethical matters that I recognize are not possible objects of knowledge? The idea is that I cannot both know some ethical truths and be confident of other ethical propositions that I recognize that I do not know. Once ethical knowledge is allowed to have some place, it may take up all the space available. If on some matters, but not all, it is acknowledged that knowledge is not attainable, even confidence in those matters lapses. This argument will be deployed on the assumption that Williams is right that reflection can destroy knowledge, and, for most of the time, on the assumption that there is at least some ethical knowledge that survives. These claims may be denied. I am not myself convinced that reflection can destroy knowledge. This may be a faulty description of the phenomena. In some cases, at least, reflection shows that what seemed to be knowledge was not so really, and perhaps what Williams calls destruction is always the uncovering of a false appearance. I shall not argue this point here, for it makes no difference provided that the second premise is true. If reflection shows that some of what I took to be knowledge is not, but some of what I take to be ethical knowledge survives it, then there is still a problem about how I can be confident without knowledge. The case is different if the second assumption is denied. If no ethical knowledge survives reflection, and this is because it is revealed that there never was any, then after this has happened it is clear that the relation to truth indicated by 'knowledge' is not a proper model in ethics. If, on the other hand, all ethical knowledge is actually destroyed, then perhaps it is here that there can be a place for confidence instead. There are further difficulties arising from other things that Williams says. He declares more than once that ethics cannot be all that it seems to be. This is obscure, because, despite confident appeals to moral phenomenology in some recent literature, it is not at all clear how we should describe what ethics seems to be. Williams is emphatic in this context that there is a contrast between ethics and science, and if there is one thing that ethics seems obviously not to be, it is science. Fortunately there is more to the idea that ethics cannot be what it seems to be than this. One element concerns the I58 J. E. J. ALTHAM consequences of understanding our own ethical thought. It may be that the content of a particular ethical conviction may survive an understanding of why one has it, but the conviction may still be recognized to involve a misrepresentation in some other way. And it may be that we cannot correct the misrepresentation while retaining the content. When I see a trompe I'oeil painting, and recognize that what looks three-dimensional is really a flat surface, the illusory appearance survives. A genuinely three-dimensional object could not have the same effect. Similarly, an ethical conviction seems to present a requirement to act which comes from outside me. In fact it comes from inside me. Any requirement that really came from outside could not have the consequences of a moral conviction, for the possibilities of rejecting any external authority are not the same as those of rejecting the authority of an ethical requirement. But also, any requirement that did not seem to come from outside could not have the grip on me that an ethical conviction has. There is insufficient space here to explore these issues in ethical phenomenology.

11 Desire need not fail when its end is understood not to be objectively valuable. Suppose Gavin wants a Porsche because he wants to excite the envy and admiration of others by appearing wealthy and stylish. He does not need to think it is objectively valuable to be envied and admired on these grounds. Objective value has nothing to do with the matter. Suppose Kevin wants to give his girlfriend a nice surprise, and manages to get hold of two tickets for the men's final at Wimbledon. This little piece of altruism is quite independent, both logically and psychologically, of whether Kevin thinks that giving his girlfriend a nice surprise, or watching tennis, has objective value. These desires stem from Gavin's and Kevin's dispositions of character, from the vanity of the one and the kindness of the other. To describe these desires in this way is already to bring them under concepts of vice and virtue, and so to characterize them as ethical desires, taking 'ethical' in this context as a classificatory term, rather than as one of favourable appraisal. But, even if we think that vanity and kindness are objectively a vice and a virtue respectively, we should not think that the dispositions to desires in which these qualities, in part, consist have to be dispositions to desire things that Reflection and confidence i59 are themselves objectively valuable. Nor should we think that the people who have these dispositions would have any tendency to lose them if they thought that the ends of their desires were not matters of objective value. There is, however, still a widespread tendency to think that some ethical conviction at least, and their associated dispositions to desire and act, are destroyed if one ceases to believe in their objectivity. Here the idea is that the loss of belief in objectivity is by itself enough to destroy conviction. This idea is often found in beginners, but not only in them. It can survive in graduate students. It is an error that needs some discussion to distinguish it from the problem I find in Williams' own position. A person in the grip of this idea typically will say something like 'If ethics is not objective, then I can do what I want.' This remark contains a firm pointer to a common source of the idea. It stems from the role of ethical conviction in checking self-love. Carlos wants to be rich, and can make a lot of money by bribing officials to secure him a business concession. If anything stands in his way, it will be a sense of justice. The picture is one of what Carlos wants for himself on the one hand, and the demands of justice on the other, purporting to restrain his pursuit of his own ends. If the principles of justice are not objectively grounded, any demand that justice can make on Carlos must really be one that he makes on himself. If he realizes this, he may wonder whether to go on making such demands on himself. From the point of view of his desire for riches, these demands are an encumbrance. From that point of view he will ask himself what making these ethical demands does for him, and will answer 'Not enough'. He will therefore have reason to try to be rid of the disposition that makes him experience justice as a demand. At least he has no longer any reason to respect it. In these circumstances, his sense of justice is likely to disappear, or at least to decay. This argument proceeds from the point of view of Carlos' desire for riches. But that may not be the only point of view available to him. If he is actually disposed to be just, then, from the point of view of his desire for justice, it may be the craving for riches that is seen as an encumbrance. Alternatively, he may see neither the desire for riches nor his sense of justice as encumbrances, and reach the stable position of pursuing wealth by just means. To give priority to the desires for riches, so as to wish to be rid of one's dispositions to be just, that desire must be assigned some kind of authority or rational l6o J. E. J. ALTHAM paramountcy. But there is no need to accord it any such status. We are not concerned here with the rather special outlook of one who thinks that, while the requirements of justice are not objective, personal wealth is objectively valuable. This would not be a scepticism about objectivity of value in general, which is currently our topic, but a scepticism specifically about justice within an acceptance of value objectivity elsewhere. If the desire for riches is thought to have authority, once a belief in the objective grounding of justice is rejected, it is likely that the ground for the thought will be that the pursuit of self-interest is rationally paramount. This is, however, dangerous territory. The miserable bog must be avoided, where the rationality of self-interest seems so powerful that even knowledge of an objectively grounded morality would provide insufficient rational motivation to with- stand it. To one stuck in this bog, the objectivity or otherwise of the demands of justice makes too little difference to rational motivation to be of much significance. If, in a conflict between morality and self-interest, the latter should rationally prevail, it does not matter whether the morality can be objectively grounded or not. The route into this bog starts with the thought that it is paramountly rational for a person to pursue his or her own happiness. It proceeds with the premise that happiness consists in states of consciousness, and adds that these states are definable independently of ethical considerations. For then it will be rational to act in ways that will promote the presence of one of those states, and, if they are independent of ethics, then it will be rational to promote them even if doing so conflicts with ethical requirements. The way out of the bog starts with the claim that the argument assumes an agent with no ethical ends. For, if an agent has any ethical ends, his or her happiness will not entirely consist of states of consciousness definable independently of those ends. If, for example, Joan wants to live an honest life, then one will not be able to specify in what Joan's happiness would consist without invoking the notion of honesty. One cannot then conclude that it would be rational for Joan to promote her happiness at the expense of ethics, since her happiness involves ethics. On the other hand, if an agent genuinely has no ethical ends, the argument can be allowed to stand, but it is not a threat. The threat was that because it is paramountly rational to pursue one's own interest, it is rational to wish for the decay of ethical dispositions that would hinder that pursuit. If there are no Reflection and confidence 161 ethical dispositions in an agent to start with, the wish for their decay has no object. More importantly, if an agent does have ethical dispositions, the argument does not lead to any rational wish for their decay. For then the agent's interest may include an interest in the exercise of those dispositions. This is not an argument that there is, after all, no conflict between self-interest and ethical considerations. Such conflict is not only possible, but common. The argument is meant only to show the failure of one attempt to show that, in the light of such conflict, it is rational to wish to be rid of the ethical considerations. Consider how the matter may appear to reflection. George has ethical desires, and he also has non-ethical ones. He reflects upon these, asking himself whether he identifies with them so as to accept his going on having them. He may conclude that he wants to have the non-ethical desires, and also to have the ethical ones, even in the knowledge that they frequently conflict. The only acceptable lives, he thinks, are ones which are vulnerable to just these sorts of conflict. The loss of desires of either kind would be an intolerable impover- ishment. I have conceded that, if an agent has no ethical ends, it is rational for her to promote the ends she does have even at the expense of ethical considerations. It must also be conceded that such an agent is a possible being. It may then be claimed that this is to concede too much, on the grounds that, if such a being is possible, one should aim to be one. The claim is implausible. The 'should' is evidently not an ethical one. The claim has no plausibility if the 'should' is instrumental. For then it says, in effect, that one will gain one's ends better by being anethical, and this is not true for an agent with ethical ends. If the 'should' appeals to a putative rationality independent of an agent's contingent ends, the argument is lacking that there is such a concept which would yield the rationality of an anethical life. The aim of providing such argument could be described as the counter-Kantian project, and has no more prospect of success, and less appeal, than the Kantian one. We have seen that, if the pursuit of self-interest as such were shown to have rational authority, the argument would prove too much. For it would then have authority whether or not ethical considerations could be objectively grounded. We have also seen that the argument, in at least one tempting form, does not succeed. Even if we concede the rationality of pursuing one's own happiness, 162 J. E. J. ALTHAM much then depends upon what one's happiness consists in. If, as may be the case, an individual cannot be happy without the exercise of ethical dispositions, then, whether or not these dispositions are objectively grounded, the agent has no reason to wish to be rid of them on grounds of happiness. An agent whose happiness is attained through an anethical life has no reason to cultivate ethical dispositions, but this remains so whether or not there is an objective grounding, and the rest of us do not have reason to want to be like that. In none of these cases is there a good general argument from rationality and non-objectivity to the decay of the ethical dispositions. It is possible that for some individuals certain of their ethical dispositions depend psychologically upon a belief that they are objectively grounded, but, upon the whole, ethical dispositions, being part of a person's character, are not even psychologically vulnerable to the results of inquiry into their general metaphysical status. This is not to deny that ethical dispositions can decay as a result of reflection, but where they do, the reflection is more specific than the conclusion that ethical dispositions in general are not objectively grounded. The main task of this chapter is to consider some more specific reflections, prompted by the thought of Bernard Williams, which may be more efficacious in modifying ethical dispositions than a general scepticism about objectivity. To that task I now turn.

in According to ELP, ethical knowledge is possible, and some people have it. It is embodied in propositions using ethical concepts whose application is both world-guided and action-guiding, concepts that Williams describes as 'thick'. They are world-guided in that the criteria for their application are such that perceptually based investigation can establish whether the criteria are fulfilled. They guide action in that they are so used to appraise actions, persons, or situations that they provide reasons for action. These concepts are also, according to Williams, in a certain respect unitary. They cannot be analyzed into distinct descriptive and prescriptive components. In some cases the world guides the application of such an ethical term only through the mediation of a false belief. For example, the proposition 'She's a typical Aquarian' depends upon a false belief Reflection and confidence 163 about the influence of heavenly bodies upon human character. Propositions whose application depends in such ways upon false belief do not express knowledge. Some thick ethical concepts, on the other hand, have applications which are not mediated by any false beliefs. Propositions involving these, claims Williams, can be true and known. People using the concepts can thus have ethical knowledge. It may look as though, if this is right, there is quite a lot of ethical knowledge. For very many concepts that are both ethical and thick are in constant use, and are applicable on vast numbers of occasions. Williams specifically mentions coward, lie, brutality, and gratitude. There are also callous, kind, generous, underhand, spiteful - the list could be continued almost indefinitely. But, according to Williams, at the reflective level there is not a body of ethical knowledge. Further, modern societies are peculiarly characterized by reflection, including ethical reflection. The claim here is that ethical knowledge is possessed at a pre-reflective level, and paradigmatically by a fiction called a 'hypertraditional society', whose members have not begun to reflect upon its practices. It is reflection that destroys ethical knowledge. Reflection is said to destroy ethical knowledge by disturbing, unseating, or replacing the concepts used to express it. Thick ethical concepts may, however, survive reflection. The conditions of their doing so are not very clear, and survival in some cases may be simply because users have not thought about them hard enough, or not in the right way. But there is no claim that enough reflection would unseat all thick concepts, and no such claim would be plausible. So the position is that unless and until reflection has made a particular thick ethical concept unusable, we may know ethical propositions expressed in terms of it, and at any given time there are likely to be many such concepts that are actually surviving reflective thought. Each of us, every day of our lives, makes ethical judgments in terms of thick concepts. We deplore one man's act of low cunning, we cannot help thinking that a certain woman is a howling bitch, we admire another's moral courage, and so on. There are so many such thick concepts, and they have so large a place in our thought about ourselves and others, that life without them would be unimaginably different, and reflection is unlikely to force such a big change upon us. So, in Williams' picture of things, contrary to first appearances, it 164 J. E. J. ALTHAM is not only the members of a hypertraditional society, who have never reflected upon their concepts and practices, who have ethical knowledge. Members of reflective societies have ethical knowledge as well, through thick concepts that have survived reflection, as well as others which may so far have escaped it — the latter perhaps a fast diminishing band. This cannot be the whole picture, however, for, if it were, there would not be the need that Williams sees for the model of ethical conviction for which he chooses the name confidence. Confidence is a post-reflective state. We may be confident of what we know, but this notion of confidence would have little point unless, in favourable circumstances, we could also be confident in ethical matters where knowledge is not available. Now if, as seems highly probable, there is a large and varied set of thick ethical concepts whose members are not unseated by reflection, this notion of confidence has no distinctive application to judgments that use them. For, even post-reflectively, these judgments can be known. If a thick ethical concept has been unseated by reflection, then the notion of confidence also has no application to judgments using it. Just because it has been unseated, we no longer make judgments in which it figures, and so are not confident of them. If we are to be confident of some ethical judgments without knowing them, the judgments must be those whose ethical concepts are thin - general concepts such as those of being right and being good. The question now is what may give people confidence in such judgments? There are two obvious possibilities. One is that a person may subscribe to an ethical theory that delivers conclusions in particular cases about the right and the good. Williams is sceptical, throughout ELP, about the present state of ethical theory, and, still more pertinently, about its prospects. One may make a notional distinction among forms of scepticism about ethical theory, between scepticism whether an ethical theory could become part of our knowledge, and whether an ethical theory could be one of which we could be confident. Williams' scepticism is at least of the former sort, so, if ethical theory were to be a source of confidence, it would be confidence of the distinctive kind, the kind supposedly available without knowledge. But this is not what ethical theory offers, and it is not easy to see that it could. Consider two kinds of ethical theory. One bases itself upon funda- Reflection and confidence 165 mental considerations - perhaps about human nature or about practical reason - that can be established independently of judgments that use thick ethical concepts. In this case the theory could correct these judgments. It would do so by putting the thick concepts in terms of which they were expressed out of use, and destroying the knowledge previously expressed in terms of them. It seems clear that an ethical theory of this kind would have to be a candidate for knowledge. Relevant considerations about human nature, for example, would have to be offered as sufficiently factual to be assessed for truth and falsity, and knowable if true rather as other general facts are. Relevant considerations about practical reason are typically offered as truths knowable a priori. Ethical theories of this kind are, therefore, not ones that can give us confidence of the distinctive Williams kind. If they were to succeed, they would give us knowledge, so that the distinctive notion of confidence would be redundant. But, in fact, if Williams' scepticism is correct, they are doomed to failure, and a failed candidate for knowledge yields no confidence in it. A second type of ethical theory is one that makes no attempt to get behind the judgments that use thick ethical concepts. It rather attempts to systematize them, and, if a general view of human nature or of practical reason emerges, it will be the product of reflection upon the thick concepts. The notions of the right and the good, of obligation, and other thin ethical concepts, will then have their application dependent upon the use of the thick concepts. In that case, however, if judgments using the thick concepts express knowledge, and the thick concepts survive, as many of them will have to in this style of ethical theory, then judgments using the thin concepts may also express knowledge. Once again, the distinctive notion of confidence has no role to play. Neither kind of ethical theory can yield confidence without knowledge, and according to Williams the former kind at least cannot yield knowledge. The confidence that he suggests as a fortunate ethical state to be in must have its place, as we have seen, in the use of thin ethical concepts, but this use must not be driven by ethical theory. However, if this use is not driven by ethical theory, it must rest upon the unsystematized judgments that use thick ethical concepts. For judgments that a person is good or that an action is right do not arise in vacua, and reflective individuals are ready to try to provide reasons for the judgments of this kind that they make. l66 J. E. J. ALTHAM Where these reasons do not derive from an ethical theory, they will appeal to thick judgments. For example, the person is good because she is brave, generous, just, wise, and a loyal friend. If the thick judgments can, even post-reflectively, be items of knowledge, then so can the thin judgments to which they give genuine support. If a thin judgment is supported neither by such thick ones, nor by an ethical theory, then it looks like the outcome of an arbitrary decision, and is not a suitable subject of ethical confidence.

IV If a role is to be found for confidence, the effects of reflection must be taken to be more radical than I have so far supported. Williams usually speaks of reflection destroying knowledge by making people no longer willing or able to go on using the concepts in which knowledge was previously expressed. If confidence rather than knowledge is to be, for us reflective people, a fortunate ethical state, it must be possible for reflection to destroy knowledge in another way. (I am continuing to assume the controversial proposition that there is such a thing as ethical knowledge). This way must also affect the thick ethical judgments, for if it did not there would still be little room for confidence as opposed to knowledge, since so many of these judgments survive, and, given the failure of ethical theory, they are the basis of every other kind of ethical judgment. Reflection must, through increased understanding of ethical life, change the status accorded to the thick ethical judgments, so that they no longer count as knowledge. We can then no longer know these judgments, but it may still be open to us to have confidence in them. Why should reflection be thought to have this effect? Here I must draw attention to a puzzle about the structure of ELP. The idea that reflection may destroy knowledge is introduced in chapter 8, 'Knowledge, Science, Convergence', but the introduction of the idea of ethical confidence as a favoured state comes at the end of the subsequent chapter, after the discussion of relativism. The puzzle is that the reflections that lead to the conclusion that some room must be found for relativism are not explicitly associated with those that lead to the introduction of the idea of confidence. For the latter hark back to the previous chapter, and the idea that reflection may unseat ethical concepts. It does not appear to be the reflections Reflection and confidence 167 associated with relativism that unseat these concepts. But perhaps the connection of thought is there, even if it is not explicitly made. Three main thoughts contribute to the sense that some room must be made for relativism. One is the consciousness, essentially a product of reflection, that our own ethical judgments and practices do not have an objective grounding. A second is the knowledge of ethical variation between cultures, and a third is some degree of awareness of the sorts of explanation there can be of such variation. These do not explain variation by showing how different cultures can be, to varying extents, suffering from ethical error. Williams thinks that one appropriate response to acceptance of these thoughts is to suspend ethical judgment in relation to some of the practices of societies that are, in a suitable sense, distant from our own. This suspension will include judgments made using some of our thick ethical concepts. Such suspension of judgment does not render the concepts unusable. We may continue to apply them confidently within our own society. So the reflection that prompts relativistic suspension does not destroy knowledge by replacing concepts. But it may destroy knowledge in another way. Suppose I read about initiation rites in some distant society, and my natural inclination is to think of them as sickeningly cruel. Then I recall the lessons of reflection, and suspend judgment. The problem is that the expression 'sickeningly cruel' is world-guided, and the guidance the world offers is that the rites are sickeningly cruel. The criteria for applying the description are satisfied. The judgment should then be true, and, if it is true, why should I not make it? It might be pointless or tactless or in some other pragmatic way out of place to make it, but that is not what is at issue. The relativistic suspension applies also to my unexpressed thoughts. I fail to believe that the rites are sickeningly cruel. It appears, then, that I know that I have adequate evidence for a belief that I yet fail to hold. This would be irrational, so to make the situation not irrational I must suppose the concept changed. The old criteria for saying that something is sickeningly cruel are no longer sufficient. What now has to be added is some description of my relations to the culture or practice within it. But this added condition is no part of my understanding of what it is to be sickeningly cruel. It changes the criteria for application without changing the descriptive character of what the term applies to. I am thus no longer just stating a truth in contexts where I do not l68 J. E. J. ALTHAM suspend a judgment of cruelty. If the judgment is appropriate, and I am aware of this, I am aware of my own social relation to the thing described. If I take the relativistic line that Williams recommends, it seems that I must take a non-objectivist view of my own activity in making ethical judgment. I see it as a socially local form of understanding, and of restricted applicability for that reason. It is, Williams says, a way of finding my way about a social world, and primarily my social world. It seems that I should, in reflection, cease to think of the belief the judgment expresses as being true or false. For there is a condition on having the belief that is no part of its content. (Williams has earlier excluded explicit relational relativism that would make relativity to a society part of the meaning of an ethical judgment.) If this is right, then the thoughts that lead to making room for relativism also lead to the destruction of knowledge, to the reflective understanding that, even where a thick ethical concept survives, the judgments made in terms of it are not knowledge, because we can no longer see them as true or false. There is a sense in which, when this happens, the old concepts become no longer usable in their original form, so that the account of the destruction of knowledge in these terms remains applicable. But to say that would obscure the fact that concepts can be displaced in at least two radically different ways. One form of displacement takes place when, after reflection, we reject a value needed to give a term its customary application, and we drop the term or change its meaning. This is displacement by ethical criticism. The displacement consequent upon relativistic reflection is displacement by meta-ethical critique. The former results from a change in ethical conviction, the latter from a change in our understanding of the nature of ethical conviction. The effects of displacement are also different in the two cases. In the former the concept is dropped, whereas in the latter it continues in use in a modified form, the change being only in its relativistic limitation of application and what is consequent upon that. Williams emphasizes displacement of the former, concept- dropping kind, and seems to bring in confidence to take the place of knowledge thus lost. I have argued, through considerations drawn from Williams himself, that confidence cannot successfully be brought in at this point, and that, if it could, there would be no need for it. The relativistic displacement, on the other hand, seems to Reflection and confidence 169 offer more hope for confidence as a replacement for knowledge. For, if the relativistic reflections make us see certain judgments as no longer expressive of knowledge, they may leave us in possession of these judgments, as made within our own society and those not too distant from it. A question that now arises is whether there are any judgments that are immune to relativistic limitation. If there are, then there is a split within ethics. The immune judgments could continue to be regarded as ethical knowledge, the others not. But this would be awkward for the latter. Ethical judgment aspires to universality, so that relativistic limitation always checks a natural tendency. There is a danger that where the tendency is checked the judgments will be seen as not the real thing. Compared to judgments that retain universality, the limited ones may come to look rather like mere socially shared tastes in human beings. This danger is avoided if all judgments are at least liable to limitation. For then the universalist aspirations would have to be taken to result from a mistake, and there would be nothing superior to show that limited judgments were not the real thing. There are reasons to suppose that, if relativistic limitation is possible at all, every ethical judgment is liable to it. The relativism is one of distance, and it seems impossible a priori to set limits on how distant another society might be from our own, and in what respects. If every kind of ethical judgment is thus liable to relativistic limitation, then, if we are conscious of this, and the considerations so far are correct, all ethical knowledge would be destroyed. The most that would remain would be confidence in judgments of potentially limited applicability.

v Williams seemed to present us with a picture in which some ethical knowledge had been destroyed, but we still had some, and in which our ethical confidence extended beyond knowledge. This now seems an unstable structure. Knowledge seems to leave no room for a confidence that goes beyond what could be known. If confidence is to be the fortunate ethical state that Williams thinks it is, it will be because the proper reflective conclusion is either that all ethical knowledge has been destroyed, or that we never had any, but have now been freed from the illusion. CHAPTER 9 The shape of a life Martin Hollis

Jim did shoot an Indian; George did not accept a research job in biological and chemical warfare. George knew that the job would otherwise go to a zealot without scruples, but still refused. Jim knew that nineteen other innocent hostages would otherwise be killed and therefore agreed to shoot one himself. When reminded of the details (in a moment), we see that each has faced a dilemma of negative responsiblity and resolved it as integrity demanded. It seems that both cannot be right. Yet, given Bernard Williams' outlook on ethics and the limits of philosophy, and granted his way of connecting ideas of action, integrity, and a meaningful individual life, George and Jim need not act in carbon copy. Serious moral choices are made in a sort of middle distance between the everyday and the eternal. In the middle distance we each have our separate projects, to be reflected on, but not to be set aside. Williams is arresting in his account both of what can give a shape to a life, and of the angle from which we must look. Although I shall dissent in the end, he captures my ethical doubts with unnerving skill and I hope to have learnt much on the way. George and Jim turn up as cases to think about in Williams' half of Utilitarianism: For and Against,1 where they illustrate what is amiss with utilitarian advice to decide moral dilemmas solely by striking a balance of utilities. Personal integrity is not to be traded off in this manner and, as Williams puts it elsewhere, we are not to treat a moral agent as 'just the representative of the satisfaction system who happens to be next to certain causal levers at a certain time'. (PCM p. 4). But the stakes are higher than whether to be for or against utilitarianism. The Kantian alternative is also rejected for miscon- ceiving moral agents and their possible motivations. In general there is no Archimedean standpoint of any kind to give universal leverage on moral problems. Plato's search for the form of the good, and

170 The shape of a life 171 Aristotle's for the proper function of a human life, failed equally, because the traditional quest of moral philosophy has been doomed from the start. Ethics and the Limits of Philosophy completes the diagnosis. It describes 'morality', meaning a systematic answer to the Socratic question of how one should live, as 'the peculiar institution' - a slavery which we must escape before we can make moral sense of life. Escape involves switching off the overhead light of pure reason and doing without its meretricious view from nowhere on a landscape without shadows. The human landscape has contours, to be seen with the help of shadows cast by subtler lighting closer to the ground. Hence the bleak skepticism of the book is 'more about philosophy than it is about ethics' (ELP p. 74). If morality is abandoned, we can attempt a situated ethics, with each life lived in its place and time by an agent whose individual meaningful life it is, thanks to projects which give it a particular shape and texture. Freed from morality, we go beyond the limits of philosophy, but without regret, because 'the only serious enterprise is living'. (ELP p. 117). The message brings together earlier themes, which might otherwise seem loose ends, and hints how the only serious enterprise might be conducted. The hints are hard to extract, because so much of the book is a demolition. The loudest message is (as in Williams' study of Descartes) that there are no absolute reasons for belief and no universal standpoints. That seems to invite a relativism of a thoroughly sceptical sort. But there is also a constructive invitation to cross the choice between absolute and relative with a choice between objective and arbitrary, and to emerge with a category of reasons which are relative, but objective. This is the raison d'Hre of the middle distance and I shall be trying it out in what follows. To focus this large theme, I begin with George and Jim and the apparent oddity of their making contrary decisions. (UFA pp. 97-118). Briefly, George has a new Ph.D. in chemistry and a family to support. He needs a job. A very good one has been offered, researching in biological and chemical weapons. It is the only good one in prospect and, were it not for his moral scruples about the work, he would have every reason to take it. His wife does not share the scruples and, to help outweigh them, there is the knowledge that the job will otherwise go to a zealot, who will push the research along with gusto. As Williams tells the tale, however, the fracture in George's life would be too great to live with. He has a moral 172 MARTIN HOLLIS character or self, which he cannot step out of. His most deeply self-definitive project cannot be tossed into the scales of utility and (for reasons which will emerge) he cannot question it radically. This moral character or self sets boundary conditions. It is not a mere bundle of preferences, and involves commitments not up for external assessment. The crux is not that the price of taking the job is too high, but that acceptance would destroy the measure of prices by which George lives. He refuses the job and thus, presumably, leaves it to be better done by the zealot. Jim is a British botanist who has strayed into the square of a small South American town at a bad moment. Twenty Indian hostages are stood up against the church wall, and an army officer in sweat-stained khaki is about to have them shot. After a brisk interro- gation, he makes Jim an offer. If Jim will shoot one hostage, the other nineteen will be released; if not, the execution will go ahead as planned. What should Jim do? The situation is as corrupt as it sounds, and the scope for Boys' Own Paper heroics nil. Yet, even though the Indians are urging him to oblige, utilitarian calculations are given pause by questions of integrity. We await a refusal like George's. But this is not what happens. Somewhat furtively, in a short sentence slipped in after twenty pages spent drubbing utilitarianism, Jim pulls the trigger. ('But if (as I suppose) the utilitarian is probably right in this case, that is not to be found out by asking the utilitarian questions' UFA p. 117.) In so doing he has stepped back from his first repugnance but, supposedly, not out of his character altogether. The demands of moral character are thus nuanced. Utilitarianism would have George take the job and Jim shoot an Indian. In Williams' eyes it fails to reckon with the projects in a person's life which govern what he can and should do. So, however, does a Kantian corrective, which would require clean hands for George and Jim alike. Yet their integrity is not so special to each of them that there is no puzzle. We are told very little about George and even less about Jim, not nearly enough to make them particular in their place and time, but yet enough to give each a course of action. Broadly, each has a project which set boundaries and tariffs for the trading off of one consideration against another. But, since George cares about his family as well as his scruples, it is not plain why he does not (should not? cannot?) pocket the scruples. Jim's projects are not specified at all, unless his 'desperate recollection of schoolboy The shape of a life 173 fiction' hints at the values which go with cold baths and a stiff upper lip. Perhaps it is simply that his projects, whatever they are, are not of a sort to block a non-utilitarian route to an utilitarian outcome. Meanwhile 'projects' is plural. There is no single universal project for all of us, like obedience to the categorical imperative, and even for each of us there is no single project which determines our character or integrity. What precisely, then, governs the scope and manner of moral choice? That question may seem to conflate 'ought' and 'is'. But I mean it to suit Williams' general line that we must set off from a moral psychology, because what one ought to do cannot be divorced from what motivations one has. The line is put very clearly in his article 'Internal and external reasons', which is a crucial preliminary to much in Ethics and the Limits of Philosophy. It is a subtle endorsement of Hume's claim that reason alone cannot be a motive to any action of will, and it aims to make us see why morality cannot be a set of external reasons for action. It has to be, in the words of the preface to Moral Luck, 'seen as something whose real existence must consist in personal experience and social institutions, not in sets of propositions'. The line is that sentences of the form 'A has a reason to ' can, on the face of it, be given an 'internal' or an 'external' interpretation, but that the latter is a delusion. Taken internally 'the truth of the sentence implies, very roughly, that A has some motive which will be served or further by his <}>-ing' (IER p. 101). Taken externally (as by a Kantian, for instance) a reason for A to can exist regardless of A's present motives (as when A allegedly has a duty to <|>, although nothing inclines him towards it). If only internal reasons can move us, then morality must take a form which recognises that a reason for A to . If their personal experience and social institutions are different enough, their moral reasons can diverge altogether. A's reasons for action depend on what is already in A's 'subjective motivational set'. But the dependence is not straightforward. It does not mean, for example, that A has the strongest reason to do what he believes will best serve his strongest current desire. Even if a ship- wrecked sailor with a raging thirst has some reason to drink sea- water, he may have a better reason not to. Even if he does not know that sea-water will send him mad, the fact that it will is still a reason for him not to drink it. The mere urgency of desire, and the mere 174 MARTIN HOLLIS presence of beliefs, are not enough to make action rational. On the other hand, there is no stepping outside one's existing subjective motivational set. Deliberation can change one's view of what one has a reason to do, but only if one has an existing motivation to deliberate from. It can even be an unconscious motivation, but it must be there. In short, reason can be a highly intelligent slave of the passions, and the 'passions' can be construed as flexibly as moral psychology allows; but in the end reason still has no other office than to serve and obey them. What started as a crisp difference between internal and external is, to my mind, fast losing definition. Revert to George, whom Williams deems to have a sufficient reason to refuse the job. Initially this might have seemed merely because George felt passionately about dirty weapons and less passionately about his wife's concerns or children's welfare. But George can no longer get away with that. He needs to have reflected on his desires, to have made sure they are consistent and, presumably, have endorsed his own current priorities among them. In reflecting, he needs to have brought rational beliefs to bear about, for instance, what the needs and cares of his family are, what effects various choices would have on them, and what the technical and social significance of the research programme is. Otherwise he may be overlooking a greater reason for him to accept the job. The more searching his reflections, and the more they are prone to rearrange his present subjective motivational set, the broader grows the smile on the face of the external reasons theorist. Where exactly is the limit which George cannot cross without losing Williams the game? The crux comes, I think, in a paragraph in IER (p. 105) on how to construe the idea of a 'subjective motivational set'. Such a set is not statically given. The process of deliberation can have all sorts of effects on it. Its elements are not only desires (except in so far as 'desire' is the formal name of any element) since it 'can contain such things as dispositions of evaluation, patterns of emotional reaction, personal loyalties, and various projects, as they may be abstractly called, embodying commitments of the agent'. Nor need these projects be egoistic; 'he will, one hopes, have non-egoistic projects of various kinds, and these equally can provide internal reasons for action'. Thus not only are George's scruples up for assessment and reshaping, but so too are his dispositions of evaluation. He can recondition his patterns of emotional reaction, rearrange his per- The shape of a life 175 sonal loyalties and projects, redirect his commitments. No doubt he cannot do it all at the drop of a hat. But his possibilities are immense. Besides, although one may hope that George will stick to non- egoistic projects, it is not plain why he should. Perhaps he could become a well-adjusted torturer, taking satisfaction in a well-paid, pensionable job among congenial colleagues and a supportive wife. There are plenty of torturers around the globe, who were not born torturers, but who trod the path gradually, until their motivational sets were transmuted. Indeed, thanks to what Hannah Arendt has referred to as the banality of evil, some function as kind husbands and fathers out of office hours without perceived incongruity. It can be done by believing that one is the virtuous agent of church or state or by seeing the victims as unpersons or in other well-attested ways. Torture, moreover, is only a dramatic case of dispositions, loyalties, and projects, which people come by in the enterprise of living, but whose morality one might seek to deny. It is crucial that Williams leaves himself internal leverage, while making motivational sets flexible. Otherwise the point that 'I must deliberate/rom what I am' (ELP p. 200) says nothing restrictive about psychology or its relation to ethics. Hume gets leverage by ascribing the same basic natural sentiments to everyone and showing how it is rational, because useful, for us to cultivate manners which give scope to, for instance, our natural sympathies. 'Mankind are so much the same, in all times and places, that history informs us of nothing new or strange in this particular.' But it takes some proving, especially when this observation from the Enquiries (vin) is immediately preceded by the remark that 'Ambition, avarice, self-love, vanity, friendship, generosity, public spirit. . . have been, from the beginning of the world, and still are, the source of all actions and enterprises, which have ever been observed among mankind.' I shall not pursue the prospect of an ethics constructed on the very varied passions of universal human nature, however, as Williams has no faith in it. It might be worth trying to argue that the torturer, along with those whose projects are marked by ambition, avarice and vanity, cannot be reflectively consistent. Just as we are inclined to say of the man who refuses the medicine which he needs, and consistently and persuasively denies any interest in preserving his health, that 'really at some level he must want to be well' (IER, p. 106), so too the torturer must at some level be at odds with himself. But, granted 176 MARTIN HOLLIS what I said before about people's powers of adjustment, the level required seems Archimedean. It is not as if everyone aspires to a fully integrated life, where the incongruity of torturing some children while loving one's own is perceived and found intolerable. Nor is it plain that an averagely selfish life cannot be, internally speaking, reflectively consistent. But, if there is an answer to these points, it will, I submit, have to be in terms of external reasons. Williams comes closest to giving himself internal leverage with a distinction in ELP between 'thick' and 'thin' concepts. 'Thick' concepts operate close to ground level, so to speak, both descrip- tively and in relation to what guides action. Examples are 'treachery 'promise'' 'brutality 'courage' (ELP p. 129) or 'coward' 'lie' 'brutality' 'gratitude' (ELP p. 140). To call a proposed action cowardly is both to describe it in a way open to empirical dispute and to put up a reason for not doing it. 'We may say, summarily, that such concepts are action-guiding' (ELP p. 140). Guidance varies with historical context, however, for reasons the most striking of which is that they can be 'unseated by reflection' (ELP p. 200). Thus honour concepts guide and give knowledge of life to those who live under an ancien regime, but lose their grip when the assumptions and institutions of an honour code come under the scrutiny of a new reflectiveness. For, as Williams puts it in an arresting phrase which I shall return to, 'reflection can destroy knowledge' (ELP p. 148). 'Thin' concepts, by contrast, are those, like the prescriptive 'ought' dear to moral philosophers, which supposedly reward the reflective search after truth and explain what it is about action, described thickly, that accounts for its Tightness or wrongness in universal terms. Thus what is wrong with breaking a promise is that promises ought to be kept. This is all philosophical illusion, however. If George has promised his daughter a teddy bear, that is why he should give her one. Not only is there nothing more universal to be said, but, besides, reflection does more to destroy George's knowledge of what to do than to underpin it. To save space, think of thick concepts as those of the foreground and middle distance, where George, Jim, and the rest of us find our everyday way about the world, and thin concepts as those of the eternal distance, where philosophers reflect on what is absolutely good or bad, just or unjust, right or wrong. The middle distance is critical. The foreground is where lives are lived in unthinking acceptance of local practices. The loyal Spartan The shape of a life 177 comes home with his shield or on it. The virtuous wife bears her husband's offspring. The thick concepts guide action in ways which raise no questions. But reflection is not ruled out, and, when thick concepts clash, cannot be avoided. Ajax and Achilles are under clear orders to fetch Philoctetes to Troy, but have made him a promise, which burdens them with his trust. Shall they betray their honour? Antigone is both the bereaved sister who must bury her brother, and the loyal subject who must not bury a traitor. The concepts remain thick, but the guidance uncertain. Reflection takes time and brings change - the dilemmas of Ajax and Antigone belong, in a sense, to the fifth century more than to their own. This is the middle distance, where people do their best with what they are given. It is, however, precarious ground. Reflection tends to the general and abstract, thus robbing thick concepts of their power to guide. Reflection can destroy the kind of knowledge needed for a meaningful individual life, which (to anticipate) has to be someone's somewhere, not, so to speak, no one's nowhere. That goes some way towards scotching any presumption that George and Jim are deliberating from a relevantly similar starting- point or sub specie aeternitatis. It also suggests interestingly that the method of Verstehen - the rational reconstruction of action so as to understand how it came about - might well proceed by finding out which thick concepts have motive power in particular people and places. But I fear that it leaves too much opaque, granted that even moderately reflective agents are not slaves to their conceptual context. The thick concepts listed earlier do not guide action precisely. 'Brutality', for instance, is not always a pejorative description and, when it is, does not have a definite price-tag for when one is wondering whether to be brutal to be kind. Thugs and criminals, Goths and conquering heroes are brutal without squeamishness - another thick concept, which, like for example, 'stubborn', 'imaginative', 'cunning' or 'nepotic', has different resonances in different circles. One recalls, too, those dry irregular verbs, like T am firm', 'you are obstinate', 'he is pig-headed', as a sign that thick concepts leave plenty of room for manoeuvre. Apply this looseness to George and Jim and the snag is plain. Was George's choice unworthy, self-indulgent and priggish? Was Jim's choice arrogant or, contrariwise, self-abnegating? Even granted that much of our moral reflection goes into trying to answer such questions, there seems to me to be simply too many of them and too much 178 MARTIN HOLLIS latitude in the thick concepts available. The middle distance threatens to become a swamp instead of an arena where orderly reflection can result in guidance for everyday action. The problem becomes one of bringing order to reflection without being forced into the traditional search after universal truths. The category wanted is that of relative and objective, as revealed by a kind of reflection which does not destroy knowledge. This category is fairly straightforward for the explanation of action. In a society where persons of one race own slaves of another or where persons of one sex are the property of the other, there will be orderly institutions which account for how people act and find their way about. Members have objective reasons, relative to their local institutions, and Verstehen can proceed accordingly. But is ethics to leave it at that? Williams hesitates, yet must, I think, press on. How else can he maintain that George was right to refuse a job which was respectable in itself and allowed him to support his family? How else can he decide that Jim, a mere tourist among alien institutions, was right to intervene? Besides, I doubt if he is content to uphold slavery or the subjection of women for any society where serious reflection on these practices would cause people to lose their way about. In that case we must think further about 'thin' concepts, despite the lack of examples in ELP. If the prescriptive 'ought' is a philosopher's fiction, does the same go for 'just', 'virtuous', and 'rational'? Are 'duty', 'honour', 'tolerance', 'generosity', 'nobility', 'hospitality', and 'kindness', which can all be cast as possible reasons for action, counted as thin or thick? One might say that it depends on context, with 'duty', for instance, being thick in Sparta and thin in Los Angeles. But take the Bible story of how Nathan the prophet convinced Kind David that he had sinned against the Lord in his treatment of Bathsheba and Uriah the Hittite. Nathan told the king a tale of a rich man with many flocks who, to provide hospitality for a visitor, took a poor man's solitary ewe lamb. David's anger was greatly kindled and he said to Nathan, 'As the Lord liveth, the man that hath done this thing shall surely die: and he shall restore the lamb fourfold, because he did this thing, and because he had no pity.' Nathan replied, 'Thou art the man' (2 Samuel 11 and 12). The exchange depends on thick concepts, but it worked because David stepped back from a middle distance, in which he was the king and so morally out of reach, into a wider context where he and The shape of a life 179 the rich man were sinners in the sight of the Lord. Thin concepts, then, can be action-guiding, too, and there is something for reason to stick to through thick and thin. A subjective motivational set, Williams said, 'can contain such things as dispositions of evaluation, patterns of emotional reaction, personal loyalties and various projects, as they are abstractly called, embodying commitments of the agent' (IER p. 105). So mixed an assortment cannot guide action without principles of reflection, which, it seems to me, might very well be governed by a project or commitment to acting virtuously, justly, and rationally. But that would return Williams to the bosom of the traditional moral philosophers. So let us make one further attempt at a category of objective and relative, which allows an ethics beyond the limits of philosophy. ELP ends with an 'optimistic belief in the continuing possibility of a meaningful individual life, one that does not reject society, and indeed shares its perceptions with other people to a considerable depth, but is enough unlike others, in its opacities and disorder as well as in its reasoned intentions, to make it somebody's' (ELP p. 20 if.). The italics belong with Williams' counterpoint between a 'somebody' who is flesh-and-blood and a philosophical 'anybody' which is an abstract nobody. The contrast appeals to me immensely. Think of George as somebody. He is at a fork in his life, one path for his taking the job and the other for his refusing it. Each path forks and forks again in the future, the whole being a set of possible individual lives for George. Some of these lives may be meaningful, some not, and only one from the whole set will be George's own. How should he live? The stress is on 'he'. 'Practical deliberation is in every case first-personal, and the first person is not derivative or naturally replaced by anyone'1 (ELP p. 68). George must deliberate from what he is. But this makes it easier to see how not to live a life than how to try for a meaningful one. Destructively speaking, there are deep Williams-style objections to setting George the problem of picking one from a number of forking paths. They emerge most clearly, if the problem is conceived as utilitarians would, with utilities assigned to each stretch of path, discounted for probability and then summed (or averaged or maxi- mined) at the ends of the paths. Some are to do with uncertainty about whether a decision at a node will lead him to the next node. Perhaps he will not get the promotions and honours which he expects entry to the military research establishment to bring. I am l8o MARTIN HOLLIS not too worried about the mere uncertainties of a long life, in so far as they are of the Cleopatra's nose variety and George can adapt the probabilities as he goes along. But there are also uncertainties arising because other people will be making decisions which affect George. Thus the utility to be assigned to a node marked 'divorce from Jane' is indeterminate without knowing more about Jane and also what comes later. The utility of a remarriage depends on who, if anyone, will have him. This threatens to demolish the whole approach, but let us suppose tolerantly that a satisficing procedure exists to deal with it. More relevant and revealing is the snag that the utilities to be assigned vary with George's personal experience. On some paths he will be made a Fellow of the Royal Society, an honour which, at present, he greatly covets. Would it please him as much in twenty years time? That depends on whether he reaches it by a path which preserves his trust in the institutions and honours of science or one where he becomes jaundiced. In other words, the nodes on the possible paths are not only events and moments of decision, but also stages of his developing subjective motivational set. On some paths he remains a committed scientist, on others he loses his enthusiasm, on yet others he comes to value a different sort of life. In effect, George's range of choice includes the measures of the value of the choices made, thus raising a thoroughly awkward question about the standpoint from which he can consider them. Suppose that his present repugnance for biological and chemical weaponry is wholly genuine, but that, were he to take the job, he would become adjusted and then enthusiastic. If he refuses the job, he may remain glad he refused. But if he accepts it, he will becorne glad he accepted it. Which weights shall he assign in making his present choice? Faced with this sort of query, we can take the idea of'a meaningful individual life' in three ways. One is thoroughly internal - any life which gives George satisfaction is a meaningful life. In that case, however, he may as well take all the easy Options by adjusting his subjective motivational set so as to be content with what he gets. But Williams would reject this as an attitude to 'the only serious enterprise', which is living, and I list it only as a conclusion to be avoided. The next is thoroughly external: For the absolute good is the cause and source of all beauty, just as the Sun is the source of all daylight, and it cannot therefore be spoken or written; yet we speak and write of it in order to start and escort ourselves on the way, The shape of a life 181 and arouse our minds to the vision; like as when one showeth a pilgrim on his way to some shrine that he would visit; for the teaching is only of whither and how to go, the vision itself is the work of him who hath willed to see. This is from Plotinus, a man who seemed ashamed to be in a body, according to Porphyry, his biographer. A meaningful life is a journey guided by a vision of the absolute good and open to anyone who has willed to see. The guidance is too external for most philosophers, modern and post-modern, because it sets the meaning of life beyond life and beyond discursive reason. But, even if we moderns believe that life has no meaning outside itself (or none which rational ethics can reckon with), the passage can still represent the pull of external reasons in its assumption that life has the same meaning for all pilgrims. That is the assumption which Williams tries to dislodge without loss of all meaning and hence of all objective reasons why a life should have one shape rather than another. Thirdly, one can start by rejecting the image of the forking paths. It is a suitable image for parametric decision problems, like planning a strategy in a game of chess. But that is because a player trying to win has a constant measure of utilities by reference to this constant aim. Williams' kind of meaningful life creates the individual whose life it is. George becomes somebody by sharing perceptions with other people linked by social institutions, and yet having his own opacities and disorder as well as his own reasoned intentions. The clear message is that choices which create a somebody are not parametric, and their meaning is not a function of their success in maximizing the antecedent values of variables. Yet what positive moral are George and Jim to draw? Granted that painting one's individual portrait in life is not a matter of capturing an antecedent likeness, what makes for a meaningful portrait? In UFA Williams introduces a notion of integrity, but, as with an existentialist's 'authenticity', it is hard to see how it gets leverage. With the self conceived as Williams does, it is no good saying 'To thine own self be true'. Sincerity does not redeem the wicked. Nor does consistency, which has the further snag that reflection can lead to disruptions of character for the better. Integ- rity, in short, demands only that, having worked out what to do, one goes ahead and does it. In ELP there is more accent on social institutions, with an individual meaningful life glossed as 'one that does not reject society' (p. 201). But, to repeat, some societies are 182 MARTIN HOLLIS viably organized around institutions and relations which one might well ask moral persons to reject. Williams is clear that George should turn down an entry to a set of viable relationships which he could have come to accept. But it is still obscure why, in learning to love the job, he would not be painting himself a meaningful portrait. George's decision, outside the limits of philosophy, remains arbitrary. My final hope rests with Jim, whose deliberations are first- personal, but guided by reflection on the lesser evil. They are first-personal in that, unlike a utilitarian, he distinguishes between his pulling the trigger and anyone's (for instance the officer's) doing so. But he still finds a sufficient reason for him to pull it. That we know so much less about Jim than about George points up the case for its being an external reason, which he makes his own by coming to recognize it. His moral psychology is, therefore, more Kantian than Williams intended, but, as I have argued elsewhere,2 IER yields enough to satisfy a Kantian on this score. In effect, we can follow the story only if Jim's dispositions of evaluation include a disposition to do whatever, on reflection, he believes to be right. His subjective motivation is so cognitive and so guided by thin concepts that, contrary to Hume, reason alone is the motive to the action of his will. There remain questions about the difference between a reason for Jim and a reason for anyone so placed; but they are no longer questions of how the causal connections work. Akrasia in persons who fail to do what they recognize best reason to do no longer shows that we need to explain how people can succeed in acting on recognition of a best reason. The onus has shifted, and there is nothing further to explain. When it comes to moral philosophy, however, Kant himself would certainly deny that Jim had recognized a sufficient reason. No one so placed should shoot an Indian after reading the Grundlegung. So where does that leave a situated ethics? Whereas Plotinus speaks of the absolute good, Kant, like the utilitarians, hoped to show pilgrims their way to the shrine by first convincing them that there is only one shrine worth visiting. There is obvious force in Nietzsche's charge that such claims to objectivity are made in the shadow of a dying God, and fall with the death of God. A situated ethics tries to retain objectivity while accepting the consequence that any meaning in human life is internal and so, it seems, stipulative. This post-modern problem is made easier - or harder - by generalizing to The shape of a life 183 all interpretative activity, and so to science too, thus denying a special difficulty about truth in ethics. But, witness George, that is no help in seeing why someone with a moral dilemma should not stipulate his way out of it on the line of least resistance which he can learn to live with. It is some help to cast the proposition that 'practical deliberation is in every case first-personal' in the first person plural. A robust notion of community, embodying a shared conception of the good and relating self to others through the expectations attached to roles, offers a situation. What we have constructed / alone cannot stipulate aside. There are then reasons for action which are external to each of us, but internal to our collective handiwork. Practical deliberation is objective but open-ended because communal reasons are indefinite, as the duties of a role are often indefinite, for particular occasions. It takes reflection to play an underscripted role, especially when one is engaged with several and it is clear only that they conflict. As with Ajax, Achilles, and Antigone, however, reflection remains objective because the problem is set by their membership of a community whose conceptions and distributions of duties is external to each of them. But this cannot be the last word on objectivity in the middle distance. It leaves the official torturer with a justified course of action, and Jim without one. We therefore have to hold out for a test to sort good or just communities from evil or unjust ones. By all means let the test focus on the social conditions of citizenship with the aid of thick concepts. But its rationale has to be that those are the conditions in which the communal moral discourse is not distorted and can be trusted as a guide to first-personal deliberation. That amounts to taking the middle distance as the place where all mankind is situated, a community, however tenuous, which includes us all. This is within the limits of traditional moral philosophy. It means, I conclude, switching the overhead light of reason back on. Yet there is still Williams' compelling thought that a meaningful individual life must be somebody's. The decision to shoot an Indian has to come out both as right and as Jim's. A situated ethics requires both that another agent could rightly have decided differently, and that Jim's decison was typically the right one. Otherwise the moral dilemma becomes either wholly universal or merely existential, the blood on Jim's hands either agent-neutral after all or wholly agent- relative. Whether an ethics which holds the tension is possible 184 MARTIN HOLLIS depends on how we construe the agent. It might be done by conceiving of agents as players of roles, which define the identity of those who play them, and the integrity of those who play them reflectively and with distance. In that case moral imperatives would never be wholly agent-neutral, but they would not be addressed to a mere personal pronoun. At any rate, if reason is to shed diffused light, it must arouse the minds of socially situated persons. By ending on a hesitant note, I do less to keep my promise to dissent from Bernard Williams' ideas of the only serious enterprise. But I am happy to have learnt more on the way from work so compellingly within limits which philosophy must embrace. There is a fine line between situating external reasons and nuancing internal ones. Yet it exists, and the attempt to argue that George and Jim both acted rightly depends on finishing one firm and full pace on the traditional side of it.

NOTES 1 J. J. C. Smart and Bernard Williams, Cambridge University Press, 1973, here abbreviated as UFA. Other abbreviated references will be to 'Persons, character and morality (PCM), and 'Internal and external reasons' (IER), both in Moral Luck, Cambridge University Press, 1981, and to Ethics and the Limits of Philosophy (ELP), Fontana Books, 1985. 2 M. Hollis, The Cunning of Reason, Cambridge University Press, 1988, ch. 5. ('External and internal reasons'). For a further word about Jim, see my 'Jim and the Indians', Analysis, 43 (1983). CHAPTER 10 Replies Bernard Williams

In thanking my colleagues who have been kind enough to write essays for this volume, I have to offer them, first, a substantial apology. It was always the design of this book that I should contribute a 'reply', and I welcomed that opportunity. Some volumes in the Festschrift format display a rather ghostly quality from the absence of their honorand: the person in whose name the guests have come is not at the party, and that must either be an insult, or be glumly explained by the party's being a wake. However, my enthusiasm at this idea was not matched by speed of delivery, and I have, by dilatoriness, condemned my friends to waiting a frustrat- ingly long time to see in print what they amiably wrote in my honour and for my comment. The explanation does not lie in indifference to their thoughts about my work — rather, the reverse. I can only say that I am sorry, and that I hope that they will not feel that their gift to me has been spoiled. As the editors have said in their Introduction, it was not the plan that I should comment on all the contributions one after the other or in equal detail. I have taken up papers, or parts of papers, on which I thought I had most to say, whether in explaining or correcting what I originally wrote, or in developing some suggestion. I hope that in doing this I have avoided the two major disasters that can affect this kind of reply, of becoming a free-floating statement of position, attached only by a cord of gratitude to the contributions, or alternatively handing out marks to the contributors and correcting what one takes to be their misunderstandings. I have grouped my remarks under various topics, which in some cases almost correspond to the discussion of one particular paper. At the end I have added some comments, less directly tied to the papers, about problems of style in philosophy, particularly in moral philosophy, that have concerned me at least since the late 1960s. 185 l86 BERNARD WILLIAMS I was taught philosophy, and first practised it, in Oxford between 1947 and 1959, in the hey-day of so-called 'Oxford' or 'linguistic' philosophy.1 One thing that I got from that experience was a stronger trust in particular philosophical inquiries than in global positions. My philosophical mentor, Gilbert Ryle, when someone in his class suggested that we should discuss logical positivism, characteristically said 'Not interested in isms: we can talk about some things positivists say.' His attitude had its limitations, indicated by a certain strain in that remark itself, but it was basically helpful and liberating. He would tell a story about a young man who approached him after he had given a lecture in Germany, and said 'Professor Ryle, I was very impressed by your lecture and would like to join your school. Unfortunately, I am a Kantian.' There has now been a great revival of philosophical isms, particularly in the philosophy of mind and in ethics, and it is at least respectable to have an overall philosophical outlook that can attract such a label. I believe, and still very much hope, that I do not have one. It is a reasonable demand that what one believes in one area of philosophy should make sense in terms of what one believes elsewhere. One's philosophical beliefs, or approaches, or arguments should hang together (like conspirators, perhaps), but this demand falls a long way short of the unity promised by a philosophical system. It is true that one can misread perception by neglecting wider implications, just as one can distort perception by letting system dominate it, but the characteristic fault of perception, at least in moral philosophy, is not that it is insufficiently linked to system, but that it is insufficiently perceptive and does not go deep enough. My concerns in philosophy do, I hope, hang together, but this is the most in that dimension that they are supposed to do.

INTERNAL AND EXTERNAL REASONS The first matter I should like to take up is my view about internal and external reasons,2 and John McDowell's criticism of it. McDowell has set out the main argument, and I shall not repeat it in detail. The central idea is that if B can say truly of A that A has a reason to , then (leaving aside the qualifications needed because it may not be his strongest reason) there must be a sound deliberative route to c(>-ing which starts from A's existing motivations. It follows that what an agent has a reason to do will be a function of what I Replies 187 called his 'S' - that is to say, the existing set of his motivational states - or, alternatively, <)>-ing will have to be an action that an agent could rationally decide to do as a result of deliberation whatever his S might be. The latter alternative McDowell agrees with me in finding unattractive. The former alternative represents what I called 'internalism' about practical reason, which, equally, McDowell does not want to accept. His aim is to leave room for an intelligible account of 'external' reasons for action: that is to say, to give a sense to 'A has a reason to (}>' that does not necessarily ground (}>-ing in A's existing S.3 McDowell's strategy is to question an assumption that he finds lying behind the two alternatives (and which, he remarks, is also made by some writers who have criticized the argument). The assumption, he claims, is that the external reasons theorist must take a certain view of the transition from the state in which A is not motivated by the alleged reason to a state in which he is motivated by it. The assumption is said to be that an external reasons theorist must see this transition as being effected by correct deliberation. It is this assumption, McDowell argues, that permits me to force a choice between two alternatives: between a deliberation that is a function of A's existing S, and a deliberation that will arrive at the desired conclusion whatever A's S may be. But, if the external reasons theorist need not assume that A would arrive at the right conclusion by deliberation, then I cannot force this choice. Now McDowell does agree that there is a sense in which to believe an external reason statement is ... to believe that if the agent deliberated correctly he would be motivated (of course not necessarily conclusively) in the direction in which the reason points. But there is no implication, as in Williams's argument, that there must be a deliberative or rational procedure that would lead anyone from not being so motivated to being so motivated. On the contrary, the transition to being so motivated is by transition to deliberating correctly, not one effected by deliberating correctly; effecting the transition may need some non-rational alteration like conversion [p. 78, his emphases]. I do not think that my argument presupposes quite what McDowell says that it presupposes, and, before moving to some more basic issues, I shall try to say where, as it seems to me, a misunderstanding has arisen. Consider the statements (R) A has a reason to . (D) If A deliberated correctly, he would be motivated to -ing. Granted this, the internalist can give a constrained sense to (D), one which - he claims - matches it to (R). But this does not imply, as McDowell's objection suggests, that the agent should be able to conduct the relevant deliberation in fact. Perhaps some unconscious obstacle, for instance, would have to be removed before he could arrive at the motivation to <|>. Someone who claimed something of this sort would have to make good on the claim; he would have to be able to make some distinction between, for instance, removing blocks to the expression of A's existing S, and adding to A's S. This may not be easy to do, but the fact that there is an onus to be discharged here just marks the point, on the internalist's view, that Replies 189 (R) has a content which distinguishes it from other things that might be said about A. How does it stand with the externalist's view? He agrees that there is a close relation between (R) and (D), and, since he does not want the internalist's constraints on (R), he does not want them on (D), either. Then it seems that (D) - the claim that if A deliberated correctly, he would be motivated to <() - means only this: (C) if A were a correct deliberator, A would be motivated in these circumstances to , where 'a correct deliberator' means someone who deliberates as a well-informed and well-disposed person would deliberate: and, in McDowell's account, this seems to be someone like Aristotle's phronimos, or, as McDowell puts it, someone who has been properly brought up. If we take (C) in this way, we face questions about such Aris- totelian conceptions, and how stable, objective, and so on they are; I shall come back to some of those questions, particularly in relation to Martha Nussbaum's paper. But it is important that this is not the only question we face when we take (C) in this way; this is one reason why the issue of external reasons, though it is closely connected with other matters, remains a special question. Let us grant for the moment that it is known, more or less, what the phronimos would be motivated to do in certain circumstances, and that it is a good thing, all round, to be a phronimos. We agree, that is to say, that we can know things of the form (G) A correct deliberator (i.e., a. phronimos) would be motivated in these circumstances to . In virtue of what the external reasons theorist has given us so far, (C) gets its content from (G). (C) merely says (G) with reference to A, where 'with reference to' means only that (G) is applied to A - A is mentioned, and is the occasion of a reiteration of (G). (G) can be applied in this way, it seems, to anybody. Given (G), the sentence form If x were a correct deliberator, x would be motivated in these circumstances to c|> will yield a truth under any substitution for x, granted only that the substitution in the antecedent yields an intelligible possibility (a point I shall come back to). It follows that on this account (R) does not make a statement distinctively about A at all. The reference to A I9O BERNARD WILLIAMS may, in a familiar way, serve to pick out some circumstances that form part of the reason and would be specified if the reason were fully spelled out, but that is the most that it can do. On the externalist account, so construed, statements of the type (R) do not relate actions to persons, but types of action to types of circumstances, and they are most revealingly expressed in the form 'in circumstances X, there is reason to (f>\ This last formula does not have to be taken in a formalistic or rule-based spirit. The formula is perhaps most familiarly invoked in connection with Kantian or prescriptivist moral theories, but it is not essentially tied to such theories. On the contrary, the externalist account I am considering gives an explanation of such formulae not in terms of rules or principles, but in terms of what a. phronimos would do. But the phronimos is an ideal type, and the fact that he is invoked does not make the formula any less impersonal, relative to particular agents. On this account, (R) invokes the person A, but none of its content is distinctively about A. This immediately raises a problem, which is as much ethical as analytical. If (R) gets its content from (C), and (C) gets its content from (G), then what A has reason to do in certain circumstances is what the phronimos would have reason to do in those circumstances. But, in considering what he has reason to do, one thing that A should take into account, if he is grown up and has some sense, are the ways in which he relevantly fails to be a phronimos. Aristotle's phronimos (to stay with that model) was, for instance, supposed to display temperance, a moderate equilibrium of the passions which did not even require the emergency semi-virtue of self-control.5 But, if I know that I fall short of temperance and am unreliable with respect even to some kinds of self-control, I shall have good reason not to do some things that a temperate person could properly and safely do. The homiletic tradition, not only within Christianity, is full of sensible warnings against moral weight-lifting. It will be no good trying to accommodate this difficulty, of squeezing the good into the right through the tubes of imperfection, by putting all A's limitations into the account of the circumstances. If the circumstances are defined partly in terms of the agent's ethical imperfection, then the phronimos cannot be in those circumstances, and (G) cannot apply at all. Rather, we shall have to modify the interpretation of (C), away from (G) and in the direction of accom- modating it more closely to the actual nature of A. The question Replies 191 then becomes, how we can do that without ending up with internalism. It is not clear to me what exactly McDowell's view is on the question that I have expressed in terms of the relations between (R), (C), and (G). On the one hand, what he says seems not to offer any content for external reason statements except considerations about what thephronimos (the person who has been 'properly brought up') would do. On the other hand, he introduces, towards the end of his paper, the example of what one might say to 'someone who cannot find anything to appreciate in, say, twelve-tone music'. I doubt that 'someone' here means 'anyone'. The fact that listening to twelve- tone music can be a worthwhile activity surely does not give everyone a reason to engage in it. If not, then 'A has a reason to try twelve-tone music' must have a content that goes beyond 'listening to this music is a worthwhile activity' - a content, one might suppose, that has something to do with A's S, such as his present taste in music. However, it may be that this example, on McDowell's view, brings in further considerations: he may not take it to be unqualifiedly part of what it is to be 'properly brought up' that one has a taste for music, or at least for twelve-tone music. I insisted in my writings about this question that no account of 'A has reason to <(>' can be adequate unless it has normative force, and I tried to explain how the internalist account (with the indeter- minacies that any realistic account demands) meets this requirement. McDowell acknowledges that this is my aim, but he claims that I have not left enough, or the right-shaped, space for normati- vity; that my account is too 'psychologistic'. I accept that the account is psychologistic, in the sense that on my view a statement about A's reasons is partly a statement about A's psychology. I do not see this as an objection, as it is (I agree) an objection to say that a theory of arithmetic is psychologistic. McDowell may think that I have drawn the line in the wrong place, and have relied on too 'procedural' a conception of the relation between the hypothetical deliberation and A's S. But some constraints on the idea of'deliberating correctly', as it appears in (C) above, are needed if statements about people's reasons, on the model of (R), are to say anything distinctive. I said earlier that it was a problem, so far, with an externalist account that (R) does not emerge as a statement distinctively about the person A. It is also, and relatedly, true that, if (R) does say anything about A, this externalist account does not suffi- 192 BERNARD WILLIAMS ciently distinguish what (R) says about him from other things that might be said about him. I agree with McDowell that it was a mistake to pick out 'irrational' as a crucial term in this connection. But I want still to press the point that I made originally, that from both an ethical and a psychological point of view it is important that (R) and its relatives should say something special about A, and not merely invoke in connection with him some general normative judgment. An externalist account that simply bases (R) on (G) cannot satisfy that condition. The difficulty we have just encountered, about temperance and the relations between the good and the right, represents just one aspect of a wider question: granted (G), what does that mean for a particular person A? Here it is very important to distinguish two different positions. One is the externalist position I have been discussing, which bases (R) simply on (G). This treats the person A merely as a point of reference or occasion for the general judgment (G). The other position also bases (R) on (G), but in association with an internalist account of (R). This invokes a substantive account of a human S, and claims that the best life, relative to such an S, is that of the phronimos. This latter is the position which, in ELP, I ascribed, with some qualifications, to Aristotle. It may be that the qualifications I made there were not strong enough.6 It was not Aris- totle's view that a human being, as such, is capable of the good life. He thought that women and natural slaves lacked distinctive capacities which enable someone to become a phronimos, and, further, that even a male who was not a natural slave, but who had not been properly brought up, could not become a phronimos: earlier it would have been possible for him to become one, but later he could not do so. The fact that these various things are not possible, or no longer possible, does not affect judgments of the good. The good life is what it is, the life of the phronimos, and the fact that it is inaccessible to women and to (natural) slaves and to the now irreversibly corrupted is, roughly speaking,7 their bad luck. But the incapacities from which these people suffer do affect what they have reason to try to do. In terms of Aristotle's own outlook, these people have no reason to try to be like a phronimos, to the extent that such a life lies beyond their competence, their understanding, and their possible motivations.8 Aristotle's own way of focusing considerations of the good life on to the individual was, if I understand him, internalist. Replies i93 However, there might be a different way of focusing on to the individual agent general Aristotelian considerations about the good life, which did not adopt the relativity to that agent's S which is the mark of the internalist view. This method would lie in taking very seriously a question that I mentioned earlier, whether the supposition of the agent's being a correct deliberator (that is to say, a phronimos) represents a real possibility. If we could make something of this, we might be able to say that (G), by itself, does not support (R); rather, it supports (R) only in conjunction with the premise that there is a real possibility that A should be a phronimos. Such an account would represent (R) as being, in a more substantive sense, a statement about A than it is if it is simply an instantiation of (G), but (at least at first sight) it avoids the path of internalism. Externalists who feel, as they should, a need for some constraints on the route from the good to the right, from (G) to (R), may want to explore this area. They will have to start from the consideration that the real possibility they need must take the form of'A could be a phronimos' and not merely 'A could have been a phronimos'; if the latter were enough for this purpose, we would be left with the sad mockery that a person who has been born defective as a result of in utero damage has reason to act as the phronimos acts. In trying to locate the kinds of people for whom it is a real possibility that they should be a phronimos, such theorists will have in mind the sobering reflection that Aristotle himself on this matter was in effect an actualist. For him, the only grown-up people who can (now) be phronimoi are actually phronimoi, and this leaves the class of those who have reason to act in the approved ways smaller than it may be for a non-Aristotelian internalist. To avoid this conclusion, neo-Aristote- lians who take this path will presumably want to reject Aristotle's own harsh opinions about what is possible for one who has been brought up badly. But can they run far with this without finding themselves, with the Kantians, in an area which they strongly and very reasonably wanted to avoid - the deep swamp of questions about free will and determination by character? Moreover, if one considers what might be involved in the kind of possibility that is at issue, it is reasonable to expect that it will revolve round the potential expressions or developments of the agent's S - that is to say, the materials of internalism. The more one considers the problems that lie in this direction, the more discouraging they seem. There are no doubt many undiscussed considerations to be pursued 194 BERNARD WILLIAMS here. I confess, though, that nothing yet has persuaded me to give up the opinion that internalism in some form is the only view that plausibly represents a statement about A's reasons as a distinctive kind of statement about, distinctively, A.

ARISTOTLE AND HUMAN NATURE Martha Nussbaum's paper discusses questions about the Aris- totelian outlook that arise at a stage earlier than this. She asks whether we can form a picture of a desirable and worthwhile life on the basis of an understanding of human nature. The answer, as she makes very clear, will affect the politics of a society that might be hoped to secure or advance such a life for its citizens. She has stated boldly, here as elsewhere, how considerations of this sort might be freed from restrictive assumptions that Aristotle himself attached to them, and thus prove more congenial to a modern politics. She has, at the same time, helpfully connected these lines of thought with other ancient Greek conceptions of the boundaries of humanity, as revealed, for instance, in the content of tragedy, and also - as she suggests in what I find a particularly illuminating discussion - in experiences that we may suppose to have been involved in watching it.9 Any attempt to recover from Aristotle's texts ethical or political ideas that are to be helpful in modern conditions immediately raises sensitive problems of the relation between historical interpretation and the demands of modern relevance. Clearly Aristotle's conceptions of nature and of the investigation of nature were importantly different from our own. Nussbaum reminds us that in inquiring into human nature Aristotle thought that he could rely on endoxa, well- established opinions shared by responsible and experienced thinkers. But Aristotle thought that he could rely on endoxa in all areas of natural inquiry, and the presuppositions of that method have not survived the emergence of what we call the 'natural' as opposed to the human sciences. It is certainly an historical mistake to take Aristotle's conception of nature as that implied by our concept of a natural science, and an even greater mistake to take phusis (the term that is translated as 'nature') to stand just for that of which the science we call 'physics' gives an account. It may seem that we must conclude that, since the view of nature which under- pins an Aristotelian ethics can no longer be ours, we must give up Replies 195 Aristotle's approach to ethics. But this is not Nussbaum's position. She wants to sustain, in a modified form, Aristotelian ethics on an Aristotelian basis, but without reinstating Aristotelian physics or cosmology. Her aim is to achieve an understanding of human nature on Aristotelian principles, an understanding which will be ethically significant. This description of the task of course leaves several things undetermined. It leaves it open to what extent the proposed understanding of human nature will need to incorporate elements of Aristotle's understanding of nature more generally; relatedly, it leaves unclear the implied boundaries between human nature and the rest of nature. What about human biology? Presumably the neo-Aristotelian enterprise cannot require us to feign amnesia about natural selection. But, if we are to accept modern evolutionary theory and also adopt an Aristotelian programme for the ethical understanding of human nature, we shall have to assume that evolutionary theory is irrelevant to such an ethical understanding. You do not need to be an enthusiast for sociobiology to wonder whether that can be altogether wise. The answer to this that I find in Nussbaum's essay is that, if one approaches the business of understanding human nature in a sufficiently hermeneutical and non- reductive way, the materials one needs to nourish ethical understanding will be obvious, and not of a kind that could be undercut by more specific or more technical inquiries. I am sure that the neo- Aristotelian answer must be on these lines, but I think that it raises a large problem, to which I shall return. Nussbaum tends to ascribe to me a more positivist account of the Aristotelian enterprise than I wanted to give. I aimed to get away from positivist assumptions, trying, for instance, to rid the discussion of the unhelpful influence of the supposed fact/value distinction. I expressed what I mainly wanted to say about Aristotle's ethics and his account of human nature in saying For Aristotle, the virtuous agent. . . could come to understand that the dispositions that gave him his ethical view of the world were a correct or full development of human potentiality. This was so absolutely, in the sense at least (Aristotle no doubt meant more) that the best possible theory of human nature and its place in the world would yield this result. (ELP, P- 52) It may be that Nussbaum would want to strike from this the suggestion that Aristotle meant more, but I do not think that otherwise she needs to disagree with it as an account of either 196 BERNARD WILLIAMS Aristotle's or her own enterprise. It seems to do better as such an account than her own suggestion that we should understand as evaluative the claim that a certain creature is a human being. The term 'evaluative', in such an employment, relies on the positivist contrasts that she and I agree in wanting to abandon or soften. Moreover, I find that I disagree with some of the substantive points that she makes under this title. She expresses some reserve about a claim of mine that 'human being' is a more robust classificatory term for ethical purposes (in matters involving termination of life, for instance) than 'person'. I still, and firmly, believe that it is. This point in itself does not perhaps shed all that much light on the ethical powers of the concept 'human being', since the concept with which it is being contrasted, 'person', is so weak in these connections. The way in which 'person' is used by some ethical theorists is near to a deceit, for at least two reasons; that it looks like a natural kind term when it is not, and that the criteria it imports are vague, relative to circumstance, and (partly as a consequence of that) very susceptible to being manipulated by powers that have an interest in the outcome.10 The concept human being does not have these weak- nesses to anything like the same extent. It is also a concept that very large numbers of human beings themselves use and have used, as Nussbaum's examples illustrate, to articulate their ethical reactions, and, even where it fails to deliver an answer, notably in the case of abortion, it is helpfully clear why this is so: concentration on the point can help us to see what assumptions are being made by various people who are willing or unwilling to kill a foetus. The powers of the concept in these respects, seem to me, as they do to David Wiggins, whom Nussbaum quotes, as rooted in its role in a classification of nature. Nussbaum says, further, that the question whether something is the same human being can be an 'evaluative' question. She wants to bring together the questions whether a certain life could be a life of mine, and whether I could possibly choose such a life. She suggests this, in particular, in relation to the argument that she cites from the Philebus:11 what told Protarchus that he would not survive in that life was just the fact that he would not consider that life worth accepting; it would not contain activities that he values so highly that he cannot regard himself as himself without them. Replies 197 The same idea is applied to Nozick's discussion of the Experience Machine and one's reluctance to be linked up to it. My reluctance in the latter case certainly involves the thought that life linked up to the machine would be no life for me, but, for that very reason, it seems not to involve the thought that the person who would be joined to the machine as a result of my choice would not be me. If, from my perspective in contemplating the prospect, the person who might be joined to the machine would not be me, then my reluctance to be joined to it would simply be the reluctance to be destroyed. But that cannot be the correct account of my reluctance. After all, I might prefer suicide to existence on the machine, and that choice would be rationally coherent. Someone might suggest that in such cases the proposed 'life' was not a human life at all, and hence that nothing existing in that state could be me, since I am a human being, that is to say, one who leads a human life. That argument would certainly provide a quick way with many problems, both personal and public. The irreversibly comatose creature in the hospital could not be anyone's father, and people who treated it as someone's father (for instance, their own) would be suffering from a misunderstanding, as would a person who feared (as she might naturally put it) ending up like that. This can hardly be a very compelling argument. But, equally, I do not think that it is Nussbaum's. In speaking of the 'evaluative' here, I think that she wants to avoid the idea of constraining arguments, or assertions to the effect that some attitude would, unconditionally, be a misunderstanding. It is rather that an attitude to certain activities, that they are worth pursuing or worthily part of my life, carries with it an attitude to the possession of certain lives or states of being: that because they do not contain or honour those activities, I could not at a level that really matters count them as mine. But, if this is the outlook in question, then it carries another and more revealing difficulty, against which Aristotle himself provides a warning. Whether I have a worthwhile life or not is not just up to me, but is in part a matter of luck; happiness is subject to contingency. But then it cannot be the case that my identity within a way of life - the question whether it is my life - can hang on my declaration or resolution that this way of life does not come up to my expectations, or does not sufficiently honour activities by which I set great evaluative store. Radical bad luck consists in being con- 198 BERNARD WILLIAMS strained to live a life that fails in such ways. I do not deny, and it is entirely obvious, that the question whether a given life would be mine is 'evaluative' in just the sense that the question is a focusfo r what I value, for what I hope and fear. But there is another sense in which it must be denied that this question is evaluative: its answer cannot itself be determined by what I value in life. Many of our values themselves, above all a sense of what would count as a loss, demand a realistic recognition that a certain life is inescapably one's own, that it cannot be abandoned except by dying.12 In this sense, I cannot believe that Nussbaum, who has devoted a book to themes of contingency in the good life, wants to say that the question is evaluative. How are these questions related to the Aristotelian issue to which Nussbaum devotes most of her paper, the issue of whether the criteria of a good or bad life are given, to some substantive degree, by considerations of human nature and of what it is to be a human being? It is clear why Nussbaum, in stressing the evaluative significance of concepts of humanity, should hope to make the Aris- totelian project seem more recognizable than perhaps it does if one thinks of 'human being' as a modern zoological or palaeontological category, referring to members of a certain hominid species. However, the concept 'human being' had better retain some classificatory and explanatory force, or the project of using it in identifying a worthwhile life will become vacuous. At the same time, the project has to avoid the well-known difficulty that, if it gives any answer at all, it is just as likely to give the wrong one.13 I am happy to accept Nussbaum's argument that — at least with regard to the relations of reason to other human functions - Aris- totle's form of expression 'an E-ikezoe' should be taken to mean a life characterized or structured by E-ing, rather than a life, or aspect of life, that consists of E-ing. As Nussbaum points out, however, the life of a wicked or self-indulgent person is equally a certain kind of life structured by reason; it is also a distinctive kind of human life. So far we still wait for the considerations that may move the idea of a life 'structured by reason' in the specific direction of a life of moderation and virtue. To get us there, if I understand her, Nussbaum uses two Aristotelian ideas. One puts firmly into the ethical equation the condition that this zoon should be politikon - the requirement of human sociability. If we are to honour the special requirements of a human life, the argument goes, we must look not just for any Replies 199 structuring use of reason, but for one that could itself form the basis of social life; the use of reason in the life of the wicked or self- indulgent person is not an example of this. This line of argument is not as simple as it may look. It assumes answers to many well-known questions, some of them modern, but some of them older than Aristotle, such as those raised in Books 1 and 11 of the Republic. Glaucon and Adeimantus, agreeing that human beings are essentially or typically rational and that they essentially or typically live in society, could still deny that human reason is displayed at its most effective in living according to the restrictive requirements of society. The Aristotelian approach takes for granted, against this, a strong view of the harmony among themselves of human capacities and needs. This assumption does, I must confess, seem to me more plausible if you can help yourself to Aristotelian cosmology, than if you regard it as an open question whether the evolutionary success of humanity, in its extremely brief period of existence, may not rest on a rather ill-assorted bricolage of powers and instincts. This is one reason why I still think, as opposed to the emphasis of Nussbaum and others, that it makes a big difference to accepting Aristotle's ethical outlook that we cannot accept his cosmology. The second idea that Nussbaum uses in mobilizing the Aris- totelian approach is one that appears also in McDowell's argument: that we should use, in putting together our picture of human life and human capacities, our own best ethical understanding, drawn from the perspective of the virtues themselves. This helps with a problem that remains even if one grants the previous step, of incorporating sociability into the picture of human flourishing: the problem that, even granted an understanding of human reason that will feed social life, an Aristotelian understanding of reason is compatible with many different forms of social life. In fact Aristotle's ideas have, more often than not, been used to justify forms of life that Nussbaum herself finds ethically uncongenial. Without a theory of error to explain what was wrong with the views of humanity taken by Aristotle himself,14 or by various medieval, or modern conservative, followers of his, the project would seem to be adrift. Nussbaum, and I think McDowell, will reply that it fails on this score only if it is understood as a top-down enterprise, moving from an account of humanity which is independently understood, to an ethical form of life; but that it should not be understood in this way. Rather, the Aristotelian approach takes our best understanding of 2OO BERNARD WILLIAMS what human beings are, and of what the virtues are, and puts them together. If our ethical understanding has moved to a point at which liberal ideals of individual autonomy are very important to us, as they are to Nussbaum, then this can be harmonized with a broadly Aristotelian understanding of human beings, even though, in relation to Aristotle's own outlook, taken as a whole, some of those ideals would have been unintelligible and some (with regard to women, for instance) ridiculous. At this point, however, there is a danger that the pendulum will swing, in the familiar way, from the unacceptable to the vacuous. When one says that our ethical outlook can be 'harmonised' with an Aristotelian outlook, how strong a relation is being introduced? Does the shade of Aristotle demand only that one's ethical demands and aspirations should be consistent with what one takes human beings to be, and, in particular, that they should allow for the fact that collectively and individually human beings take thought about what to do? This, obviously enough, determines far too little. Every social aspiration takes itself to pass this test; even Utopian planners offer an account of what they take human beings to be, and of what they believe human beings might become. Every actual social system inevitably passes the test, in the sense that human beings must have whatever capacities it takes for that system to exist. History has not been so sparing to man, that it made him barely a rational and social creature, and left it to Aristotle to work out how rationality might be expressed in society. If the Aristotelian is to avoid this looming vacuity, then she must surely provide some constraints on what can count as the best understanding of human powers, and those constraints must be, to some degree, independent of the distinctively ethical demands and hopes. It cannot be enough for her to say simply that the traditional Aristotelian understanding of what women are like, for instance, should be rejected because it does not fit with liberal aspirations; what is needed is some account of what, apart from that, is wrong with such views. This is the sense in which I said earlier that Nussbaum could not let it stand that accounts of human nature were 'evaluative'. I do not now want to claim that the Aristotelian enterprise requires a 'top-down' derivation of ethical conclusions from a scientifically respectable account of human beings. In some things I have written I think I have suggested that this would be necessary, Replies 201 and this needs correction. I grant that the enterprise may be understood in coherentist or hermeneutical terms. I also grant that we can understand Aristotle himself as having seen it in some such terms. However, I do not think that we should infer from those two points that, because we can see it in such terms, and so did Aristotle, we can see it just as Aristotle saw it. This is because our understanding of nature and of what is 'natural' does not bear the same relation to (what we call) hermeneutical or social understandings as Aristotle did, and that is why, even granted these points, I still think that, in leaving behind Aristotle's cosmology, the modern world has left behind elements necessary to making his style of ethical theory as a whole plausible, however many useful thoughts we can, quite certainly, gain from it. Nussbaum refers to a 'tension' in my attitude to Aristotelian ethics, and I think she is right. On the one hand, it seems to me to fail, for the kind of reasons that I tried to express in Morality and in ELP, and (in a way that I hope does something to meet my critics) here. On the other hand, it still seems to me, as one might put it, the only colourable attempt to provide a foundation for ethics. I choose that adjective in order to register a couple of distinctions. I do not mean that it is the only intelligible such attempt. I think that Kant's is intelligible, except to the extent that it is not - that is to say, to the extent that Kant himself admitted that, resting on the theory of noumenal freedom, it was not. Nor do I mean that Aristotle's attempt is uniquely plausible, in the sense that there is a lot, but not quite enough, to be said for it. On the contrary, I am inclined to think that in its original form it simply could not, now, be acceptable.15 The point is rather that it uses, in its attempt, the kind of material that one needs to consider in arriving at any sensible view of the status of ethics, namely the richest account available of human powers and social arrangements. It is conceivable that, as Nussbaum hopes, some attempt of that sort might succeed, although Aristotle's own attempt failed. I allowed for that possibility in ELP. But if, on the contrary, we come to the conclusion that all such attempts are likely to fail, then, just because the attempt could be taken seriously, we have come to understand something significant. If the possibility of such a 'foundation' for ethics could be immediately dismissed a priori on the strength of some such consideration as the fact/value distinction, then the lack of any such foundation would not be 2O2 BERNARD WILLIAMS interesting — the whole idea would rest on a simple misunderstanding. But, if reflection on Aristotle can give us both an idea of what such a foundation might have been like, and of why it is not available, we have a better sense of what we have not got, and of why we might have wanted it.

NATURALISM AND MORALITY So far in these remarks, as in ELP, I have laid most stress on the differences between Aristotle's world and ours, differences that discourage the attempt to carry his outlook into modernity. But I am also impressed, particularly since working on Shame and Necessity, with the thought that Aristotle's hopes were special even in anti- quity. There were conceptions of human nature and of society available to the ancient Greeks themselves that give less comfort to the prospects of human fulfilment in virtue than Aristotle's do. In relation to these, Aristotle's psychology, despite its richness and elaboration, can seem ethically superficial. These other conceptions imply a psychology that is less moralized, less adapted already to the demands of the ethical, than the theories of the mind offered by Plato or Aristotle. It is very important that the idea of a less moralized psychology is not that of a 'value-free' psychology, in the sense favoured by behaviourists and others who have tried to bring the social sciences as close as possible to natural science. One style of moral psychology that particularly cultivates the scientistic image, that of utilitarianism, is in fact adapted to ethical purposes, those of an economic and administrative morality. A non-moralized, or less moralized, psychology uses the categories of meaning, reasons, and value, but leaves it open, or even problematical, in what way moral reasons and ethical values fit with other motives and desires, how far they express those other motives, and how far they are in conflict with them. Thu- cydides and (I believe) the tragedians, among ancient writers, had such a psychology, and so, in the modern world, did Freud. As Charles Taylor remarks, there is a Nietzschean interest here, which relates both to the interpretation of the ancient world and to questions about the psychological character of the ethical. Among my critics in this book, Taylor's position is perhaps closest to my own, at least on the matter he discusses in his paper (though manifestly his description of it as a 'footnote' to a chapter of mine is a Replies 203 friendly understatement). We agree on many points about what I have called 'the morality system', for instance on its damaging isolation both from other ethical ideas and from the rest of life more generally. I particularly agree with him on a point that he makes here, and has developed in his book Sources of the Self, that it is typical of modern moral theories that they lack the resources to display their own ethical appeal. Utilitarian theory appeals implicitly, and sometimes explicitly, to a sort of self-denying generosity, or, again, to a scientific spirit of objective rationality, the ethical power of which it itself is not very well placed to capture. Again, the 'dignity', for Kant, of the moral law and of the moral agent who wills that law, is a value that cannot be captured in the terms provided merely by that law itself. Taylor himself would like to mobilize these richer ethical resources in terms of conceptions of the good, which are partly to be understood in historical and psychological terms - kinds of understanding (again we agree) that moral philosophy needs to use if it is to have any hope of understanding itself. In Taylor's own work, the idea of what it is 'good to love' is at least as important as ideas of what it is good to do or to be. It is in terms of what it is good to love that he has formulated notions of the power of the good, notions which carry Platonic or Christian resonances. Some time before this point is reached, Taylor will rightly have expected me to part from him, suspicious of the 'siren songs of old metaphysical bird-catchers', in Nietzsche's words, calling 'you are more, you are higher, you are of a different origin!' In these connections his views, for me, are too removed from naturalism.16 In referring to 'naturalism', I do not have in mind the concerns of the so-called 'naturalistic fallacy', but the conceptions underlying a naturalistic psychology of ethics. I do not pretend that this notion is at all perspicuous. At any general level, it is not at all clear what is involved in a conception of ethics that will be, to an acceptable degree, continuous with our understanding of human beings in other respects. On the one hand, as I have already said in discussing Nussbaum's paper, we should not be looking toward a reduction of the ethical consciousness in terms of the natural sciences. On the other hand, we cannot simply say that everyone's account of ethics, from Aristotle to Prichard, is equally naturalistic since each finds in nature what he needs for ethics. In the face of such difficulties, I in fact rather doubt whether one can give general conditions for naturalism. But it is possible, at least, 204 BERNARD WILLIAMS to adopt an appropriately suspicious rule of method: never explain the ethical in terms of something special to ethics if you can explain it in terms that apply to the non-ethical as well. As the suggestion that this is a 'suspicious' rule implies, it is not motivated so much by general canons of scientific economy, as by a sense that there is good reason for suspicion: that sophisticated and reflective observers have always had good reason to think that stories human beings tell themselves about the ethical tend to be optimistic, self-serving, superstitious, vengeful, or otherwise not what they seem to be.17 It is important, once again, to recognize that what is at issue here is not a mechanistic reductionism, an attempt to represent ethical thought in terms supposedly appropriate to the natural sciences. Most who have that aim do not understand any human phenomenon, whereas the representatives of the style I have in mind — writers such as Thucydides, Diderot, Stendhal, Nietzsche - are marked in the first place by their powerful understanding of human motivation, in particular of relations between ethical and non- ethical motivations.18 There are certainly some such writers, particularly in the Epicurean tradition, who have described themselves as having reductive, scientific aims, but that is not a necessary or even a typical feature of the style. What is in question is not the reduction of the human to the non-human, but the placing of the ethical among human motives. The Aristotelian conception of the ethical life as the most truly and thoroughly human, and Platonic notions of self-fulfilment in terms of openness to the good, are anti-reductionist, but what attracts suspicion to them is not this, but rather their faith in opposite values, as Nietzsche put it, their failure adequately to recognize the interpenetration of (to use their own terms) the higher and the lower, of good and evil. The morality system, as I called it, is overdetermined in these respects (this helped to make it a particularly vulnerable object of Nietzsche's scepticism). Its purism and its self-sufficiency mean that it is structured not to hear any considerations that might limit its own. In characterizing the special phenomenon of 'morality', I in fact picked on, not just the general idea that it was specially concerned with obligation, but on what might be seen as a rather technical matter, that it interpreted the conclusions of moral practical reasoning as obligations: as I tried to explain in ELP, this does express some of the most typical features of this peculiar and familiar version of the ethical. It is worth rehearsing this point, in particular in order to repeat that 'morality' is not simply another term for the Replies 205 'deontologicaP, in the sense in which the standard classifications contrast deontological ethical theories with consequentialist ones. I suggested that Utilitarianism, which counts as a consequentialist theory if any does, in many of its forms is an example of morality; it involves a shorter list of input obligations than most other theories, but is no less determined to generate obligations as output. Against morality, I urged, among other things, that obligations are never final practical conclusions, but are an input into practical decision. They are only one kind of ethical input, constituting one kind of ethical consideration among others. I also made some suggestions about the kind of consideration they are, to the effect that they are one way - a way that runs through the deliberations of socialized agents - of securing the protection of important interests. This was not meant to be a novel idea. It is broadly in the spirit of Hume, and, like Hume's account, it deals in everyday, but still important, kinds of importance — the importance to each person of not being killed or despoiled. It is interesting that Taylor, starting from this point, should move rapidly in the direction of a more deeply ethical kind of importance, one that corresponds to what he calls 'strong evaluation', and to value rather than to mere desire. There is surely room for such notions, and for further inquiry into the kinds of value involved when we take various activities and states as being worthwhile, worth preserving or cultivating. We can also relate these sentiments to obligation, in recognizing, for instance, the kind of self-respect that is invested by some people in being an agent under obligation - the value that Taylor has rightly said lies, for Kant, behind the blank imperative of duty. But I would be reluctant to ground obligations themselves in what is important in such a sense. Though obligations are thought to be the object of very high sentiments, and have attracted some very refined casuistry, they relate basically to needs that are very everyday. It is just because the needs involved are so elementary that the psychological mechanisms designed to meet those needs are demanding; and it is because those mechanisms are demanding that the theory which grows around them becomes so dense and oppressive.

THICK ETHICAL CONCEPTS, KNOWLEDGE AND REFLECTION I tried in ELP to get beyond certain distinctions that seemed to me to have distorted and oversimplified the discussion of objectivity and related ideas in ethics. In particular I accepted that, in the case of 2O6 BERNARD WILLIAMS what I called 'thick' ethical concepts, 'factual' and 'evaluative' elements cannot be separated: to understand how such a concept can be applied to a new sort of situation it is likely that one will have to grasp its evaluative point, the outlook of the people who use it. I also agreed that users of such concepts, in correctly applying them, could properly be said to have knowledge. To this extent, certainly, the fact/value distinction should be overcome. (I argued as a separate point that the form which the distinction had taken in linguistic philosophy, though it was notably clear, was specially implausible.) I do not think, however, that all the concerns that underlie and motivate the fact/value distinction thereby disappear. Many of those concerns still have to be taken seriously, though in ways that allow more for historical and cultural differences between modern and other societies than appears in discussions that address only semantic or metaphysical questions.19 This is not to deny that conclusions shaped by historical and cultural concerns had better make semantic and metaphysical sense. I tried to make sense of them in ELP by combining, in what many critics have thought an unstable blend, a number of ideas each of which seems to me to make a powerful claim. Since it is perfectly obvious — whatever one makes of it at a reflective level - that the thick concepts used in different societies and at different times are not all the same, and since we can understand to some extent the thought, language, and experience of people who differ from us in this respect (which is how we know, among other things, that they differ from us), it cannot be the case that in order to understand such a concept we have to share it. Since in understanding it we must grasp its evaluative point, we can grasp evaluative outlooks we do not share. This is a platitude to ethnogra- phers,20 but is sometimes forgotten in philosophy. Statements made by people using thick concepts that are not our own are not, then, unintelligible to us. Nor can I see any reason for saying that they are, one and all, false. There seem to be perfectly good grounds for saying that some of them are what, in local terms, they are taken to be, namely true; and, since the people who use them satisfy other relevant conditions, we can say that those people have some knowledge under these concepts. But this is not knowledge that we share, since we do not share those concepts. One way, at least, in which such concepts go out of use is that people become more reflective about them; so some knowledge, at least - that knowledge - can be lost under reflection. Replies 207 At this point two distinctions are needed which I did not make clearly in ELP, and which I am grateful to Altham's paper for helping me to see. If the argument so far is right, then there is knowledge that can be lost, but not by its being forgotten: I can recognise as knowledge something which I could not, granted my outlook, myself share. This already raises some philosophical problems. It is a further question, how far and in what ways this result is brought about by reflection. It is a further question again, how far the lost knowledge may be replaced by other knowledge, in particular by other knowledge expressed in other thick ethical concepts. It is this last possibility that Altham has particularly emphasised. In ELP I made an historical claim, which is independent of the general structure, that modern societies tend to use less thick concepts and rely more on the thin concepts, such as 'good', 'right', 'obligatory', that are, significantly, the materials of typically modern ethical theories. I also claimed, for philosophical reasons, that, to the extent that our ethical outlooks rely simply on those concepts, they will not give us knowledge. We shall need, rather, something I called 'confidence'. Altham has asked why we need confidence. I agreed that some thick concepts would survive reflection; we can add that new ones may emerge. So what is done by confidence? The answer I had in mind was that, granted the nature of modern societies, we would face a good number of ethical tasks with the help of unsupported thin concepts,21 and, since there was not going to be knowledge in that connection, it would be as well if we had confidence. Pointing out that we shall have some knowledge under such thick concepts as survive reflection, Altham asks how we could combine with this a confidence that does not take the form of knowledge. Part of an answer might be found in getting clearer about what is involved in a thick concept's surviving reflection. I did not mean by this that the knowledge which is available under such a concept when the concept is properly deployed is turned by reflection into reflective knowledge: it will not have been validated, as under the Aristotelian aspiration our ethical concepts would be integrated into a dense and comprehensive understanding of nature, or at least of human nature. It survives reflection just in the sense that we would not have encountered any considerations that led us to give it up, lose hold on it, or simply drift away from it, as modern societies in the past two centuries or less have, for instance, done one or more of those things in relation to the concept ofchastity. While we 208 BERNARD WILLIAMS shall have the knowledge that comes with the deployment of our surviving thick concepts, we shall still not have any knowledge to the effect that we have a definitively desirable set of such concepts. However, unlike the inhabitants of the fictionally pre-reflective society, we do have the thought that other people have had different concepts, and that people may come to do so in the future. So we are aware, when we think of it, of something that less reflective people were not aware of, that these concepts are not simply given, and this leaves a space where confidence, again, is indeed, confidence in seeing the world in these evaluative terms. The thick concepts under which we can have some pieces of ethical knowledge are not themselves sustained by knowledge, but by confidence. As Altham remarks, there is no reason why our view of others' thick concepts should take a radically relativistic turn. We shall probably have resources within our own ethical outlook to think about other practices, as, in Altham's example, we can think of human sacrifice as 'sickeningly cruel'. I have no reason to deny that this thought is appropriate, or that it is true. But it does not stand in any very revealing relation to our understanding of why others had this, or any other, cruel practice. It does not in itself contribute to a theory of error for alien practices and beliefs, and, since in general we do not have much of such a theory, we do not have any very robust sense of there being a collective cognitive enterprise in ethics, such that we can represent our rejection of alien concepts and our use of our own as in itself an advance in knowledge. Again, this does not mean that we need be hesitant or double-minded in using our own; if we are not, this just shows that we can sustain them with a certain measure of confidence. It was in connection with such considerations that I claimed that there was an asymmetry between ethics and natural science, as we can now best understand each of them. The contrast between ethical understanding and natural science - our natural science, not Aris- totle's - is not at all the same as the discredited fact/value distinction, but it was part, I suggested, of what those who upheld the distinction might reasonably have had in mind. In trying to give an account of the asymmetry, I used the idea of 'the absolute conception of the world', a conception of the natural world expressed in terms maximally free from peculiarities introduced by our own, or any other observers', local perspective. I shall not try to discuss the problems of these ideas here,22 beyond making three very compressed comments. Replies 209 First, it has been claimed by some critics - not, perhaps by any represented in this book - that my use of the idea of the absolute conception of the objective of science involves an absurd, effectively pre-Kantian, metaphysical realism, under which we are supposed to compare our descriptions of the world with a world that presents itself independently of any description. Certainly that was not my intention, and the way in which I originally introduced the idea in my book about Descartes emphasized that it was meant to help us avoid a dilemma, of which one horn consisted precisely of that objectionable metaphysical picture. However - and this is the second point — I do accept that some formulations of the idea have implied a poorly considered causal theory of knowledge. I believe that the general idea is independent of such a theory, but certainly it needs to be shown in detail how this is so, and how notions such as 'world-guided convergence' may be best understood. What is important to the use that I have made of these ideas is that our agreement on truths under thick ethical concepts, or any other perspectival concept, can be coherently understood as subject to local constraints, while it is the aim of scientific theory, or at least of the most general scientific theory, to overcome local constraints. If Hookway is right, and there is no such contrast, then certainly I am wrong. But he has not yet persuaded me that I am wrong about this overall contrast, as opposed to more specific formulations of it. He points out that we find acts cruel or despicable 'because they are cruel or despicable', but, as I said in discussion of Altham, our understanding of this does not in itself tie in very closely with our best understanding of why we (in contrast, perhaps, to some other human beings) use these particular notions in structuring our ethical experience. In the scientific case, on the other hand, our understanding of what gives us reason to use some theoretical concept rather than another is tightly related to our understanding of why certain statements using that concept are true; for instance, to take a simple but central type of case, our understanding of what gives us reason to use the concept oxygen is part of our understanding of the truth that there is such an element as oxygen. To some extent, the asymmetries between ethics and theoretical science are results of the unsurprising truth that it is the aim of science to explain our experience, including experiences generated by the activity of science itself, such as the results of experiments; while it is not the primary point of ethics to explain experience at all, 2IO BERNARD WILLIAMS and our ethical experience is explained, often, in terms that make less claim on the validity of our own ethical experience. (The earlier points about naturalism come into focus again here.) But, beyond that, science's aim of explaining our experience can take the form of explaining, up to a point, why the physical world should present itself to creatures like us in the way that it does; whereas there is no world - certainly not the physical world, but also not an ethical world - such that it is the aim of ethical thought to explain why it should present itself to us as it does. There is certainly much more to be done in giving an illuminating account of such asymmetries, but I still think that the asymmetries are real and important, and that a notion like that of the absolute conception helps us to understand what they are - and so to see that the relation of honest ethical thought in the modern world to modern science cannot be the same as the relations between ethical and scientific inquiries in earlier times. This is true and important, even when we have laid aside the fact/value distinction, and also the scientistic distortions that often feature in contemporary supposedly disenchanted accounts of ethics.

INTEGRITY In A Critique of Utilitarianism, I told two brief stories about characters called George and Jim, whose situations Martin Hollis recalls in his paper. Ingenuously, I asked about each of them, 'what ought he to do?' There must have been some sense, perhaps more than one, in which this was a sensible question to ask. Since the question appeared in a text, its being a sensible question must have meant, at any rate, that it was a sensible question to ask a reader. Granted that a writer is not writing a letter, to which he awaits an answer; granted in particular that he is a philosophical writer, what questions (I now ask) might it be sensible for him to ask his readers? Questions, I suppose, that it is sensible for his readers, at his suggestion, to ask themselves. This is what basically, if not perhaps undistractedly, I had in mind in offering those stories and the discussion of them. Of course, I did more than simply invite a reader to reflect on the stories. I asserted a number of things, in particular that a direct Utilitarian would give certain answers to those questions, for certain reasons, and would not find them very difficult questions to answer. (This Replies 211 seemed to me, as it still seems, certainly true; to make it more obvious, I tried to head off some considerations that a Utilitarian might bring in to complicate the issue.) I brought into the discussion a particular notion, integrity, which I supposed, as others have supposed, might apply in such situations, and I claimed that the application of that notion would stand in the way of the Utilitarian argument and might well stand in the way of the Utilitarian conclusions. I suggested that for someone who used this notion, Utilitarian- ism's conclusion might seem downright unacceptable in George's case, though not so, perhaps, in the other. Over all this, or round it, and certainly at the end of it, there should have been heard 'what do you think?', 'does it seem like that to you?', 'what if anything do you want to do with the notion of integrity?' A considerable amount has been written about these stories and my discussion of them, but in much of it, it seems to me, those questions have not been heard. If they have not been heard, or not heard in the right way, that is in good part my fault, because I did not see steadily enough what I wanted the discussion to do. In association with that, there was a stylistic failure, an excess of polemical assertiveness. Besides that, however, I had not allowed enough for certain preconceptions of moral philosophy. It turned out that I had invented something called 'the integrity objection' to Utilitarianism, and had started a discussion on the question of how, if at all, that objection might be met. I can hardly deny that I produced an integrity objection in the sense that I made an objection and it centered on the notion of integrity. (I certainly did not invent it - it has been encountered by any number of people who have resisted courses of action commen- ded on consequentialist grounds, or have admired others for doing so.) The objection did not, however, take the form of my trying to disprove a theory by counter-example, as much of the discussion has assumed. If the stories of George and Jim have a resonance, it is not the sound of a principle being dented by an intuition. Some of the objections I have to the counter-example model in such connections are very general. I do not think that we need reject the notion of an ethical intuition altogether, so long as we do not assimilate it too closely to a linguistic intuition,23 and, given such intuitions, they can be useful in giving trouble to theories. But the counter-example model standardly goes beyond this and implies that when a theory has been damaged by an intuition it should be 212 BERNARD WILLIAMS replaced by an improved theory, and I have not kept it a secret that in ethics I do not think that this is a useful way to proceed. This is one reason why I said (ELP p. 200) that it was a misunderstanding to ask, in the way ethical theorists often ask, 'what alternative' one has to their formulations. They mean, what formulation does one have as an alternative to their formulations . . . and there is none. Quite apart from these more general considerations, there are special reasons for not understanding the discussion of George and Jim in terms of a counter-example provided by an ethical intuition. If that were the point of the discussion, the content of the intuition in each of the two cases would have to be specified, and it was not specified. The intuitions would be ethical thoughts available to George or Jim, and to specify those thoughts one would have to state fully the characters' reasons, not merely against the consequentialist course (reasons which I did sketch), but for preferring the non- consequentialist course, reasons which I did not give, as Hollis remarks. I hoped that I had deflected24 a further misconception, that their reasons were to be stated in terms of integrity itself— that these people would be motivated (at least if they were reflective enough) by thoughts about their own integrity. Reflections on one's own integrity are unappealing in any case, but integrity as a supposed criterion of action seems seriously confused as well, for what will the integrity be to? Sometimes it seems to be suggested, in discussion of such examples, that integrity can be a criterion, but one that gives the wrong answer. 'What about the integrity of the fanatical Nazi?' is a familiar question. Behind such a question is perhaps a thought to the effect that the fanatical Nazi goes on being a fanatical Nazi because he uses integrity as a criterion of action, and if he gave up that criterion, he would stop - what? Being a Nazi? Or just being a fanatical one? (An old and muddled literature comes to mind here, which used to discuss which thing one ought to do, what one ought to do or what one thinks one ought to do.) When I brought in integrity, it was not as a supposed criterion of action, but as a quality that many people prize and admire. It is in such ways that people put the notion to ethical use. My claim was that if people do put it to ethical use, they cannot accept the picture of action and of moral motivation that direct utilitarianism requires - and here were two stories to remind them, perhaps in different ways, of that truth. The argument turned only on the idea that Replies 213 integrity is a value. It did not imply that it was the only value, or the only admirable quality; or that things would always go better if people had integrity than if they did not. There are good questions about what integrity, as a quality, is. Everyone, I took it, agrees that George, if he refuses, shows integrity: if one thinks he would have done better to accept, then one thinks that integrity can do harm. Perhaps, but less clearly so, the same may be said of Jim. The fanatical Nazi who refused bribes which he was offered to save Jews was thought, even by his enemies, to display integrity; the enemies reasonably thought also that if there were going to be Nazis, it was better that they lack integrity. A less fanatical Nazi who was moved not by bribes, but by the pleas of the Jews, would be thought by fanatical Nazis to lack integrity, but probably not by the humane, perhaps because they entertain an idea (they certainly do not want to discourage it) to the effect that this was not a lapse or a weakness, but a rediscovery. The structure of integrity, as of its relative, sincerity, is complex in such ways. The discussion of George and Jim did not, certainly, uncover any of that complexity. It relied just on one feature of that structure, that there would be no use for the notion unless there were a value to be recognised in a particular person's sticking by what that person regards as ethically necessary or worthwhile. The claim was that direct Utilitarianism could not leave a space for that value. Martin Hollis, in his sympathetic discussion of the examples and their relation to my views, helpfully recognizes that I do not say the same about the two stories. They are different, and, if I had not thought that they were different, I would not have told two stories. Hollis perhaps chides me, mildly, for rather belatedly and obliquely indicating what I think Jim should have done. But, as I have already said here, my answer to the question about Jim, indeed any answer to that question, was not the main point. The main point lay in an invitation to think about the notions and styles of thought that one can bring to bear on such questions. This invitation Hollis himself has very constructively taken up. However, the framework in which he conducts his discussion does, I must confess, leave me rather puzzled about what he takes me to be trying to do, and what, by implication, he himself thinks such discussions can do. At one point, for example, he introduces the idea of 'objective reasons' grounded in local institutions, and asks whether ethics can 'leave it at that'. So far as I am concerned, he says that I must 214 BERNARD WILLIAMS press on. How else can [Williams] maintain that George was right to refuse a job which was respectable in itself and let him support his family? How else can he decide that Jim, a mere tourist among alien institutions, was right to intervene? The force of this, I take it, is that I could not offer the answers I do offer to the question about George and Jim if I accepted the following: (I) The right thing for any agent to do is what is in accordance with the local institutions. (I) would rule out the suggested answer in George's case, because either course of action, and certainly acceptance, would satisfy this condition; while in Jim's case there is a real question about what would count as 'local'. I agree. But why should Hollis think that I might be disposed to accept (I), or anything like it? Moroever, why should he think that I am concerned whether ethics can leave it at something like (I)? Ethics will be dealing in proposals such as (I), surely, only if it is offering a theory-based criterion of the right. But Hollis well knows that I do not think that ethics should be in such a business; to re-quote myself, it is a misunderstanding to ask what formulation one has as an alternative to the theorists' formulations. The path that has led Hollis to this point starts from my ingenuous question, as I called it earlier, 'what ought he to do?', and it runs, perhaps, through an assumption which Hollis does not state but which must, I think, be shaping his discussion, to the effect that the ingenuous question means 'what has he reason to do?' If that, in turn, expresses the notion of which I gave the internalist account, then it seems that I have not provided enough material to answer it. But the notion expressed in the question cannot be the internal reasons notion. For, in the sense in which the ingenuous question, 'what ought he to do?' is asked about George, it is quite clear that the direct Utilitarian can appropriately answer it by saying that he ought to take the job, and if George's S does not support that conclusion, that is a problem with George's S. So if the ingenuous question always and irreducibly means 'what has he reason to do?', then there must be external reasons, and it must be quite obvious that there are. I do not suppose that the ingenuous question always and irreducibly means 'what has he reason to do?' I use the clumsy formula "always and irreducibly" not (I hope) evasively, but because there is nothing to stop an English speaker using those words in all sorts of Replies 215 situations if he or she wants to, and securing some appropriate uptake. The point is that, in any given type of situation, there are some other words that might have secured the same uptake, but the same words are not the appropriate replacement in every situation. If we take the word 'ought' as applied to persons, it can sustain many different readings, from one that introduces reasons in a completely internal sense, through obligation, legal requirement, and so forth, to statements of desirability or mere appropriateness: 'she ought to have been a ballet dancer', we may say of a graceful young person, without suggesting that she had a reason to become one.25 In various different contexts 'ought' sustains, clearly, one or another of these readings. But there is a further possibility, which applies to the ingenuous question 'what ought he to do?' This is that the question, while it can sustain a whole range of readings, aims at the strongest reading (in internalist terms) that it can get: as one might put it, it sustains an internal reasons reading for as long as it can. Much of the time, we consider other people's conduct and comment on it under the general assumption (in certain contexts, one might better say 'fiction') that they are people with whom one might share deliberation, that one can set out from their S and arrive at the same action, at some relevant level of description, in this situation as one might setting out from one's own S. We can move to various distances from this assumption. 'If I were you . . .' is the appropriate expression for launching oneself into someone else's personal deliberation, and the supposition expressed in those words can settle down intelligibly at any number of points between two extremes, one at which it simply represents me as the actual you, and the other at which I am merely telling you what might be done by the actual me. At the farthest extreme from internalist identification, the thought conveyed by 'you ought. . .' may merely be to the effect that this is what someone quite different from you, a decent person, would do. Of course, one may insist on expressing this thought, too, (and a fortiori all the thoughts on the path to it) in terms of reasons for acting in this way which one claims that the agent has. But then we are back with the questions that came up earlier in connection with McDowell's paper. What is gained, except perhaps rhetorically, by claiming that A has a reason to do a certain thing, when all one has left to say is that this is what a phronimos, a decent person, or some such would do? That is a question that I put to McDowell, and would want to put to Hollis as well. 2l6 BERNARD WILLIAMS But there is a further question, for none of this yet has found a place, so far as I can see, for Hollis' concerns with what ethics can do. I seem to hear in Hollis' paper - though I may well be wrong - a suggestion that there is an important connection between the matter of external reasons, and the aspirations of ethics. I do think, as I said at length in ELP, that it would make a difference to ethics if certain kinds of internal reason were very generally to hand: the fate of the Aristotelian programme, as I understand that programme, is an example. But what difference would external reasons make?26 Suppose that the ingenuous question about George and Jim did necessarily sustain a reading in terms of external reasons. Would this make a difference to the aims of ethics? Should we suppose that, if genuine external reasons were to be had, morality might get some leverage on a squeamish Jim or priggish George, or even on the fanatical Nazi? (I hasten to say that Hollis does not say this, and the fanatical Nazi is not a member of his cast. I am merely trying to find a string that might tie together a line of argument that I continue to find puzzling.) I cannot see what leverage it would secure: what would these external reasons do to these people, or for our relations to them? Unless we are given an answer to that question, I, for one, find it hard to resist Nietzsche's plausible interpretation, that the desire of philosophy to find a way in which morality can be guaran- teed to get beyond merely designating the vile and recalcitrant, to transfixing them or getting them inside, is only a fantasy of ressen- timent, a magical project to make a wish and its words into a coercive power.

STYLE In their modest way, the stories of George and Jim - and, more interestingly, the ways in which they have been taken - touch on a problem that I mentioned at the beginning of the first book I published, a problem in finding a style for moral philosophy. The problem is to reconcile the general or conceptual concerns of philosophy with a sense of concrete detail, a sense that must be there in writing about ethics is going to be realistic or interesting.271 am still concerned with that conflict, but I do not now think that there is one problem to which, if one were lucky or ingenious enough, one might find a general solution. There is one philosophical problem, a pervasive or structural difficulty that concerns the existence of moral Replies 217 philosophy, but it generates a series of different problems in writing, which can be solved, if at all, only in terms of the particular text. To the extent that my work has seemed, even to myself, on some occasions to have solved or at least got round such a problem, it has sometimes been criticized as being destructive or 'purely negative'. Often, some theory has been under criticism, and the more particular material has come in to remind one of the unreality and, worse, distorting quality of the theory. The material itself, by the standards of good biography, some history, or certain kinds of nineteenth-century fiction, is itself extremely schematic, but it is at least nearer to psychological and social reality than the theory, and helps to bring out the basic point that, very often, the theory is frivolous, in not allowing for anyone's experience, including the author's own. Alternatively, the theory does represent experience, but an impoverished experience, which it holds up as the rational norm - that is to say, the theory is stupid. The procedure of confronting that kind of theory with more concretely realized considerations is a kind of philosophy, and the arguments that make it work, if it works, are philosophical arguments. It is hard to deny that, as the critic says, it is destructive. Is it, as the critic next says, purely negative? I would rather say that it is impurely negative. If the complaint is that one has taken away a thing of a certain sort and not replaced it with another thing of that sort, then usually this is true, but misses the point; the aim was not (once again) to replace that formulation with another such formulation. If the complaint is that something has been removed and has not been replaced by anything of any sort, that is not always true. It may have been replaced - to repeat, if the procedure works at all - with some understanding of why it is that no such theory is needed. In any case, it is wrong to discuss this as though the only thing at issue were the question of how much has been taken away or replaced in philosophy. If one tries to destroy a philosophical theory (I speak all the time of theories in ethics, but the point goes wider), and says at the same time that it would be a misunderstanding to look for a better philosophical theory to replace it because there is no useful place for such theories at this point, then one should have some account of how all this is related to our understandings outside philosophy. Otherwise, the enterprise will be self-confessedly rather futile. It is not true in general that denouncing a futile enterprise must itself be futile, but it is fairly futile to denounce an enterprise as 2l8 BERNARD WILLIAMS futile simply in its own terms and in order to contribute simply to that enterprise. This is the basic failing of a certain vulgar Wittgensteinianism, which makes an academic philosophy out of denouncing academic philosophy. In the famous and apposite words of Karl Kraus about psychoanalysis, it is the disease for which it is itself the cure. The complaint does not apply at all to Wittgenstein himself, for two quite different reasons. He suffered from the sense that his philosophical anxieties might be merely his problem, and prayed to get away from them; at the same time, he thought that there was one problem that was everyone's problem, an emptiness and cruel superficiality of everyday thought, which a better philosophy certainly could not cure, but which it might stand against, if only in his own, and perhaps someone else's, mind. To adopt that view of philosophy as the framework of a supposedly self-justifying academic enterprise is wonderfully perverse, rather like setting up as a Kierkegaardian bishop. If philosophical destruction of theories in ethics can provide a liberation, it had better do more than liberate philosophy from those theories. Its hope, rather, must be to liberate a reader, someone who wants to think about such questions, from distortions or misunderstandings involved in his or her own experience, and the aim of this (it applies as much to the writer as the reader) is the Socratic aim of improving that experience. The destruction of theory is able to play a role in this because a powerful philosophical theory can be an effectively articulated expression of those distortions. It should not be thought that philosophy itself is, on this view of things, the cause of the distortions. Some modern philosophy has certainly given an account of itself by which orthodox philosophy has itself played the role of distorting force, figuring as the priests and kings in a replay of an eighteenth-century myth of the fall of natural man. Some of Wittgenstein's remarks encourage that picture, and some of Moore's; and Austin lent support to it, in a more conservative, Oakeshottian tone, with the story he sometimes offered of the inher- ited wisdom of ordinary practical people in comparison with which the philosophers and their theories were mere johnnies-come-lately. This account, by which there is a better theory at the bottom of the well of common sense, really is open to Russell's remark about common sense being 'the metaphysics of the stone age'. In ethics, certainly, the rejection of theory does not rest on the idea that Replies 219 common sense would do very well by itself if only philosophers would not bully it with theory. The idea is rather that philosophical theory may be one expression of misconceptions in our ethical consciousness, which can be exposed to some extent by philosophy (of that other, destructive, kind), but are not to be explained primarily by a mysterious power of philosophical theory to invade healthy practice. They are explained, like our other personal and cultural discontents, by many kinds of considerations, including, no doubt, some other kinds of theory.28 This point answers a familiar charge, that, inasmuch as moral philosophy acts in this destructive way, it is an ally of prejudice in blocking radical reconsideration of our morality. On the contrary. The destruction may well turn out to be the destruction of part of our moral consciousness, and to offer (if one is lucky) some liberation from illusion. Above all, the project takes seriously the idea not only that we are deceived, but that we are deceived by forces worth worrying about, such as our own fears and resentments, our misunderstanding of social representations, and the effects of tradition. The idea, by contrast, that philosophy's contribution to the radical reinterpretation of our life lies in producing some more, and more radical, ethical theory merely postpones (or, to put it less benignly, conceals) the questions of power, the power both of the discontents and of their cure. It represents philosophy both as less than it really is, and as more. To understand any power that philosophy may have over our ethical outlooks, we need a good deal more than philosophy itself.

NOTES In these notes the following abbreviations are used: ELP: Bernard Williams, Ethics and the Limits of Philosophy London: Fontana/ Collins, 1985. Morality: Bernard Williams, Morality: An Introduction to Ethics New York: Harper and Row, 1972; Harmondsworth: Penguin, 1973; Cambridge University Press, 1976 (and, with a new preface, 1993). Moral Luck: Bernard Williams, Moral Luck: Philosophical Papers igy^-ig8o Cambridge University Press, 1981. Shame and Necessity: Bernard Williams, Shame and Necessity Berkeley, CA: University of California Press, 1993. 1 The style of philosophy most associated with these labels is probably Austin's, but others were equally influential in Oxford at that time: principally Ryle, and a more directly authentic voice of Wittgenstein, 22O BERNARD WILLIAMS chiefly represented by Elizabeth Anscombe who salutorily tried to teach one that there was more to philosophy than being quick. I am not sure that I personally learned a great deal from Austin (I set down some reactions to his philosophy in a review of two of the posthumous books {Oxford Magazine, 6 December 1962)). Aspects of Austin's public per- sonality were only too familiar to someone who had been to an English grammar school; it was something of a revelation to discover that to a person from a different culture he could seem almost exotic, a perception recorded in Stanley Cavell's 'Austin at Criticism' (reprinted in Must We Mean What We Say, New York: Charles Scribner's Sons, 1969). One point to be made about philosophy in Oxford at that time is that it should not be equated with a theoretically inert observation of ordinary language, rejecting philosophical argument: Strawson's Individuals was published in 1959. 2 See 'Internal and external reasons' in Moral Luck, and 'Internal reasons and the obscurity of blame,' Logos (USA) 10 (1989) 1—11, which offers some clarifications to the earlier piece. There are connections between this view and other aspects of my work, as Herbert Hart pointed out in a review of ELP in The Mew York Review of Books, 17 July 1986. The question recurs at various points in these comments, particularly in my discussion of Hollis' paper. 3 For the purpose of the present discussion, I shall accept McDowell's formulation, under which internalism requires an agent's reason to depend on the contents of his S. Matters are more complicated than this implies, above all because of the case of Kant. Hollis, in his paper, takes it as evident that Kant is, in my sense, an externalist, because an agent, for Kant, can have a reason to do a certain thing 'regardless of [his] present motives'. But I do not agree that this in itself is enough to make Kant an externalist (cf. 'Internal reasons and the obscurity of blame', and my reference there to the paper by Christine Korsgaard to which McDowell refers). The issue is difficult, in ways which relate to the discussion that follows here. Kant thought that a person would recognize the demands of morality if he or she deliberated correctly from his or her existing S, whatever that S might be, but he thought this because he took those demands to be implicit in a conception of practical reason which he could show to apply to any rational deliberator as such. I think that it best preserves the point of the internalism/externalism distinction to see this as a limiting case of internalism. 4 In particular to exclude boring counter-examples in which the motivation to c|> is merely a causal consequence of the activity of deliberating. 5 Neo-Aristotelians will want to consider whether temperance is an item that they want to transport unchanged from the fourth century to the present time. They should ask how much of Aristotle's apparatus can really have applied to people (who? how many?) in his own time. It is hard to believe that Athens can have changed quite so dramatically Replies 221 from the world known to Sophocles and Euripides, Thucydides and Plato, as Aristotle's restful endoxa about virtue seem to imply. It is important, on the other hand, that Aristotle himself did not believe that everyone had reason to act as the phronimos would: see below. 6 If not, I have righted the balance (neo-Aristotelians will probably think, tipped it in the wrong direction) in chapter 5 of Shame and Necessity. 7 Only roughly speaking: while it is my bad luck if I have been badly brought up, it is not, on Aristotle's views, X's bad luck that X is a woman, or (I take it) a natural slave - though he leaves it, unsurpris- ingly, vague what we are supposed to think about the latter case. I take up such issues in chapter 5 of Shame and Necessity. 8 This is why I said in ELP that Aristotle's theory did not offer a reason to each person. The question does not present itself to Aristotle in exactly this form. In part, this is connected with the complex question of the supposed audience of his treatises on ethics; as Martha Nussbaum points out in her essay, it is important to remember the role of these books as guides to rulers or educators rather than to the ethically perplexed. 9 The same question is very interestingly discussed, in a different context, by Ruth Padel in her In and Out of the Mind (Princeton, NJ: Princeton University Press, 1992). 10 I thus stand by what I said about certain proposals made on this basis by Michael Tooley, ELP p. 114. That particular passage perhaps does not do enough to bring out the sheer frivolity of such proposals, which lies in their refusal to engage with the only two things that matter: the politics of trying to make rules for such situations, and the experience of people engaged in them. Because of this refusal, the philosophy that produces such proposals is, in the bad sense, deeply 'academic'. It would be comforting to believe, with Hume, that it was therefore merely ridiculous. 11 Philebus 2 ic. It is not entirely clear which of several points the passage is making: it may mean that the life of pleasure without intellect is not much of a life, or that it is not a life at all, or even that it is not a life with all that much pleasure in it, since it lacks the pleasures of anticipation and recall. In any case, the point stands that Protarchus does not want it - Robert Nozick, Anarchy, State, and Utopia, Oxford: Basil Blackwell, f 12 All this includes what one has done. This is one strand in the relation of these themes to tragedy: so Oedipus' reflection at Colonus (which I discuss in chapter 3 of Shame and Necessity), that what had happened to him was that he had done these things. 13 As I suggested it might in the section of Morality called 'Moral standards and the distinguishing mark of man' (a title, incidentally, both parts of which I would now want to change, for different reasons). 14 Some of Aristotle's less congenial opinions, notably about slavery, have 222 BERNARD WILLIAMS been claimed to be produced by ideological distortions of his own best ideas. The question is complicated; I have discussed it in chapter 5 of Shame and Necessity. I may perhaps add, though I shall not try to argue the point, that I find the suggestion that Aristotle took natural slaves to be mental defectives (granted that we know what that might mean) to be quite implausible. It is hardly helped by the comparison with barbarians, to which Nussbaum refers: Aristotle's discussions of barba- rian forms of government, for instance, hardly fit with his supposing something to the effect that all barbarians were mentally defective. 15 Michael Dummett once said to me 'Aristotle's theory of the heavens turned out to be false, but it might not have done'. This seemed to me, and still does, a strange idea. Aristotle's world was one in which the arrangement of the heavens had a meaning in human terms, and to a modern outlook it is not a contingency that this should not be so. It is an interesting question, what the difference is between those who by disposition share the attitude expressed in Dummett's remark, and those who share mine. The members of the two parties may well disagree among themselves on other things: I am sure that Ayer, who had little time for Aristotle and none for Christianity, would have agreed with Dummett. It is not so much a matter of metaphysical theory as of one's attitude to history. 16 In this area, some of Taylor's concerns come close to those of Iris Murdoch, which she has now expressed in philosophical terms, more fully than before, in her book Metaphysics as a Guide to Morals, London: Chatto and Windus, 1992. Nietzsche: Beyond Good and Evil, section 230. 17 I have said a little more about this in a recent article, 'Nietzsche's Minimalist Moral Psychology', European Journal of Philosophy, 1:1 (i993)- 18 I count Hume as such a writer. A frequent rhetorical device of his is to put 'moral' and 'non-moral' motivations together without a change in tone; there is an example in a passage that Hollis quotes, about 'ambition, avarice, self-love, vanity, friendship, generosity, public spirit'. To learn from Hume, we do not have to share his very strong view, which Hollis discusses, about the universality of human sentiments. 19 Here my approach has something in common, once more, with Taylor, and also with Alasdair Maclntyre. If Taylor and Maclntyre will forgive my putting them into a mere cartoon sketch, one set of relations between our positions might perhaps be put like this: Taylor and Maclntyre are Catholic, and I am not; Taylor and I are liberals, and Maclntyre is not; Maclntyre and I are pessimists, and Taylor is not (not really). A word that seems to me not to help in describing any of these positions is 'communitarian'. 20 I stressed the importance of'the ethnographic stance' in my response to a review of ELP by Simon Blackburn: Philosophical Books 27:4 4 (October 1986). Replies 223 21 Thin concepts such as 'right' can operate, in familiar ways, as superve- nient on, or supported by, thick concepts. Susan Hurley (in Natural Reasons, Oxford University Press, 1990) rejects, as I do, the doctrine she calls 'centralism' - the doctrine that thin concepts are primary - but treats the idea that they can operate independently of a thick basis as just a philosophical mistake. This seems to me to ignore the historical developments I refer to. 22 One reason is space: there is a lot to be said. I apologize to Hookway and to Jardine, who have written on these questions (as also to Elster and to Sen, whose very interesting papers are on topics which I have not been able to incorporate into these comments). But the other reason for not taking up the absolute conception here is that I plan to complete in about three years' time a study, part historical and part philosophical, of truthfulness and its relations to truth, in which I hope to approach issues of perspectivism, convergence and scientific realism in a different way, which will perhaps improve on the rather fragmentary treatment I have given to these things in the past. 23 See 'What does intuitionism imply?' inj. Dancy, J. Moravscik, and C. C. W. Taylor (eds.), Human Agency: Language, Duty and Value, Stanford University Press, 1988; and ELP pp. 93f. In current graduate school practice, 'intuition' (as in 'my intuition is') often means no more than an undefended theoretical position. 24 To some degree in the original text, but at any rate in 'Utilitarianism and moral self-indulgence', reprinted in Moral Luck. 25 Cf. 'Ought and moral obligation', reprinted in Moral Luck. The suggestion that follows generalizes to some extent a proposal made in 'Internal reasons and the obscurity of blame'. 26 It is possible that Hollis has in mind a programme in the style of Kant, which he would call externalist and I would not (see note 2 above). If that is so, then we are tied up in merely verbal differences; I myself have claimed that the success of that programme would make a difference to ethics. But I do not really think that this is Hollis' point, in particular because he moves in the direction of external reasons not by providing a Kantian or similar basis for them, but by arguing, if I understand him, that internal reasons do not get enough grip or leverage. Hollis and I are in agreement, I am sure, that a Kantian conclusion cannot be reached in that way: one cannot conclude that pure practical reason issues the moral law, just because other sources seem to lack authority. 27 Morality, Preface. A point I did not make then, but would make now, is that this second demand, is in some part, a product of history. Kant hardly felt the need for concrete particularity, but his work cannot be said to have lacked an impact. But it is a feature of Kant's philosophy itself and also of its relation to its time, that it could hope to rely on a certain kind of abstractness. 28 I have become increasingly aware of the extent to which we need to understand our ethical ideas and their failings in terms of their history, 224 BERNARD WILLIAMS

which is in good part a contingent history. Most of those who have thought history relevant to philosophy have taken it to be necessary, while most of those who have recognized it as contingent have thought it irrelevant to philosophy: this is just one connection in which philosophy needs to give up its long obsession with necessity. Name index

The editors would like to thank Peter Kail for preparing these indices

Achilles 107, 109, 177, 183 Blackburn, S. 8306, 222n2O Adeimantus 199 Brock, D. i8n26, 126m 7 Aeschylus 97, 129050 Broome, J. 13, 14, I7n22 Ajax 177, 183 Burkert, W. 126m 8, I26n23 Altham, J. E. J. 2, I3in65, 207-8, 209 Annas, J. 131065 Cavell, S. 220m Anscombe, G. E. M. 84014, 220m Cebes 94 Antigone 177, 183 Chernoff, H. 2802, 2907 Ardent, H. 175 Chipman, J. S. 2902 Aristophanes 129050 Cleopatra 180 Aristotle 86-95, Ioo> IO3~4> IO6~7i 110-24, Collins, H. 44m 2 I24n5, I25n8, 126m 7, i26n20, I26n26, Coombs, D. H. 28n2, 2907 I27n29, I27n3i, i27n32, 129^0, Cooper, J. 115, 116, i27n38, I3in62, 131064, 136, 143, 148, 149, 171, 128046, 128047, I28n49, 128049 189, 192, 193, 194-202, 203, 208, 22on5, 22in7, 22in8, 22ini4, 222015; Dancy, J. 85022, 223023 Nicomachean Ethics 8307, 86, 90, 91, 98, Davidson, D. 31 n 17, 83n 1 o, 84014, 84017 102—5, no-20, i25ng, I28n46, I28n47, Davies, R. L. 28n2, 2gn7 I3on5i; Eudemian Ethics 116, 12505, Democritus 147 127030, I27n39, I28n47, 1301151; Derrida, J. 153 Politics 92, 103, 106-10, 128047, '301151, Descartes, R. 147, 171, 209 1301153; Z>e Afofa I24n2, I27n38, 128031; Detienne, M. 126m 8 Magna Moralia 127030; Parts of Diderot, F. 204 Animals 128040, 128047; De Dummett, M. A. E. 84ni2, 222ms Anima 128045; Topics 128047; History of Durkheim, E. 44nio Animals 128047; The Generation of Dworkin, R. 154019 Animals 128047, 130051; Metaphysics 128n4g; De Motu Elleo, R. F. 4408 Animalium 130051 Elster, J. 2, 1603, 223022 Arrow, K. J. 28n2, 29n2, 2907, 2gn8 Euripides 97, 129^0, 22in5 Anderson, S. B. i8n24 Austin, J. L. 218, 219m Farrar, C. I26n23 Ayer, A. J. 222ni5 Firth, R. 44015 Fishburo, P. C. 2902 Bacon, F. 146, 154016 Foot, P. 8309 Baier, A. 84022 Foucault, M. 4409, 153 Bar-Hillel, M. 1502 Frege, G. 77 Bathsheba (OT) 178 Freud, S. 202 Bentham, J. 136, 145, Benveouste, E. 126m 9 Gallop, D. 126016

225 226 NAME INDEX

Gay, R. 84m 5 Korsgaard, C. M. 84021,22003 Geertz, C. 141, 154118 Kraus, K. 218 Gilbert, H. 17m 6 Glaucon 199 Larson, E. B. I7m6 Green, J. 28m Leibniz, G. 6, i6n7 Grotius 143, 148, 1551119 Leooble, R. 154ml Levi, I. 28m, 2902, 2gn7 Habermas, J. 137, 147, 148, 150, 151, Little, I. M. D. 28n2 Locke, J. 143 Hacking, I. 41, 451119 Luce, R. D. 2gn2 Hahn, F. 28111 Lukes, S. 45m 9 Hampshire, S. 154m 9 Lycan, W. 45n2O Hansson, B. 29112 Lysias 129050 Hare, R. M. 138, 146, 150, 153112, 154117, 1551122, 1551124 Maclntyre, A. 12402,222019 Harrison, R. 2, 30m 1, 83m Mackie, J. 15406 Hart, H. 22on2 Majumdar, T. 2801, 2gn2, 30013 Havigurst, CM. 17m 7 Marschak, J. 2gn7 Hecuba 97 Marx, K. ng, 131057 Heil, J. 441115 Mathieu, D. i6n8, 17010, 17m 8, i8n26 Hemlick, J. S. 181124 Mauss, M. 44010 Heracles 96, 131 n66 McDowell, J. 2, 45017, 186-94, '99> 2I5> Heraclitus 123, I3in62, I3in63 22003 Herodotus, 1291150 Mill, J. S. 149, 155020 Herzberger, H. 2gn2, 29117, 29n8 Milligan, M. 131057 Hobbes, T. 144, 154m 9 Moore, G. E. 218 Hofstee, W. K. B. i8n24 Moravscik, J. 223023 Hollis, M. 2, 45ni9, i84n2, 210-16, Mozart, W. A. 134 22On2, 22On3, 222ni8, 223n26 Murdoch, I. 145, 154015, 222ni6 Holwerda, D. 126m 9 Homer 97, 107, 127032, 127033, I28n5o Nagel, T. 84018, 127038, 128043, 155019 Hooker, B. 84m 5 Nash, J. F. 2802, 2907 Hookway, C. 2, 209, 223n22 Nathan (OT) 178 Houthakker, H. S. 28n2 Needham, R. 44010 Hume, D. 71, 82, 173, 175, 182, 205, Newman, N. R. 108, 127032, 127033 22imo, 222ni8 Nietzsche, F. 151, 153, 182, 203, 204, 216, Hurley, P. 84022 222m6, 222017 Hurley, S. 223n2i Nozick, R. 11, 28m, 197, 22inn Hurwicz, L. 2gn2 Nussbaum, M. 2, 126m 7, 127026, 131065, 189, 194—202, 203, 22in8, Irwin, T. H. I24n2, 12405, 128043 Isaeus Oakeshott, M. 154010 Isocrates 129050 Oakley, F. I54nio Odysseus 97 Jardine, N. 2, 223022 Oedipus 221 n 12

Kahn, C. 126m 9 Padel, R. 22iog Kant, I. 2, 132, 135, 136, 144, 145, 147, Page, B. 17n 12 148, 15304, 154014, 182, 201, 202, 205, Pauly, M. V. 171118 22003, 223n26, 223027 Philebus 98, 100-1 Kelly, J. 2gn2 Philoctetes 177 Keyt, D. 115, 127038, 128046, 128047 Peirce, C. S. 47, 49, 51, 52, 67m, 6702, Kiog David (OT) 178 6704 Kiog, W. M. 17017 Plato 93, g4, 101, 103, I2gn5o, 147, 170, Kirk, G. S. 126:118 202, 22105; Laws 128047; Phaedo 93, 105; Name index 227

Philebus 98-102, 105, no, 111, 116, 22inn; Republic 1261121, 199 Plotinus 181, 182 Porphyry 181 Stewart, J. A. 127^8, I28n42, Powell, E. 25 Strawson, P. F. 84ni4, 220m Prichard, H. A. 203 Suzumura, K. 2gn2, 29n7, 2gn8 Protarchus 98—101, 105, in, 196, 221ml Putnam, H. 12 71128 Talmon, J. Taylor, C. 2, I3in65, 202-5, 222ni6, Radner, R. 29117 222nig Raiffa, H, 29112 Taylor, C. C. W. 2231123 Rawls, J. 4, 5, 109, 1271136, 1271137, 145, Thrall, R. M. 28n2, 2gn7 146, 149, 154m 9, 1551120 Thucydides I2gn5o, 202, 204, 22in5 Reason, D. 44n8 Tooley, M. 22inio Richter, M. K. 29112 Tucker, R. Robertson, J. A. i6n8 Roemer, J. i6n3 Uriah (OT) 178 Rorty, A. 83117, I24n2, 127^8, 1301152 Rosner, F. 17ms Vernant, J.-P. I26ni8 Rothschild, E. 28m Vidal-Naquet, P. 126m 8 Rousseau, J.-J. 136, i55nig Vlastos, G. 126n25, 126n26 Russell, B. 218 Ryle, G. 154116, 186, 2igm Walzer, M. 12, i6n4 Waugh, E. 13 Samuelson, P. A. 20, 28n2, 29116, 30m 2 Wiggins, D. 51, 53, 66, 6703, 67n4, 67ns, Scanlon, T. M. 84^2, i55nig SchafFer, S. 44m 2 2, 196 Sen, A. 2, 4, 29n8, 127m 7, 1311165, Wilkes, K. V. 127^8, I28n42, 3 Winkler, J. 126023 Shapin, S. 44n 12 Wingrave, O. 30ml Sheehy, G. 154ng Wittgenstein, L. 218, 219m Siramias 94 Wollheim, R. Skorupski, J. 44nio Smart, J. J. C. 2, 2gn2, 184m Xenophanes 96, I26n2i, Socrates 93, 94, 98, 99, ioo, 105, 106 Sonnenschein, H. F. Yaari, M. I5n2 Sophocles i2gn5O, Stendhal 204 Zeckhauser, R. 28m Subject index

absolute conception 32, 55-60 fact/value distinction 195, 201, 206, 208 akrasia 182 fallibilism 57, 59, 66 allocation 5-18 folk taxonomy 34-5 axiomatic choice theory 19 Frankfurt School 42 freedom 143-7, '51 bios 115— 16, I28n47 Buridan's ass 23 George (from Utilitarianism: for and against) 170-2, 174-5, 176—8, 179-80, causality: and knowledge 32, 36, 38, 47, 181-2, 183—4, 210-16 209; and perception 36-7, 47; and god 138, 143, 144, 147, 182 world-guidedness 32,47 character 43, 158—62, 174; see also hedonism gg, 118, 144 dispositions human, nature see nature Chernoff condition 20-1 choice 19-31 imagination 70 Christianity 142, 146, 190 imperatives, categorical 152 claims 14 importance 133 confidence, ethical 43, 61, 156-69, incentives 8—9 206-10 increment principle 6-8 consensus 31,47—50,60,62 integrity 170, 184, 110-20 consistency 19—31 internal reasons 30m 1,68-85, 173-6, 182, convergence 32, 42-3, 46-55, 56-67; 186-94, 2I5> 22on3 world-guided 35—38; precedence-guided intuitionism 135, 149 33. 38-42, 43 intuitions 138, 140, 146, 211-12 conversion 74 Jim and the indians 170-3, 176-8, 179, desire 68-70, 78, 158, 159-60, 173-4 181-2, 183—4, 210—16 disagreement 46, 66 justification: causal 32, 38; by precedent 33, discourse ethics 148 38-42 dispositions 36,43, 174-5, !&2 distribution of goods; and justice 4-18; of Kantianism 135-6, 147-9, 170, 182, 190 educational resources 4—18 knowledge: ethical contrasted with division of labour 63-4 scientific 32-45, 46-67, 205-10; and reflection 33,47, 156-69, 176, 177, endoxa 88, 100, 194 205-10 envy 8, 10—11 equality 4-18 level principle 6—8 ergon 112—15 liberation 142 error: explanation of 41-2, 49, 57, 208 lotteries 14-15 essentially contested concepts 64, 66 luck 197 eudaimonia 103-5, IIH!i 116-18 external reasons see internal reasons menus 24-6 228 Subject index 229

merit 4, 8 reflection 32, 41, 43, 48, 61, 156-69, 176, morality: contrasted with ethics 132, 177, 205-10; and confidence 43, 61, 204-5 156-9, 207; and the destruction of ethical motivation 68—85, r44> l73~6> x 86-94 knowledge 33, 47, 156-69, 176, 177, 205-21 o; and relativism 166-9, 208 naturalism 137—8, 140-3, 145—6, 202—5 reflective equilibrium 5 nature: Aristotle's conception of relativism 66, 166-9, ll1> 2°8 (general) 194; Aristotle's conception of reliability 38 (human) 86—131, 194-202; human 156, risk 12 165 need 4, 8 science: history of 35, 36, 38-40; contrasted with ethics 32-45, 46-67, 102, 157, objectivity and the absolute conception 32, 208-10; and absolute conception 32, 55-60; ethical 32-45, 46-67, 60, 159-61, 56-60 182-3; ethical contrasted with secondary qualities 34: analogy with thick scientific 32—45, 46-47, 102, 157, 208-10; ethical concepts 32-3 precedence guided 33, 38—42; world slaves 122 guided 32-5, 35-8 style 185, 216-19 obligation 132-5, 137, 140, 142, 145—6, subjective motivational set 69, 70, 173, 165, 186-205 174, 179, 180 paideia 97 teleology 102 paternalism 8, 9-10, 144 thick ethical concepts 32—4, 141, 162-6, perception, ethical 65, 66 167, 168, 176-9, 205-10 personal identity 91-5, 101, 105-9 thin ethical concepts 32—3, 35, 165, persons 89, 91-5 176-9 philia go, 103-4, 117 truth 49, 50, 51, 54, 81, 183 philosophy 185-6, 216-19 phronesis 148 utilitarianism 6, 8, 109-10, 132, 135-6, phronimos 189-94, 2I5 138, 144-5, 146—7, 149' l7°~2i '^2, 203, phusis 95, 194 205, 210; and integrity 170, 210-16 political nature 102—10 utility 6, 170-2, 179, 180, 181 poverty 9 practical reasoning 19-31, 68—85, I0I> vitiation 38-40 118-19, "48, 152, '56, 165 pragmatism 13, 54 weak axiom of revealed preference 19, 20, prescriptivism 138, 190 22-4 procedural conception of ethics 147-52 welfare 6 projectivism 80 women 122 psychologism 77-82, 191 Zoe 113—16, 2i8n5O reasons see internal reasons