Moral Particularism Edited by BRAD HOOKER and MARGARET OLIVIA LITTLE
CLARENDON PRESS • OXFORD
MORAL PARTICULARISM
This book has been printed digitally and produced in a standard specification in order to ensure its continuing availability
OXFORD UNIVERSITY PRESS
Great Clarendon Street, Oxford OX2 6DP Oxford University Press is a department of the University of Oxford. It furthers the University's objective of excellence in research, scholarship, and education by publishing worldwide in Oxford New York Auckland Bangkok Buenos Aires Cape Town Chennai Dar es Salaam Delhi Hong Kong Istanbul Karachi Kolkata Kuala Lumpur Madrid Melbourne Mexico City Mumbai Nairobi Sao Paulo Shanghai Taipei Tokyo Toronto Oxford is a registered trade mark of Oxford University Press in the UK and in certain other countries Published in the United States by Oxford University Press Inc., New York © The several contributors 2000 The moral rights of the author have been asserted Database right Oxford University Press (maker) Reprinted 2003 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, without the prior permission in writing of Oxford University Press, or as expressly permitted by law, or under terms agreed with the appropriate reprographics rights organization. Enquiries concerning reproduction outside the scope of the above should be sent to the Rights Department, Oxford University Press, at the address above You must not circulate this book in any other binding or cover and you must impose this same condition on any acquirer ISBN 0-19-823883-5
CONTENTS
1.
Introduction
vii
List of Contributors
xii
Moral Particularism: Wrong and Bad
1
BRAD HOOKER
2.
Particularizing Particularism
23
ROGER CRISP
3.
The Truth in Particularism
48
JOSEPH RAZ
4.
Ethical Particularism and Patterns
79
FRANK JACKSON, PHILIP PETTIT, AND MICHAEL SMITH
5.
6.
Ethics as an Inexact Science: Aristotle's Ambitions for Moral Theory T. H. IRWIN
100
The Particularist's Progress
130
JONATHAN DANCY
7.
Ethical Particularism in Context
157
DAVID BAKHURST
8.
Particularity and Principle: The Structure of Moral Knowledge
178
JAY GARFIELD
9.
Against Deriving Particularity
205
LAWRENCE BLUM
10. Why Practice Needs Ethical Theory: Particularism, Principle, and Bad Behaviour 227 MARTHA NUSSBAUM
11.
Unprincipled Ethics
256
DAVID MCNAUGHTON AND PIERS RAWLING
12.
Moral Generalities Revisited MARGARET OLIVIA LITTLE
276
Contents
vi Bibliography
305
Index
313
INTRODUCTION Brad Hooker and Margaret Olivia Little
Moral particularism is currently one of the most widely discussed—and hotly contested—issues in ethical theory. Spurred in large part by reactions to the writings of John McDowell and Jonathan Dancy, philosophers continue to divide between those who find particularism's claims insightful and those who find them exaggerated or wrong-headed. Indeed, philosophers continue to divide over how best to interpret what claims 'moral particularism' is meant to represent in the first place. In this collection, we present a dozen new essays by theorists who take up the controversy.' The collection begins with those who are sceptical of moral particularism. In Chapter 1, Brad Hooker argues that adherents of the doctrine are overly impressed with the dangers of moral principles. He contends that, while the search for such generalizations has at times led to crudeness in theory, the particularist's response of jettisoning such principles introduces dangers that are far deeper. Starting from the premise that certain 'non-trivial general rules seem overwhelmingly sensible, Hooker argues that, once we isolate what particularism must claim in order to count as a distinctive thesis, we will see that the arguments proffered in its favour are unpersuasive. Indeed, faithful followers of particularism would precisely fail to display the reliability we seek to develop in—and hope to rely on in—moral agents. Next, Roger Crisp in Chapter 2 distinguishes amongst various forms of particularism and argues that the true forms are uncontroversial and the controversial ones false. After arguing that self-described generalists can accommodate important insights about the incommensurability of values and the ineliminable need for judgement, Crisp criticizes the idea that underlies Jonathan Dancy's radical particularism, namely, the idea that a full specification of the reason for acting in some way can, in another context, fail to constitute a reason or even constitute a reason for acting otherwise. Crisp argues that such variance means that the reason cannot then be complete. Just as we do not in science take an explanation as complete if the I Two of the pieces, Joseph Raz's and (an expanded version of) Martha Nussbaum's, are also appearing in other fora.
viii
Introduction
factors cited could in another case lead to the opposite result, we should not do so in ethics. Since Dancy himself demands that a difference in considerations be cited when comparing two cases, Crisp urges, there is no good reason for stopping there. Joseph Raz continues to press on the particularist's model of explanation. A sensible 'intelligibility principle' requires that there must be an explanation for the difference between a good and a bad action. If we agree that we must be able to cite a difference in pair-wise comparisons, we should agree that such a difference must be found more generally, and we recover pressure towards exceptionless principles. More specifically, on Dancy's view, a complete specification of an agent's reason for acting as he does can cite considerations that, in a different context, count as reasons against so acting. This means, Raz points out, that what determines the moral status of an action must extend beyond what the agent's reason for acting is. But this claim, Raz urges, 'drives a wedge' between the evaluative and the guiding functions of reasons, distorting what it is for an agent to be guided by reason in the first place. Frank Jackson, Philip Pettit and Michael Smith continue the sceptical press on particularism's notion of explanation, this time with a semantic argument. While particularism is compatible with the doctrine of moral supervenience (the doctrine, roughly, that any moral difference must be accompanied by some nonmoral difference), it must reject the idea that there are patterned nonmoral differences underlying attributions of moral properties. Yet this second claim, they urge, is essential to making sense of semantic competence with ethical concepts. The explanation of the consistency in our use of our evaluative concepts has to find pattern in the natural. By abandoning the commitment to pattern at the natural level, the particularist renders mysterious how we could learn or justify our use of moral concepts and terms. In 'Ethics as an Inexact Science, T. H. Irwin argues that a close examination of Aristotle's texts indicates that he cannot, as it is often thought, be pressed into service as an ally to particularists. For Aristotle, some moral generalizations are, in certain respects, normatively prior to particulars in explanation, justification, and knowledge. Looking closely at Aristotle's acknowledgement that morality includes 'usual' generalizations, Irwin argues that Aristotle takes some moral principles with exceptions to be natural norms, not mere statistical frequencies. Irwin goes on to argue that, as in science, these moral principles are 'usual' not because the exceptions cannot be specified, but because an exhaustive specification of them would be irrelevant to the normative function of the principles.
Introduction
ix
Next, in Chapters 6, 7, 8, and 9 comes a series of essays by theorists identified as broadly sympathetic to the particularist enterprise. Jonathan Dancy continues to advance the radical thesis that every consideration is capable of varied moral salience; his main concern, though, is not to defend this claim against possible exceptions but 'to break the stranglehold' of the generalist conceptions of how moral reasons must function. Emphasizing that particularism is an outcrop of holism, he argues that the latter allows us to understand why explanations of an action's moral status can be complete without guaranteeing the same result in another context. Here he defends and expands the scope of his project to defend holism in the realm of epistemic and practical reasons, and in the realm of values and choice. He argues, intriguingly, that the availability of another alternative may change not just an agent's bottom-line decision, but a prior ranking of other alternatives. Dancy also argues that such a view is still compatible with the possibility of full ordering of values. In Chapter 7, David Bakhurst argues that the contextualism set forth in Alasdair MacIntyre's After Virtue can be adjoined to Dancy's particularism to yield a more satisfying picture of the moral domain. A moral person must, as it were, have certain lingering commitments or concerns—she must set herself in favour of certain sorts of things and against others. Unless we can make sense of this, we can have no account of the structure of moral personality. To cash out these enduring concerns in terms of principles conceived as rules of thumb, as particularists sometimes try to, is, Bakhurst thinks, implausible. So he tries to show how a particularist can think of such concerns as being directed to certain morally significant features, and yet to do so in a way that does not contradict Dancy's idea that, in any particular case, the features in question may not be morally relevant per se (that is, they do not there contribute to the overall value of the case). Jay Garfield argues that if we take seriously certain broadly Aristotelian and Wittgensteinian lessons alluded to by John McDowell, we will see that particularism is not merely defensible, but superior as an account of moral epistemology and moral psychology. Exploring the issues through a sustained examination of Onora O'Neill's writing, he distinguishes two kinds of rules. Some rules are capable of relatively mechanical application. But other rules, like moral principles, require experience to learn, judgement to apply, and admit of ever increasing expertise. Garfield defends a Wittgensteinian account of why consistency need not be found at the natural level, and argues that, far from straining our understanding of the moral life, such a view provides a better explanation of moral motivation and competency.
x
Introduction
Lawrence Blum explores in detail one of the most important threads related to the debate over particularity and generality, namely, the role of partiality in the moral life. He applauds the renewed appreciation of partiality in moral literature, but finds that claims to accommodate partiality's importance often, on closer examination, still turn out to judge the moral life by impartiality's lights. In this chapter, Blum distinguishes amongst, and argues against, different versions of this move, discussing in detail one such prominent attempt by certain consequentialists. The final three chapters are by theorists who, identified in their writings as sympathetic to particularism's lessons, seek here to refine or relocate those lessons by reconsidering the proper roles of generality. Martha Nussbaum responds to the charge, common in some circles, that moral theory, especially in its Enlightenment versions, is needless and dangerous. She argues that such charges are misdirected. Invoking the Stoics' tripartite distinction amongst theories, rules, and concrete judgements, she argues that objections to moral theory are at best objections to the idea that such theory could reduce to a system of rules (a reduction, she adds, that no major historical figure has advanced). Rules of action, while useful, do have limitations; they admit, for instance, of exceptions, and they set aside concern over the psychology of those who act. But theory, with its explicitness, abstraction, and generality, is precisely the arena that supplements these limitations: for instance, by making perspicuous the 'point and purpose' of a given rule, it allows one to see where exceptions to it are warranted. Indeed, she argues, the real danger is presented by those who advocate the overthrow of theory; without abstraction and generality, we could not have made the strides we have in the battle against injustice. In 'Unprincipled Ethics, Piers Rawling joins with David McNaughton, one of the ground-breaking authors on particularism, to defend a brand of moderate particularism. They distinguish amongst different versions of `intuitionism' by the types of properties to which one might deny invariant moral import. They agree that nonmoral features carry variant moral relevance, since such features can enter moral principles only if they are understood as carrying evaluative riders (conditions that cannot be spelled out in purely nonmoral terms); but the same is not true of thick moral properties such as justice, which should be seen as carrying invariant moral valence. A more thoroughgoing particularism, according to which even thick moral properties are accorded variant moral valence, cannot happily explain the role that such concepts play in learning moral competency and justifying subsequent beliefs.
Introduction
xi
In the final chapter, Margaret Little defends the model backing moral particularism as distinctive and persuasive, but revisits the implications that the model carries. She argues that the doctrine is both more and less radical than many suppose; most pointedly, she argues that its implications for epistemology and explanation have been largely misunderstood by particularism's critics and supporters alike. Moral generalities play an indispensable role beyond important heuristic and pedagogic functions, for they are crucial to justifying moral beliefs and explaining actions' moral statuses. But particularists who believe otherwise are in the grip of a picture their own argument is designed to dispel. The target, she urges, has been misidentified: the model backing particularism issues in objections, not against moral generalizations, but against a certain picture of what those generalizations have to be like in order to do their work.
CONTRIBUTORS
is Professor of Philosophy at Queen's University at Kingston. He is the author of Consciousness and Revolution in Soviet Philosophy (Cambridge University Press, 1991) and co-editor (with Christine Sypnowich) of The Social Self (Sage, 1995). In addition, he has published articles on Russian thought, philosophical psychology, and ethics. He is currently at work on a book on culture and mind. LAWRENCE BLUM is Professor of Philosophy and Distinguished Professor of Liberal Arts and Education at the University of Massachusetts, Boston. He is the author of Friendship, Altruism, and Morality (Routledge, 1980), and Moral Perception and Particularity(Cambridge University Press, 1994), and the forthcoming 'I'm Not a Racist, But . . .': Race and Racism Through a Moral Lens. ROGER CRISP is Tutor in Philosophy, and Fellow of St Anne's College, Oxford. He is the author of Mill's Utilitarianism (Routledge, 1977), editor of Utilitas, and a member of the Analysis Committee. JONATHAN DANCY is Professor of Philosophy at the University of Reading, UK. His main interests lie in moral theory, especially that part of it that borders on epistemology. His first publications on particularism were two articles in Mind in 1981 and 1983. His main publication in ethics is Moral Reasons (Blackwell, 1993); Practical Reality is published by Oxford University Press in 2000. JAY L. GARFIELD is Professor of Philosophy at Smith College, Professor in the graduate faculty of Philosophy at the University of Massachusetts, and Director of the Hampshire/Five College Tibetan Studies in India programme. His books include Belief In Psychology, (MIT Press, 1998), Cognitive Science: An Introduction, (MIT Press, 1995), Western Idealism and its Critics, (Pyrrho Press, 1998), Fundamental Wisdom of the Middle Way: Nagarjuna's Mulamadhyamakakarika (OUP, 1995), and several anthologies on topics in the foundations of cognitive science and ethics. BRAD HOOKER works in the Philosophy Department at the University of Reading, UK. He is the author of Ideal Code, Real World: A Rule-consequentialist Theory of Morality (Clarendon Press, 2000). DAVID BAKHURST
Contributors
xiii
TERENCE IRWIN is Susan Linn Sage Professor of Philosophy and Humane Letters, Cornell University. He is the author of Plato's Gorgias (translation and notes) (Oxford University Press, 1979); Aristotle's First Principles (OUP, 1988); Classical Thought (OUP, 1989); Plato's Ethics (OUP, 1995); Oxford Reader in Classical Philosophy (OUP, 1999); and Aristotle's Nicomachean Ethics (translation and notes) (Hackett Publishing Co., 2nd edn., 2000). FRANK JACKSON is Professor of Philosophy in the Research School of Social Sciences, and Director of The Institute of Advanced Studies, Australian National University. He is the author of Perception (Cambridge University Press, 1977), Conditionals (Blackwell, 1987), The Philosophy of Mind and Cognition (with David Braddon-Mitchell) (Blackwell, 1996); and From Metaphysics to Ethics (Oxford, 1998). He has held a number of visiting positions, including Senior Humanities Council Fellow at Princeton, and John Locke Lecturer at Oxford. MARGARET OLIVIA LITTLE 15 Associate Professor of Philosophy at Georgetown University, and a Senior Research Scholar at the University's Kennedy Institute of Ethics. She is author of works in both metaethics and normative ethics. Her book entitled Abortion, Intimacy, and the Duty to Gestate is forthcoming with Oxford University Press. DAVID MCNAUGHTON is Professor of Philosophy at the University of Keele. He writes mainly in ethics and philosophy of religion. He is the author of Moral Vision (Blackwell, 1988), and President of the British Society for Ethical Theory. He and Piers Rawling are jointly writing a book on agent-relativity and deontological ethics. MARTHA NUSSBAUM is Ernst Freund Distinguished Service Professor of Law and Ethics at the University of Chicago. She is appointed in the Philosophy Department, Law School, Divinity School, and the College, is an Associate in Classics, a board member of the Center for Gender Studies, and an Affiliate of the Committee on Southern Asian Studies. PHILIP PETTIT is Professor of Social and Political Theory at the Australian National University and a regular visiting professor at Columbia University, New York. Among his recent books are The Common Mind: An Essay on Psychology, Society and Politics (Oxford University Press 1993, 1996); Republicanism: A Theory of Freedom and Government (Clarendon Press, 1997, 1999); and (with Marcia Baron and Michael Slote) Three Methods of Ethics: A Debate (Blackwell, 1997). PIERS RAWLING is an Associate Professor of Philosophy at the University of Missouri-St Louis. He writes in a wide variety of areas in
xiv
Contributors
philosophy, including decision theory, ethics, and philosophy of science. The essay 'Unprincipled Ethics' (in this collection) is the eighth in a series of articles that he and David McNaughton have jointly written on the themes of agent-relativity and deontological ethics. They are working on a book on these topics. JOSEPH RAZ is Professor of the Philosophy of Law, Oxford University, and Fellow of Balliol College, Oxford, and a regular visiting professor at Columbia University. Among his books are Practical Reason and Norms (2nd edn. 1990, reprinted by OUP, 1999); The Morality of Freedom (OUP, 1986); Ethics and the Public Domain (OUP, revd. edn. 1995), and Engaging Reason (OUP, 1999). MICHAEL SMITH 1S Professor of Philosophy at the Research School of Social Sciences, Institute of Advanced Studies, Australian National University. He is the author of The Moral Problem (Blackwell, 1994), and the editor of Meta-Ethics (Dartmouth, 1995).
1
Moral Particularism: Wrong and Bad Brad Hooker
In this chapter, I focus on the view I think is most often meant by the title `moral particularism. This view is held by some of the most morally admirable people I know. Yet, as I shall explain here, the view seems wrong. I shall also argue that, whether or not it is wrong, the view would be a bad one for society to accept. What is Moral Particularism?
To be distinctive, moral particularism has to stick to some thesis that other moral theories reject. What thesis is distinctive of particularism? There is a thesis distinctive of particularism, and I shall say what it is later. First, I want to focus on theses often misidentified as particularism. Consider the thesis that what is permissible is so situation-relative that the only predicate satisfied by all and only morally permissible actions is the predicate 'is morally permissible'. This thesis will immediately require qualification since all and only morally permissible actions must also satisfy equivalent predicates, such as 'is not wrong. (In the previous sentence, I assume that all morally right actions, even ones that are obligatory, are morally permissible.) But I shall henceforth ignore the clutter necessary to accommodate conceptual equivalents. Particularists most certainly do accept the thesis that the only predicate satisfied by all and only morally permissible actions is the predicate 'is For helpful discussion on the arguments here, I am grateful to Jamie Ball, John Bishop, Emma Borg, John Broome, Roger Crisp, Jonathan Dancy, John Gardner, Berys Gaut, Peter Goldie, James Griffin, John Heil, Dudley Knowles, Gerald Lang, Jimmy Lenman, Al Mele, Tim Mulgan, Richard Norman, Derek Parfit, Philip Percival, John Preston, Geoffrey SayreMcCord, John Skorupski, Tom Sorell, Philip Stratton-Lake, Christine Swanton, Elizabeth Telfer, Alan Thomas, Jay Wallace, Peter Vallentyne, James Williams, Martin Wilkenson, and Nick Zangwill.
2
Brad Hooker
morally permissible'. And certainly many generalists reject it. For example, maximizing act-consequentialists reject it. They hold instead that all (and only) morally permissible acts not only satisfy the predicate 'is morally permissible' but also the predicate 'would maximize the good' (or, alternatively, the expected good). Likewise, satisficing act-consequentialists hold that all and only morally permissible acts satisfy the predicate 'would produce "good enough" consequences, or, alternatively, 'has high enough expected value'. And different varieties of rule-consequentialists hold that all morally permissible acts are permitted by the set of rules with the greatest value or expected value. Meanwhile, Kantians think that all morally permissible acts satisfy the predicate 'passes the categorical-imperative test. And contractualists think that all morally permissible acts satisfy the predicate 'is allowed by rules no one could reasonably reject as the basis for informed unforced general agreement'. Finally, many who think of themselves as virtue ethicists hold both that there is a general feature that makes certain settled dispositions virtues, and that all morally permissible actions satisfy the predicate 'is what those virtues would allow one to do'. 1 So while particularists all accept the thesis that there is no predicate that all and only morally permissible acts satisfy in common except the predicate 'is morally permissible, many different kinds of generalists reject it. Then why can we not think that this is the thesis distinctive of particularism? The answer is that there are some generalists who agree with particularists about the thesis. One very familiar kind of generalism holds that morality is composed of an irreducible plurality of principles that do not come in a strict order of priority. 2 (Such theories can, and should, hold that there are some rough 1 I have in mind that virtue ethicists would offer some such account as that what makes a disposition a virtue is that it conduces to the flourishing of the individual or the species. This account of what makes something a virtue is perfectly general. And the virtue ethicist's account of permissible action is framed in terms of these general virtues. So, although some particularists say warm things about virtue ethics, the two theories as I understand them are incompatible. 2 W. D. Ross, The Right and the Good (Oxford: OUP, 1930), ch. 2; Isaiah Berlin, Four Essays on Liberty (Oxford: OUP, 1969), esp. Introduction and Essay 3; Donald Davidson, `How is Weakness of the Will Possible?', in J. Feinberg (ed.), Moral Concepts (Oxford: OUP, 1969), 93-113, at 105-106; J. 0. Urmson, 'A Defense of Intuitionism', Proceedings of the Aristotelian Society, 75 (1975), 111-19; Thomas Nagel, 'Fragmentation of Value', in his Mortal Questions (Cambridge: CUP), 128-41; Stuart Hampshire, Innocence and Experience (Harmondsworth: Penguin Books, 1992); Bernard Williams 'Conflicts of Value', in Alan Ryan (ed.), The Idea of Freedom: Essays in Honour of Isaiah Berlin (Oxford: OUP, 1979); Ethics and the Limits of Philosophy (Cambridge, Mass.: Harvard University Press, 1985), 93-119, 185-7; 'What Does Intuitionism Imply?', in his Making Sense of Humanity (Cambridge: CUP, 1995); James Rachels, The Elements of Moral Philosophy (New York:
Moral Particularism: Wrong and Bad
3
priorities, on which I shall say more later.) The pluralist generalists I have in mind point to the principle against physically harming others except when necessary to defend either others or ourselves from physical harm. And they point to principles against stealing or destroying others' property, breaking our promises, and lying. They also point to principles about giving special weight in our decisions (about the allocation of our own resources) to the welfare of those with whom we have special connections. Some principle about generally helping others also appears, as do principles about promoting justice, being grateful to one's benefactors, and making reparation to those one has wronged. 3 According to this familiar kind of generalism, whether an act is morally permissible depends on the interaction of all these principles. 4 Let me call this form of generalism Rossian generalism. Some other generalists will join with Rossians in endorsing the principles about not harming others, not stealing or destroying others' property, not breaking promises, being loyal, benefiting others, and so on. This is true of, for example, rule-consequentialists, Kantians, contractualists, and (I think) virtue ethicists. But these other generalists will oppose Rossian generalism concerning a deeper question. They oppose Rossian generalism over the question of whether there is some further principle underwriting the principles about not harming others, not stealing or destroying others' property, not breaking promises, being loyal, benefiting others, and so on.
Harper Collins, 1993), 114-16, 120-38, 180-93; Berys Gaut, 'Moral Pluralism', Philosophical Papers, 22 (1993), 17-40; Robert Frazier, 'Moral Relevance and Ceteris Paribus Principles', Ratio, 8 (1995), 113-27; D. D. Raphael, Moral Philosophy, 2nd edn. (Oxford: OUP, 1994); David McNaughton, 'An Unconnected Heap of Duties?', Philosophical Quarterly, 46 (1996), 433-47; 'Moral Intuitionism', in Hugh LaFollette (ed.), Blackwell Guide to Ethical Theory (Cambridge, Mass.: Blackwell Publishers, 2000); Philip StrattonLake, 'Can Hooker's Rule-consequentialist Principle Justify Rossian Prima Facie Duties?', Mind, 106 (1997), 751-8. See also Simon Blackburn's 'Securing the Nots' in W. SinnottArmstrong and M. Timmons (eds), Moral Knowledge? (New York: OUP, 1996), 82-100, especially 97-9; and his Ruling Passions (Oxford: Clarendon Press, 1998), 308-10. For an influential discussion of this sort of pluralism, see Rawls, A Theory of Justice (Cambridge, Mass.: Harvard University Press, 1971), 34 ff. 3 If a duty of self-improvement is added, this list looks much like Ross's (The Right and the Good, 21). I think a duty of self-improvement should not be added; there should instead be some qualification about how much one is required to sacrifice for others. See my `Intuitions and Moral Theorizing', in P. Stratton-Lake (ed.), Essays on Moral Intuitionism (OUP, 2001 forthcoming). 4 For important complexities about how these principles interact, see Shelly Kagan, 'The Additive Fallacy', Ethics, 99 (1988), 5-31.
4
Brad Hooker
Rossian generalism maintains that these various principles about harming others, stealing or destroying others' property, breaking one's promises, and so on cannot themselves be usefully conjoined together to make one long principle. If all the various principles came in a strict order of priority, then really there would be just one multi-faceted, hierarchical principle.5 It could take the following general form: Never do A; never do B unless necessary to avoid doing A; never do C unless necessary to avoid doing A or B; never do D unless necessary to avoid doing A, B, or C; never do E unless necessary to avoid doing A, B, C, or D; etc. If such a principle were correct, then every morally permissible act would satisfy a predicate other than the predicate 'being morally permissible. The further predicate all permissible acts would satisfy is the predicate 'does not offend against the hierarchical principle "Never do A; never do B unless necessary to avoid doing A; . . . " (Note that a strict hierarchy of duties would resolve all conflicts between different kinds of moral consideration.) In holding that general moral considerations do not come in a strict order of priority, Rossian generalists hold that none is necessarily always overriding. Rather, each is capable of being overridden by the others. In just this sense, general moral duties (general moral considerations) are in Ross's terminology 'prima facie'. 6 Perhaps a better term is pro tanto.7 The idea is that a duty or consideration is overridable, not that it appears at first glance and yet on closer inspection may prove to be an illusion. Rossian generalists do subscribe to the general principle that one should always do what the balance or mix of moral considerations demands in the case. But such a general principle is hardly unique to the kind of generalism now under discussion. Nor is such a principle informative. A principle about which moral considerations override the others would be informative. Thomas Nagel, 'The Fragmentation of Value', 131. Ross, The Right and the Good, 22, 34-5, Foundations of Ethics (Oxford: Clarendon Press, 1939), 79; Bernard Williams, 'A Critique of Utilitarianism', in J. J. C. Smart and B. Williams, Utilitarianism: For and Against (Cambridge: CUP, 1973), 77-150, at 90; Gaut, `Moral Pluralism', 35; Audi, `Intuitonism, Pluralism, and the Foundations of Ethics', 103; McNaughton, 'An Unconnected Heap of Duties?'; for discussions of the very notion of prima-facie duty, see Jonathan Dancy, Moral Reasons (Oxford: Blackwell Publishers, 1993), ch. 6; Frazier, 'Moral Relevance and Ceteris Paribus Principles'; Russ Shafer-Landau, 'Moral Rules', Ethics, 107 (1997), 584-611, 585-7. C. D. Broad, Five Types of Ethical Theory (London: Routledge, 1930), 282; Shelly Kagan, The Limits of Morality (Oxford: Clarendon Press, 1989), 17, esp. fn. 13; S. L. Hurley, Natural Reasons (New York: OUP, 1989), 130-5. 5
6
Moral Particularism: Wrong and Bad
5
But Rossian generalism denies that any general consideration that is informative must always be overriding in determining right from wrong. This may need qualification. For perhaps we can think of some general moral reason that overrides whenever it appears. Consider, for example, the strength of the moral reason to refuse to do any act that would eliminate forever all consciousness in the universe (except where this really is the only way to prevent an eternity of universal misery). I cannot see what could outweigh this reason. But let me set aside this special case. Even if no other general moral consideration is always overriding, some of these considerations may typically be stronger than others. I shall return to this point. Rossian generalism implies agents will need to weigh moral reasons against one another. Rossian generalism thus maintains that there is an ineliminable role for judgement in order to resolve some conflicts between moral considerations. There will also of course be questions of interpretation. Interpretation is needed sometimes to ascertain whether an act would constitute breaking a promise, or whether an act would constitute destroying property, or stealing, or lying, or whether an event would make someone worse off, or whether someone's connection to you entitles her to special weight in your practical reasoning. 8 In light of these points about Rossian generalism, consider this passage from Dancy: Particularism claims that generalism is the cause of many bad moral decisions, made in the ill-judged and unnecessary attempt to fit what we are to say here to what we have said on another occasion.... It is this sort of looking away that particularists see as the danger in generalism. Reasons function in new ways on new occasions, and if we don't recognize this fact and adapt our practice to it, we will make bad decisions. Generalism encourages a tendency not to look hard enough at the details of the case before one, quite apart from any over-simplistic tendency to rely on a few rules of dubious provenance. 9
But Rossian generalism hardly tells us that the details of cases are always unimportant. On the contrary, Rossian generalism holds that the details of the case can be crucial to the question of which generally important moral properties are instantiated. Furthermore, which of the instantiated properties is most important can depend on the details in the case. 8 Carritt, A Theory of Morals (London: OUP, 1930), 114, and Shafer-Landau, 'Moral Rules', 601, discuss this in terms of the need for judgement to apply even just one rule. See also Samuel Scheffler, Human Morality (New York: OUP, 1992), 43-51; and T. M. Scanlon, What We Owe Each Other (Cambridge, Mass.: Harvard University Press, 1998), 199, 225, 246, 299. 9 Dancy, Moral Reasons (Oxford: Blackwell Publishers, 1993), 64.
6
Brad Hooker
Dancy acknowledges this. 10 Dancy agrees with Rossian generalists that the details of the particular case determine which moral properties are instantiated and which of these are most important in the case. But Dancy goes further. He maintains that whether a given property even counts morally for or against an act instantiating it depends on the circumstances. Generalists hold that some properties, whenever they are instantiated, always count morally in favour of an action, and that other properties, whenever instantiated, always count morally against. Particularists hold that the very same properties may count morally in favour in some circumstances and against in other circumstances." The Form of Moral Argument A pervasive view is that moral argument essentially involves comparing cases in order to think about how a consideration would apply if circumstances were different in various ways. Suppose you are considering whether to do x to me. Would you think it morally permissible for me to do x to you? If not, what is the relevant difference between our cases? Admittedly, particularists accept that, if something is permissible in one case but not in another, then there must be some other difference between the cases. They do not deny that moral properties supervene on nonmoral properties. What they do deny, or at least what Dancy denies, is the attempt to determine a moral conclusion about one case by appeal to a conclusion about another case. He refers to this kind of moral reasoning as a 'switching argument'. He thinks switching arguments are unsound and dangerous. This is a revolutionary thought, for switching arguments are absolutely ubiquitous in ethics. In fact, if we want to change someone's mind about the moral relevance or force of some fact in a case under discussion, the normal way of trying to do so is to compare this case with others. But this form or argument depends on particularism's being false. As Simon Blackburn writes: In trying to discover what to do, we imagine different actions, and register their good and bad features. It is essential to this process that these features are reliably extracted from any contexts or total situations in which we have come across them, and carry some moral import when transplanted into the new hypothetical situation. . . . If these features lost their moral import just as soon as they were
10
Dancy, Moral Reasons, 56-7.
11 Dancy, Moral Reasons, 60.
Moral Particularism: Wrong and Bad
7
abstracted from other cases, in which they had been marinaded with others to give some holistic moral gestalt, then this process would be totally unjustified.' 2
So a standard form of moral reasoning relies on a view opposed to moral particularism. This fact in itself will be enough to make most of us reluctant to give up the view. Still, the fact that a great deal of our moral practice depends on a certain idea does not prove that the idea is right. We must therefore consider other arguments.
Counterexamples to Particularism?
Is particularism refuted by counterexamples—that is, by general properties that always count morally in favour of an action, or by general properties that always count morally against? Consider the property of producing pleasure. One familiar idea is that the presence of this property counts in favour of any act that has it. If this idea is right, we have at least one counterexample to particularism. However, particularists point out that, while the property of producing pleasure makes an act better in some circumstances, this property makes an act worse in other circumstances. That an act would give pleasure to the sadist is not merely an overridden positive feature of the act. Rather, sadistic pleasure actually makes the act morally worse than it would be if it didn't afford sadistic pleasure. As Dancy comments, the pleasure people get from watching hangings makes letting them watch morally worse. I do not see how particularism can win on this battlefield. Rossian generalists try to ascertain a generally good-making property. Particularists then acknowledge that the property mentioned does often make acts morally better. But particularists go on to point to a context in which the property described counts against rather than for an action. At this point, Rossian generalists have two obvious options: 3 One is to stick to the line that (for example) producing pleasure is always a pro tanto moral plus, even if the pleasure comes to a wicked and undeserving person. The other option is to start making distinctions. For example, there may be a distinction between benefit to a person and moral value. Thus sadistic pleasure might constitute some benefit to the sadist even if it has no moral value, indeed even if it has moral disvalue.
12 3
Blackburn, 'Securing the Nots', 97. Shafer-Landau, 'Moral Rules', 590, 593-4.
8
Brad Hooker
Many philosophers would say that not just the moral but also the selfinterested benefit of pleasure depends on its source. Pleasure based on the truth might constitute a greater self-interested benefit than does pleasure based on illusion. And pleasures accompanying more complex true beliefs or more complex achievements might constitute greater self-interested benefits than pleasures derived from simpler beliefs and simpler achievements. 1 4 We might nevertheless hold that even the most base or vindictive pleasure constitutes some benefit to the agent. This is why we think that wicked pleasure is an unjust benefit to the person who gets it. Even the wrong kind of pleasure constitutes some benefit to the agent. But the moral status of pleasure depends on what kind of pleasure it is. Nonsadistic pleasure is always a moral plus. Sadistic pleasure, even where a self-interested plus, is always a moral minus. To be sure, these moral pluses and minuses can be outweighed by other moral considerations. Thus, all things considered, an act can be impermissible although it gives someone nonsadistic pleasure. Likewise, all things considered, an act can be permissible although it gives someone sadistic pleasure. Nevertheless, the moral polarity of sadistic and of nonsadistic pleasures never changes. If (but not only if) we can run the above line, particularism is in trouble. For if we can run this line, then we can point to at least one general property—that is, 'would produce nonsadistic pleasure'—that always counts on the same moral side. When we point this out, we point to an informative general principle in normative ethics. Dancy himself obviously hates sadistic pleasure.' 5 It is tempting to hold that sadistic pleasure is never a moral plus. But that tempting thought might be too hasty. I do not pretend to confidence about this matter. What I am confident of is that generalists are right to say that at least all nonsadistic pleasure is a moral plus.
Other Counterexamples to Particularism
Giving others nonsadistic pleasure is not the only thing that always counts as a moral plus. The fact that an act would benefit others, even if not by bringing them nonsadistic pleasure, always counts as a moral reason to do 4 Mill, Utilitarianism, ch. 4; William Frankena, Ethics, 2nd edn. (Englewood Cliffs, NJ: Prentice Hall, 1973), 89-92. 15 See his Moral Reasons, 56, 61.
Moral Particularism: Wrong and Bad
9
it. This reason can be overridden by other moral reasons. But it is a moral reason always anyway. Likewise with promoting justice. How could the fact that an act would promote justice ever be anything less than a pro tanto moral plus? Again, justice may not be always overridingly important. But, where it is at stake, it always counts on the same side. Now consider promise breaking. Does the fact that some act constitutes my breaking a promise count always on the same moral side? Perhaps centuries ago promise breaking was considered always a moral evil (even if sometimes less of an evil than the alternatives). Our modern understanding of promising, however, usually takes certain kinds of promises to carry no moral force. For example, morality exerts no pressure on agents to keep immoral promises. 16 An immoral promise is a promise to offend against one's other obligations. So it would be a promise to physically harm someone, to steal, to destroy others' property, to break another promise to someone else, to promote injustice, or to ignore one's special responsibilities for those to whom one has special connections. Our modern understanding of promising also dismisses promises that were made under certain conditions. Promises obtained under coercion or deception are without force. As Judith Jarvis Thomson writes, 'anyone who thinks a word-giving whose source is coercion or fraud does nevertheless give a claim [a claim right to the person who was promised] is excessively 17 respectful of what goes on in a But what counts as coercion? We must be very careful not to count having been in a weak bargaining position as having been coerced. For if we did insist that binding exchanges of promises cannot be executed by parties in weak bargaining positions, then many mutually advantageous deals could not be struck, because the stronger party would see that the weaker party's promise would not be binding. To undermine the possibility of exchanging binding promises between unequal parties would be bad all round. Remember that sometimes someone in a weak bargaining position wants desperately to be able to exchange trusted promises with
6 For more careful discussions, see Henry Sidgwick, Methods of Ethics (London: Macmillan, 1st edn. 1876, 7th edn. 1907), 305, 308; Judith Jarvis Thomson, The Realm of Rights (Cambridge, Mass.: Harvard University Press, 1990), 313-16. 17 Thomson, Realm of Rights, 311. Cf. Sidgwick, Methods of Ethics, 305-6; Charles Fried, Contract as Promise (Cambridge, Mass.: Harvard University Press, 1981), esp. ch. 7; Scanlon, What We Owe Each Other, ch. 7. Fried's account of what counts as a coerced promise seems correct to me.
10
Brad Hooker
someone in a stronger bargaining position, which would be impossible unless the promises are both binding: 8 My claim here is merely that differential bargaining power between the parties need not prevent their exchanging morally binding promises. Unequal bargaining power does not entail coercion. I accept, however, that promises obtained by fraud or coercion have no moral force. Let me summarize the main points about promise keeping. We have to be careful to state our principle about promise keeping. And we must admit that promises are not always the most important moral consideration in play, not even if they are important promises: 9 Still, we can state a general principle about promising: the fact that an act would involve keeping a morally permissible promise that was elicited from you without coercion or deception always counts morally in favour of your doing the act. Now, consider the moral polarity of stealing. Clearly, stealing is a moral minus. Yet there can be circumstances in which it is nevertheless justified. Suppose stealing someone's change is the only way I can call an ambulance. I'm stealing the change, not borrowing it, because (let us suppose) I have no idea whose it is, and will never be able to find out. Although stealing in such circumstances is morally right, the stealing as such has not become morally neutral, much less morally positive. Rather, it is a moral minus, though heavily outweighed by the other considerations in play in this situation. Breaking promises to others and stealing or destroying their property are normally harmful to them, but not always. Furthermore, of course, there are many ways to harm others other than stealing from them or breaking promises to them. Hence the need for an independent principle against harming others. But is harming the guilty as part of justified punishment any moral minus at all? Is harming someone in self-defence any moral minus? What about in the defence of innocent others? What about harming someone when this is not part of one's ends or means? What about harming others with their consent? What about harming their interests simply by outperforming them in some morally permissible competition? As before, generalists have two obvious options here. One is to stick to the simpler line that, yes, harming others is always a pro tanto moral minus. In many cases, this minus is outweighed by opposing moral pluses. These 18 Here I am supposing that widespread acceptance of a morality should, among other things, enable mutually beneficial practices. That a morality should be evaluated in terms of how well it performs any particular role, however, is not universally accepted. Many particularists, for example, would reject this supposition. 19 I will discuss a case in point in the final section of this chapter.
Moral Particularism: Wrong and Bad
11
might be the pluses of channelling disapproval of and deterring crime, or of protecting the innocent, or of respecting others' autonomy, or of fostering the goods that come out of competition. The generalists' other option is to fine-tune their principle about harm. It might become 'there is a pro tanto duty not to harm others except in the course of justified punishment, or of protecting the innocent, or of respecting their informed wishes, or of pursuing morally legitimate competition.' Whichever option generalists take, they can think that their principle about harm gives them another weapon with which to beat particularism.
Particularism about the Content of Evaluative Concepts
Suppose generalists propose that the general property of producing innocent pleasure is always a moral plus. The particularist replies that we will not be able to set out in purely naturalistic terms what counts as innocent. Particularists could take the same line with respect to justice. They might say that we cannot specify in purely naturalistic terms what counts as just. The first point I want to make about this line of thought is that there certainly are at least some principles linking entirely natural properties to pro tanto duties. For example, that an act would bring about the involuntary death of a self-conscious human being who is not threatening others' physical security is always a moral minus. Perhaps we do not even need the qualification 'who is not threatening others' physical security'. For perhaps bringing about the involuntary death of a self-conscious human always has something morally against it, though this consideration is often outweighed when that person threatens others' physical security. Furthermore, I think we should hope to find some moral principles picking out natural properties. For to the extent that questions of justice and rights are not tied down by concrete terminology—indeed, by the use of terms with at least fairly clear naturalistic truth conditions—people can have difficulty forming stable expectations. I shall come back to this. Even if the extension of evaluative properties cannot be specified naturalistically, this hardly entails particularism. Not all generalists are naturalists. Take, as just one example, those consequentialists who think the good is not just pleasure but also significant achievement, important knowledge, and the appreciation of true beauty, and who admit that we cannot naturalistically define significant achievement or important knowledge or true beauty. Although such consequentialists might tie the right directly or
12
Brad Hooker
indirectly to the good, they do not think the good can be defined or described in purely naturalistic terms.
Examples of Essentially Particular Good Practical Reasons?
Suppose we accept that generalism succeeds in showing that at least some moral considerations always have the same polarity—that some things always count morally in favour and some other things always count morally against. But what generalism needs to show, according to Dancy, is that every moral consideration must retain its polarity wherever it appears. Why? Once we admit that some moral considerations can switch their polarity depending on the circumstances, we have no reason to think of any given consideration in front of us now that it operates in the same way in all cases. Dancy suggests that, if some moral reasons are not general, then, as far as the logic of reasons goes, any reason before us may be merely particular, that is, count in this case in a quite different way from in other cases. If a reason is really a particular one, we need not worry about how it would apply in other cases. But are there any essentially particular practical reasons? Consider nonmoral reasons: For instance, that there will be nobody much else there is sometimes a good reason for going there, and sometimes a very good reason for staying away. That one of the candidates wants the job very much indeed is sometimes a reason for giving it to her and sometimes a reason for doing the opposite. 2°
Very plausibly, all normative, or good, reasons for action are either moral or self-interested. 21 If this is correct, we should try to interpret the example of 20
Jonathan Dancy, 'The Particularist's Progress', in this volume, 132-3. See also Dancy's
Moral Reasons, 60-6. 21 The locus classicus for this view is of course the end of Sidgwick's Methods of Ethics. See also Hooker, 'Parfit's Arguments for the Present-aim Theory', Australasian Journal of Philosophy 70 (1992), 61-75; Roger Crisp, 'The Dualism of Practical Reason', Proceedings of the Aristotelian Society, 96 (1996), 53-73; Derek Parfit, 'Reasons and Values', Proceedings of the Aristotelian Society, suppl. vol. 71 (1997), 99-130. I myself think that there are a limited
number of kinds of self-interested benefit, or components of well-being, and a limited number of kinds of action-guiding moral considerations. But this is not to espouse a `checklist' theory of practical reasoning, since the agent obviously cannot run through the entire list before every decision. Even when the agent could run through the list, often this would be undesirable. To take but one example, it's best if I spontaneously take a strong interest in my children, i.e., without this interest being the result of self-interested or moral reasoning. And of course we all do have many spontaneous affections and enthusiasms. But when asked to provide normative reasons underwriting our behaviour and attitudes, we
Moral Particularism: Wrong and Bad
13
a nonmoral practical reason about the secluded spot as an example of an incomplete self-interested reason. 22 You have self-interested reasons to pursue pleasure and knowledge for yourself. Suppose going to the secluded spot in the bright sunshine will bring you pleasure. Suppose going there will also give you some time away from distractions so that you can think through some important questions. Suppose that thinking them through will increase your knowledge. If going to the secluded spot will increase either your pleasure or your knowledge or both, then you have self-interested reasons to go there. Now suppose going to the same secluded spot in the middle of the night is traumatizing. You have a good self-interested reason to avoid things that are traumatizing. Thus you have a self-interested reason not to go there at night. In these cases (as elsewhere) the operative reasons are general. That some act would benefit you is a reason to do it. That an act would harm you is a reason not to do it. Of course whether going to a secluded spot would benefit or harm you depends on further facts. But this hardly lends support to particularism. Turn now to Dancy's example about the job candidate. The fact that someone very much wants the job could be evidence that this person would work conscientiously (either out of gratitude for getting the job or at least out of a desire to keep it). It could also be evidence that the person is so ill-suited to employment that she can't find any other job. If a job should be done, then the fact that some candidate would work more conscientiously than others is always a reason to appoint that person, though of course this reason can be outweighed by other ones.
Dancy's Argument from Holism about Normative Reasons
Dancy's most general argument for particularism about reasons for action is that this is but an instance of a general holism about normative reasons. There are normative or good reasons for action, and normative or good reasons for belief. Dancy argues that all good reasons for belief are obviously holistic (context-dependent), and he supposes we have no powerful must turn to morality and self-interest. For powerfully argued views opposed to this, see Scanlon, What We Owe Each Other, 228-33; Raz, 'The Central Conflict: Morality and Selfinterest', in R. Crisp and B. Hooker (eds.) Well-Being and Morality: Essays in Honour of James Griffin (Oxford: Clarendon Press, 2000), 209-38. 22 Compare Shafer-Landau's discussion of handshaking (`Moral Rules', 593).
14
Brad Hooker
reason for believing that good reasons for action should be different, i.e., general. But if all good reasons for action are holistic, then they are particular in Dancy's sense. So Dancy's argument could be cast as: P1 All good reasons for belief are holistic—i.e., context-dependent, particular. P2 If all good reasons for belief are holistic, then all good reasons for action are too. Thus: All good reasons for action are holistic (context-dependent, particular). How are good reasons for belief supposed to be holistic? Under normal conditions, when I have the visual experience as of seeing red in front of me, this is a good reason for me to believe there is something red in front of me. But suppose I believe I have taken a drug that makes blue things appear red to me. Then my visual experience of red gives me a reason to think the thing in front of me is blue. I find the argument set out above utterly unpersuasive. Consider its first premise—that all good reasons for belief are particular. I'm not at all confident this is correct. Perhaps a full specification of our reason for belief about a thing's colour points not only to our visual experience but also to assumptions about context. Consider the case where I do have good reason to believe the thing is red. The reason here is that I have the visual experience of red and I assume I am in 'standard conditions'. I assume, that is, that I am in normal lighting, have no mind-altering drugs in me, and no naturally blown fuses in my head. Likewise, in other circumstances, my good reason to believe the thing is instead blue is that I have the experience as of seeing red and I believe I have taken a drug that makes me see red where there is really blue. If we believe that features of the context (or beliefs about them) should be included in the full specification, we can still believe that usually they do not need to be mentioned in everyday reference to reasons for belief. They normally do not need to be mentioned if they are so standard that we simply presume they are in place unless we are warned otherwise. The opposing view, which Dancy prominently favours, is that features of the context must be left outside the full specification of a reason for belief. They are to be relegated to 'enabling conditions' or 'defeating conditions'. I myself lean towards the view that these features of context should be included inside the full specification of the reason for belief. If this is right, then we could hold a unified theory of normative reasons, according to which all normative reasons are general.
Moral Particularism: Wrong and Bad
15
But I am much more confident that no argument starting from a premise about good reasons for belief could be part of a compelling argument for such a startling thesis as particularism about good reasons for action.23 Admittedly, an argument of that form could reasonably convince us of an answer to a question to which we had no default answer. But what Dancy tries to do is use particularism about reasons for belief as a premise in an argument for particularism about reasons for action. He does this despite the fact that particularism abut reasons for action is highly counterintuitive and thus has very widely been seen as dead in the water. 24 To make fly what seems highly counterintuitive we need an argument employing overwhelmingly plausible premises. This rules out any argument with the premise 'If all good reasons for belief are holistic, then all good reasons for action are too.' Particularists may accuse me of having begged the question here. They might say that the argument in my previous paragraph points to a presumption against particularism, when particularism is precisely the thing at issue. How can I reason fairly from a presumption against particularism, given that particularists start with a presumption in favour of it (to say the least)? In this section, I am not trying to persuade particularists to abandon their theory. This section has a different purpose. Its purpose is to point out to particularists that they shouldn't expect anyone initially inclined against particularism (which after all is the vast majority of ethicists) to be persuaded by the argument from holism. Particularism Versus the Value of Predictability I think particularists won't be able in the end to give an adequate account of moral education. But let me set that point aside to discuss a related point about grownup moral agents. 23 I made a similar point in 'Parfit's Arguments for the Present-aim Theory', 68-9. Incidentally, William Child contends that the best model we have for a particularist theory of what we have most reason to believe is John McDowell's particularism about what we have most reason to do. See Child's Causality, Interpretation, and the Mind (Oxford: Clarendon Press, 1994) 58-60. But if the argument for particularism about theoretical reason relies on particularism about practical reason, then the argument for particularism about practical reason cannot—except circularly—rely on particularism about theoretical reason. 24 Dancy himself acknowledges that western moral philosophy and practice has overwhelmingly been generalist.
16
Brad Hooker
One of the things a shared commitment to morality needs to do is provide people with some assurance that others won't attack them, rob from them, break promises to them, or lie to them. Providing people with such assurance is of course one of law's most important functions. Ideally, perhaps, people's moral commitments would be adequate to provide this assurance, without the reinforcement of legal sanctions. But realistically we recognize the need for legal sanctions to protect persons and property from others, and to enforce contracts. 25 Yet there are some things which both (a) we want people's moral commitments to ensure that they do, and (b) we do not want law to get involved with. 26 An example may be that it is desirable for morality to pressure people to keep their spoken promises to their spouses, but we don't want the law to poke its big nose into such matters. 27 Andevwhrlasoundetickso,wdpreaintlization of moral restrictions is clearly needed. For knowing that others have certain firm moral dispositions can give us added assurance about how they will behave. Now if shared commitment to morality should, among other things, create settled expectations about how others will behave, how does particularism look? Imagine we knew of other people only that they were committed moral particularists. This is all we know of them—the particularist content of their moral view and their strong moral commitment to live by it. Would we have enough confidence that they'd virtually never attack us, rob from us, break their promises to us, and so on? Some of my best friends think of themselves as moral particularists. These people are as dependable as anyone could reasonably desire. If one of them made a promise to me, I would certainly trust it. Experience has taught me these particular people are trustworthy. Is this empirical confirmation that moral particularism does provide enough assurance about how believers in it will behave? But what morality people espouse and even think they follow might not be the morality they really follow. (Many people who espouse kindness and sincerely think of themselves as kind are actually uncaring, vindictive, ruthless, and so on.) Since 'actions speak louder than words, if people who call themselves particularists act reliably and consistently in ways that 25 See H. L. A. Hart's classic discussion in his The Concept of Law (Oxford: Clarendon Press, 1961), 189-95, esp. 193-5. 26 'There is a limit to the amount of law enforcement which any society can afford, even when moral wrong has been done,' (ibid., 162). 27 Though where spouses go to the trouble to have written contracts with one another, the law should treat these contracts like any other.
Moral Particularism: Wrong and Bad
17
accord with Rossian generalism, I may certainly have good reason to think they won't attack me, steal from me, break promises to me, or lie to me. My experience with them gives me good grounds for predicting their future behaviour. In fact, I may mostly ignore their own reports of their moral beliefs. I may think that these people are sincerely describing what they think they believe, but that they are mistaken. To answer my question about particularism and reliable expectations, we need to consider not merely someone who thinks of himself or herself as a particularist. We need to consider someone whom others think of as consistent with his or her particularist ethics. Suppose Patty is such a person. All you know of her is that she really does live by her particularist beliefs. Now imagine that you can strike a deal with Patty. She asks you to help her get in her crop now in return for her promising to help you get yours in next month. Half her crop will spoil if you don't help her. This would drive her to bankruptcy. That is why she is willing to promise to help you later in return for your helping her now. Likewise, you must have help with your crop later if you are to avoid going bankrupt yourself. That is why she thinks you might be willing to accept the deal she proposes. 28 Suppose you have no direct or indirect experience of Patty. Nor do you have time to ask others how trustworthy she is. All you have to go on is her self-description as a particularist. If you had some means of forcing her to keep her side of the deal, then you wouldn't need to rely on her moral attitudes to make her do it. You might have the means to force her to keep her side if you could bring in legal sanctions—like suing her if she doesn't keep her side. Or you might be able to prove to the world she is unreliable if she doesn't keep her side, and this damage to her reputation would cost her more than keeping her side of her deal with you would cost her. But suppose neither the threat of legal sanctions nor the threat of ruining her reputation would be enough next month to get her to keep her side of the deal. That is, suppose legal enforcement mechanisms and reputational effects are for some reason ineffective here. Assume that the only thing that could possibly make her keep her promise is her moral 28 Hart identifies the 'division of labour and perennial need for co-operation' as the factors that make the exchange of binding promises 'necessary in social life'. 'Where altruism is not unlimited, a standing procedure providing for such self-binding operations is required in order to create a minimum form of confidence in the future behaviour of others, and to ensure the predictability necessary for co-operation. This is most obviously needed where what is to be exchanged or jointly planned are mutual services' (Concept of Law, 192-3).
18
Brad Hooker
outlook. Assume there is no doubt about the strength of her moral motivation. She is completely committed to behaving morally, according to her moral outlook. The question is only whether her moral outlook, if consistently particularist, gives you enough assurance now. As a particularist, Patty thinks that there are no considerations that always retain their moral polarity. She thinks a consideration (such as the fact that she promised to do something) might be a reason for keeping her side of the deal in one situation, but a reason against keeping it in another situation. So, will she think that having made a promise to you gives her any reason to do what she promised? Not necessarily. As a particularist, she can and might attach no weight whatsoever to the promise when the time comes to keep it. And, as a particularist, she can't point to any general considerations that mark off the situations in which a promise would be morally binding from the situations in which it wouldn't. As a particularist, she also thinks that any fact can be morally relevant, depending on the circumstances. So she thinks any fact would conceivably interfere with the moral status and force of the promise. If Patty would really live by such beliefs, how much could you trust her? We might think that someone would have to misunderstand the very nature of a promise in order to think that an informed uncoerced promise whose content is morally innocent could lack moral force. We might also think that an ineliminable part of being trustworthy is being disposed to attach weight to every one of one's promises as long as they were informed, uncoerced, and had morally innocent contents. If this claim about the concept 'trustworthy' is right, then a true particularist cannot meet the necessary conditions for being trustworthy. Rather than rely on this conceptual argument against trusting the particularist, I want to point to a practical argument. How much trust would you put in the particularist Patty in my example? Would you trust her enough to make the deal with her? If the answer is no, then both of you would be worse off than if both made the deal and then both kept their sides of it. But if you knew nothing of her except the information specified, you would have little assurance that you could predict how she would behave. If the example about promise keeping is correct, the point generalizes to other moral considerations. If agents accept that there are general pro tanto duties, they are not particularists. If they are particularists, they think that the fact that some adult human beings have neither killed nor threatened others nor asked to be killed themselves need not be any moral reason to
Moral Particularism: Wrong and Bad
19
avoid killing them. Particularists will believe that no moral reason for action always emerges from the fact that an informed uncoerced morally innocent promise has been made, or from the fact that some property is owned by others, or from the fact that a statement would be a lie, or from the fact that some act would cause suffering. How safe would you feel in a society of people who lived by these beliefs? Particularists might try to defend their view by pointing out that Rossian generalism is not absolutist. Absolutist generalism about promise breaking holds that promise breaking could absolutely never be morally permissible. Rossian generalism gives promises only pro tanto force. In other words, Rossian generalism holds that promise breaking is always a moral minus but can sometimes be permissible, or even morally required, if there are important moral reasons on the side of breaking the promise. Then particularists might claim that, in order to trust Rossian generalists, we would need to trust their exercise of moral judgement about when to keep a promise. If we can rely on Rossian generalists to exercise good moral judgement about when to keep promises (particularists might retort) then we can likewise rely on particularists. Let us compare our particularist Patty with a Rossian generalist, Gerry. Gerry believes that physically harming others is a serious moral minus, and that stealing or destroying others' property, promise breaking, and lying are moral minuses. He also believes that promoting justice, helping others, and expressing gratitude are moral pluses. But he believes each of these considerations can be overridden. Just like Patty, Gerry promises you that, if you help him get his crop in this week, he will help you with your next month. As is the case with respect to Patty, you cannot depend on either law enforcement or concern about reputation to get Gerry to keep a promise to you. As with Patty, the only thing that might make him keep his promise is morality. You know what morality he is committed to, including which general principles he subscribes to, but you have not had experience with him before and can't ask others how he behaves. In sum, as is the case with respect to the particularist Patty, the moral convictions of our generalist Gerry are the only thing that could induce him to keep his promise, and his morality is the only thing you know about him. You know Gerry subscribes to the general principle that promise breaking is always a moral minus, unless the promise is extracted by means of coercion or deliberate deception, or is itself an immoral promise. None of these general defeating conditions obtains in the case at hand. So Gerry would hold that breaking his promise to you is a moral minus.
20
Brad Hooker
Admittedly, Gerry does not think that breaking his promise to you would necessarily be, all things considered, wrong. He is not an absolutist about the wrongness of promise breaking. He admits that, in certain circumstances, he should and would break his promise to you even though your financial ruin would result. In particular, he would break his promise if necessary to save the life or limb of anyone for whom he has special responsibility (because of some very special connection he has with this person). 29 So Gerry would break his promise if necessary to save the life or limb of his parent, or family member, or friend. He would not break his promise to you, in order to work for Oxfam for the week (though this predictably would prevent more suffering and loss). For he believes the people whom Oxfam aims to rescue are not ones for whom he has any special responsibility. 3° So, to decide whether you should depend upon Gerry, what you'd need to know is the probability of his deciding he has to break his promise in order to protect the life or limb of someone for whom he has special responsibility. Now we should admit, I think, that it isn't exactly clear for whom Gerry thinks he has special responsibility. And even with respect to someone who falls squarely within the circle, it isn't exactly clear how high the risk to that person's life or limb would have to be in order to induce Gerry to abandon his promise for the sake of protecting this person. A onein-a-million risk is obviously too little. A one-in-two risk is obviously enough. But where is the threshold between too little risk to justify breaking the promise and enough risk to justify breaking it? For all that, you have vastly less to worry about with the generalist Gerry than with the particularist Patty. First, Gerry necessarily attaches some moral weight to the promise; Patty does not necessarily attach moral weight to the promise. Second, there are only a very limited range of facts that might interfere with Gerry's deciding to keep his promise. Admittedly, these facts may require interpretation and the exercise of judgement. Still, there are limits for Gerry. But for Patty, any fact can become pivotal to whether she will take her having promised as any moral reason at all for her to do what she promised. Given these points, and given the limited information you have about the Rossian generalist Gerry and the particularist Cf. Scanlon, What We Owe Each Other, 299. Some philosophers may think Gerry has his priorities wrong. Why should he keep his promise to you when only your business is at stake, and when he could instead devote the time to saving the very lives of others? Indeed, whenever we have the opportunity to help save lives at risk, how can morality permit us not to take this opportunity? I shall not take up this crucial issue here, because the issue is not one on which the generalism vs. particularism debate turns, or at least not in a way that favours particularism. 29 3()
Moral Particularism: Wrong and Bad
21
Patty, clearly you should think Gerry would be more likely to keep his promise than Patty would. Let me be clear that the argument above does not beg the question against particularism by assuming the particularist is going to make more moral mistakes than the Rossian generalist. My argument was that, whether or not particularism is likely to lead agents to make moral mistakes, the Rossian generalist seems in the circumstances more likely than the particularist to keep the promise. To see this, compare the absolutist about promise keeping with the Rossian generalist. The absolutist will be more trustworthy than the Rossian generalist. This is true even though (as most of us believe) absolutism is mistaken. Since we believe absolutism about promise keeping is mistaken, we must believe that an agent who always complies with it could sometimes act wrongly. Such a person will keep promises even when something else is more important. Such a person is maximally trustworthy but not morally right. So the plausibility of a moral theory and the trustworthiness of an agent following it may part company. I am not assuming that it is because particularism is mistaken that someone trying to follow it will be less trustworthy than someone following some other theory. My argument does not rely on the premise that particularism is mistaken. On the contrary, let us make the entirely non-question-begging assumption that Rossian generalism and particularism are initially equally plausible. My argument is that, given this non-questionbegging assumption, collective public commitment to Rossian generalism would lead to considerably more trust amongst strangers than would collective public commitment to particularism. Absolutism
vs. Rossian Generalism vs. Particularism 1 1 Absolutism is implausible. Assume both Rossian and particularist views are plausible. Absolutist agent more Rossian more trustworthy trustworthy than Rossian. than particularist. Implausibility of absoAs between vews iews equally lutism too great for the plausible in other respects, theory to be rescued by the the fact that one of them fact that absolutist agents enables trust counts heavily in its favour. would be maximally trustworthy.
v
22
Brad Hooker
The example about promising generalizes. You know of generalists that they take a limited number of certain general features to count morally in the same way every time they occur. You know of generalists that they believe that any of these features can be outweighed only by one or more of this limited number of features. Neither of these things is true of particularists. In so far as they reject general moral principles, particularists leave us unable to form confident expectations about what they will do. Let me address one final possible objection to my argument. This is the objection that the truth about morality is one thing, and what would result from our believing that others believe that truth is another matter. According to this objection, a view such as particularism might be right— even if the consequences of its acceptance, and of public awareness of its acceptance, would be bad. This objection seems to me an overgeneralization. Belief in the best moral view could have bad consequences because of interference from some evil demon. To take a simple example, the demon might rain misery on the world if this moral view is believed. Concerning this sort of special case, we might reasonably hold the correctness of the moral view immune from contamination by the bad consequences of believing it. But in the everyday world where demons aren't trying to bully people into rejecting some moral view, a moral view does seem unattractive if widespread awareness of its widespread acceptance would have very bad net effects on human well-being. And that is just what I think is the case with particularism. I might be accused of begging a question against particularism by appealing to a premise that particularists will not accept. This is the premise that the social internalization of morality has some purposes, a central one of which is to increase the probability that people will conform with certain mutually beneficial practices. This is a premise shared by many kinds of consequentialists, by contractualists, by many Kantians, and by many natural law theorists. Admittedly, some defenders of particularism will reject the premise—especially when they see where it leads. But this does not mean that my argument in this section is worthless. The overall plausibility of a moral view is seriously impaired if it denies that one of the points of morality is to increase the probability of conformity with certain mutually beneficial practices.
2
Particularizing Particularism Roger Crisp
Occasion by occasion, one knows what to do, if one does, not by applying universal principles, but by being a certain kind of person: one who sees situations in a certain distinctive way.' John McDowell The general is dark, uncommunicative, if it is not realized in a concrete image. . . . The particular is prior . . Martha Nussbaum . 2
Although particularism has come under a good deal of scrutiny in the last quarter of a century, particularizing it is not easy, since the arguments and positions advanced by those known as particularists vary considerably. In this essay, I shall discuss three different but related forms of particularism, concerning, respectively, rules, reasons, and motivation. My conclusion will be that particularism has fewer implications for normative ethics than have sometimes been claimed, 3 because its true forms are largely uncontroversial. I shall end with a brief defence of a form of Rossian generalism.4
I am grateful for comments on previous drafts to Jonathan Dancy, Francis Dunlop, James Griffin, Brad Hooker, Brian Klug, Andrew Mason, Andrew Moore, Mark Nelson, Thomas Norgaard, Derek Parfit, Scott Shuchart, and Philip Stratton-Lake. 1 J. McDowell, 'Virtue and Reason', in R. Crisp and M. Slote (eds.), Virtue Ethics (Oxford: OUP, 1997), 141-62, at 162. 2 M. Nussbaum, Love's Knowledge (New York: OUP, 1990), 95. 3 See e.g. McDowell, 'Virtue and Reason', 161-2; J. Dancy, Moral Reasons (Oxford: Blackwell, 1993), ix. 4 `Generalism' is now the standard term for the contrary to particularism.
24
Roger Crisp Particularism about Rules
It is hard to deny that morality has something to do with rules. Some moral codes are expressed in the form of rules, and moral education consists partly in instructing children to abide by certain rules. But before considering the details of particularism about rules, three questions. First, is particularism here a view concerning all rules, or some subset? This depends on the argument. Wittgensteinian arguments concerning rule-following, for example, are usually intended to apply to all rules. Most often, however, it is moral rules that concern particularists. 5 Secondly, are particularist claims intended as descriptive of how rules are in fact applied by moral agents, or as prescriptive recommendations of a certain ideal of moral agency? There is certainly no use of hard empirical data in particularist arguments. But I suspect that most particularists would claim to be pointing to certain limits on the actual application of rules in moral reasoning, and to be basing their recommendation of an ideal of moral agency on their appreciation of those limits. This brings me to my final question. Are moral rules to be given any role at all in the model of ideal moral agency? Sometimes it seems not. Dancy approvingly interprets McDowell's view of the virtuous person as follows: This virtuous person is not conceived of as someone equipped with a full list of moral principles and an ability correctly to subsume each new case under the right one. There is nothing that one brings to the new situation other than a contentless ability to discern what matters where it matters. 6 5
See e.g. D. McNaughton, Moral Vision (Oxford: Blackwell, 1988), ch. 13.
Dancy, Moral Reasons, 50. Because the virtuous person brings nothing to the new situation, Dancy's first sentence should perhaps be understood without 'full'. It is worth noting two things here. First, we should remember that McDowell himself sees his project as that of (approvingly) elucidating Aristotle, and it is clear that Aristotle has a place for moral rules, as I shall demonstrate below. Secondly, though McDowell does say certain things that might lead one to think that Dancy has indeed captured his understanding of Aristotle, he also says things tending to the opposite conclusion. For example: 'A virtuous person has a correct conception of what doing well is (in general, if you like), and applies it—puts it into practice—in "decision" and action on particular occasions ... Aristotle's scepticism about general truths in ethics implies that the content of this general conception cannot be definitively written down, in a shape suitable for deduction of particular practical conclusions.' (J. McDowell, 'Comments on "Some Rational Aspects of Incontinence", by T. H. Irwin', Southern Journal of Philosophy, suppl. 27 (1988), 89-102, at 93.) So one might read McDowell as allowing that the virtuous person does bring something with him to each new situation, namely, his general conception of eudaimonia (and this may be seen, perhaps, as including a list, albeit one less than full). McDowell's concern is to stress that, according to Aristotle, this conception cannot be definitively written down, and so what 6
Particularizing Particularism
25
Other particularists, however, speak rather of the posteriority of rules to, for example, sensitivity to particular cases.' Uncodifiability: Wittgenstein
The most common version of particularism about rules consists in the claim that morality is essentially uncodifiable, at least in the sense that moral rules are not sufficient on their own to provide moral guidance: As Aristotle consistently says, the best generalizations about how one should
behave hold only for the most part. If one attempted to reduce one's conception of what virtue requires to a set of rules, then, however subtle and thoughtful one was in drawing up the code, cases would inevitably turn up in which a mechanical application of the rules would strike one as wrong—and not necessarily because one had changed one's mind; rather, one's mind on the matter was not susceptible of capture in any universal formula. 8
Before I consider this Aristotelian position, I must address the argument McDowell offers after this passage against those whose 'deep-rooted prejudice about rationality' prevents their accepting the Aristotelian position. The prejudice is the view that, if morality were uncodifiable, no moral outlook could be rational, since rationality requires consistency, and this must itself be understood as guidance by an explicit universal principle. McDowell's argument against the prejudice is based on Wittgenstein's so-called 'rule-following considerations. McDowell uses Wittgenstein's example concerning the extending of a series of numbers to suggest that learning what to do in particular circumstances emerges not so much from the application of an objective rule, as from a sensitivity grounded in our shared 'whirl of organism'. 9 Rational consistency, that is to say, requires not codified principles, but the capacity to see how to go on. Removing a prejudice that constitutes merely an obstacle to accepting some argument need not, of course, provide any consideration in favour of D. J. Allan called a `rule/case' conception of practical reasoning cannot be ascribed to Aristotle. Nor need it be the case that the virtuous person can say nothing of his conception. McDowell allows that he can 'gesture' at it, rather as Aristotle does in his fairly detailed accounts of the characters of those who possess particular virtues. And the virtuous person can speak discursively in justification of his actions, whether he is looking forward or backward. But there is a point at which these justifications run out, and that is, as McDowell puts it, when he can say only, 'You have to see it.' More on this in my text below. See e.g. M. Nussbaum, Love's Knowledge, ix, 66, 70. 8 McDowell, 'Virtue and Reason', 148. McDowell makes reference in his footnotes to Aristotle, Nicomachean Ethics, 1.3 and V.10, esp. 1137b19-24. 9 McDowell, 'Virtue and Reason', 151.
26
Roger Crisp
accepting that argument, or any consideration that might itself play a part in that argument. This is worth stressing, since the rule-following considerations, or McDowell's use of them, are sometimes taken to provide positive support for an ethics of virtue. But since the rule-following considerations apply to all rules, they cannot in themselves be taken to favour any particular rule—such as the rule that one should be virtuous— over others. In connection with particularism, they leave untouched the question whether ethics is in fact uncodifiable in the ordinary sense. Some areas of human life do seem uncodifiable. Consider practical skills. A good carpenter will not only work without the constant application of rigid rules, but be able to respond correctly to unforeseen circumstances—that particular knot, in that particular plank—which could not have been predicted and captured in any prior rule. But adding two is not like this. If you ask me to continue a series of numbers by adding twenty-nine, the most effective way for me to do that is to apply the rule, 'Add twenty-nine. I do not just 'see' that 203 is 174 plus 29. Nor is any unpredictable mathematical circumstance in the ordinary sense going to cause me a problem. Now, I may start continuing my sequence 261, 319 . . . and so on, and assure you that I am continuing in the same way. At this point, you may begin to reflect on the rule-following considerations, and conclude that going on in the same way is ultimately determined only by our whirl of organism and not by following objective 'rules as rails. Or you may not. But whatever you conclude, your conclusion does not bear on the question whether, when it comes to the application of rules in the ordinary sense, morality is more like carpentry or more like mathematics. It may be suggested that a narrower version of the rule-following considerations, applied to ethics in particular, explains why morality is in fact more like carpentry than like mathematics. This version might focus on the way in which we learn ethical concepts in particular, noting how we gradually develop dispositions to see situations as falling under a particular concept, such as 'cruel, rather than learning strict intellectual rules of application—as perhaps happens in some disciplines such as mathematics. This version, however, since it rests on pointing out certain aspects of moral education stressed by Aristotle, brings us close to the Aristotelian conception of uncodifiability.
Particularizing Particularism
27
Uncodifiability: Aristotle
Let me assume that McDowell has indeed removed the prejudice in the way of accepting uncodifiability, and consider that notion on its own terms. Here, as we have seen, Aristotle rather than Wittgenstein is the influence. What we must recognize, according, for instance, to Martha Nussbaum, is that 'practical matters are mutable, or lacking in fixity'. i° Any system of rules can be based only on past experience, and the complexity and unpredictability of human affairs are such that circumstances will arise in which the rules are inappropriate for determining what we should do. The virtuous agent will need not only to be able to work out what to do on each occasion as it arises, but to be sensitive to an 'evolving and situation-relative list of virtues'." As an example, let me take one of Aristotle's own. Aristotle repeatedly warns us against wasting time on excessive detail in constructing our ethical theory. We must seek the proper degree of exactness, 'so that side-issues do not dominate the task in hand'. 12 But this is not to say that some generalizations cannot profitably be offered. One example might be: 'We should return a benefit instead of doing a favour for our companions." 3 This is usually true, but circumstances may arise in which the rule is mistaken. Imagine that someone has lent you money, and is now asking for it back. Ordinarily, of course, you should return it. But your father has been kidnapped by pirates, who are demanding a ransom, and in this case your obligation to your father overrides that to your creditor. You might modify your rule, of course. Indeed Aristotle would encourage you to do so: 'we should decide the issue as best we can'. 14 But what if one group of pirates has your father, a complete villain, on one ship, and on another ship a second group has your virtuous twin brother, and your creditor? You can ransom either your father, or your brother and your creditor. The rule has run out: rules about action are no more definite than their subject matter. It is important to note that, in this example, the rule concerns the repayment of debts, and not the virtue of friendship, the main subject of Aristotle's discussion. The rule is a practical generalization, an aid to 10 Nussbaum, Love's Knowledge, 71; cf. 157. See also McDowell, 'Virtue and Reason', 148 (cited in my text above); D. Wiggins, Needs, Values, Truth, 3rd edn. (Oxford: Clarendon Press, 1998), 229; McNaughton, Moral Vision, 193. " Nussbaum, Love's Knowledge, 71. ' 2 Aristotle, NE 1098a32-3. 13 Aristotle, NE 1164b31-2. 14 Aristotle, NE 1165a35. Aristotle means that we should seek to refine our practical generalizations to as great a degree as is reasonably possible, in the knowledge that properly deciding what to do in particular cases will require non-rule-governed perception.
28
Roger Crisp
doing what is always right, that is, to be virtuous, or to live in accordance with the virtue of friendship. This is indeed standardly the case with examples of mutability. The rule, 'Do not lie, is a generalization to assist in honesty, and it is easy to think of examples in which it gives the wrong answer. But it is not easy to think of cases in which one is required not to be honest.' 5 The claim that practical generalizations run out may be accepted by ethical theorists of any stripe. 16 Consider act utilitarianism, a common example of an allegedly non-particularist position.' 7 It may appear that any act utilitarian, because she advocates the principle of utility-maximization, will recommend that we consciously attempt at all times to maximize utility, and is thus more likely to run into problems with uncodifiability than an advocate of the virtues. But this appearance is deceptive, and will not survive even a cursory glance at the utilitarian tradition. To claim this about utilitarianism is only as plausible as to suggest that a virtue theorist must insist that we aim consciously to be virtuous in our actions. There are two issues in ethics that must be kept apart. 18 First, what is the truth about how we should live or act? The virtue theorist will claim that we should live virtuously, the utilitarian that we should live so as to maximize the overall balance of happiness over unhappiness. Secondly, how should we think, morally? Here, a virtue theorist is likely to put weight on the importance of virtuous dispositions, grounded in a solid moral education, and on the notion of a sensitivity to the morally salient features of situations. And something like this will probably be recommended by the utilitarian, and by any plausible moral theorist. Indeed a utilitarian who appreciates the utility of the virtues may advocate exactly the same ideal of ' 5 Dancy suggests that 'honesty can be important' (Moral Reasons, 68). But it is not clear whether he means: (a) honesty, if present, is important, but it may not be present; or (b) honesty, if present, may be important, or it may not. His particularism suggests (b), but as I shall demonstrate in the main text below the examples he uses to support his particularism do not concern virtues themselves. Someone who believes that the virtues conflict is likely to accept that a dishonest action may be required in certain circumstances. In section iv below, I shall argue that the virtues do not conflict. 16 W. D. Ross calls these generalizations axiomata media; see his Foundations of Ethics (Oxford: Clarendon Press, 1939), 190. 17 See e.g. Nussbaum, Love's Knowledge, 70. 18 See e.g. H. Sidgwick, The Methods of Ethics, 7th edn. (London: Macmillan, 1907), 413; R. E. Bales, 'Act-utilitarianism: Account of Right-making Characteristics or Decision-making Procedure?', American Philosophical Quarterly, 8 (1971), 257-65. The distinction is well known in moral philosophy, if often ignored by writers opposed to utilitarianism and indeed Kantianism. I mention it here because its implications for uncodifiability have not been appreciated.
Particularizing Particularism
29
moral agency as do McDowell and other particularists. 19 Utilitarians have standardly recommended 'secondary principles' 20 to assist agents in maximizing utility over their lives, and, because utilitarians may recommend virtue as a model for moral agency, there is nothing to stop any utilitarian agreeing with whatever plausible generalizations are suggested by any advocate of virtue ethics. Such a utilitarian may even allow, with Nussbaum, that virtues are situation-relative, and that the appropriateness of any virtue depends on its outcome in the particular social milieu in which it is encouraged and acted upon. Mutability is not the only notion used in arguments based on uncodifiability for particularism about rules. Two others, related to mutability, are indeterminacy and uniqueness. First, we must note that rules do not apply themselves, nor do they usually contain or come with a package determining when they are applicable.21 And of course any such package would require another package, and so on ad infinitum. One requires a sensitivity to the morally salient features of particular situations even to see which duties might be relevant within them. Secondly, even when one has recognized that a situation is one in which, say, justice or courage is called for, because of the complexity and unpredictability of human life discussed above, no rule can on its own, without some independent understanding of the nature of justice or courage, and some capacity to judge what they require in any particular case, satisfactorily guide action: [E] xcellent choice cannot be captured in general rules, because it is a matter of fitting one's choice to the complex requirements of a concrete situation, taking all of its contextual features into account. A rule, like a manual of humor, would both do too little and too much: too little, because most of what really counts is in the response to the concrete; and this would be omitted. Too much, because the rule would imply that it was itself normative for response (as a joke manual would ask you to tailor your wit to the formulae it contains), and this would impinge too much on the flexibility of good practice. 22
Both of these claims about the indeterminacy of rules are correct. In any circumstance, one will require judgement to discern not only the applicability of a certain rule, but also its requirement in that particular case. And 19 See my 'Utilitarianism and the Life of Virtue', Philosophical Quarterly, 42 (1992), 139-60, esp. 154-9. 20 J. S. Mill, Utilitarianism [1861]; ed. R. Crisp (Oxford: OUP, 1998), 2.24.37-8. 21 See Nussbaum, Love's Knowledge, 156; cf. McNaughton, Moral Vision, 197; Wiggins, Needs, Values, Truth, 236. 22 Nussbaum, Love's Knowledge, 71-2.
30
Roger Crisp
these points apply both to practical generalizations, and to ultimate moral rules or principles themselves—such as the principles that one should live virtuously, be honest, maximize the good, or whatever. But once again, because these are general points about all rules, they bear no special implications for normative ethics. A utilitarian, whether she advocates constant and conscious application of the utilitarian principle, or whether—as indeed all utilitarians have done—she recommends the use of some practical generalizations, may, indeed should, make room within her account for a conception of judgement that allows for the applying of moral rules, or the acting in (often non-deliberate) accordance with them. At this point, it is worth mentioning incommensurability. It is common to find a commitment to incommensurability among particularists, and arguments against the use of rule-based theories of practical reason based upon this commitment. Consider Wiggins: No theory, if it is to recapitulate or reconstruct practical reasoning even as well as mathematical logic recapitulates or reconstructs the actual experience of conducting or exploring deductive argument, can treat the concerns an agent brings to any situation as forming a closed, complete, consistent system. For it is of the essence of these concerns to make competing, inconsistent claims. (This is a mark not of our irrationality but of rationality in the face of the plurality of ends and the plurality of human goods.) 23
Wiggins' suggestion might be illustrated as follows. I have to decide how to spend my weekend. I bring to that decision at least two concerns: a concern for fulfilling professional responsibilities (I have a lecture to prepare), and a concern for enjoyment (I have been offered a sailing trip by a friend). Those concerns ground competing and inconsistent claims: if I write the lecture, I shall not sail, and vice versa. The concerns cannot be seen as forming a complete system, that is, a set of unambiguous principles or rules of priority, on the basis of which I may make my decision. A rule to rank work over play would of course be absurd. And there will be certain aspects of this particular choice situation which could not have been predicted, and the force of which could not have been caught under any serviceable set of principles: my sailing friend cannot go without me, she is feeling down, and the trip will take me to some new waters; another friend, to whom I have certain obligations, has asked me to do what I can to cheer the sailing friend up; and as the sailing friend is a philosopher, I will be able to make a start, anyway, on thinking about my lecture through talking to her about the issues. And so on. These salient features, of course, introduce 23
Wiggins, Needs, Values, Truth, 231; see also Nussbaum, Love's Knowledge, 66-7.
Particularizing Particularism
31
further concerns, such as a concern for loyalty to friends. Wiggins' point is not, I take it, that choices such as these cannot be made. That is, he is not making a claim of incomparability. 24 Rather, he is suggesting that deciding which concerns are relevant, how they are related, and how to judge between them cannot be done on the basis solely of any set of rules. Nonrule-governed judgement is also required. As I see it, this incommensurability argument for non-rule-governed or particularist decision-making is closely related to the Aristotelian arguments for uncodifiability. Indeed, in a footnote, Wiggins refers to Aristotle's explanation of 'why we cannot expect to lay down a decision procedure for adjudication in advance between claims, or for prior mediation'.25 Wiggins is not to be taken as suggesting that we must accept that the weight of concerns varies wildly, or for no good reason, between cases. His point is that how much various concerns matter will vary from case to case, and that, since these concerns are not reducible to some common value measurable on any single scale, judgement will be called for in any particular case. As I have suggested, this is a claim any moral theorist should accept, whether a monist or a pluralist. Now for uniqueness. Nussbaum approvingly interprets Aristotle as recognizing 'the ethical relevance of non-repeatable components' of a particular situation. 26 Example: A moderate diet for Milo the wrestler will be excessive for you or me, because of contingent facts about Milo's weight, intentions, and so on. Of course, we might say that we have here a universal principle with a single instance, but: [T]his would not be the sort of universal principle that would satisfy most devotees of principles, since it is rooted in the particulars of Milo's historical context in such a way that it could not have been anticipated with precision in advance; and perhaps (indeed, very likely) will be of no further use in the future. An ethical science with 'principles' this context-specific would have to have a vast and infinitely extensible series of principles; and this is not a science that will satisfy those who are looking for science. 27
Nussbaum does not name here any of the devotees of principles she has in mind.28 But there is no doubt that she is right, and that such a devotion 24 See Ruth Chang, 'Introduction' to her (ed.), Incommensurability, Incomparability, and Practical Reason (Cambridge, Mass.: Harvard University Press, 1997), 2. 25 Wiggins, Needs, Values, Truth, 232, n. 8. 26 Nussbaum, Love's Knowledge, 72; cf. Aristotle, NE 1106a36—b4. 27 Nussbaum, ibid., Cf. McNaughton, Moral Vision, 192, 197. 28 A failure to identify opponents by name is common among particularist writers. As I have already implied, this is at least sometimes because those opponents are made of
32
Roger Crisp
to principle would be quite absurd. Practical generalizations are useful, and `we should decide the issue as best we can'—but only to the point at which imprecision and unpredictability require reference to judgement, sensitivity, or practical wisdom. Once again, however, all this good sense is quite consistent with all the views of normative ethics presently under consideration in moral philosophy. To conclude this section. Wittgenstein's rule-following considerations have no implications concerning which normative ethical theory we should accept, or whether an ideal moral agent will attempt always, never, or sometimes to follow rules. Aristotelian uncodifiability will be accepted by most moral theorists, since it allows for the usefulness of practical generalizations while insisting on the importance of ethical judgement.
Particularism about Reasons
Holism and Reasons
The writer with whose name particularism has become most closely linked in recent years is Jonathan Dancy. Dancy distances himself from particularism about rules, understood as the view that 'no set of principles will succeed in generating answers to questions about what to do in particular cases'. 29 He sees that particularism about rules is consistent with a Rossian view according to which we have several prima facie duties, the job of deciding what is our duty sans phrase in any case being left to judgement, and Ross is, according to Dancy, 'the classic generalisf. 30 As we have seen, other views, including act utilitarianism, are consistent with particularism about rules. 'Particularism, if it is go beyond this, must give a stronger sense to the thought that the moral relevance of a property in a new case cannot be predicted from its relevance elsewhere: 3 ' Dancy offers us this stronger sense in his version of particularism about reasons: The leading thought behind particularism is ... that the behaviour of a reason . . . in a new case cannot be predicted from its behaviour elsewhere. The way in which the consideration functions here either will or at least may be affected by other considerations here present. So there is no ground for the hope that we can find straw. They would have to be quite unaware of manifest facts about the nature of human capacities, habits, instincts, dispositions, judgement, skills, education, and training. Could the views of anyone so ignorant, even if he or she existed, be worth discussing? 29 Dancy, Moral Reasons, 56. 30 Ibid. 31 Dancy, Moral Reasons, 57.
Particularizing Particularism
33
out here how that consideration functions in general . . . nor for the hope that we can move in any smooth way to how it will function in a different case. 32
Before moving on, I must clarify which species of reasons for action I take to be the main concern of particularism about reasons. I shall not be directly concerned with straightforwardly explanatory reasons, those which may concern human actions seen as events in the same class as the colliding of two billiard balls. Examples: What is the reason for the red ball's moving? It was hit by the white. What is the reason for her having broken the glass? The knocking at the door gave her a shock. Nor, at first anyway, shall I be concerned with motivating reasons. Examples: Why did she go to the fridge? She wanted a Coke, and believed she would find one there. 'Why did you go the fridge?' I'm thirsty.' Or perhaps: 'Why did you give her that pill?' I believe that people should, other things being equal, alleviate suffering when they can. Giving her the pill seemed the way to do that here.' Here, the motivating reason is understood as belief in a grounding reason (see below). 33 Motivating reasons are standardly distinguished from so-called justifying reasons. Example: 'What justification do you have for giving her that pill?' She was in pain.' But the distinction has to be made carefully. It may be tempting to think that her being in pain is what justified me in giving her the pill. But that is not so. Imagine that I had not known she was in pain, and had still given her the pill. Her being in pain would have been no justification of my action. Or imagine that she is not in fact in pain, but I had very strong reason to believe that she was. My action would then be justifiable, and it becomes clear that my justifying reason was my belief that she was in pain. 34 Behind the notion of justifying reasons and beliefs lies the way things are. Here we find the kind of reasons I shall primarily be discussing, and which I think are of most interest and importance in ethics. I shall call them 32 Dancy, Moral Reasons, 60. Derek Parfit pointed out to me that it is worth distinguishing the claim that any ultimate reason need not function in the same way in all cases from the claim that many reasons that have been taken to be ultimate are in fact not, and must be understood holistically. The second claim may be defended by particularists even if the first is refuted. (For more on the ultimate/non-ultimate distinction, see the main text below.) 33 It may be that a desire is also required for motivation. I shall return to this issue in the third section of this chapter. 34 It may be objected that it does not follow from the claim that beliefs sometimes justify that it is always beliefs that justify. Perhaps, when she is in pain and I am aware of it, it is her pain that justifies. The onus will then be on the objector to explain what is wrong with the more systematic and parsimonious account I am suggesting.
34
Roger Crisp
grounding reasons. In my example, the grounding reason for my giving her the pill, when she is in pain, is that doing so will decrease (her) suffering. It is not a belief I have about my action, but a property of that action. Our question concerns how such properties behave as reasons. Generalism and Rationality
Generalism, of course, is the denial of the leading thought behind particularism mentioned by Dancy above. Dancy claims that our pre-theoretical intuitions are particularist, and that the assumption of generalism that a property cannot make a difference in one place without making the same difference everywhere else is 'completely unmotivated'. 35 The first claim rests on an appeal to certain examples, which I shall discuss shortly. But the second needs addressing now. As I see it, the generalistic assumption, far from being unmotivated, underlies not only the rationality of ethics, but— on one plausible conception of it—rationality, and its exercise in enquiry, as a whole. 36 Consider, again, causal explanation. You may be happy with my explanation of the red billiard ball's moving by reference to its being hit by the white. But now it happens that the white hits the red, and the red fails to move, since it is made of lead. That will lead you to return to the original case: a fuller explanation of the movement will be that the white hits the red, and that the red is sufficiently light to be moved by the white. Now imagine a third case, in which the white and the red are both light in weight, the white hits the red, but the red fails to move. The search will be on to find some further difference between the third case and the first two, and on discovering that difference one's understanding will be further advanced, and one's explanatory principle more genera1. 37 It will not do for ss
Dancy, Moral Reasons, 62,104. Cf. McNaughton, Moral Vision, 191. It has often been claimed that ethics should not be understood on the scientific model. Now if ethics is not to be irrational, it must seek general principles of ultimate justification just as science seeks general principles of explanation. But, as we have already seen, it need not seek principles of sufficient precision to guide action in every case, without the use of judgement. Likewise, a comprehensive science need not supply precise principles of prediction (explanation may be post hoc, in the terms of some general theory already accepted), and a philosophy of science should not forget the importance of human capacities for judgement in its account of how explanations function in practice. 37 It will of course be more detailed, but generality—in this sense—is not the same as imprecision. One of the disadvantages of failing to respect the distinction between universality and generality, explained so clearly by Hare, Freedom and Reason (Oxford: Clarendon Press, 1963), 38-40, is that one tends to think of general principles as non-specific. 36
Particularizing Particularism
35
a rational enquirer to accept that it is just a brute fact that in some collisions the second ball moves, and in others it does not. To accept such ideas is deeply irrational, and would undermine the philosophical credentials of ethical particularism were it found to be implicit within it. 38 Dancy is not unaware of these facts about explanation. 39 But he does not accept the implication that the regulative ideal in seeking ever more general explanations is a principled theory which will explain any event. There seem to be two related problems for Dancy here. First, the physical sciences are based on the assumption that such a theory should be available, and there is no conclusive evidence yet to suggest that this assumption is mistaken. Certainly such theories are themselves stated in generalist terms. Consider Newtonian mechanics. 4° This theory can provide an explanation of any event in terms of the principle F= ma. When some mass accelerates, there must be some balance of forces that produces it. This principle is in a sense an ultimate explanatory reason, analogous to the ultimate grounding reasons of which we seek an understanding in moral philosophy. The second problem for Dancy is the lack of parsimony in his account, or, to put it another way, a slippery slope. Dancy allows that when some cause is sufficient for an effect in one situation, and it fails to bring about that effect in some other situation, ' [o] f course ... there will be some explanation of it'.41 But once one is constructing general explanations, there seems no good reason to think they should suddenly stop at some point, where we can only shrug our shoulders. Dancy appears to find it obvious that shoulder-shrugging is out when we initially compare the two cases. 38 Note that I do not wish to deny that what appear to be ultimate reason-giving properties in a certain case may not function with the same force when combined with other properties. But different functioning is a sign that one's ethical theory is not yet complete. G. E. Moore's account of ethics may serve as an example. Moore believed that properties can combine in organic wholes of greater value than the combined value of their constituents taken individually. Consciousness of a beautiful object has greater value than the sum of the value of the object, unseen, and the value of conscious experience in isolation from the beautiful object (Principia Ethica (Cambridge: CUP, 1903), 28). But, having discovered this, we may conclude that consciousness of a beautiful object will have the same value in different circumstances, unless we are able to think of some further organic whole in which its contributory value may be less. And if we believe, with Moore, that rightness consists in the maximization of good, we may believe that consciousness of a beautiful object will make the same contribution to rightness wherever it occurs. Our aim, then, is to locate reason-giving properties of which we can safely assume the 'additive assumption' to hold true; cf. S. Kagan, 'The Additive Fallacy', Ethics, 99 (1988), 5-31. I am more optimistic than is Kagan about the possibility of success in such an enterprise (see Kagan, 'The Additive Fallacy', 21, n. 9). 39 Dancy, Moral Reasons, 24. 4 () I owe this example, and the analogy based on it, to Scott Shuchart. 41 Dancy, Moral Reasons, 24.
36
Roger Crisp
But if it is out at this point, why not everywhere? And if it is not out at any level, why should I not just shrug my shoulders in the initial comparison? 42 (Ishalretunomsidcungpartlsmboivtion, on page 44 ff.) Ultimate Reasons
The first and main support Dancy seeks to provide for 'holism' in the theory of reasons (the 'leading thought behind particularism') comes in the form of examples: 43 Case 1: My having borrowed a book from you is a reason for my returning it to you. Case 2: My having borrowed a book from you is not a reason for returning it to you, since it turns out that you have stolen it from the library. Contraband Case 1: That my claim will be a lie counts as a reason against making it. Case 2: We are playing 'Contraband', in which the aim is to smuggle goods past a customs officer. That my claim will be a lie counts in favour of it. Traditions Case 1: That we did this last time is a reason for doing the same this time. Case 2: That we did this last time is a reason for doing something different this time. The Book
42 There is an analogy here, pointed out to me by Mark Nelson, with Gilbert Harman's view that what appears to be simple enumerative induction is best understood as inference to the best explanation (G. Harman, 'Inference to the Best Explanation', Philosophical Review, 74 (1965), 88-95). For example, the fact that the first ninety-nine swans you see are white does not give you reason to think that the one-hundredth swan will be white unless it also follows from the best explanation of the phenomenon that the one-hundredth swan will be white (you have not, for example, been told that on this farm Old Macdonald has ninety-nine white swans and one black one). In Dancy's cases, he expects some explanation of why a property functions in one way in one situation, and another way in another, and that expectation itself rests on the idea that, ultimately, things do not just happen. A real particularist would not have that expectation. There is a further point here: if particularism is true, we should perhaps be a little surprised that certain considerations dominate our ethical thinking, whereas others do not. An extreme particularist can offer no account of why, for example, causing serious harm to people seems to matter so often, and why its being Tuesday when the harm is caused never matters. (I owe this point also to Mark Nelson.) (Dancy allows for the possibility of what he calls a 'default' position, according to which killing, for example, is always wrong in the absence of justification. To that extent, then, his particularism is not extreme. See Dancy (Moral Reasons, 230).) 43 Dancy, Moral Reasons, 60-2.
Particularizing Particularism
37
Case 1: That an action is fun is a reason for doing it. Case 2: The fun experienced by hunters is a reason against hunting. 44 Pain Case 1: That my action will knowingly cause pain counts as a reason against it. Case 2: That my action will knowingly cause pain counts as a reason for it, since the sufferer deserves pain. The Journal An author's having published two papers on a topic in a journal counts both for and against publishing a third. 45 Illegality My action's being against the law is both a reason against doing it and a reason for doing it, since this sort of behaviour should not be regulated by law. Hunting
All of these examples have a common factor. To demonstrate it, I must draw a distinction between ultimate and non-ultimate reasons. 46 Take first The Book. The fact that my having borrowed a book from you can both sometimes count and sometimes not count in favour of my returning the book to you suggests that we are not dealing here with an ultimate reason, that is, a reason that we can rest satisfied with as grounding the actions in question. To rest satisfied here would be like accepting it as a brute fact that sometimes one billiard ball moves another, and sometimes it does not. Dancy's example is reminiscent of that used by Socrates in the first book of the Republic to show that justice cannot be identified with returning what one has borrowed. 47 If one has borrowed a machete from someone who subsequently goes off her head, it would not be just to return it. One of Socrates's points here is that an action's being just is an ultimate reason in favour of it. If I can demonstrate the justice of my action, you cannot go on to ask whether I have any ground for doing it, for justice is such a ground. Thus, in Dancy's example, I have a reason to return the book to you in the ordinary case because it would be just; but it would be unjust to return a stolen book to the person who stole it. Justice, then, is an ultimate reason. The fact that I have borrowed the book from you, in an ordinary case, can be described as a reason for returning it to you; but this is only because that is what constitutes justice in this particular case. As Dancy 44 This case is similar to the 'worm-crushing' case discussed by Dancy (Moral Reasons, 56), in which my enjoying treading on a worm counts as a reason against my doing it. See also McNaughton's circus example (Moral Vision, 193). 45 This case is similar to Traditions, except that the same property counts both in favour of and against the same action in a single case. 46 It is consistent with what I say below to claim that non-ultimate-reason statements are incomplete versions of ultimate-reason statements. 47 Plato, Republic, 331c1-9.
Roger Crisp
38
suggests, and as anyone would presumably admit, it will not function as a reason in certain other cases. A similar account is available for Dancy's other examples. In ordinary cases, it is dishonest to lie, and so I have a reason not to do so. A claim's being dishonest always provides a reason against making it. But there is nothing dishonest about playing Contraband. To think that there is is to misunderstand honesty in the way that William Brown does in Richmal Crompton's William the Good, insulting several of his parents' friends by being utterly truthful, 48 or in the way that Cephalus in the Republic misunderstands justice. Traditions may be understood without reference to any virtue, though for the sake of argument let me refer to prudence. In an ordinary case, it may be that doing the same this time as we did last time will save both of us time. It may sound grand, but it is nevertheless the truth, that our ultimate reason for doing the same thing as we did last time is that our lives will go better for us if we do. This, in other words, is what prudence requires. But maybe we are beginning to become bored by always doing things in the same way, in which case a change would be pleasurable. Increasing the balance of pleasure over pain in one's life, in an ordinary case in which the pleasure is not tainted (more on this below), is also prudent. Two of Dancy's cases concern pleasure and pain. In an ordinary case, the fact that some activity will be fun counts in favour of it, and it would be prudent to engage in it. Pursuing that fun is what prudence, in those cases, consists in. But sometimes pursuing fun involves cruelty, and we then have at the very least a reason against such pursuit (its being cruel), and perhaps no reason of prudence to pursue it (it may be claimed that some pleasures are not worth having, or even that they are worth not having, or that prudence, as a virtue, requires us to restrict ourselves to morally acceptable options). That deals with Hunting. Pain is perhaps best understood as involving both benevolence and justice, respectively in each case. Not causing pointless pain is benevolent, or perhaps rather not malevolent. Imposing a deserved punishment, however, is just, and not malevolent. The Journal is another case which may be understood without reference to the virtues, but let me again refer to benevolence and assume that the object of an editor is to promote the understanding and the enjoyment of her readers. Those who have already learned from and enjoyed the author's first two papers will be able to find further understanding and pleasure in reading the third. But those who have not found what they wanted in those first two articles are unlikely to do so in the third, and may be able to find 48
Being 'brutally' honest is not brutal. Likewise, being cruel to be kind is not cruel.
Particularizing Particularism
39
it in some other paper instead. Justice, perhaps, might count against a third publication in this way, and also if one thinks that someone so far unpublished in the journal should be given a crack of the whip. Illegality can again be understood with reference to justice. Breaking the law may be said to instantiate one kind of injustice (legal?), campaigning through disobedience for its reform to instantiate another (social?), or at least to demonstrate a virtuous concern for such justice. As far as I can see, the above analyses are fairly uncontroversial, though of course I am ready to admit that there may be better ones available. They all rest on Aristotle's point that virtue consists in getting it right.49 It is not enough to tell someone to be prepared to give away their money, or not to be angry. One must give away one's money at the right time, to the right people, for the right end, and so on, and likewise with anger and the other emotions. Each virtue can be seen as concerned with some central sphere of any human life—anger, fear, handing over money—and virtue consists in feeling and acting rightly in that sphere (one vice, or rather set of vices, will consist in feeling and acting wrongly in that sphere, and another in failing to feel and act rightly). We can now see that the fault in particularism about reasons is analogous to one of those in particularism about rules. Some versions of particularism about rules focus excessively on the media axiomata, the practical generalizations with exceptions which Aristotle and others recommend as helpful to the moral agent, as if these are the only ethical rules we have, and as if they are the property of the particularist. Since these rules are mere generalizations, they will often be inapplicable, and may even lead to error. But they are likely to be helpful, and we have seen that even an arch generalist, such as an act utilitarian, may recommend them. In a similar way, we find particularism about reasons focuses on non-ultimate reasons. The act utilitarian will be able to come up with all sorts of ways in which returning borrowed items tends, in ordinary cases, to promote the balance of pleasure over pain. But she will also be able to explain why, with the same goal in mind, theft should not be condoned. So even an act utilitarian can accept that a non-ultimate reason functions in favour of an action in one case, and against it in another. I suggest that particularists about reasons have a view of ethics that is quite the same, as far as reasons are concerned, as those philosophers commonly thought to be generalists. I hope that my analysis of Dancy's examples in terms of virtue demonstrates this. Dancy, along with other alleged particularists, believes in the virtues as constituting an important, perhaps 49
Aristotle, NE 1106b16-24.
40
Roger Crisp
the whole of, morality. 50 These particularists accept that an action's being just, kind, honest, generous or whatever always counts in favour of it, and an action's being unjust, unkind or cruel, dishonest, mean or whatever always counts against it. In other words, particularism about non-ultimate reasons will be accepted by all, while particularism about ultimate reasons is accepted by no one. 51 Universalizability
As Dancy notes, It] he rejection of universalizability is part and parcel of the rejection of generalism in the theory of reasons'. 52 According to Dancy's understanding of the universalizability thesis, if some action is judged to be right, then any relevantly similar action must likewise be judged to be right. It is clear how denying universalizability is likely to commit one to particularism about reasons, so let me briefly consider whether it should indeed be denied. Wiggins, influenced by Peter Winch, entertains seriously the prospect of its denial. 53 Following Winch, he employs the example of Captain Vere's decision to hang Billy Budd for killing Claggart, in the story by Herman Melville. Winch suggested that, though he himself could not have acted as did the captain, `Vere did what was for him the right thing to do: 54 50 See Dancy, Moral Reasons, 221-4; McDowell, 'Virtue and Reason', 141; Nussbaum, Love's Knowledge, ch. 2; McNaughton, Moral Vision, 116; Aristotle, NE, 1098a16-18. 51 Thomas Norgaard has suggested, plausibly, that the debate concerning particularism about reasons may be related to that concerning ethical naturalism. Non-naturalists may be impressed by the 'shapelessness' of any set of natural properties said to be identical to or to constitute some moral property. In the same way, particularists are impressed by the fact that some consideration plays one role here, another there. In both cases the solution is to postulate moral properties which always speak in favour, or always against, actions. Dancy does not directly discuss ethical naturalism (though cf. his 'In Defense of Thick Concepts', in P. A. French et al. (eds.), Midwest Studies in Philosophy, 20 (1995), Moral Concepts), but it may worry him that the argument for non-naturalism based on shapelessness results in generalism. It may also be suggested that the main claim of particularists is that principles guiding action via the identification of natural properties, such as pleasantness, are not available. This seems an empirical question, the answer to which depends contingently on which is the correct moral theory (as well as on a good account of naturalness). Certainly some utilitarians—those who advocate an objective list of well-being, for example—are unlikely to accept it. 52 Dancy, Moral Reasons, 57. It is of course possible that there may be other versions of the universalizability thesis which make it consistent with particularism, but I shall not consider them here. 53 Wiggins, Needs, Values, Truth, 166-73. 54 Cited in Wiggins, Needs, Values, Truth, 167. Wiggins himself is more ready to criticize Vere's decision; ibid., 171-3.
Particularizing Particularism
41
The natural claim to make in response to Winch, if we are not to accept relativism, is that if Vere's action was right 'for him, then it must have been right because of some property of him such that a universalizable claim could be made on the basis of that property. Winch allows that Vere, in his moral thinking, may be discovering some property of himself: [I] t seems ... that what one finds out is something about oneself, rather than anything one can speak of as holding universally ... What a man finds out about himself is something that can be expressed only in terms of the moral ideas by consideration of which he arrives at his decision. 55
But the question is whether these moral ideas themselves may be universalizable, and it is hard to see how they could not be, unless, that is, Vere is permitted to claim that his very identity is morally relevant. Winch suggests that the sheer fact that one of the agents is prepared to give precedence to one moral idea, such as, say, clemency, while the other is prepared to prioritize some other idea, such as justice, may be significant: [But] if such dispositions as this have to be taken into account in applying the notion of 'exactly the same circumstances', surely the last vestige of logical force is removed from the universalizability thesis. 56
Here Winch has a point: allowing such dispositions into the circumstances does indeed diminish the force (though not the logical force) of any substantive universalizability thesis. But it does not show it to be mistaken, which is what is required for particularism to follow. In other words, we may allow the following: [I] t matters in a special way . . . what Vere makes of the considerations in front of him—how things strike him. For while many ways of being struck by the situation are no doubt excluded as morally misguided . . . what we always may lack reason to assert (if we follow Winch's doctrine of the primacy of the agent's perspective . . .) is that there is just one way it must strike him if he is sincere. 57
On this view, if we accept that Vere's decision was right, then we are committed to the claim that the same action of any other agent in a relevantly similar situation, who is struck in the same way by the considerations confronting her, is also right, as well as to the claim that the same action of an agent who was struck differently, in otherwise relevantly similar circumstances, may have been wrong. Nevertheless, there does seem something odd about including the way a person views a situation as part of the Cited in Wiggins, Needs, Values, Truth, 168. Cited in Wiggins, Needs, Values, Truth, 169. 57 Wiggins, Needs, Values, Truth, 170-1. 55
56
42
Roger Crisp
account of the morally relevant features of that situation. This is borne out by the phenomenology of moral judgement itself: Vere would not have felt that he was engaged in an enterprise of self-discovery. What is striking about the cases used in arguments against universalizability is the fine balance of considerations for and against each action. (Another common example is Sartre's famous case of the student who must decide whether to look after his sick mother, or leave her to fight for the Free French.) If we wish to avoid allowing the agent's moral dispositions to be part of the morally relevant circumstances, we may claim that in these cases the considerations are in fact (roughly) equally balanced, so that Vere is in fact permitted to go either way. Particularism about reasons implies the falsity of the universalizability thesis. Since, even if it is not practically important, that thesis seems plausible, we have here a further argument against particularism.
Particularism about Motivation
Dancy believes that his account of motivation, since it is structurally analogous to his account of reasons, provides support for particularism. I shall argue, of course, that because it is structurally analogous it fails for the same sorts of reasons. But it must be discussed briefly, in case it be thought that the account provides independent support for particularism about reasons. 58 Dancy suggests that beliefs alone can motivate. 59 Imagine that I come across a friend in tears, and acquire the belief that she is miserable and requires comfort. That is one representation. Another representation is of my friend cheered, having been comforted by me. I recognize, in other words, that comforted by me she would obtain what she needs and be happy again. That gap between how the world will, or might, be, and how it is, leads me to act, and my being so motivated is what constitutes desiring. It is not the case that what motivates me is some independent desire to be a good friend, to comfort others, or whatever, since what motivates me are my beliefs, and they do so in their own right. But can we not imagine a case in which I have the two representations, and yet am not motivated, a case, perhaps, of moral weakness? Dancy 58 This section is a development of material in my review article on Dancy's book, `Motivation, Universality, and the Good', Ratio, 6 (1993), 181-90. 59 Dancy, Moral Reasons, chs.1-2. Since writing the book, Dancy has changed his view, thinking that neither beliefs nor desires constitute motivational reasons.
Particularizing Particularism
43
agrees that we can, but denies that this requires him to introduce an independent desire to explain why I act in the case in which I do. He draws an ingenious distinction between distinctions: first, we can distinguish essentially from contingently motivating states; secondly, we can distinguish states that motivate in their own right from those that do not. In the case of my friend, Dancy will argue, my representations motivate in their own right, but not essentially. Here, then, we find the particularism in Dancy's view of motivation. Just as a consideration may be a grounding reason in one case and not in another, so a representation, or a pair of representations, may motivate in their own right in one case and not in another. But just as we should not be satisfied with particularistic groundings for actions, so we should not be satisfied with particularistic explanations of motivation. It is tempting, in other words, to include the presence of a sufficiently strong will as part of the account of my being motivated. 60 As we have seen, Dancy is not unaware of the availability of this move. He is not ready just to shrug his shoulders when faced by cases in which the same representations motivate in their own right and fail to do so: 'The general line is that the ability of a consideration to motivate can be affected by background conditions which are not themselves motivators.' 61 Dancy relies on a distinction similar to that between a cause and a background causal field or set of causal conditions.62 'Why did the fuel explode at t?"Because a lighted match was dropped into it.' Here the lighted match functions as cause, but it can so function only, for instance, in the presence of oxygen, or in the absence of God's extinguishing all lighted matches before t. What falls on either side of the distinction between cause and causal field, however, is dependent on the case in question, and the knowledge of the person seeking understanding of the cause. There is usually oxygen around, so when I ask about the explosion I assume its presence. Likewise, to my knowledge God was not expected to put all matches out before t. But imagine a world in which oxygen has become scarce, and is available only in canisters. Then some reference would be required to the presence of oxygen in the explanation of the explosion. And were we to have been expecting God to interfere, reference to his failure may also be relevant in a causal explanation. Causes explain. A full explanation—that is, an explanation not relative to the interests of any enquirer or to any set of assumptions60 I am not ruling out the possibility of belief's being able to motivate in its own right, merely requiring that any account of motivation not be particularist. 61 Dancy, Moral Reasons, 24. 62 See J. L. Mackie, The Cement of the Universe (Oxford: Clarendon Press, 1974), 34-5.
Roger Crisp
44
will make reference both to the presence of oxygen and to the absence of God's interference. The same is true of action: the strength of my will is part of the explanation of my comforting my friend. It may be said that this somewhat Millian account is likely to result in the collapse of the distinction between cause and causal field, and that this consequence is not tolerable. But this is not so, since in most if not all cases the cause of some event may be picked out as the cause of that event within some field or other. What is important to note is that the causal field itself has a role to play. As Dancy himself said in the quotation above, the causal field can affect the ability of a consideration to motivate—and affecting is a causal notion if anything is. Dancy's argument here suffers from an equivocation on two kinds of reason: reasons that explain, and reasons that an agent takes to justify her action: I might buy an ice-cream because I expect to enjoy it. But if I were to accept that ice-creams damage one's teeth, I would not buy one. This should not be taken to show that among my reasons for buying the ice-cream was the fact that I did not believe that it would damage my teeth. 63
In this case 'my reason' for acting is the not the same as 'the reason why' I act. When I offer my reason, I shall present the consideration that, as far as I can see, led me to act in this particular case: the prospective enjoyment. But if you are seeking a full explanation of the action, you might wish to know why I—someone usually careful of my health—was so keen on eating something that you know to be damaging to teeth. And here one of the reasons for my buying the ice-cream is indeed the fact that I do not believe that it will damage my teeth. So we can conclude that causes are not to be understood in a particularist fashion, and that any argument for a particularist account of reasons resting on an analogy with particularist accounts of motivation will fail. A full explanation of an event will consist of a full account of the cause of that event. And such a cause will be sufficient to bring about an event of the same kind in any other case.
In Defence of Rossian Generalism
Let me assume that the ethical view gestured at above, and attributed by me to particularists, is correct, that is, that being moral consists in living or acting virtuously. In one sense, then, there is a single ethical principle: act vir63
Dancy, Moral Reasons, 24.
Particularizing Particularism
45
tuously, or rather, perhaps, be virtuous or live virtuously. But this apparent simplicity, of course, hides a plurality of principles: act justly (be just), act kindly (be kind), act courageously (be courageous), and so on. On the most plausible account of the virtues, they cannot conflict. 64 Justice,forxampldnvecoitflwhkndes.Coict is ruled out by the idea that what is virtuous in any situation is what the virtuous person would do, that person being able to do only one thing. 65 But consider the following case. You are invigilating a test in a school classroom. You notice that one of the pupils taking the test is cheating, through consulting a cribsheet. She is clearly behaving dishonestly and unfairly, taking advantage of the other pupils' honesty. As an invigilator, it may be said, your duty is to confront her. But you know that this particular student is usually quite conscientious, and that her parents are in the process of breaking up. It is more than likely that this is what lies behind her having failed to prepare properly for the test. Surely, it may asked, in this case it would be unkind to follow the requirements of justice? So here we appear to have a case of conflict between kindness and justice. This interpretation of the case lacks appreciation of the notion that the virtuous person sets the standard for what is right, and that the virtuous person, because she possesses all the virtues, will do and feel what is right in those spheres of human life that are governed by the virtues. If the right thing to do in this case is to keep quiet, then it would be wrong to speak out. Indeed speaking out, because it would be going wrong within the sphere governed by justice, would be a kind of injustice. But this is not to say that we should not make room for conflict here, of a kind which will issue in deliberation itself. The considerations that favour speaking out are real enough, and they do indeed conflict with those favouring keeping quiet. We can capture their force using counterfactuals. Had the girl not been facing problems at home, justice would have required you to speak out. 66 64 I include prudence within my list of virtues. The prudent person will know what weight to attach to her own interests vis-a-vis those of others. There is, then, (at least) a dualism of practical reason, but it is not irresoluble. See my 'The Dualism of Practical Reason', Proceedings of the Aristotelian Society, 96 (1996), 53-73. 65 To be more precise, one should do what the virtuous person would advise, since there are some situations the virtuous person would never allow herself to get into. One might also allow for a range of options as acceptable to the virtuous person. 66 Why should we not say that the virtues do conflict, and that the virtuous person is the one who can judge how to resolve those conflicts? An action's having a virtuous quality is plausibly said to make it good, in that respect. The opposite goes, of course, for vicious qualities. If we allow a conflict of the virtues, we shall be forced to accept that there can be 'laudable injustice', or indeed reprehensible kindness. See Mill, Utilitarianism, 5.37.
46
Roger Crisp
Our duties to be kind or just, then, may be described as what Ross calls `prima facie duties, in the following sense. 67 We have an overriding duty to do what is required by virtue in any situation. But we have a prima facie duty to do what kindness would require in the absence of countervailing considerations, what justice would require in the absence of countervailing considerations, and so on. And these considerations do provide real reasons for acting in cases of conflict. 68 This version of the view that there are prima facie duties is generalist: if the considerations in this case are such that justice would count in favour of a certain action, then justice will count in favour of any such action in such circumstances. Indeed, it seems that any view of ethics that includes non-particularism about reasons can be seen as a theory of prima facie duties in this sense. Although, as Dancy himself notes, such a form of generalism escapes many of his strictures, he does provide an argument against it as stated by Ross. 69 The argument focuses on Ross's definition of a prima facie duty: I suggest 'prima facie duty' or 'conditional duty' as a brief way of referring to the characteristic (quite distinct from that of being a duty proper) which an act has, in virtue of being of a certain kind (e.g. the keeping of a promise), of being an act which would be a duty proper if it were not at the same time of another kind which is morally significant. Whether an act is a duty proper or actual duty depends on all the morally significant kinds it is an instance of. 7° Ross recognizes that it may be objected to his view that it offers no principle to work out what our actual duty is in any situation. 71 But he rightly points out that no rival (pluralistic) theory is in any better position. What does Dancy find to object to in it? His central point is the following: Ross's definition [cited above] tells us only what effect 'being a prima facie duty in virtue of property F' has when F is the only morally relevant property the action 67 See W. D. Ross, The Right and the Good (Oxford: Clarendon Press, 1930), 18-19; Foundations of Ethics, 84. In the preface to The Right and the Good, ix, Ross points out his debt to Prichard. See H. A. Prichard, Moral Obligation (Oxford: Clarendon Press, 1949), 21. 68 They are not, that is to say, 'silenced', as McDowell puts it (`The Role of Eudaimonia in Aristotle's Ethics', in A. Rorty (ed.), Essays on Aristotle's Ethics (Berkeley: University of California Press, 1980), 359-76, at 369-70). I would myself prefer to avoid the language of duty altogether, speaking only of reasons. Thus: I have a (prima facie) reason to promote the welfare of all, and a reason to promote my own welfare. These spheres are the concern of benevolence and prudence respectively. In conflicts between my welfare and that of others, when the welfare of all provides stronger reason, the benevolent course of action will be to act on that reason, and it will not be prudent to further my own interest, though I may have a reason to do so grounded on the furthering of my self-interest. Dancy, Moral Reasons, 97-8. 70 Ross, The Right and the Good, 19-20. 71 Ross, The Right and the Good, 23.
Particularizing Particularism
47
has. It has nothing to say at all about what it is like to 'be a prima facie duty in virtue of property F' in any other situation. 72
Ross's definition, however, appears to tell us exactly what Dancy claims it does not. According to Ross's definition, if I have made some promise, and I can keep it only by (for example) leaving someone lying ill on the pavement, we can understand my prima facie duty to keep promises as being a duty that is such that, were this person not lying on the pavement, it would my duty proper to keep it. Rather than say nothing about situations in which F is not the only morally relevant property, Ross explains how F and G function in cases in which they conflict by reference to cases in which they function alone. There seems nothing more mysterious here than the distinction between what one has reason to do, and what one has reason to do overall. And this, I suggest, is equivalent to the distinction between ultimate reasons, and one's strongest ultimate reason. 73 To conclude. Normative ethics may be seen as the search for the correct principles of prima facie obligation, and an account of our duty sans phrase and how we are to decide what this is in our everyday lives. The Rossian generalist schema is such that many views in normative ethics—including, for example, Rossian pluralism, virtue ethics, and act or rule utilitararianism—may be captured within it. Any such view is likely to give some role to rules in its account of ideal moral agency, but also to allow that judgement is required for the application of any rule and for those occasions on which rules run out. The principles arrived at will be universalizable, since they will describe ultimate grounding reasons. In other words, any ultimate reason that counts in favour of any action counts in favour of any action in which it is instantiated. The central question in ethics is what those ultimate grounding reasons are, and that question is left largely untouched by the debates over particularism.
Dancy, Moral Reasons, 97. It is true that organic wholes may complicate the story, in the sense that the moral weight of various considerations may vary according to the presence or absence of other considerations. But if there are organic wholes, the implication for the Rossian view is that one should speak of their having prima facie weight as wholes, and of their constituents as individually having certain weights in the absence of the other constituents. 72 73
3 The Truth in Particularism Joseph Raz
When reflecting on the way we come to do what we do, or to refrain from what we refrain from doing, it is natural to be torn between a particularist and a generalist tendency. We tend towards a generalist view when we feel that at least sometimes we are guided by general precepts and are inclined to believe that it is always possible to be so guided. 'I must do this'—I may say—`My son is relying on me: Or, 'I cannot do this. It would be cheating.' I am, and am conscious of being, guided in my decision by general precepts: 'One should not cheat: `One should not let one's children down.' These principles may be too simple. They may have to be refined and qualified to be exceptionlessly true. But surely such refinements are forthcoming if we are minded to look for them. How else can we tell what to do and what not to do? But sometimes such reflection seems out of place. I do what I do because I know (or believe I know) that that is what should be done in the circumstances. I see a toddler about to step into the road and I stop him. I have no general principle in mind. Indeed, the moment I think of any general principle examples which refute it come to mind (the adult accompanying the toddler is keeping a close eye on him, etc.). No generalization free from exceptions, however complex, seems possible. Moreover, I do not and should not consult any such generalization before acting. I spot a mistake in a student's essay and I remark on it. I do not think: 'I should correct all the mistakes I spot, and therefore, having spotted this mistake, I should correct it.' It would be absurd to let such a generalization mediate between my perception of the situation and my action, or intention to act. In any case I do not believe in this generalization, nor in any other. Sometimes it I am grateful to Martin Stone, Ben Zipursky, John Hyman, and Anthony Kenny for comments which helped me improve this chapter. This chapter first appeared in my Engaging Reason (Clarendon Press, 1999).
The Truth in Particularism
49
would be better not to correct this or that mistake. To correct it would be to appear unduly severe, or critical, would be beyond the student's ability to understand, or to remember, etc. There is no exceptionless generalization about when to correct mistakes which I believe. These are but examples of some of the kinds of contexts in which it would be natural to be drawn in a generalist or a particularist direction.' But where do these directions lead? Do they lead to two theoretical positions about the nature of reasons or of practical deliberation, between which one should adjudicate? Or do they remain, as above, vague generalizations each possibly expressing some partial truth? Would the appearance of conflict be dissolved once we carefully distinguish the question of how we (properly) deliberate about what to do from the question of the nature of the considerations which may be relevant to establish what we should or should not have done? My aim is not typological and I will not essay a characterization of 'ethical particularism. My aim is to examine some of the thoughts which may tempt in one direction or another. In particular I will suggest that Dancy and Winch help us locate two problems in understanding reasons for actions, which I will describe, and to which I will offer a solution. The first section examines the case for a generalist position as I understand it, and believe it to be true. Whatever is true in particularist claims has to be compatible with it. The following sections examine three particularist tendencies: the case against principles (section 2), Dancy's thesis that what is a reason in one situation may be no reason in another (sections 3 and 4), and the argument, deriving from Winch, about the difference between the first- and the third-person perspective (section 5).
1. The Generalist Tendency: Intelligibility and Supervenience
Throughout the discussion I will hold to two considerations which any account has to accommodate. First, the (evaluative) properties of actions 2 whicmaketgodrb,ihwngadterfoswhic 1 I tried to describe these examples as they may be described by people who are not committed to any theoretical account of these cases. They are meant to provide a nontheoretical way of locating the whereabouts of the issues to be considered. I suspect that the expression 'ethical particularism' is used to refer to a whole family of views. I will continue to talk vaguely of 'particularist directions' to avoid singling out any one of them, or the implication that they constitute a family united by important theoretical common features. 2 Including the properties of the context of the actions and of their relations to other acts or events.
50
Joseph Raz
we have reason to perform or to avoid, serve both to judge the wisdom, morality, desirability, etc., of actions, and to guide agents in forming intentions, and in deciding what to do. I will refer to them as the evaluative and the guiding function of reasons and of evaluative properties. 3 I will assume that while possibly the two come apart in some cases, in general reasons and evaluative properties cannot serve one role unless they can serve the other. Second, the domain of evaluative properties and of reasons is intelligible. I can only give a very vague description of this point. Some of its more precise implications will emerge below. The intelligibility of the domain of value means that nothing in it is as it is just because that is how it is; there is nothing `arbitrary' 4 in the domain of value. It is, after all, the domain of reason(s). There is an explanation for everything, an explanation for why what is good is good, what is bad is bad, etc. 5 The intelligibility of value exerts a strong pressure towards some sort of a generalist view. In particular, it implies that regarding any two situations such that some evaluative concept applies to one of them but not to the other there is some further difference between them which can be helpful in explaining why. How is this point to be understood? One suggestion is that it reflects the supervenience of the evaluative on the non-evaluative. In any case, it may be argued, an examination of such supervenience claims holds the key to an understanding of the debate between generalists and particularists. I doubt, however, whether there is much we can learn from any general supervenience thesis we know of today. Before we proceed with the debate about particularism I will deviate to explain the irrelevance of supervenience to that debate. 3 Some evaluative properties figure not so much in reasons for or against performing actions but in reasons for or against performing them in one manner or another: if you are to do A do so skilfully, courageously, etc. Some evaluative properties provide what I called `conditional reasons', i.e. reasons which presuppose other reasons: if you have reason to sit down choose a comfortable chair to sit on. The evaluative properties which feature in reasons for action are varied. They may show the action to be sensible, sensibly cautious, wise, generous, just, right, and so on. To simplify I will refer to reasons as showing actions to be good or right, and will disregard the wide range of dimensions in which they can be good. This may give a regrettably artificial air to some of the points below, but I believe that it does not affect any of the arguments. 4 This is a rhetorical way of making the point. But be warned that it begs the question: something is arbitrary if it flouts reasons, or is brought about in disregard for reasons, where reasons apply. If a domain is not intelligible through and through then those aspects of it which escape reason are not arbitrary. Some people assume the intelligibility of the evaluative domain while denying that it follows that everything in it can be explained. If they are right, and something more will be said on the issue later on, that is enough for my purposes, provided that the dependence of intelligibility on conceptual thought is preserved.
The Truth in Particularism
51
To begin we should note that, as is generally agreed, there is no reason to think that the evaluative predicates of English (or of any other natural language) supervene on its non-evaluative predicates. Nor is there any reason to think that the evaluative concepts available to English speakers (or to members of any other group) today supervene on non-evaluative concepts available to them. 6 In fact my doubt is more general. It concerns the truth of global evaluative supervenience theses in general.' Applied to predicates the general evaluative supervenience thesis says that evaluative predicates (of a given language at a given time) supervene on (its) non-evaluative ones if and only if necessarily8 for any two situations, if there is an evaluative predicate which applies to one and not to the other then there is a non-evaluative predicate which applies to one and not to the other. An analogous thesis asserts the supervenience of evaluative concepts on nonevaluative ones. Which predicates belong to a language at any given time is a contingent matter, a product of its historical development to that time, and so is the number and identity of the concepts available to any person, or any cultural group. Hence if predicate or concept supervenience obtains at any given time that would be a surprising contingent matter, of no philosophical interest, unless there is some factor or mechanism which makes it necessary9 that it should exist. An argument which, if sound, would establish somel° form of evaluative supervenience would show that one cannot master evaluative predicates or concepts, and cannot have the ability to use them correctly, unless one has command of non-evaluative predicates or concepts such that one's skill in correctly applying the evaluative predicates or concepts consists in establishing which of the non-evaluative ones apply and concluding on that basis whether the relevant evaluative predicates or concepts apply. This, however, is not so. We commonly learn to apply evaluative predicates and 6 This is no surprise to philosophers upholding the supervenience of the evaluative on the non-evaluative. As will be seen in what follows their theses concern the supervenience of evaluative properties on non-evaluative ones. Global because it refers to the supervenience of all evaluative predicates, concepts, or properties, on all the non-evaluative ones; specific supervenience theses relate to the supervenience of one evaluative predicate, concept, or property, or of a class of them, on some non-evaluative ones. 8 A weaker form of supervenience will be confined to how things are rather than how they are necessarily. As our interest is in a claim that supervenience is of the essence of the evaluative the stronger form is relevant here. 9 Even natural necessity may endow supervenience with a philosophical significance. 10 Directly the argument shows the correctness of some specific, rather than global, supervenience theses. I will suggest below that in general explanatory needs call for specific rather than global supervenience theses, if any.
52
Joseph Raz
concepts by example, and through their association with other predicates or concepts, mostly with other evaluative predicates and concepts. Examining people's command of such predicates and concepts reveals that their ability to identify contexts to which they apply in evaluative terms far exceeds their ability to describe them in non-evaluative terms. Often there is no alternative way of determining the application of evaluative concepts, except by reference to other evaluative concepts. Supervenience, as McDowell pointed out, leaves this possibility open." If evaluative predicates and concepts supervene on non-evaluative ones that is not because people's understanding of them and their ability to apply them correctly presuppose knowledge of how they or the contexts of their application can be characterized or identified by non-evaluative predicates and concepts. The alternative arguments I can think of, when given their strongest form, support a different thesis. Take for example an argument which says that since evaluative discourse is objective, admitting of the possibility of mistakes, and of criteria of correctness, evaluative predicates and concepts must supervene on non-evaluative ones, for ultimately only naturalistic discourse can be objective. Any other domain of discourse can be objective only to the extent that it depends on naturalistic discourse. As stated this suggestion for an argument flies in the face of the facts. A good deal of reasoning proceeds with the use of evaluative predicates and concepts. It presupposes an understanding of the implications of the applications of such predicates and concepts, but—to repeat the point made above—such understanding commonly expresses itself in terms which involve the use of other evaluative predicates and concepts. To strengthen this argument it has to include claims about what are alleged to be more fundamental conditions of objectivity, for example, that without supervenience there will be no guarantee of convergence of inquirers under ideal conditions; or that without supervenience the existence of evaluative facts will not be the best explanation of any beliefs in evaluative propositions; and that these (i.e. convergence and the best explanation thesis) are preconditions of objectivity. I do not think that either of these is a " 'It does not follow from the satisfaction of this requirement [i.e. that of supervenience] that the set of items to which the supervening term is correctly applied need constitute a kind recognisable as such at the level supervened upon. .. . Hence there need be no possibility of mastering, in a way that would enable one to go on to new cases, a term that is to function at the level supervened upon, but is to group together exactly the items to which competent users would apply the supervening term. Understanding why just those things belong together may essentially require understanding the supervening term.' NonCognitivism and Rule Following', in Mind, Value and Reality (Cambridge, Mass.: Harvard University Press, 1998), 202.
The Truth in Particularism
53
condition of objectivity.' 2 But this is beside the immediate point. What is relevant here is that even if they are based on sound premises such arguments cannot establish the supervenience thesis we are considering, for they do not address the difficulty for the thesis raised by the contingency of the range of available predicates and concepts. At best such arguments would establish supervenience in some ideal or extended languages, not in languages as we have them. The same is true of other possible arguments. Some will argue that the understanding of predicates and concepts which I relied on above, while real enough, is defective and incomplete. A complete understanding of evaluative concepts includes understanding how the evaluative properties they refer to supervene on non-evaluative properties. If sound such an argument will show that no one today has a complete understanding of many evaluative concepts. Some people would regard this as a reductio ad absurdum of the argument, and conclude that it cannot be sound. My point here is different. It is that if sound the argument establishes supervenience not under present conditions but under ideal conditions. Ideal conditions of what? A natural suggestion is to turn to extensions of English and other natural languages. For example, it may be claimed that at any time it is possible to enrich English with additional non-evaluative predicates so that the evaluative predicates English has at that time will turn out to supervene on its non-evaluative predicates. But this route does not seem very promising. For one thing it seems to be compatible with the reverse thesis. At any given time, the counter-thesis goes, if the evaluative predicates then in English supervene on its non-evaluative predicates it is possible to enrich English with additional evaluative predicates so that its evaluative predicates will no longer supervene on the non-evaluative ones. You may well feel that I have been led up the garden path by approaching supervenience via a confused thesis about the intelligibility of the evaluative domain. The case of supervenience depends on metaphysical considerations about the nature of what can exist (and some would add epistemic considerations about what can be known).' 3 The arguments sketched or hinted at in the preceding two paragraphs are metaphysical in character and therefore cannot be taken to establish supervenience of predicates or concepts. They are theses about the world, or more specifically about the properties which exist in the world. They are about the 12
See 'Notes on Value and Objectivity', in Brian Leiter (ed.), Objectivity in Morality and
in the Law (Cambridge: CUP, 1999). 13 Relying e.g. on Benacerrafs 'Mathematical Truth', The Journal of Philosophy, 70 (1973), 661.
54
Joseph Raz
supervenience of evaluative properties on non-evaluative properties, not about predicates or concepts. Properties manifested in the world are, as the saying goes, part of the fabric of the world and their identity and number does not depend on the current state of this language or that, nor on that of any conceptual repertoire of any person or group. The difference between concepts and properties can be exaggerated. While the criteria of identity of properties differ from those of concepts, they too depend, however indirectly, on our conceptual repertoire, actual or possible. This is consistent with several of the theses about the identification of properties. Suppose, for example, that they are identified by reference to the causal powers of things. Different conceptual schemes offer different ways of individuating the causal powers of things, and therefore, according to that view of properties, different principles for the individuation of properties. This casts no doubt on the objectivity of properties (or of concepts), it merely reminds us that different aspects of reality are accessible to different creatures, and this depends on their conceptual reach as much as it does on their sense organs, intellectual ability, and other factors. Claims that evaluative properties supervene over non-evaluative ones have, therefore, either to be relativized to particular schemes for the individuation of properties, or to apply to all sets of properties according to all possible principles for their individuation. It is tempting to think that if the global supervenience of the evaluative on the non-evaluative is a cogent thesis at all it must apply to the relations between evaluative and non-evaluative properties under any scheme for identifying them. However, the motivation either way seems unclear. My doubts stem from the fact that the explanatory advantages of the thesis of the supervenience of the evaluative on the non-evaluative seem to depend on the availability of more specific explanatory theses. We are in need of some understanding of which non-evaluative properties specific evaluative properties depend on, and why. We seek to understand how evaluative events (i.e. events in which evaluative properties figure, like murder, or the performance of kind acts) fit in our understanding of causal explanations of events, and with our tendency to regard causal explanations as the primary mode of explanations of events. The puzzles about the way evaluative facts and events fit in our scientific world-view are not solved by one global thesis of the supervenience of the evaluative but by specific theses which are not directly entailed by it. The problem is not only that the specific theses we need to generate the explanations we are looking for are not entailed by a global evaluative supervenience thesis. The problem is that the global evaluative superve-
The Truth in Particularism
55
nience thesis does not even guarantee the possibility of these additional theses. Just as we have no reason to think that any set of concepts available to us at any time is such that its evaluative concepts supervene on the rest so we have no reason to think that in any set of properties we know of at any given time the evaluative ones supervene on the rest. If any such set can be expanded by adding non-evaluative properties so that in the expanded set the evaluative properties supervene on the rest then it may also be possible further to expand the expanded set by adding additional evaluative properties so that the new set will not meet the condition of the supervenience of the evaluative.' 4 It is possible that the general evaluative supervenience thesis is true in virtue of properties we cannot in principle come to know, and that therefore cannot figure in any explanation of the properties we do know. These remarks show how little the general thesis gives us. It does not have much, if any, explanatory power. But the less explanatory power it has the less reason we have to believe in it. If some form of evaluative supervenience obtains, then for it to serve a significant explanatory role we need specific rather than merely global evaluative supervenience theses, but no such theses are available. I will, therefore, disregard the question of supervenience for the remainder of this chapter. Putting supervenience on one side, let us return to the thesis about the intelligibility of the evaluative. It implies that whenever two situations differ in some evaluative property there is an explanation of that difference. Inevitably such an explanation points to a difference (that is, another difference) between the situations which accounts for the fact that the evaluative property applies to one and not to the other. Typically explanations will be couched in evaluative terms. They will not all conform to one explanatory pattern. They will meet different puzzlements in different ways. Nor will they ever end, nor even strive to reach a bedrock of final indubitability and transparency. They will aim at answering present questions, not to put to rest all possible ones. Is there an interesting 'particularist' thesis compatible with the intelligibility of value? Does not the intelligibility of value compel the rejection of particularism? The case for an affirmative answer to this second question may lie in the thought that since, regarding any evaluative concept and any two situations, if it applies to one and not to the other there is an explanation for this difference, it must be in principle possible to amass all the points which all these explanations may rely on and formulate one principle which sets a 14
I am assuming that the number of properties is infinite.
56
Joseph Raz
comprehensive and exceptionless rule for the use of that evaluative concept. That, the assumption is, is inconsistent with particularism. The discussion below will bear on the argument from the availability of explanations of differences in pair-wise comparisons to the existence of comprehensive and exceptionless principles. First, we may use the assumption that such principles are inconsistent with particularism to launch an examination of some specific claims advanced in the name of particularism.
2. Putting Principles in their Place
Particularist tendencies are aroused by reflection on (some) examples of concrete situations in which it would be at best misleading, and possibly outright false, to say of people who acted well that they acted on the basis of a principle, even though they acted intentionally and for a reason. Writers on the topic concentrate on particularism regarding ethics. But as there are no context-independent nor theoretically significant boundaries between ethical matters and other evaluative or normative issues I will disregard this limitation on the particularist case: 5 Outside the domain of morality the temptation to think of intention or action as guided by principles almost disappears. Where the issue is essentially instrumental, that is, about the way to achieve a set goal, it seems that principles are out of place (though rules of thumb may be a great help). 16 They do not figure when deciding about a menu which will be tasty, varied, and low in fat content, nor in choosing the best investment, nor in deciding whether to fly or to take the train to one's destination, etc. Nor, however, do principles figure when the decision is based on the intrinsic value of different options, as when one chooses which novel to read, or where to take a holiday (among equally expensive options), and so on. Many moral actions are no different. You see children torturing a cat and you stop them. No principle figures in your decision. In this just as in the non-moral cases the action is intentional and is taken for a reason. The reasons the agents would cite if asked (and the reasons which figure in their
15 For my reasons for doubting the significance of the boundary see 'On the Moral Point of View', in J. B. Schneewind (ed.), Reason, Ethics and Society: Themes from Kurt Baier (Chicago: Open Court, 1996), 58. 16 See on rules of thumb D. Regan, law's Halo', Social Philosophy and Policy, 4 (1986), 15.
The Truth in Particularism
57
deliberations where their actions are preceded by deliberation) are some prominent aspect(s) of the situation, as they saw it: that the cat is being tortured, that Ibiza would afford a very relaxing holiday, that this novel is excellent in its portrayal of a doctor confronting moral dilemmas, etc. Were the reasons which figure in deliberation, or those cited by agents when they are asked, roughly at the time of action, why they do it, the only reasons for which they act, completely described, then we would be within sight of a particularist thesis: something about reasons for an action typically being capable of being also reasons against it. To quote Dancy: The fact that an action will give pleasure can be a reason for doing it or for approving of it when done. But it can also be a reason for disapproving of it. If I tread on a worm by mistake, my action is perhaps morally indifferent. But if I tread on it with pleasure or to give you pleasure, my action is the worse for it. 17
Even before we examine such a thesis we can endorse a milder antiprinciple thesis. Since explicit deliberation does not usually take the form of identifying and following a principle, there is no reason to think that an augmented account of being guided by reasons, one which takes note of the way we are guided by reasons which do not figure in deliberation, to the extent that we are so guided, would take the form of identifying principles and following them. Possibly, when rational one is guided by considerations which can be expressed as a principle. But it is hard to think of a reason for claiming that one's reasoning, explicit or implicit, conscious or subliminal, must consist in identifying principles, and following them. The natural assumption must be that it takes the form that it appears to take in much explicit deliberation, that is establishing the considerations for and against available options, and evaluating their relative merits. This does not mean that principles do not play any role in guiding action. Rather, it would seem that they play the role which has often been assigned to them: 8 They may point out an important consideration to be taken into account, so that being guided by such a principle is the same as giving the reason to which it points due weight in one's reasoning. In this sense, 'Do not disappoint your friends' can be said to be a principle. Other principles do not mention any consideration to bear in mind. Instead they tell their subjects what to do (e.g. 'Do not be late for your appointments'). Such principles can be valid if they reflect what one has to do in the Dancy, Moral Reasons (Oxford: Blackwell, 1993), 56. The examples that follow assume that we are dealing with stand-alone principles, and disregard their role within inter-related and inter-dependent systems of rules and principles, such as the law. 17 18
58
Joseph Raz
circumstances given all the conflicting reasons which apply. Roughly speaking they are valid if they represent the correct outcome of rational deliberation about the merit of the various reasons which bear on such decisions: 9 Some principles fall in between these two kinds. They indicate that the consideration they point to is conclusive, or just very stringent, in itself, so that even while they point to a single consideration they also point to a decisive reason. 'Always treat the humanity in others as an end in itself, and never as a means only' may be a principle whose supporters typically think of it in that way. This familiar view of principles can explain why they are important in moral education, as well as why they are more often mentioned in moral contexts than in others: many people believe that some moral considerations are in and of themselves conclusive reasons for action, regardless of what reasons point the other way, perhaps barring some exceptional circumstances. If 'moral particularism is the view that general moral principles play less of a role in moral thought than has often been claimed' 20 then these considerations are ample vindication of it. 21 However, as the quotation from Dancy makes clear, some understand particularism to endorse a much stronger thesis about the nature of moral reasons, and by implication of reasons generally. Is Dancy right in claiming that what is a reason in one context need not be a reason for the same action in another?
3. Reasons and Justification
There is a superficial way of understanding Dancy's thesis. When so understood it is true but uninteresting. What people cite as their reason in a particular situation they may cite as a reason for another, even a contradictory action or behaviour on other occasions. That the action will give you plea19 I have suggested that typically such principles function as mandatory norms which themselves constitute protected reasons for action, which while being dependent on, also displace the reasons whose verdict on the situation they reflect. See for the general analysis Practical Reason and Norms (Oxford: Oxford University Press, 1999), and for its application to the special context of norms set by authority The Morality of Freedom (Oxford: OUP, 1986). 20 As per Roger Crisp in 'Moral Particularism', Encyclopaedia of Philosophy, ed. E. Craig (London: Routledge, 1998). 21 Another reason militates against assigning principles a major role in practical philosophy. The term covers distinct normative phenomena, and its philosophical use invites misunderstandings, and futile dispute about meaning. It can of course be given a technical meaning within this or that theory, but given its profligacy it is hard to avoid using it in its natural meaning even when that deviates from the technical definition.
The Truth in Particularism
59
sure may be cited by your friend as a reason to perform it. It may, however, be a reason not to perform it (e.g. for someone in charge of punishing you, or, as in Dancy's example, where the pleasure is unworthy). We do not regard this fact as remarkable for we assume that in both cases there is more to reasons for action than what people would typically cite as their reasons. Your friend mentions your pleasure as his reason for he assumes that the other factors which along with it make a reason are known, or could be learnt from the context, or are not of interest, to whoever he is addressing. When we think of his reason in its entirety we realize that it cannot be a reason for the opposing action. Dancy does not have the superficial reading of his thesis in mind. His thesis applies not to what people would typically cite as a reason, but to reasons as they are. And it applies not to aspects of reasons taken in isolation, but to complete reasons. 22 Reasons, in his usage and mine, need not be conclusive. They are conclusive when they prevail over all conflicting reasons. Dancy's thesis is not that one and the same reason may be conclusive in one situation and defeated by other reasons in another. Rather it is that the very same features of a situation which are a reason in the context of that situation may not be a reason at all in another, or even be a reason for a conflicting action. Another of his examples illustrates the first possibility: 22 A terminological point is called for here. In part this chapter strives to contribute to the explanation of the concept of 'a reason'. The phrase 'a complete reason' is, however, given a somewhat stipulative definition, aimed at helping with the explanation of reasons. A complete reason consists of all the facts stated by the non-redundant premises of a sound, deductive argument entailing as its conclusion a proposition of the form 'There is a reason for P to V (where P stands for an expression referring to an agent or a group of agents, and V for a description of an action, omission, or a mode of conduct). A similar definition can be provided for reasons of belief, emotions, etc. An examination of the use of expressions such as 'this is a different reason', 'this is the same reason', and similar expressions, will show that this notion of complete reason captures an important aspect of our understanding of reasons. It is partly stipulative in regimenting the use of expressions such as 'same reason' and 'different reason', which in their ordinary use often invoke differing, contextdependent, standards of completeness. In another partly stipulative deviation from ordinary usage I refer to any fact stated by any proposition which can be a non-redundant premise in a sound argument of the kind just described as a reason (or as part of a (complete) reason). Some aspects of a complete reason are more readily perceived as reasons than others. Contrast 'I am thirsty' with 'this liquid will not harm me' as statements of reasons for drinking the liquid in front of me. Ordinary usage allows reference to many aspects of complete reasons as reasons when the context is appropriate (e.g. 'this is water' is often cited as a reason for drinking it). Appropriateness is a matter of sensible assumptions about what the addressees of various remarks know, and what they do not, what would puzzle them and what would be taken as self-evident, etc. Generally, any aspect of a complete reason can be cited as a reason in some circumstances, and I will use the term 'reason' to refer to any such aspect.
60
Joseph Raz
that I borrowed a book from you is, often, a reason to return it to you. But if having borrowed it I discover that you stole it from the public library I have no reason to return it to you. What would show him to be right or wrong? To refute him it is not enough to show that there is a difference between situations where the reason points one way and situations in which it points some other way. This is readily allowed by Dancy. His example of the way giving pleasure can be good or bad comes with an explanation of the difference. His thesis depends on the fact that the features in which the situations differ which provide the explanation are not parts of the reasons for action. Why not? As was pointed out at the beginning of the chapter, the value of actions serves both an evaluative and a guiding function. It is the basis by which they are judged, as well as serving as a reason for their performance. While the two can come apart in certain cases, they cannot drift too far apart without conflicting with our understanding of the notions of guiding and of evaluating actions. A major weakness of Dancy's thesis is that it drives a wedge between reasons for action and the evaluation of those actions. 23 Herisonwayfptgheoin.Iavkrstobefau of the action they are reasons for. They are generic features of action-types capable of being instantiated on various occasions. That an action would be the returning of a borrowed book is such a feature. According to Dancy, the same feature can be a reason for the action of which it is a feature in one context and against it in another. How can that be? This cannot be an arbitrary brute fact. The intelligibility of value means that there must be a difference between the context of the two instantiations which explains why the same fact (e.g. that the action is a returning of a borrowed book) is a reason in one and not in the other. Yet, the difference will not figure as part of the reason. If it did then it would not be true that the same fact is a reason in one situation and not in the other. Therefore, according to Dancy, the considerations which determine what is right (or good, etc.) extend beyond that reason itself. It follows that not everything relevant for the evaluation of an action is part of the reasons for or against the action. We know that this is the case in some special contexts—for example, that some values can be achieved only by actions which are not taken in order to achieve them. But Dancy's thesis is meant to be a general thesis, arising out 23 It is not clear whether Dancy is fully aware of this. He says 'the fact that . . . can be a reason for doing it or for approving of it when done. But it can also be a reason for disapproving of it' (Moral Reasons, 56). He seems to apply his particularist thesis not only to reasons but also to values generally. To follow him would fly in the face of the intelligibility of value, to which, as his discussion of all his examples shows, he is committed.
The Truth in Particularism
61
of the nature of practical rationality itself, and not limited to special types of reasons. The question inevitably arises: if certain factors are relevant to the evaluation of an action why are they not also part of the reasons for or against it? Should we not do the right (wise, sensible, etc.) thing and avoid wrong (foolish, etc.) actions? Does that not mean that the factors which determine the rightness (or wisdom, etc.) of an action are the reasons for it? An affirmative answer refutes Dancy's thesis, for it denies (special cases aside) the general gap between reason and evaluation on which his thesis depends. One route for avoiding an affirmative answer is through raising the guiding problem. We can accept the principle that in any given situation only what can be someone's reason in that situation (i.e. can be a reason for which they then act) can be a reason. Only those evaluative considerations which can serve a guiding function are reasons for action. Relying on this principle it can be argued, in support of Dancy's thesis, that while most evaluatively relevant considerations can be (part of) someone's reason for action, in any given situation no one can be guided by all the evaluatively relevant factors present in it, or at least that that is not normally, or commonly possible.
4. The Guiding Problem
The issue turns on whether there is more to people's reasons than those factors which figure in their deliberations, and which they cite as reasons when asked, and on how much more is part of their reasons. We can only judge the matter against a picture, however rough and sketchy, of the way we are normally guided by reasons. In outlining such a picture I will disregard the cases where we act out of a belief in a reason which is false, where the reason does not exist. I assume that such cases are to be analysed by reference to the normal case, the case in which no such mistake occurs. There are two ways in which the reasons we act for are not the reasons which figure in our deliberations, nor the ones we will avow when asked. We may be mistaken about the reasons for which we act. We may think we act for reasons which are not genuinely our reasons, while rejecting the thought that our reasons are what they really are. We also may simply be aware of only some or some aspects of our reasons while being unaware of others. We act for reasons whenever we act intentionally, but only on some of these occasions do we deliberate shortly before acting. 24 What marks 24 Earlier deliberations and advanced planning and decisions are central to human rational capacity. I exclude them at this point because they do not pinpoint the time of action,
62
Joseph Raz
intentional action as guided by reasons is not that it follows deliberation, but that it is undertaken in light of an appreciation, accurate or faulty, of our situation in the world (an appreciation of ourselves as well as of our environment). That appreciation is embodied in beliefs formed over time, as well as in what we currently learn of our immediate circumstances at the time of action. That action for reasons is action informed and shaped by our appreciation of ourselves and our circumstances allows both for mistakenly thinking that a certain reason exists when it does not, and for being ignorant of the fact that certain reasons exist. For our purpose here a third kind of mistake is pertinent: we may be mistaken about the reasons for which we act. We may believe that we act for reasons which we do not in fact act for, and we may be unaware of the reasons for which we do act. Action for reasons presupposes an appreciation of the situation we are in, but it does not presuppose self-knowledge. It does not presuppose knowledge of our own beliefs, and of which of them inform us of the reasons for which we act. But we have to tread carefully here. For creatures capable of self-knowledge, having beliefs involves having the capacity to be non-inferentially aware of them.25 Mistakes and ignorance have to be explained. The onus in explaining mistakes is particularly severe. Ignorance can be due to no more than inattention, failure of recall, etc. Mistakes about what one believes, or desires, or feels call for stricter explanations, especially in the many cases in which they involve self-deception. In acting for reasons we can be wrong not only about how things are in the world and about ourselves. We can also be mistaken about our reasons, i.e. the facets of ourselves and of our situation (of which we know), which guide our actions. Such mistakes would typically occur when we are affected by motivated irrationality, subject to wishfulfilment, to the distorting effects of a sense of guilt or shame, or other factors which lead to self-deception. I mention these factors not because they are central to an evaluation of Dancy's thesis. They are not. I mention them because once we allow that we may be mistaken about the reasons for which we act we must allow that we can act for reasons which do not figure in our deliberation, and which we would not cite and therefore do not lead to the formation of an intention which uniquely identifies the act performed. 25 Such a capacity is consistent with externalism about mind and meaning. While creatures capable of self-knowledge are capable of being non-inferentially aware of their beliefs, that does not imply that they always have complete knowledge of the content of their beliefs. More needs to be said about externalism and self-knowledge.
The Truth in Particularism
63
when asked (by ourselves or others) for our reasons. The centre of the argument about Dancy's thesis rests with the second way in which there is more to reasons than that: the case of ignorance of our reasons, rather than of mistakes about them. We may be unaware of our own reasons. We act for reasons we know of, be it through our general stock of beliefs, or through what we come to believe about the situation we are in at the time of action. Either way we know more than we can articulate. No one can spell out all that he knows, and no one can detail all that he perceives, or even just sees at any moment. There may be nothing we know which cannot be stated, and nothing we see which cannot be described. But it does not follow that we can state all that we know, and all we see. While not all that we know, nor all of our perceptions, are tapped by us when they are relevant, while we may fail—as we say—fully to grasp the implications of some of our knowledge, our rational responsiveness to their implications does not require coscious reflection or deliberate articulation. We can rationally respond to what we see and act as we do because of what we know, without being aware of that knowledge at the time. Hence the reasons for which we act need not be reasons about which we deliberated prior to action. There are various contexts in which we act for reasons without deliberation. There are the many actions we perform regularly which are embedded in automatic routines. Standing in the kitchen I decide to get something from the fridge. A whole series of actions follows as it were automatically. I am barely aware that I perform them. They are all done for reasons, but these have been instilled in me and I need not reflect on them or re-endorse them. The reason-guided character of the actions is manifest in the fact that I monitor them, and will abort them if the situation changes, or is revealed to have changed. Some rules, like 'wash your hands before eating, are internalized and lead to semi-automatic action similar to actions embedded in automatic routines. I will say nothing about these types of cases, nor several others. To the extent that they are relevant to our purpose they share the characteristics of the situations which more obviously serve in particularist arguments. One kind of case is exemplified by the toddler example: you see a toddler about to step into the road and you reach out and stop him. Your action is swift. You do not stop to ask yourself what you should do. You react to the situation as you see it. I will refer to such actions as swift actions. The second type of situation I have in mind is very different. You agonize over a question for a long time. Let us suppose that I cannot decide whether to
64
Joseph Raz
take up a job offer or not. 26 I do deliberate and survey all the reasons for and against that I can think of, but am unconvinced either way. I let the matter reset for a while, and a week later when I think of it again the decision is immediately clear: I should accept the offer. At that stage, the stage when I actually reach a decision, I do not deliberate. Nor am I able, once the decision is taken, to adduce any reasons to justify it that I did not have before. I have no new arguments. I just made up my mind, apparently on the basis of the old arguments. Let us call such cases opaque cases. Action without deliberation is not action without awareness of its reasons. Self-conscious beings are capable of describing their intentional actions in one way or another (if only in their thoughts). They may act semi-automatically, or swiftly, or opaquely, but in all cases some description of their action, as intentional, is available to them at the time. If they do not actively think of what they are doing the thought can be triggered by questions. Moreover, whenever a self-conscious being acts intentionally a description of the action which, as he sees it, shows it to be desirable is available to him.27 That means that people have citable reasons for their intentional actions, i.e. they act intentionally for reasons which are available to them if asked at the time. This does not mean that the reasons for which people act are (identical with) the reasons available to them when they act. The case for mistakes was made above, and there are further reasons to doubt that identity. We have to rely on counterfactuals to establish people's citable reasons, that is the reasons available to them at the time of action. But what sort of counterfactual condition is relevant here? People will give different answers to different questions. Perhaps we should say that their reasons are the totality of the reasons they will mention in reply to all the (infinite number of) questions they can be asked. However problems remain. Once we move beyond the most obvious aspects of the situation people's claims about their reasons will depend on their frame of mind at the time. They may be more or less attuned to the significance of different aspects of their situation as they know it. You may say that that does not matter. Their reasons depend on their frame of mind at the time. If because of their mood at the time they do not then know of certain reasons, or if their knowledge
26 For an analogous moral case think of a person agonizing whether to volunteer for a food convoy for the relief of starving people during the war in Bosnia. Matters may progress exactly as in the job example. 27 All this is subject to the possibility of mistakes mentioned above, and to various pathologies we need not examine in this brief sketch.
The Truth in Particularism
65
remains untapped, if it does not influence their conduct, not even counterfactually, then they do not act for these reasons. While true, when given as a reply to the objection this answer assumes without justification that people's frames of mind affect what reasons they act for, and does not affect their ability to articulate them or to become aware of them. Experience teaches us that that is not so. In certain moods our ability to acknowledge our concerns and to become aware of what we know is affected and no longer matches the reality of our concerns. Nor does it mean that we do not respond to the knowledge of which we remain unaware at the time. For example, we may avow willingness to act out of revenge when we are not actually willing to do so, and vice versa. One cannot escape the difficulty by making the counterfactual by which we identify the reasons for which we act relative to answers we would give when in a cognitively ideal frame of mind. To do so is to make the opposite mistake of thinking that our frame of mind affects only our ability to become aware of our reasons, but not our reasons themselves. It follows that sometimes what we would cite as reasons if asked are not our real reasons. Therefore, our reasons cannot be identified with our citable reasons. While this argument relies on the possibility of making mistakes it does not assume that in the situations in which the counterfactual test will yield the wrong results we are actually making a mistake about our reasons. In most of those cases the reasons we would cite if we asked ourselves appropriate questions are not elicited and are not ones we have in mind when we think of our reasons for our actions. Hypothetical mistakes are not real mistakes. The argument suggests, however, an approach to the identification of the reasons for which people act, namely they are the reasons they would avow if asked appropriate questions at the time, provided their answers are not distorted by mistakes. Where they are so distorted, understanding the mistakes and the reasons for them may lead to the real reasons for those actions. Needless to say, if this is a correct test for identifying the reasons for which self-conscious creatures acted then the notion is very vague. There are other reasons to think that it is vague so this should not be taken as an objection to the test. Let us return to Dancy's thesis. According to it, the very facts which are a reason a person has for one act on one occasion may be, for the very same person, a reason for another, even conflicting action on another occasion. As we saw, this yields the result that reasons for people's actions do not determine whether their actions are right or wrong, wise or foolish, etc. To
66
Joseph Raz
know that one needs more than the reasons people have. 28 Reasons, however, are general, according to Dancy. They are features of situations, or of actions, which can be instantiated on an indefinite number of occasions. His view is particularist in the sense that the rightness of an action is not determined by the reasons which apply to the agent. There can be a rival particularist thesis. It can be illustrated by thinking of both swift and opaque actions. In both cases it is natural, when asked for one's reasons, to point to the situation as a whole: 'in these circumstances that was the right thing to do. The situation in all its concreteness is what one reacted to. That explains how swift intentional action is possible. The absence of deliberation shows, so the argument might go, that one reacted to the perceived situation as a whole, not to any part of it. The opaqueness of opaque cases shows that no feature of them satisfied the agent as decisive, or determinative. Rather the agent reacted to the situation as a whole. In some ways this view that reasons are concrete is more attractively particularistic than Dancy's thesis. On this view reasons are particular, not general as per Dancy. Being particular they cannot justify any other action (assuming that they are conclusive reasons for the action taken). But, unlike Dancy's reasons, they justify the action they are reasons for. No gap is allowed between the evaluative and the guiding functions of evaluative considerations. These advantages notwithstanding, Dancy's thesis is the better one. The concrete reasons view appears to do justice to the fact that we can have the skill to discern reasons without analysis and reasoning. But in fact it goes overboard, and undermines its own case. It claims not that we can without deliberation distinguish relevant from irrelevant features in a concrete situation, as indeed we can, but that we do not need to discern any features of a situation to know that it is a reason for this action rather than that. That is to deny the intelligibility of reasons, for intelligibility depends on generality. Nor is it reasonable to deny the generality of reasons in the case of either swift or opaque actions. If we ask people who act swiftly and without deliberation they would still be able to distinguish relevant from irrelevant features of the situation: that the toddler (who was stopped from stepping into the road) was wearing green trousers was irrelevant. That the road was busy with traffic was relevant, and so on. There can be
28 Two clarificatory reminders may be helpful here. First, in this chapter 'reason' refers to good or valid reasons. Second, when discussing the possibility that people's reasons, i.e. the reasons which apply to them, in any given situation determine whether what they do or may do in that situation is right or wrong, etc., I am referring to the totality of all the reasons applying to them in that situation, and not merely to a single reason.
The Truth in Particularism
67
no general argument that people are always mistaken when they so identify the features which they took to be reasons. Preserving the connection between reasons and intelligibility is an advantage of Dancy's thesis. Moreover, the approach delineated above for identifying the reasons which guide people's actions appears to support at least a weak form of Dancy's thesis. Is there any general reason to suppose that, if prompted by all the pertinent questions, every agent would cite enough features of his situation as his reason for the reason he cites to be incapable of being in different circumstances a good reason for a conflicting action? Unfortunately for Dancy's thesis the answer is yes. Possibly whenever what people would cite as their reason (when duly prompted) can be a reason for a conflicting action then their understanding of their own reason is incomplete. If their understanding of their reasons were adequate they would not cite factors which can fail to be reasons for the same action in different circumstances. That is, whenever people act for a particular reason, and because they do so in those circumstances, their action is justified, their reason includes all the relevant evaluative factors which show their action to be justified. 29 That (sometimes) the factors people would cite if asked for their reason fail to meet this condition does not show anything about the nature of reasons. It shows that those people's understanding of their reasons is imperfect. If so then the gap between guidance and evaluation that Dancy's thesis assumes does not exist, and the thesis is mistaken. It gains an aura of plausibility because sometimes people's understanding of their own reasons is incomplete. If one identifies people's reasons too closely with their understanding of their reasons Dancy's thesis appears plausible. This reply to Dancy is supported first by reflection on some of his examples, and secondly by a general consideration of the relation between one's reasons and one's statement of one's reasons. Some of Dancy's examples are problematic. Several involve the notoriously complex question about the nature of pleasure and pain. To the extent that his particularist thesis is made more plausible if one supposes, as Dancy does, that pleasure is not always a reason for, and pain not always a reason against, action, this may 29 This is true even when additional evaluative features of the situation were not part of their reason, but bear on the justification of the action. If these additional features tell against the action then the reason for which the agent acted will show them to be defeated in the circumstances. If the additional features tell in favour of the action they will be additional reasons for performing it which were not the agent's reasons, or evaluative features which cannot serve as reasons (they will be self-defeating if relied on, etc.). But the action is justified even without taking them into account.
68
Joseph Raz
be an argument against the supposed view of pleasure and pain, rather than in favour of particularism. But take the book loan example mentioned above. Most likely when asked people would say their reason for returning the book was that they borrowed it, or promised to return it. But if asked at the time would the fact that the person from whom it was borrowed had the right to possess it, that he did not steal it, etc. be relevant to their reason (i.e. was their reason that they borrowed from someone entitled to lend them the book), most people would say yes. Regarding those people the example fails. Their reason was not one which applies in cases of a borrower who stole the book. If they are right it would follow that a more complete statement of the reason for which people who borrow a book should return it is that it was borrowed from someone entitled to have it. If so then those who would deny, when appropriately asked, that the fact that the book was not stolen was part of their reason would be under suspicion that they simply misunderstand the reason for which they act. They may have acted for a good reason, but they have only an incomplete understanding of it. Dancy objects to counting the absence of conditions like the death of the promisee, which he calls defeaters, among the elements which make up a reason: 'there are just too many potential defeaters for the absence of each one to count among our original reasons'. 3° But this remark confuses articulation and knowledge. There are too many potential defeaters for us to be able to mention all of them or to think of all of them. But it does not follow that we cannot know all of them. 31 And those who know all of them may be guided by all of them (i.e. by their absence). That is, for those who know them all, their absence can be part of the reason. Still, the reply to my promising story applies only to this example. For any statement of a reason, however expanded, an example can be found of a situation in which that reason is no reason (not, that is, unless expanded further). There is no general argument to show that any such new example can be countered by further aspects of their reasons elicited from those who acted for that reason where it was a good reason. So let us imagine a case. Suppose that John promises his neighbour to look after his cacti when he goes on holiday. When you ask John for his reason he would cite his promise. Suppose that as a matter of fact the promise remains binding even if the neighbour dies during his holiday, but only until an executor is appointed to deal with his property, whereupon it lapses. However, when Moral Reasons, 81. There are other reasons to think that we cannot know all of them. The argument below allows for that possibility. 30 31
The Truth in Particularism
69
we ask John whether his reason includes the fact that his neighbour is still alive, he does not give this answer. Imagine first that he gives no answer. He simply says that he does not know whether the promise remains binding after the death of his neighbour. The question is: what was John's reason for looking after the cacti? It seems to me that he acted for the right reason (that is because he promised, and the promisee has not died, etc.) though his understanding of it is incomplete. There is no direct argument here against Dancy's suggestion that he acted for the reason that he promised tout court. It cannot be refuted by the fact that in pleading ignorance John denies that he takes a promise to be a reason regardless of whether the promisee is alive. To take this as an argument against Dancy is to beg the question against him. But that was not the purpose of the argument. Its purpose was to show that there is an acceptable alternative to Dancy's position. The reason for preferring it is that it avoids the gap between evaluative considerations and reasons which Dancy's position opens. But there are additional reasons for preferring my suggestion. First, at the level of describing the phenomena, it is closer to the way we understand reasons. We may assume that people disagree about the precise conditions under which a promise expires, or about who can make promises, and in what ways and under what conditions binding promises are made, etc. Are we to say that people who disagree with each other do not act for the same reason when they keep a promise because they (rightly) believe that it was binding? In fact we regard all of them as having the same reason, though they differ in the way they understand it, and at least some of them understand it imperfectly. At a more abstract level my suggestion relies on the fact that, since reasons are objective, one can refer to them without understanding them well, even while being mistaken about some of their aspects. 32 All these possibilities arise out of general features of reference, and there is no reason to think that reference to reasons is the exception. Indeed it cannot be a total exception or we will not be able to explain how it could be that one acts intentionally and for a reason, without knowing anything about it, except that one knows that it exists for one was assured of this by reliable people. 32 As would be the case in our example were John of the view that the promise binds until his neighbour's heir releases him from it or that it lapses immediately upon the death of his neighbour. The relative opacity of reference throws up various intriguing questions in some problem cases. Some would be cases where people try to act for a reason and think they did, but through their misunderstanding of it they did not succeed. Another general result is that in explaining agents' actions one would sometimes need to refer not only to their reasons, but also to some aspect of the reasons which they knew and thought relevant.
70
Joseph Raz
It is not clear whether Dancy can account for the fact that our reasons need not be totally transparent to ourselves.
5. The Personal Perspective
The attractions and the interpretations of particularism are many. I will end by examining just one other claim, based on the distinction between first-person and third-person practical judgements. The claim I have in mind was put forward by Winch whose view 'puts a certain class of first person moral judgements in a special position as not subject to the universalisability principle'. 33 Winch uses Melville's Billy Budd to illustrate his view. In the story Vere finds himself confronting a conflict between his private conscience and the 'imperial one embodied in the military code. He sees it as a moral conflict between two moral 'oughts, each of which is, in Winch's view, universalizable. In deciding to condemn Billy Budd and execute him Vere acts on a moral judgement as to what he should do, given his situation. This judgement, says Winch, is not universalizable. Winch finds that were he confronted with the same situation I could not have acted as did Vere;.. . I should have found it morally impossible to condemn a man 'innocent before God' under such circumstances. In reaching this decision I do not think that I should appeal to any considerations over and above those to which Vere himself appeals. It is just that I think I should find the considerations connected with Billy Budd's peculiar innocence too powerful to be overridden by the appeal to military duty. (p. 163)
Nevertheless, Winch tells us, 'The story seems to me to show that Vere did what was, for him, the right thing to do' (pp. 63-4). It is merely that it would have been wrong for Winch to do the same. This example illustrates Winch's general claim that if A says 'X is the right thing for me to do' and if B, in a situation not relevantly different, says 'X is the wrong thing for me to do', it can be that both are correct. That is, it may be that neither what each says, nor anything entailed by what each says, contradicts anything said or implied by the other. (pp. 164-5)
Winch is at pains to emphasize that he is talking of genuine judgements, and that he, Vere, or anyone else making such judgements can be mistaken. Whether his account of the conditions under which we will be mistaken is 33 'The Universalisability of Moral Judgements', in his Ethics and Action (London: Routledge and Kegan Paul, 1972), 159.
The Truth in Particularism
71
adequate is immaterial to our purpose. 34 My discussion of the example assumes that all the normal ways in which people can make mistakes apply to such judgements as well. The crux of the matter is not in the ways the claim can be wrong, but in the factors which can make it right. Winch explains that what is puzzling about the judgements he is concerned with is 'that they seem to span the gulf between propositions and expressions of decisions.... [T] he deciding what to do is, in a situation like this, itself a sort of finding out what is the right thing to do' (p. 165). This is the key to the explanation of the judgements under discussion. Two elements combine here. First, sometimes what makes an action right for me and wrong for you is something about me, and about you. Second, the process of discovery is the process of decision. They are inseparable. Both elements are problematical. What follows is my gloss on them. It does not necessarily coincide in all respects with Winch's view. If what makes the action right for me is something about me, it is trivial that it is not necessarily right for everyone. This would show how it is possible that the two propositions (right for me and wrong for you, in the same situation) can both be true. But is this construal true to Winch's (and Melville's) description of Vere's and his own reasoning? Vere considered Budd's innocence, the situation of the navy, etc. He did not consider himself as a relevant factor. Winch, as we saw, tells us that 'I do not think that I should appeal to any considerations over and above those to which Vere himself appeals. It is just that I think I should find the considerations connected with Billy Budd's peculiar innocence too powerful to be overridden by the appeal to military duty.' If so then Winch does not rely on any consideration to do with who he is. But, if they are both right because they are different people does it not mean that they rely on different considerations: Vere on who he is, and Winch on who he is? This reply misunderstands my first point. It was not that Vere's character or personality, or moral sensibilities, or anything like that figures among his reasons. At least it does not figure as such, under that description. It was merely that it is part of what makes the decision right for him. Sometimes people may rightly take their own personality as a reason: `Given my nervous disposition, I may say, 'embarking on this course of action will cause me such anxiety that I had better avoid it.' But nothing of this kind is relevant to Winch's case. His case is that when faced 'with two conflicting sets of considerations, the one man was disposed to give 34 Wiggins in his illuminating discussion of Winch's article has shown them to be too narrow. See 'Truth, and Truth as Predicated of Moral Judgements', in his Needs, Values, Truth, 3rd edn. (Oxford: OUP, 1998), 166-84.
72
Joseph Raz
precedence to the one, and acquit, the other to give precedence to the other, and convict' (p. 169). The fact that they had these dispositions makes their conflicting decisions right for them, but they are not among their reasons. How are we to understand this situation? To make sense, and to avoid attributing a mistake to either of them, we must assume that Vere and the Captain of the Marines each takes his reason to be that, in the circumstances, one set of considerations is decisive for him, while conceding that if another person reaches a different decision they would not necessarily be wrong. Not every judgement of the relative stringency of different considerations is like that. Typically, cases like that of Winch and Vere occur when impersonally judged there is no answer to the question of which set of considerations must prevail. 35 Typically, that is, impersonally the conflicting considerations are incommensurate. In such cases the fact that the decision is Vere's or Winch's may make a difference. But if impersonally the conflicting reasons are incommensurate what is there for agents to discover through their decision? What was it that Winch discovered when he asked himself what he should have done in Vere's situation? Winch describes his own discovery: 'I could not have acted as did Vere; I should have found it morally impossible ...'. It is clear here that it is in a sense about himself. He still believes that Vere did no wrong. He simply discovered that he, Winch, could not do likewise, and that is the same as discovering that it is wrong for him, that for him one set of considerations (Budd's innocence) overrides the other. The impossibility is not a result of a belief that it is wrong. Nor is the belief that it is wrong a result of a feeling that one cannot perform the fact. The two are one and the same. There is a belief about what is right for me which is an aspect of knowing what is normatively 36 possible or impossible for me to do. The two are one, or rather aspects of one phenomenon. It is significant that, as Winch points out, we tend to talk of impossibilities in these contexts. Impossibilities do not normally accompany judgements that of two conflicting reasons one is stricter than the other. Impossibility occurs in the context Winch discusses because one's judgement involves, in part, a discovery about oneself, about what one can or cannot do.
35 It is not logically necessary that it be so. But this being the typical context we need not here consider others. 36 Here too I do not believe that the impossibility need be moral. It could be an impossibility to allow that a pig's heart be transplanted to replace one's failing heart, or many others. They need not be irrational superstitious qualms. They may not be based on any false beliefs, nor need they display any other signs of irrationality.
The Truth in Particularism
73
We are still dealing with the difficulty I started with: how can the decision be right, or even right for me, because of something about me, when my condition was not part of my reasons? Locating the phenomena among cases where of the conflicting considerations neither is superior to nor more stringent than the other, cases where reason underdetermines the result, where the two best options are incommensurate with each other, helps us towards a better understanding. It can explain, for example, how it could be that personal factors may matter, even though the choice is between life and death for another person. It cannot be that it was right to convict Billy Budd because Vere is such and such a person. But Winch does not say that it was right to convict Billy Budd (impersonally speaking, i.e. without regard to who is convicting him). He says that Vere's decision to convict was right for him. I added that this can only be if when impersonally judged neither course of action is dictated by reason, only if reason underdetermines the outcome. That is why it is impossible for Billy Budd, on Winch's analysis, to complain that he was wronged, though he was unlucky not to have Winch or the Captain of the Marines as his judge. But if impersonally neither action was dictated by reason, how could one be the right action for one and the opposite action be the right action for the other? Winch does not mean that both courses of action were permissible. He is saying something quite different, which is captured by the invocation of impossibility: that it would have been impossible for him to convict (or for Vere to acquit), when the impossibility is not one of weakness of resolve. Winch's example is unusual. In most cases where the choice between options is underdetermined by reason people just follow their inclination, or follow a momentary desire, or just choose. How can such cases, paradigmatic of unhindered choice, be determined by what it is not possible for one to do? Whether people are aware of this or not, during their life, through myriad decisions and actions, people develop their personality, and create their own distinctive tastes and dispositions. Emphasizing that people's personality and some of their basic dispositions are self-determined is consistent with the fact that people's character is fundamentally affected by inherited characteristics, and by the impact of their environment on them. That goes without saying. Nor should we think of the self-determining aspect of character formation as intentional. It rarely is. We make our choices with an eye on the occasion. Yet such choices reveal aspects of ourselves to ourselves, and they create precedents, set and consolidate trends. The patterns of our lives help us make sense of our lives and of ourselves. We may fight them, and reverse them in future actions, accept them with pleasure, drift
74
Joseph Raz
along with them without appreciating their significance, or follow some other course. One way or another our past actions and decisions form us. They make us into who we are. My purpose is not to explain this process, but merely to point to it. The point needed for understanding Winch's claim is the special importance of choices which are not determined by reason for the formation of our personality. When we follow reason, or fail to follow it, we reveal and we mould our executive virtues or failings. It is, however, primarily where matters are underdetermined by reason that we reveal and mould our distinctive individuality, our tastes, our imagination, our sociability, and many of our other, including our moral, characteristics. Winch's claims concern the role of our moral character (though I would repeat again that similar considerations apply to various other aspects of character). Notoriously the morality of right and wrong is not exhaustive. Many moral acts are supererogatory. In these cases the demands of morality are incommensurate with some nonmoral reasons. When this is so regarding, say, charitable giving, or volunteering to help with various good causes, we are not rationally required to choose the moral option, but if we do we prove ourselves generous with our time or money. People's choices and their habits of giving determine how generous they are. Less often acknowledged by philosophers are the many occasions in which the demands of morality themselves are indeterminate (many cases of supererogation illustrate this indeterminacy as well). Often the demands of justice and of mercy are as we say 'finely balanced, that is incommensurate. 37 Some prove themselves, and make themselves, merciful by generally choosing the side of mercy. Others turn into stern and unforgiving people. These are crude and simple examples. The complexity of moral life defies such simple descriptions but confirms the general view they exemplify. How do these features of moral psychology express themselves normatively? Impersonally judged since both options open to Vere are supported by incommensurate reasons neither is wrong. 38 It is possible that people when faced with such a decision would find that it is impossible for them to do anything but acquit (or convict). Does that impossibility have normative force? I think that it does. People do violence to themselves if they 37 For the reasons why cases like these are rarely cases in which the conflicting considerations are exactly equal in strength see Raz, Morality of Freedom, ch. 13. 38 Those who think that there is a weak sense of 'right' in which it is synonymous with `not wrong' would say that both options are right. It seems, however, that 'right' connotes more than that, so that saying that, impersonally judged, both convicting and acquitting Billy Budd would have been right is at least misleading.
The Truth in Particularism
75
go against the grain, and act in a way which offends their moral character. Their integrity and self-respect are transgressed when they do so—unless, of course, they should do so. They may come to realize that their moral character is corrupt, or just that it leads them to the wrong option on a particular occasion. But we are assuming that that is not so, that impersonally the considerations which apply to the case underdetermine its outcome. In such a case it is right for people to act as their moral character tells them to act. But their reason is not that that is what they are disposed to do, or that this is more consistent with their past decisions. It is that they can do no other. 39 They cannot but prefer one set of considerations to the others; for them it is the more important or stringent set of considerations, even while knowing that impersonally speaking they are incommensurate. 4° So much for my attempt to make sense of the way in which an action may be right for me and wrong for you, in the circumstances Winch discusses. If these remarks are along the right lines they help with the second element in Winch's story: the reason the process of discovery is the process of decision, and the sense in which first-person judgements are different from third-person judgements. Note that Winch does not mean actual decision. He discovers what is right for him by imagining himself to be in Vere's shoes. Sometimes we may well feel that we cannot really know what we will do if faced with a decision until we are actually faced by it. The pressure of reality may prove our imagined response wrong. But this psychological possibility is beside Winch's point. His point is that we should think of the problem personally, rather than impersonally. Think not of what is right or wrong for one, but for us. The impersonal question engages our understanding of right and wrong. It should play a part in any decision39 Even though, had they thought that their disposition led them to the wrong conclusion, they could have fought it and could have gone against it. 4" These cases are interesting to compare with the normative effect of commitments: people's personal commitments to various pursuits affect the reasons confronting them. After graduation when I considered becoming a teacher the merit of that activity, its impersonal importance, was an important consideration in my decision. A couple of years later, when I was a struggling beginner teacher, the value of teaching to me changed. It no longer was just the impersonal value of teaching. Now teaching is more valuable to me. Its value now is in part a product of my commitment to a teaching career. This 'extra' value is not merely the result of the economic or psychological investment. It is a result of the way my life became involved with teaching, the way success in teaching has become contributory to the success of my life, and walking away from it, or failing in it, a contribution towards the failure of my life. So commitments are ways in which people's biography affects what is right or good for them, while not affecting the impersonal value of the options. Yet commitments need not affect people's character, and they are themselves additional reasons, reasons for the committed.
76
Joseph Raz
making. But in cases like Vere's decision we should also confront the personal question, imagine ourselves in the situation and ask what is right for us to do. Only thus do we let our moral character fully express itself. In what sense is a discovery involved? Suppose that I know myself well, suppose that I am a judge in a morally tainted system and not unfamiliar with ambiguous situations. Is there still something for me to discover? Do I not just know that the right thing for me is to decide this way? Note that what I know about myself may be known by others as well. If I can rely on that knowledge so can others. Will that not give the lie to the suggestion that first-person judgements are privileged? The answer is that 'discovery, in the sense relevant here, need not be of something unknown or surprising. It can be no more than reaffirming what one thought to be the case any way. Indeed, many scientific discoveries are discoveries in that sense: experimental confirmation of a theoretically predicted result. In practical decision-making the element of discovery is in holding oneself open to what one may find, that is not prejudging the case. In some attenuated sense the same may be said of reflection on the merits of alternative options impersonally considered. But it is particularly important when the personal perspective predominates. The question is not, given my moral character what shall I do? To put it thus is to foreclose the possibility of a change in one's personal perspective, and to deny the self-determining, the self-creating aspect of decision and action. The question is, given these (impersonally valid) considerations what should I do? What is the right thing for me to do? This question allows my moral sensibilities to express themselves in evaluating the relative merits of the impersonally valid considerations. It also allows a continuous process of self-determination, for it leaves open the possibility, however unlikely, that my response will surprise me, that it will not confirm my own previously formed idea of myself. Winch's example establishes the privileged standing of first-person judgement in this limited range of cases. This does not mean that only the people whose action it is can know what is right or wrong for them. The privilege has two aspects: first when Winch, as he does, asks what was right for Vere he is putting himself in Vere's shoes and asking the question as Vere would ask it, given his, i.e. Vere's, moral character. Winch may find the right answer, but he does it from Vere's point of view. Secondly, and this somewhat qualifies the first point, decision is part of discovery in the sense that it is open to development, and is part of the process in which we are part authors of our own character. In that sense but for the fact that Vere has already decided, Winch could not have known what was
The Truth in Particularism
77
right for him. The answer must come from the person whose decision it is. Only they, by going through the process of confronting the issue in life or in their imagination, can discover/determine what is right for them.
6. How Much Particularism?
How much particularism do these reflections affirm? They reject the thought that morality or any other significant domain of practical rationality consists in principles, or that conformity to reason within it consists in following principles. Moreover, they suggest that our knowledge of reasons exceeds our ability to articulate them, and that to be guided by reasons we do not need to be good at articulating them. This conclusion is of great practical importance, though it is difficult to draw general operational instructions from it. It suggests that we know more than we know we know, and more than we can explain. Some of our knowledge can be brought to bear only in the concrete situation of decision and action. Any attempt to limit the freedom of judgement by restricting it to the application of general principles inevitably excludes our ability to tap the inarticulate fund of knowledge at our disposal. On the other hand, critical examination of our views requires (even if it is not exhausted by) articulation and explicit argument. When should we trust principles, which have been exposed to explicit rational scrutiny, and when our inarticulate knowledge, is not a question admitting of an easy general answer. Sometimes we can tell that we or others are good at judging matters of a certain kind by the results of our judgements. That would suggest that we, or they, should be trusted even when they cannot explain their judgements. This is especially so when understanding of matters in that area is slight. But often no easy guides like this one are available. The most radically particularistic conclusion is indicated by the discussion of Winch's article. My conclusions apply in a restricted domain, typically where the impersonal reasons are incommensurate. But within this domain they are radical. As Winch indicates, they show that reasons for action are not universalizable. This is consistent with the intelligibility of reason, practical reason included. What makes it right for Vere to decide as he did, and for Winch to take the opposite decision, can be explained, and the explanation, relying on the difference in their moral character, and in the concrete fact that they did decide as they did, is universalizable. Whenever two situations are evaluatively different the difference can be
78
Joseph Raz
explained in universalizable terms. But the explanations are not themselves reasons, and they need not refer to factors which are reasons. In this case they refer to the agents' moral character, and to the fact of their decisions, neither of which are reasons for these agents.
4
Ethical Particularism and Patterns Frank Jackson, Philip Pettit, and Michael Smith
Information couched in descriptive terms plays a major role in determining our moral judgements. Perhaps we learn that an action involves breaking a promise and respond by forming the view that it is wrong. Later we learn that it was necessary to break the promise in order to save a life, and retract our earlier judgement and decide that the action was right. Later still we learn that, although a life was saved, many more were lost as result of the promise-breaking and we return to our original judgement. Here we have a simple example of the role of descriptive information in leading us to a moral judgement, and of how our judgement may change as more descriptive information comes to hand. A familiar question in meta-ethics is the status of the passage from the descriptive to the evaluative, from, as it is so often put, an 'is' to an 'ought. Is it an entailment? How might it be justified, if at all? Is it sui generis? Is it to be understood as some kind of rationally defensible adopting of an attitude? And so on. However, our focus will be on the question whether there is a pattern in the transition, rather than on the status of the transition itself. We will be concerned with whether there is a pattern to the way descriptively given information determines moral conclusions, and, more generally, with whether there are patterned interconnections between the non-evaluative and the evaluative. Utilitarians say that there are relatively simple patterns, and that they know what they are; they say, for example, that if the ethical conclusion is that X is right, the pattern is given by the rubric: X is right if and only if X maximizes expected happiness. Others say that the pattern for rightness is given by: X is right if and only if X is what an agent who exemplified all the virtues would do; or by: X is right if and only if X satisfies a certain We are indebted to Richard Holton, Rai Gaita, and, especially, Jonathan Dancy, for their many helpful comments and conversations.
80
F. Jackson, P. Pettit, and M. Smith
weighted sum of prima facie duties better than any available alternative to X; or by: X is right if and only if X is what an ideal agent would desire to desire to do.' Still others are agnostic about what the pattern is but are confident that there must be one; perhaps they hope to find the pattern during their next study leave and tell us what it is in a future article or book. Much of the history of normative ethics is the history of attempts to find and state the pattern in some set of more or less complicated principles. Let's call the consensus that lies behind this history principle-ism. Our concern in this chapter is with a major challenge to this consensus that often goes under the name of particularism. Particularism's best known contemporary defenders are perhaps John McDowell, Jonathan Dancy and David McNaughton. Margaret Little has recently offered a particularly clear and concise account of the doctrine, and in what follows we have been much influenced by her formulations of the issues. 2 We should, though, emphasise that our focus is on the radical, interesting view we find common to their writings, not on textual fidelity to any individual presentation. According to these theorists, the relationship between descriptive or non-evaluative information, on the one hand, and a moral or evaluative verdict, on the other, is not merely complex—pace, say, utilitarianism—it is irreducibly complex. There is no codifiable pattern to be found in the passage from the descriptive to the ethical, and vice versa. Little puts the basic idea in a number of ways: 'There is no way of cashing out propositionally the ways in which non-evaluative properties contribute to the evaluative natures of situations, actions, characters.' 'The particularist's claim is that the good-making relation cannot be cashed out in propositional form.' [Particularists] share the intuition that moral properties are, to use Simon Blackburn's felicitous phrase, "shapeless" with respect to the nonmoral: `To understand the real lesson of particularism is to understand that there is reason to doubt the existence of any codifiable generalities linking moral and nonmoral properties.' A familiar objection to particularism is an epistemological one: we need moral principles to arrive at and justify our moral judgements. 3 ' We are thinking, of course, of versions of these views that cash out their key notions in descriptive terms. 2 John McDowell, 'Virtue and Reason', The Monist, 62 (1979), 331-50; Jonathan Dancy, `Ethical Particularism and Morally Relevant Properties', Mind, 92 (1982), 530-47; Jonathan Dancy, Moral Reasons (Oxford: Basil Blackwell, 1993); David McNaughton, Moral Vision (Oxford: Basil Blackwell, 1988); Margaret Little, 'Moral Generalities Revisited', Ch. 12 this volume. Quotations in this paragraph of text from Little's chapter are from 283, 285, 279, 288 respectively. 3 Russ Shafer-Landau, 'Moral Rules', Ethics, 107 (1997), 609.
Ethical Particularism and Patterns
81
Particularists talk of coming to know the moral landscape by discernment, or in similar terms; principle-ists hear this as a refusal to engage properly in the business of justifying one's moral judgements. They argue that we cannot properly adjudicate between competing discerned judgements except by noting the principles that one or the other discerned judgement falls under, or fails to fall under. However, our line of objection to particularism turns more on semantic and metaphysical considerations, as will emerge. But, first, it will be helpful to address some preliminary matters.
Preliminaries
The first preliminary concerns the distinction between, on the one hand, the descriptive, non-evaluative, factual, natural etc. and, on the other, the evaluative, ethical, normative, moral etc., that figures centrally in statements of the issue between particularism and principle-ism. All we can do here is presume some reasonable conception of the distinction and note that particularists do likewise. The statements of the particularist credo we gave above, and the statements and arguments to be found in the particularist literature, would all be a nonsense if there is no viable distinction to be drawn. Indeed, whether or not the relation between the descriptive and the moral is codifiable is not even a subject for discussion if there is no viable distinction between the descriptive and the moral to start with. Though we will try, as much as possible, to frame matters so as to avoid begging controversial questions about the nature of the distinction, our own view is that it is probably best to think of the distinction as one between vocabularies rather than properties or states of affairs as such. For example, those who think that moral properties and states of affairs are descriptive or natural properties and states of affairs, still think that there is an important question concerning the relations between matters framed in the language of morals and matters framed in the language of the natural sciences. The question as to whether there are principles, properly socalled, of the form 'D E, where 'D' and 'E' are descriptive and ethical sentences, respectively, is a question of interest independently of whether or not one should think of the sentences on each side of the conditional operator as concerning different properties or states of affairs, or as concerning the same properties and states of affairs but picked out in different terms. We will, though, sometimes speak loosely in the interests of brevity of, for example, moral properties and situations, when strictly it would be best to talk of properties and situations picked out in moral terms. Also,
82
F. Jackson, P. Pettit, and M. Smith
although we will conduct the discussion in terms of the 'thin' moral terms—indeed, we will mainly focus on the term `right'—what we say could be said mutatis mutandis about the thick moral terms like 'courageous' and 'generous. Our case for saying that there are patterned interconnections between the descriptive and the ethical is independent of whether the ethical is thought of as thin or thick. The second preliminary concerns the bearing of 'hedged' generalizations of various kinds on the debate. It is not in dispute that many acts of deliberate torture are wrong. It is not in dispute that when they are wrong, very often the reason that they are wrong is that they are acts of deliberate torture. Or at least these facts had better not be in dispute. It might, accordingly, be suggested that we can non-controversially say 'Typically or other things equal or . . . an act of deliberate torture is wrong, and that particularists can only be objecting to ethical theories that offer neat, exceptionless generalizations; their objection cannot be to patterns and principles linking the descriptive and the moral per se; it must be to ones that seek to avoid terms like 'other things equal' and 'typically. However, this would be to misunderstand the radical, and radically interesting, nature of the particularists' proposal. They are much more than pluralists about value who insist that, when we try to state how the different values stack up, we cannot avoid the hedgers' usual suspects. Nor do they belong to the party which insists that these usual suspects are ultimately vacuous, that all you can ever really mean by 'As are typically Bs' is that As are Bs when they are Bs. They have too much respect for commonsense to hold that self-denying position. Their view is that we cannot understand the hedge terms descriptively, that they do not capture something about the descriptive way things are. Perhaps the key point can be best grasped via a simple example. Consider the following raft of true conditionals connecting facts about particular heights with facts about who is taller: If x is 180cm and y is 190cm, then x is shorter than y If x is 185cm and y is 190cm, then x is shorter than y If x is 180cm and y is 170cm, then x is not shorter than y and so on. There is an obvious pattern in the antecedents, and, once you have grasped it, you have grasped what it is for someone to be shorter than someone else. What is required is that you latch on to the right way to go on, that you see what the 'and so on' comes to. There are contentious issues here; namely, those discussed under the heading of the rule-following
Ethical Particularism and Patterns
83
debate. All the same, whatever sceptics may say, we do grasp what is meant by 'and so on' in these kinds of contexts, and, in doing so, we grasp a similarity among the states of affairs specified in the antecedents. The contention of the particularists is that, when given a list of conditionals of the form If D, then E where the Di are various descriptive states of affairs in which some particular moral claim E is true, no matter how long and varied the list may be, we can never say 'and so on. The problem is not the term 'and so on' as such. As we have just seen, that can play a perfectly legitimate role. The problem, according to particularists, is that there is no projectible pattern in the As to latch on to. There is no pattern in the Dis, the grasping of which would enable you to write down new members of the list. The same applies to terms like 'typically' and 'other things equal. When biologists say what hearts typically do, they say what, among the hearts, is typical; the problem, according to particularists, with saying that torture is typically wrong is not that it is false, but that what is typical is not typical among the relevant descriptive circumstances: you cannot find the 'shape' if you restrict yourself to the descriptive facts alone. The third preliminary concerns what particularists mean by holding that the relation between the descriptive and the moral is irreducibly complex; that the moral is shapeless. The doctrine is not that there is a pattern in the descriptive facts that underlie an act's being, say, right, but it is a highly complex, difficult-to-spot one. That view would simply be a principle-ism that maintained that the principles are complex and hard to spot. The doctrine is that there is not even a highly disjunctive commonality or pattern that unites the right acts when described in descriptive terms. It is not, for example, like Wittgenstein's famous example of a game and, more generally, of family resemblances. In these cases, it can be difficult to spot or state the pattern, but the fact that, given a large enough diet of examples, we can say of some new case whether or not it is, say, a game (or, perhaps, that it is indeterminate whether it is or not) shows that there is a pattern we can latch on to; our ability to project shows that we have discerned the complex commonality that constitutes the pattern. As a final preliminary, we should mention that the contention that there is no pattern in or among the descriptive facts underpinning some given moral category is occasionally expressed in an unfortunate way. Sometimes particularists express it by saying that we could not grasp the relevant
F. Jackson, P. Pettit, and M. Smith
84
pattern unless we had the relevant moral concept. 4 However, this is something that analytical descriptivists in ethics accept. Consider, for example, analytical utilitarians who hold that 'X is right' means 'X maximizes expected happiness. According to them, the relevant descriptive similarity among right acts is maximizing expected happiness, and you cannot grasp that without grasping the relevant moral concept because that is the relevant moral concept. But, of course, analytical utilitarianism is an extreme version of exactly the kind of doctrine that particularists oppose. The key issue, therefore, is not whether you can grasp the descriptive similarity without grasping the moral concept, but whether there is a descriptive similarity to be grasped. Supervenience Conditionals Particularists typically grant the supervenience of the ethical on the descriptive, but insist that it is consistent with their view. We agree that supervenience, in and of itself, is compatible with their view, but will argue that considerations that take off from the fact of supervenience raise serious problems for particularism. Supervenience is the thesis that descriptively identical situations, actions, characters and so on are evaluatively identical. It comes in two versions in discussions in ethics. One is a global thesis, and one is an intraworld thesis. The global thesis says that descriptively identical worlds are morally or evaluatively identical; the intra-world thesis says that descriptively identical acts, states, etc. within a world are morally identical. From the global thesis, the version that will mainly concern us here, it follows that there are necessary truths that take us from the descriptive way things are to the moral way they are; if the moral nature of a world cannot vary independently of its descriptive nature, then descriptive nature fixes moral nature. In particular, any complete specification of the descriptive nature of a possible world—a specification that is true at that possible world and at all possible worlds that are descriptively exactly like that world—necessarily determines whether or not, say, X is right in that world. There will, therefore, be a raft of necessarily true conditionals whose antecedents are complete specifications of the descriptive nature of a world and whose consequents say that X is right, and another raft of necessarily true conditionals whose antecedents are complete specifications of the descriptive nature of a world and whose consequents say that X is not right. 4
See, e.g., Dancy, Moral Reasons, 78, and his discussion of McDowell on p. 79.
Ethical Particularism and Patterns
85
To simplify the discussion, let's focus on the various complete descriptive specifications of ways things might be that determine that X is right. Supervenience tells us that there is a raft of conditionals of the following form: If Di , then Xis right If D2, then Xis right. We can write this as a single conditional, thus If DI or D2 or . . ., then X is right. Here we have a conditional that takes us from the descriptive to the moral. But particularists are right to urge that this fact is, in itself, no vindication of principle-ism. The reason is that, for all that supervenience says, there need be no pattern in the dependence of the moral on the descriptive reflected in this conditional, or, equally, in the raft from which it was constructed; there may be no pattern unifying the Di. The conditional per se does not constitute a principle of the sort in which the principle-ist believes. For all that supervenience says, the assignment of moral properties among the various complete descriptive states of affairs could be essentially random. Provided only that identical descriptive states of affairs are assigned the same moral predicates, supervenience will be respected. As the point is important, we will labour it with a simple example; also the example will be useful later in the paper. Suppose we construct a machine that flashes a light only when objects of certain shapes are placed in front of it. Suppose, further, that we programme which shapes will, and which shapes will not, trigger the flash of light by using a table of random shapes and the following rule: a shape triggers a light flash if and only if its first appearance in the table is at an oddnumbered place. If, per impossible, every possible shape appears in the table, we will have two conditionals of the form If a presented object has shape . . . or . . . or . . ., the light will flash If a presented object has shape . . . or . . . or . . ., the light will not flash whose antecedents between them cover every shape. It will then be true that identity in triggering light flashings supervenes on identity in shape. However, there will be no pattern in the connection between shapes and light flashings. Or, more precisely, there will be no pattern in the shapes
86
F. Jackson, P. Pettit, and M. Smith
themselves. There will, of course, be an extraneous pattern: the shapes that trigger the light will all share the property of having made their first appearance at an odd-numbered place in the random table, and the shapes that fail to trigger the light will have made their first appearance at an evennumbered place in the table. The upshot is that, although supervenience tells us that there are necessarily true conditionals that take us from descriptive ways things might be to moral ways things might be, it is a separate question whether there are moral principles in the sense of patterned connections between descriptive ways things might be and moral ways things might be. This sounds like good news for particularism, but, in fact, when we look at the various ways that supervenience might be respected without there being the kind of patterned connections between the descriptive and the moral that principle-ists affirm and particularists deny, we find serious problems in each way—or so we now proceed to argue. As we said earlier, supervenience in itself is compatible with particularism; it is considerations that take off from it that cause the trouble. We start by considering the suggestion that the connection between the descriptive and the moral is essentially akin to that between shapes and light flashings in our example—that is, that the reason that there is no pattern in the connection between the descriptive and the moral, over and above the minimum required to respect supervenience, is that we are dealing with what is, at bottom, a random phenomenon.
Could the 'Connection' be Random? We suspect that few particularists will want to embrace this suggestion, but we place an objection to it on the table, nevertheless. It is important that it be clear that the suggestion is bizarre and, hence, that a major question for particularists is how their view differs from it. Also, the thought behind the objection will play an important role in later sections. We can diagram the suggestion as follows (Figure 1). The 'randomness' suggestion is that there is no pattern uniting what lies inside and outside the circles in either case. In neither case could you say for a new case—a new shape or a new descriptively specified action—based on the answer for as many old cases as you care to nominate, whether it fell inside or outside the circle. The basic objection to this suggestion is a semantic one. We use words to mark divisions. Tables are different from chairs, and we mark this by using
Ethical Particularism and Patterns
87
descriptively specified right acts
Figure 1
different words for them. In the same way, wrong acts are different from right ones—how else could it make sense to care which we did? And we use the moral terms to tell each other about the difference; the word 'right' is (and had better be on pain of not knowing what papers and discussions in ethics are about) a good word for talking about right acts. What, then, marks off the acts we use 'right' for from the acts we use 'wrong' for? Or, equivalently, what do the right ones have in common that the wrong ones lack? Particularists cannot answer that what unites right actions is simply the fact that we properly apply the predicate 'is right' to them. The problem with this answer can be variously put by saying that there is no such thing as bare predication, that predicates apply because of how things are, or that predication supervenes on nature. They might say that all that the right actions have in common is that they belong to the set of right actions. Grasp of the predicate 'is right' simply consists in a grasp of the various Di which constitute that set. But this cannot be all that unites the class of right actions. There must be some commonality in the sense of a pattern that allows projection from some sufficiently large subset of the Di to new members. If there isn't, we finite creatures could not have grasped through a finite learning process (the only sort there is) the predicate 'is right'. So, there must be a pattern or commonality—in the weak sense operative in this paper of that which enables projection—uniting the set of right acts. It might be objected that, pace what we said before in the preliminaries, Wittgenstein's example of family resemblances shows that this line of thought is mistaken. A diet of examples, or putative examples, can give us understanding of a term, can allow us to grasp a concept, without its being the case that there is a pattern exemplified by the examples, namely, the pattern whose grasp underlies our ability to say of new cases whether or not they fall under the concept. What shows this is that, in the case of family
88
F. Jackson, P. Pettit, and M. Smith
resemblance concepts, new cases often call for decision—perhaps arbitrary, perhaps guided by 'external' considerations. 5 But then, the argument might continue, there is no pattern, because if there were, no decision would be called for. However, if there is no pattern in the diet of examples, every new case would call for decision, and any decision would be as good, semantically speaking, as any other. Sceptics about meaning can perhaps embrace this conclusion, but meaning scepticism is a high price to pay for particularism in ethics. We can all agree that there are cases where it is indeterminate whether or not some concept or term applies (and this is consistent with there being a pattern, because it can be indeterminate whether or not a pattern is exemplified), but only a meaning sceptic accepts wholesale indeterminacy. But if there must be a pattern uniting the right acts, either it is a descriptive one, in which case particularism is false, or it is one which cannot be understood in terms of the presence or absence of the descriptive—something unanalysable and non-natural, as G. E. Moore put it when discussing goodness. 6 If this is the particularists' view, however, then we think that they can fairly be accused of false advertising. Under examination the new and exciting thesis that there are no moral principles collapses into the jejune doctrine advanced by Moore at the turn of the century: moral properties are sui generis, and hence are not to be found among the descriptive. It might be objected that there is another possibility. The pattern uniting the right acts might be neither descriptive nor a sui generis Moorean one; it might be something like being something that there is a good reason to do or a pattern capturable in terms of the thick moral concepts. However, being something that there is a good reason to do, along with being generous and the like, supervene on the descriptive in exactly the same way that rightness does: two descriptively identical acts cannot differ solely in that one is something that there is a good reason to do whereas the other is not; two descriptively identical acts cannot differ solely in that one is generous whereas the other is not. This means that the same line of argument applies to them. Our language for talking about them—the predicates 'is something that there is good reason to do' and 'is generous'—must apply because of the nature of what they apply to. There must, therefore, be a pat-
5 H. L. A. Hart's example of the distinction between the legal and the illegal may well be one where new cases often call for decision, but there are external considerations in the sense of general ethical and practical considerations relevant to which decision should be made. 6 G. E. Moore, Principia Ethica (Cambridge: CUP, 1929), ch. 1.
Ethical Particularism and Patterns
89
tern that makes it intelligible how we could have mastered that language. But, given their supervenience on the descriptive, the choice is then as before: either there is a projectible descriptive pattern, or else the unifier is sui generis.
Uncodifiable Patterns?
It might be objected that we moved too quickly when we said that if there is a descriptive pattern, particularism is false. Particularists can and should allow that there are patterns in the way the moral connects with the descriptive. Their claim is rather that these patterns are uncodifiable or escape propositional capture. But what does this amount to? Perhaps to be codifiable or propositionally capturable is to be expressible in language. The particularists' claim would then be, on the construal now in play, that there is a pattern in the way that the moral connects with the descriptive, but it is one we cannot, as a matter of principle, express in words; there is, for example, a descriptive pattern uniting the right acts but we cannot say what it is. One way this might be true is by virtue of the descriptive patterns being ones we cannot know: if we cannot know what the patterns are, we can hardly capture them in words. But unknowable patterns present similar problems to non-existent ones. We noted earlier that if the connection between descriptive ways things are and moral ways things are is a random one, then it is impossible to see how we could have come to grasp moral concepts by exposure to, or reflection on, a finite number of cases. The same is true if, as far as we can tell, the connection is a random one. We might know that some descriptive similarities or other were germane to questions about the distribution of moral properties, but if we could not know which they were, we would not know which similarities we could properly regard as germane. Moreover, the suggestion is an open invitation to scepticism: if there is a descriptive pattern that settles what is right but we cannot know it, we cannot know what is right. The view is an invitation to scepticism in the same way as one that holds that what is right is settled by what God approves of, at the same time as holding that we cannot know what God approves of. Therefore, particularists who hold that there is a pattern should allow that it can be known. But then it is hard to see why we could not capture it in words. There are many things we know that we do not have words for— for example, the number of colours we can recognize substantially outruns
90
F. Jackson, P. Pettit, and M. Smith
our colour vocabulary—but it is quite another thing to say that we could not have words for them. Surely, if the pattern that connects matters described in descriptive terms with matters described in moral terms can be discovered, then, after we have made the discovery, we can tell people about it. Indeed, typically, to discover a pattern is to discover how to conceptualize it and, thereby, how to capture it in language. For example, when presented with a sequence of numbers that exemplifies the pattern: the number at the first place is 1 and the number at any later place is the sum of the numbers at all earlier places, spotting that this is the pattern is seeing how to conceptualize it. Particularists might appeal at this point to the possibility of partial knowledge of a pattern. Perhaps we know that there is a pattern, perhaps we have some rough idea of the cases that fall under it and roughly why they do, but what escapes us, as a matter of principle, is full knowledge of the pattern. One way to flesh this idea out is to pretend that there is a superbeing, God, who can comprehend very complex patterns. She alone grasps in full the pattern in the way that moral matters connect with descriptive ones. What we are doing, the suggestion might run, is trying to latch on to the pattern She sees. We succeed to a limited extent, which means that the connection between matters descriptive and matters moral does not look like a complete 'mess' to us, but we cannot succeed in full. This, it might be suggested, is the sense in which principle-ism is false. The trouble, however, is that different super beings will see different patterns, and, more generally, that a partial grasp will leave open a number of candidates to be that which is being grasped (that is what makes it partial). So the key question becomes: Which is the right pattern for moral rightness, goodness, and so on? How should we choose among the candidates? Theists might appeal to the pattern privileged by being the pattern discerned by God—the all-knowing, all-good, all-powerful creator of everything—but this is of little help to us atheists. Followers of Moore might insist that the right pattern is given by the distribution of sui generis moral properties of various kinds, but, as we noted earlier, this would be to turn the new and exciting doctrine of particularism into the jejune doctrine of Moorean non-naturalism. All that remains, as far as we can see, is that it should be something about us that settles which patterns are the right ones. Something about the rules we follow, explicitly or implicitly, in our use of the moral vocabulary settles the connections between the descriptive and the moral. After all, it is we who are using the moral terms to draw various distinctions. How such an account might be spelt out is a hard and controversial
Ethical Particularism and Patterns
91
question which we have addressed elsewhere in a little detail.' What is immediately to the point at hand is that if it is something about our pattern of usage of the moral vocabulary that determines the patterns, then finding the right patterns cannot be impossible in principle. Think of what grammarians say about our classifications of sentences into the grammatical and the non-grammatical. They note that the classifications do not happen by magic; they note that there is obviously a pattern in the data to be discerned somehow or other; and they note that we are finite beings, so that, although the pattern may be complex, it cannot be impossibly complex. The task that remains is to find the pattern in our usage of the terms 'grammatical' and 'non-grammatical'. We say the same about the classifications we effect using moral language. Of course, it does not follow from this alone that we can capture the patterns in the distinctions we mark using moral terms descriptively. The reason for thinking that, in addition, we had better be able to capture them descriptively lies in the supervenience of the moral on the descriptive and the problems for appeals to the sui generis. Although we insist that there must be a pattern and that it must be codifiable in principle, we take no position on whether it is sometimes, always, or never codified, in the sense of being 'before the mind, when a moral judgement is made. Some particularists insist that the right way to make a moral judgement in a given case is to consider the case in all its particularities and follow one's intuitive moral response, allowing for the possibility of subsequent revision, perhaps as a result of criticism by others or more thoughtful reflection. They oppose the view that we should make moral judgements by first seeking acceptable ethical principles and then seeing whether the case at hand falls under them. Our view, as it happens, is that this is a case by case matter. In the case of judgements of ungrammaticality, sometimes we do best to go by the fact that the sentence 'looks funny, and sometimes by the fact that it violates a principle. We think a broadly similar, case by case approach is best in ethics (while differing about exactly why this is so). But we do not argue for this view here. Patterns and Responses
We have argued that there must be discoverable descriptive patterns unifying cases that fall under the same moral classification. There are, however, two different, possible descriptive patterns to distinguish here. Michael Smith, The Moral Problem (Oxford: Blackwell, 1994); Frank Jackson and Philip Pettit, 'Moral Functionalism and Moral Motivation', Philosophical Quarterly, 45,178 (1995), 20-40.
92
F. Jackson, P. Pettit, and M. Smith
Consider a familiar question that arises in discussions of colour. One thing that unites red things is that they typically look red. It is plausible that this is part of our concept of red, which is why it is plausible that something (something) along the lines of 'xis red if and only if x typically looks red in normal circumstances' is a priori. However, we can be confident that something further unites red things. The fact that colour vision evolved—in part, it seems, to help detect what is edible—tells us that the unity in our response to red things—their looking red in normal circumstances—connects to some underlying common feature, albeit one which is moderately disjunctive and of interest only to creatures with certain kinds of physiological makeups. There are, that is, two patterns: one in the nature of our response, and the other in the nature of what our response is to. The same is true quite generally of the classifications we effect with terms like 'comic, 'attractive, and 'calming'. For example, comic situations are united in what they typically tend to do to us, their tendency to make us laugh; this is a conceptual fact about the comic. 8 But they are also united in what explains what they typically tend to do to us. It is the latter that essays on the essential nature of comedy are usually directed to elucidating. One suggestion, for example, is that comic situations are united by the fact that they make us expect a connection between two ideas when we know perfectly well that no such connection really exists; it is this feature that underwrites the tendency comic situations have to make us laugh. Now, many have held that the concept of rightness is like the comic, colour, and so on, in having a response-dependent element. Just as part of what makes something comic is its effect on us, so part of what makes something right is its effect on us. For example, one suggestion is that part of what makes something right is that it ideally tends to attract us, and this is the sense in which internalism in ethics is true. 9 If this, or something along these general lines, is right, and we think it is, there are two questions to ask about right actions: Is there a descriptive pattern essentially involving our responses to right actions? and: Is there also a descriptive pattern in the right actions themselves? We might call the view that there is no descriptive pattern in the right actions themselves restricted particularism. It holds that all that unifies the 8 This is putting the point very roughly. There is, of course, a normative element to be acknowledged. Some things that tend to make us laugh ought not to do so, and are thereby disqualified from being comic. 9 Response-dependence in this sense is a matter of certain responses being part of what makes a term or concept apply to something. Response-dependence in a quite distinct (and, we would argue, much more ubiquitous) sense is a matter of certain responses being in part responsible for a term or concept having the content that it does.
Ethical Particularism and Patterns
93
right (and, for that matter, the good, the bad, and so on) lies in something about our responses. This can be given descriptively, and so, in one perfectly good sense, there are principles, properly so called, running to and from the descriptive and the moral, and the moral is not shapeless with respect to the descriptive. However, on this view, its shape can only be discerned when you step back and see its effects on us. Obviously, restricted particularism is a substantial retreat on what some particularists want to say—the moral is shaped, albeit that its shape comes from our responses— but we hazard, all the same, that restricted particularism is, at bottom, the view of many who call themselves particularists.10 Restricted particularism is not subject to the semantic argument we brought against full-blown particularism. By allowing that there are descriptive patterns unifying the situations, acts, characters, and so on that fall under some given moral classification, it respects the supervenience of predication on nature without thereby being committed to the existence of a class of Moorean sui generis moral properties to provide that nature." It is, however, very hard to believe that the only way to unify right acts is by looking at the descriptively specified, response-dependent role that they play—and hence that there is no descriptive unity in the acts themselves. The reason is that if this were true, the acts themselves would have to be as randomly related as the set of shapes that trigger the light in our example earlier in the paper. When we look at an act itself, independently of the response-dependent role it plays, it is all a 'mess. But this would violate the platitude about moral argument that, in debating controversial moral issues, a central role is played by various similarity claims, claims of the form 'Given you say that about this case, the onus is on you to explain why you do not say the same about this other similar case,' where it is often clear 1 " Our qualified suggestion is that restricted particularism is the view McDowell defends in 'Virtue and Reason'. Our evidence for this is McDowell's assimilation of values to secondary qualities, as opposed to primary qualities, in Ted Honderich (ed.), 'Values and Secondary Qualities', Morality and Objectivity (London: Routledge & Kegan Paul, 1985). Our suggestion must be qualified, however, because in the latter paper McDowell does appear to back away from the idea that we could give a purely descriptive characterization of moral features as dispositions to elicit responses in us: 'we make sense of fear by seeing it as a response to objects that merit such a response' (p. 119). We must admit that this does sound like the postulation of a sui generis Moorean moral property of being meritworthy. The problem, of course, is that if McDowell does postulate such a property then it is difficult to see what the assimilation of values to secondary qualities, as opposed to primary qualities, really amounts to. 11 For the record, Mooreans can allow that there are descriptive patterns in the right acts themselves. They are simply committed to denying that the descriptive pattern provides us with a semantics for 'right'. As proof of this, witness the example of Moore himself who was, after all, an ideal utilitarian, and hence, in one perfectly good sense, a principle-ist.
94
F. Jackson, P. Pettit, and M. Smith
that the similarities in question are descriptive ones in the acts themselves, as opposed to similarities in the response-dependent role they play. For example, defenders of abortion are challenged to explain why they oppose infanticide; those who oppose contraception on the grounds that it is unnatural are asked to explain why they do not oppose the wearing of spectacles; and meat eaters who oppose sexism and racism are asked about speciesism. While the force of this kind of challenge is always open to debate in any given case, it is incredible that there is something in principle wrong with making it. It is, surely, a platitude that, in any discussion that counts as being about morality, one who claims that acts of a certain sort are right while claiming that acts of a similar but not identical sort are wrong is required to justify themselves; it is a platitude, that is, that descriptive similarities and differences in acts are relevant to moral similarities and differences in acts. But, if the connection between the descriptive nature of acts themselves and the moral is random, it is random. To think that descriptive similarities and differences in acts have something to do with moral similarities and differences would be like thinking that, in our shape example, shape similarities and differences are relevant to similarities and differences over whether or not they trigger the light. (Moreover, we had better have some idea of which descriptive similarities matter, otherwise we would be at a complete loss to know which to appeal to.) Particularists sometimes appeal to the idea that there can be patterns that only become visible at certain levels of generality. They say, in effect, that similarities can emerge, and cite the famous Putnam 'round hole, square peg' example. They argue that when a square peg fails to go through a round hole whose diameter equals the side of the peg, although there will always be an explanation in terms of proton positions, what unites the phenomena is invisible at the level of protons and their positions. This is hard to believe. Surely what unites the phenomena at the level of proton positions are how far apart various protons are from one another and the rigidity of the lattices they make up. It is true, of course, that what unites the phenomena will be harder to spot if we are restricted to framing our information in terms of proton positions, but the question concerns what is possible in principle, not degrees of difficulty. But, in any case, the crucial point here is that if it really were true that what unites the phenomena is invisible as a matter of principle at the level of protons, then it would be a fundamental mistake to argue from similarities at the level of protons to similarities at the level of round holes and square pegs. If it really is a complete 'mess' at the proton level, the best one could ever do would be to argue from identity at the level of proton positions, to identity in behaviour
Ethical Particularism and Patterns
95
of the macroscopic objects. In the same way, the doctrine that moral similarities emerge from the descriptive facts would not help the restricted particularist explain the relevance of descriptive similarities in acts to ethical debates; rather, it would mean that such similarities were irrelevant except in the case where they amount to exact similarity in every descriptive respect. It is this that is so hard to believe. A variant on restricted particularism which allows that there are pertinent descriptive similarities in the acts themselves holds that items that fall under some moral classification form regions that display intra- but not inter-descriptive similarity. The situation is diagrammed below in Figure 2 for the case of right actions. Inside the circles are items that are suitably descriptively similar—how similar might well vary from one case to another—but there is no similarity between items in different regions. Would this possibility be one where there was no pattern in all the right acts? It depends on whether there are indefinitely many such regions. If there are only finitely many, we have the situation envisaged by some pluralists about value and we have automatically a pattern—a pattern made up, in effect, by a finite number of disparate disjuncts. It is a pattern because it is projectible—it is like the pattern we grasp when we grasp how to use the phrase 'is a rock or a number or a tiger'. Moreover, to refuse to count it as a pattern would simply reduce particularism to a version of pluralism about value which has principles linking the descriptive and the ethical that have, on the descriptive side, a finite disjunction with disparate disjuncts. If, on the other hand, there are indefinitely many such regions, we may have a case where there is no pattern in all the right acts. We would, nevertheless, have a view which respects the platitude about moral argument that similarities in acts are relevant to moral conclusions. This is because, for any right (wrong, etc.) act, there is a region of descriptive similarity around that act which contains only right (wrong, etc.) acts.
Figure 2
96
F. Jackson, P. Pettit, and M. Smith
However, in view of the fact that we are finite beings, it is very hard to see how the needed unifier for all the right acts—the unifier we need by the semantic argument—could consist solely in some response to those acts. There is a limit to how many differences we can register. Sooner or later, the differences between the regions cease to matter and we can lump all regions after that point is reached together to get a finite set of 'super' regions, which returns us to the case just discussed. Before we leave the question as to whether there is a descriptive pattern in the right acts over and above the pattern given by the responsedependent role they play, we should emphasize that this question is separate from the question as to whether we can see the interest or point of the pattern independently of the role played. Dancy sometimes seems to be arguing that there need be no 'relevant shape or similarities' in the `resultance base' (his term for the relevant descriptive information) for some moral property; sometimes that the shape or similarity would only be visible to one who had the relevant moral concept; and sometimes that the point or rational interest of the shape or similarity would be unavailable to anyone who lacked the relevant moral concept.' 2 We can agree with the last claim. Although we hold that there is a pattern in the resultance base, we can agree that much of its interest lies in the role it plays. Analogy: there is a pattern in the class of comic situations over and above their effect on us, but much of this pattern's interest lies in its effect on us. Only those who know about this effect understand the point of going to see a Chaplin film.
On the Particularists' Argument from Holism about Moral Reasons
We have argued that there must be patterned connections between the descriptive and the moral, that the moral cannot be shapeless with respect to the descriptive. If our argument succeeds, there must be something wrong with any argument that there is no such pattern, including the particularists' argument from the holism of moral reasons: 3 Can we, though, say something more illuminating than this? Holism about moral reasons holds that any reason R for A being the right thing to do can be defeated by setting it in a wider context. Moreover, 2 See, e.g., Dancy, Moral Reasons, 79 f. The middle claim is essentially the one we argued earlier does not express a claim distinctive of particularism on the ground that it is part of, for example, analytical utilitarianism. 3 See, e.g., Dancy, Moral Reasons, ch. 4.
Ethical Particularism and Patterns
97
it need not be the case that R retains its earlier 'valency'. It may be that, in the wider context, R is now a reason against doing A. In this kind of case, the wider context does not deliver additional reasons that weigh more heavily against A than does R in A's favour; it delivers reasons that turn R into a reason against A, that reverse R's valency, as it is put. Thus, it is argued that the pleasure arising from an action is often a reason for doing it, but if we learn that the action is torturing someone, then the pleasure becomes an additional reason against doing it. The pleasure is not a reason for that gets overwhelmed by the dreadful nature of torture; it makes the torture even worse. Let's grant this (undeniably attractive) way of viewing the matter. Little observes that a similar situation obtains with inductive reasons. E may be a reason for H, while E conjoined with E* may be a reason against H. But it need not be the case that E is a reason for Hthat gets overwhelmed in the sense that E* is a stronger reason against H than E is for H. It may be that, in the context of E*, E is a reason against H. Formally, the situation may be represented by the following inequalities: Pr(H/E) > Pr(H); Pr(H/ E.E*)
EwPr(w/A).V(w), which obtains if and only if V(A.R)>V(A). On this model, it is easy to get cases where R I is a reason for A, R 1 .R2 is a reason against A, but not because R2 outweighs R I . It may be that, in the context of R2, RI is a reason against A. The situation may be as follows: V(A.R1)>V(A); V(A.R1.R2)
98
F. Jackson, P. Pettit, and M. Smith
How can this happen? If the value in the A worlds is mainly in the A worlds which are also R 1 worlds, R1 will be a reason for A. This is consistent with the value in those A.R1 worlds mainly being in the worlds where R2 is not the case, so that R 1.R2 is a reason against A. Further, provided that R 1 .R2 is an unlikely way for either R 1 or R2 to be realised, this in turn is consistent with the R 1 .R2 worlds being especially bad, so that R 1 is a reason against A in the context of R2. Here is a diagram (Figure 3) illustrating the key point. The first number in each region is the value of that region given that A obtains; the second number is the probability of that region obtaining given that A obtains. V(A) = 3 x 0.5 + 4 x 0.2 + 1 x 0.1 + 2 x 0.2 = 2.8; V(A.R1)= (4x0.2+1x0.1)÷0.3=3; V(A.R1.R2) = 1 x 0.1 ÷ 0.1 = 1; V(A.R2)= ( 1 x 0.1 + 2 x 0.2) ÷ 0.3 =1.67. These figures deliver the inequalities.
Figure 3
So, holism about moral reasons in the sense of variable valency is compatible with patterned connections—in particular, the patterned connections endorsed by utilitarianism—between the descriptive and the moral. What is true, though, is that patterned connections are incompatible with what we might call unrestricted holism about moral reasons. Unrestricted holism maintains that, no matter the quantity and nature of the descriptive information you have that provides a reason for some moral conclusion, say, that X is right, more may come to hand that leaves the previous information undisturbed and yet, when combined with it, provides a reason against X being right. Believers in principles that run from the descriptive to the moral must deny this kind of holism. For they believe that there are necessary truths of the form 'D and if `1:)E' is necessary, so is
Ethical Particularism and Patterns
99
`D.I E'. Once you have to hand the descriptive information that makes up the antecedent of some necessary principle that runs from the moral to the ethical, the only way that new information can undermine the moral conclusion is by undermining the original information. What we principle-ists must say, and are happy to say, about the particularists' argument from the holism of moral reasons is that, sooner or later, as more and more descriptive information of the right kind comes to hand, the phenomenon they point to washes out.
Conclusion Particularism is the view that the evaluative is shapeless with respect to the descriptive: there is no descriptive pattern unifying the class of right acts. Against this we argued that, absent a Moorean appeal to the sui generis, particularism falls to a semantic argument. The only plausible explanation of our capacity to use evaluative predicates to mark distinctions in the ways things are is the existence of a descriptive pattern unifying all cases of right action. On our view, an important part of that pattern is provided by the response-dependent role played by moral features. This allows us to separate out a more restricted version of particularism, a version which holds that right acts are only united by the response-dependent role that they play; there is no descriptive unity in the right acts themselves. Against this, we argued that it contradicts the platitude that it is always appropriate to ask why descriptive similarities and differences in the right acts themselves fail to match up with moral similarities and differences. Restricted particularism makes the appropriateness of such appeals inexplicable.
5 Ethics as an Inexact Science: Aristotle's Ambitions for Moral Theory T H. Irwin
Modesty in Ethical Theory
At the beginning of the Nicomachean Ethics,' (hereafter EN), Aristotle warns that we must not demand too much exactness in ethical inquiry (1094b11-14). The difficulty of finding an exact account in ethics results not from our lack of diligence, but from an ineliminable feature of the subject matter. The variability of fine, just, and good things makes it impossible to reach necessary and universal principles; we must say something `roughly and in outline, contenting ourselves with 'usual' (or 'for the most part': hos epi to polu) rather than necessary principles (1094b14-27). Some readers take these passages about the inexactness of universal claims about ethics to support a 'particularist' interpretation. According to a particularist view, Aristotle does not regard general rules as prescriptive guides to action, and does not seek general principles that will guide agents in deciding that one course of action is morally right and another is wrong. Some of Aristotle's later remarks have been taken to support particularism. In his view, what we can say about particular cases is even less exact than what we can say in universal accounts; particular cases must be decided by the agents themselves, who must consider what is opportune (1103b34-1104a2). We cannot be expected to give a full account of particI have received helpful comments from audiences at the Oxford Philosophical Society, the Universities of Nebraska and Toronto, and MIT. I have especially benefited from remarks by Richard Kraut, Jennifer Whiting, David Gallop, David Charles, Lindsay Judson, Pekka Vayrynen, and Roger Crisp. 1 As D. Devereux points out (`Particular and Universal in Aristotle's Conception of Practical Knowledge', Review of Metaphysics, 29 (1986), 483-504, at 499-502), this explicit emphasis on inexactness is absent from the EE (Eudemian Ethics). It is also, as far as I can see, absent from the MM (Magna Moralia), which Devereux does not consider.
Ethics as an Inexact Science
101
ulars. Though we make our account more complete and exact by providing the sorts of details that specify the doctrine of the mean, we cannot make it altogether exact (1109b12-23). The gap that is left open by advice stated in general terms has to be filled by 'perception' (1109b12-23, 1126a35—b4; cf. 1112b34-1113a1). This role for perception reappears in Aristotle's account of prudence (phronêsis). When he claims that prudence is concerned with the 'last thing, namely the particular, he takes it to include some perceptual element (1142a23-30). He connects a perceptual capacity with prudence in his contrast between practical and theoretical understanding (nous) (1143a32—b5). Since prudence does not confine itself to the sorts of general remarks that Aristotle offered in Book II, it must advise us about the particular situations that are difficult to capture in a general account. Since particulars are the province of perception, prudence must include the sort of perception that allows us to make the appropriate judgements about particular cases. According to a particularist view, these passages imply that particular perceptual judgements are the foundation of moral virtue, and that they neither need nor allow any further defence that appeals to general principles. Generalizations, then, are simply incomplete summaries of the considerations that the virtuous person recognizes. Aristotle's accounts of the individual virtues may appear to confirm these claims about perception. He provides a series of sketches of people with the different virtues and vices. These sketches do not include the detailed instructions that we might expect if we thought we were being given a set of general rules for achieving the mean. But in the light of Aristotle's claims about inexactness, these sketches are (according to a particularist) all we ought to expect. 2 If we want to understand the virtues, we need to grasp what the virtuous person grasps in particular perceptual judgements. We cannot grasp that by learning generalizations that are only incomplete summaries of what the virtuous person grasps. We learn to identify roughly the people who are recognized as virtuous, and we learn from them by imitation, so that we eventually acquire the same perceptual capacity that they possess. 2 This point is derived from McDowell, 'Comments on "Some rational aspects of incontinence" ', Southern Journal of Philosophy, 27 (1988) suppl. pp. 89-102, at 93 and n.7: `Aristotle's scepticism about universal truths in ethics implies that the content of this general conception [namely, of what doing well is] cannot be definitively written down, in a shape suitable for deduction of particular practical conclusions. No doubt it can be gestured at, perhaps by way of reminder to people who share it (cf. 1095b4-6), by listing virtues and giving character sketches of their possessors, as Aristotle of course does in Books III—V.'
102
T H. Irwin
If we argue in this way, we attribute a 'particularist' view to Aristotle, in so far as we claim that, in his view, perceptual judgements about particular situations are normatively prior to general rules. Martha Nussbaum, 3 John McDowell,4 and Jonathan Dancy, 5 among others, have defended this posi3 See M. C. Nussbaum, Love's Knowledge (Oxford: OUP, 1990), 68: 'We must notice that rules could play an important role in practical reason without being prior to particular perceptions. For they might be used not as normative for perception, the ultimate authorities against which the correctness of particular choices is assessed, but more as summaries or rules of thumb, highly useful for a variety of purposes, but valid only to the extent to which they correctly describe good concrete judgments, and to be assessed, ultimately, against these. On this second picture, there is still room for recognizing as ethically salient the new or surprising feature of the case before us, features that could not have been anticipated in the rule, or even features that could not in principle be captured in any rule. If Aristotle's talk of rules is of this second kind, there need be no tension at all between his evident interest in rules and definitions, and his defense of the priority of perception. I shall now argue that this is, in fact, the situation, and explore his reasons for giving priority to the particular.' Nussbaum quotes passages from Aristotle referring to the judgement of perception in particulars, and takes these passages to show that Aristotle 'explicitly claims that priority in practical choice should be accorded not to principle, but to perception' (p. 68). See also The Fragility of Goodness (Cambridge: CUP), 299. 4 McDowell, 'Comments on "Some Rational Aspects of Incontinence"', p. 94 and n.12: `If having the right conception of doing well . . . cannot be identified with acceptance of a set of universal rules or principles, something whose correctness we could try to make out independently of their application to particular cases, then there is really nothing for it to be except the capacity to get things right occasion by occasion: that is, the perceptual capacity . . . that singles out the right minor premiss ... There is a link, which these words should suggest, between the unwillingness of many commentators to take Aristotle's particularism fully seriously and the prevalent idea that the concept of eudaimonia is supposed to promise a validation of the particular ethic which Aristotle thinks correct, from outside its own distinctive valuations.' In a later paper, McDowell may qualify the strong claim in 'there is really nothing for it to be'. See 'Some Issues in Aristotle's Moral Psychology', in Cambridge Companion to Ancient Thought: Ethics, ed. S. Everson (Cambridge: CUP, 1998), ch. 5, at p. 111: 'Having the right conception of the end is, at least, a state of one's motivational susceptibilities.' 5 See Jonathan Dancy, Moral Reasons (Oxford: Blackwell, 1993), 50 (reference letters added): `(a) The virtuous person is not conceived of as someone equipped with a full list of moral principles and an ability correctly to subsume each new case under the right one. (b) There is nothing that one brings to the new situation other than a contentless ability to discern what matters where it matters, an ability whose presence in us is explained by our having undergone a successful moral education.' Dancy claims (correctly, to judge by 'If having . then there is really nothing' in the passage from McDowell quoted in fn. 4) that McDowell attributes the conjunction of (a) and (b) to Aristotle. Claim (a) is fairly weak; it simply denies that the virtuous person has a 'full' set of moral principles and an ability to subsume 'each' new case under the right principle. We would agree with this if we supposed that people may be virtuous even if they have less than a full set of principles, and lack the ability to subsume every single new case under one of these principles. If, for instance, we think virtuous people must have a large, though less than full, set of principles, and that they must be able to subsume most, though not all, new cases under these principles, we will agree with (a), but we are not committed to particularism. Our admission that the principles are incomplete is quite compatible with
Ethics as an Inexact Science
103
tion—both as an account of Aristotle and as the truth, and in speaking of particularist views I have their views primarily in mind. 6 `Normative priority' is difficult to describe clearly, but it seems to capture one significant issue. For particularists need not deny that there are true, even useful, general principles in ethics. Nor need `universalists' deny that the trained perception of the virtuous agent is useful, even indispensable, for acting rightly and living well. The dispute seems to be about priority—not about temporal priority in learning, but about theoretical normative priority. Are the true principles true to the extent that they summarize the particular perceptual judgements of virtuous agents, or are the particular judgements correct in so far as they conform to true general principles? Critics who have presented the most careful and detailed case for treating Aristotle as a particularist are also particularists themselves. It is sometimes difficult to tell whether they are actually ascribing an argument for particularism to Aristotle himself, or offering their own arguments for Aristotle's conclusions. This difficulty is especially acute when one reads McDowell, who is committed to particularism on the strength of more general Wittgensteinian arguments that are difficult (for those who lack the insistence that they are normatively prior to (or not normatively posterior to) perception of particulars. Claim (b) introduces particularism. If virtuous people bring nothing to new situations except a contentless ability to discern what matters where it matters, they cannot be guided by normatively prior general principles: for if they were guided by such principles, they would have a contentful ability, whose content is given by generalizations that are always needed, though not always sufficient, for a correct decision. The claim that is rejected in (a) and the claim accepted in (b) do not exhaust the possibilities; if we are to agree with McDowell and Dancy, it must be clear not only that they are right in rejecting the interpretation of Aristotle that they reject (the one rejected in claim (a) ), but that they are right in accepting the interpretation they accept (the one accepted in claim (b) ). To show that McDowell's view is Aristotelian, Dancy adds: 'Anyone who has read Aristotle's Nicomachean Ethics will discern the Aristotelian style of this account of virtue, both in its refusal to see moral judgment as the subsumption of a new case under a previously formulated moral principle and in its stress on the role of moral education' (p. 50). Dancy adds a cautionary footnote to this sentence: 'This claim is, I fear, an exaggeration' (p. 58). I am not sure whether, in Dancy's view, the exaggeration consists in the suggestion that anyone who has read Aristotle's Ethics will discern the things Dancy mentions, or in the suggestion that these things are really present in Aristotle. 6 Nussbaum mentions McDowell as holding a view similar to hers. See Love's Knowledge, 36-7. Particularism is discussed further by R. B. Loudon, Morality and Moral Theory (Oxford: OUP, 1992), 88f, 103f; 'Aristotle's Practical Particularism', in Essays in Ancient Greek Philosophy, iv, J. P. Anton and A. Preus (eds.), (Albany: SUNY Press, 1991). A moderate version of particularism is attributed to Aristotle by N. Sherman, Making a Necessity of Virtue (Cambridge: CUP, 1996), ch. 6 (see esp. pp. 262-4 for the particularist theses that she regards as 'broadly Aristotelian in spirit', and pp. 268-72 on the usual).
104
T H. Irwin
appropriate perceptual capacity) to see in Aristotle.' It is hardly surprising that my own doubts about the truth of particularism influence me in doubting whether it is to be found in Aristotle. Still, we can raise a question about Aristotle's view that is distinct from any question we might raise about the truth of particularism. We refute the case for regarding Aristotle as a particularist if we show that we have no good reason to believe that he regards perception of particulars as prior to acceptance of general principles. To show this, it is not necessary to show Aristotle is a 'universalist' or a `generalist' rather than a particularist; that is to say, he might deny priority to particulars without assigning it to general principles. As I understand particularism and universalism, they are mutually exclusive, but not exhaustive. If we decide that Aristotle is not a particularist, we may then consider whether he is a universalist, or he rejects both particularism and universalism. I am not trying to answer this further question; an answer to it would require us to settle several issues that we can leave open if we are simply considering particularism. I emphasize this point to make it clear that it would be illegitimate to argue from the moral insufficiency of general principles to the truth of particularism. The insufficiency of general principles might (if it is the appropriate kind of insufficiency) constitute an argument against generalism, but it is not sufficient to prove the stronger thesis of particularism, and it is this stronger thesis that I discuss. My argument is purely exegetical; it seeks to decide whether Aristotle can be shown to be a particularist, not whether particularism is true. Still, it may have some philosophical relevance. For it is probably not an accident that particularists appeal to Aristotle. We might assume that if we accept some familiar and plausible Aristotelian claims about generalizations and perception in ethics, we ought to find particularism at least plausible and intuitively appealing. If this assumption is false, we lose some intuitive support for particularism.
The Wittgensteinian arguments are presented in 'Virtue and Reason' in Mind, Value, and Reality (Cambridge, Mass.: 1998), ch. 3.
Ethics as an Inexact Science
105
The Inexactness of Ethics 8
In EN i 3 Aristotle connects the inexactness of ethics with the 'usual' character of ethical generalizations. 9 He intends this discussion for the audience at his lectures; his hearers are to recognize inexactness in the theoretical claims and generalizations that he puts forward (1094b19-1095a13). He is not directly concerned, in these remarks in Book 1, with all moral agents or with the generalizations that should be presented to them when they face particular choices in particular situations. In saying that ethical generalizations are not exact, Aristotle means that (1) they are not about subject matter that allows the truth of unqualified generalizations, and that (2) they are not sufficiently qualified to take account of all the relevant exceptions to unqualified generalizations. This second feature, however, still raises a question. When Aristotle says 'we would not seek the same degree of exactness in all sorts of arguments alike, any more than in the products of different crafts, we might understand him to make either of two claims: (a) The material we are dealing with simply does not allow exact treatment. (b) Even if we could treat it exactly, exact treatment would be inappropriate. Aristotle gives some examples to clarify his claim that ethical generalizations are inexact and merely usual (1094b14-19). He mentions difference and variation in fine and just things, and notices that some people have inferred from this difference and variation that things are just and fine merely by convention. 10 8 Passages on inexactness are discussed by G. Anagnostopoulos, Aristotle on the Goals and Exactness of Ethics (Berkeley: University of California Press, 1994). 9 I add numbers and letters to mark the stages of Aristotle's argument: `(1) Our discussion will be adequate if its degree of clarity fits the subject-matter; for we would not seek the same degree of exactness in all sorts of arguments alike, any more than in the products of different crafts. (2) (a) Now fine and just things, which political science examines, have much difference and variation, so as to seem to be by convention, but not by nature. (b) Goods, however, also have the same sort of variation, because harm comes to many people from them; for it has happened that some people have been destroyed because of their wealth, others because of their bravery. (3) Since these, then, are the sorts of things we argue from and about, it will be satisfactory if we can indicate the truth roughly and in outline; since [that is to say] we argue from and about what holds good usually, it will be satisfactory if we can draw conclusions of the same sort. (4) Each of our claims, then, ought to be accepted in the same way [as claiming to hold good usually], since the educated person seeks exactness in each area to the extent that the nature of the subject allows; for apparently it is just as mistaken to demand demonstrations from a rhetorician as to accept [merely] persuasive arguments from a mathematician,' (1094b11-27). 10 'Difference and variation' (diaphoran kai planen) could be taken in two ways: (a) Aristotle appeals to subjective variation. Fine and just things cause many differences and
106
T H. Irwin
Aristotle answers the inference from variation to convention by pointing out that goods are also subject to variation. It would be quite implausible to claim that since some goods sometimes result in harm to their possessor, what is good for someone is simply a matter of convention. The fact that wealth or bravery sometimes results in harm does not make it a matter of convention that it is good for us on some occasions and in some circumstances. Similarly, if food is sometimes bad for us, it is not therefore simply a matter of convention whether food is ever good for us. Variation does not support conventionalism." Though ethical generalizations are usual and inexact, they are neither useless nor unimportant.
The Usual
To see what Aristotle might mean by saying that ethical generalizations are usual, we must distinguish two kinds of regularities that he calls 'usual. I will call these 'frequencies' and 'norms, resulting in 'frequent' and 'normal' regularities. 12 (1) Sometimes, when Aristotle says that F is usually G, he means simply that Fs are Gs more frequently than they are not-G. Human beings, for instance, usually acquire grey hair as they age. This does not happen in every case, but it happens more often than not (APr (Prior Analytics), 32b4-10). (2) Sometimes, however, 'F is usually G' indicates that the natural way for F to be is G, though sometimes F fails to be G. This is why nature includes the usual as well as the necessary and invariable ( GA (De Generatione Animalium), 770b9-17, 777a16-21, Met. (Metaphysics), 1027a8-15). In a given species organs and limbs are formed and work together for the good of the organism. The teleological regularities applying to these organs and limbs are the basis for understanding the behaviour of the species, and indeed for counting particular things as members of the species. In this case exceptions to the general rule are 'freaks' (terata) or `deformities' (pêro mata), or at least 'against nature' (GA 770b13-27), since variations in people's opinion about them. (b) He appeals to objective variation. Fine and just things vary in themselves; what is fine or just in some circumstances is not fine or just in others. In this case Aristotle may be referring to examples such as the one in Republic i (331cd) about justice and paying back what we have borrowed. I am assuming (b). 11 This interpretation requires us to take toiautên de . . . kai' as adversative. See J. A. Stewart, Notes on the Nicomachean Ethics (Oxford: Clarendon Press, 1892), ad loc. 12 See J. Barnes, Aristotle: Posterior Analytics (Oxford: Clarendon Press, 2nd edn., 1993), 192; L. Judson, 'Chance and Always or for the Most Part" in Aristotle', in Aristotle's Physics, ed. Judson (Oxford: Clarendon Press, 1991), ch. 4. Since my use of 'normal' refers to norms, it is not the most common use in current English.
Ethics as an Inexact Science
107
they lack something that belongs to the proper interconnection of limbs, organs, and vital processes. The difference between these two sorts of usual regularities is important; for only variation from the natural norm implies contrariety to nature. If some people live longer than usual, or their hair falls out before it goes grey, that does not mean that any of the teleological regularities specifying their natural course of behaviour has failed; Aristotle does not limit the natural so strictly that every variation from the most frequent counts as a freak or a deformity.' 3 The generalizations that express these different sorts of regularities are differently related to observation of particular instances. The generalization that people go grey more often than not must simply be established (let us suppose) by observation; nothing in our general theory about human nature or the functions of human organs makes grey hair especially suitable. By contrast, our view that a human being normally has lungs allowing the appropriate sort of breathing does not rest on mere observation of particular instances; it rests on a more general conception of a human being as a goal-directed organism and of the contribution of lungs to a goal-directed system. The claim that this is the normal human condition would not be falsified even if it happened that most people lived in polluted air and suffered from diseases of the lungs. Since some of Aristotle's usual generalizations have this normative component, they are not simply the result of observation of statistical frequencies. 13 Once we see that the normal is not identical to the merely frequent, we can take a further step that Aristotle does not take, and notice that greater frequency is not necessary for normality. (This point is discussed by Judson, p. 97.) Aristotle's use of the expression 'hos epi to polu' suggests that he has in mind what happens more often than not; even if he recognizes that it includes more than mere frequency, he does not clearly recognize the possibility that the normal might not be the most frequent case. But such a possibility suggests itself. Selective breeding of a certain variety of dog, for instance, might cause freakish specimens of the kind to be the most frequent specimens of it; but it would not be any the less true that they were freaks and deviated from the norm for the species. Aristotle has epistemological and metaphysical reasons for insisting that in nature the normal case is also the most frequent case, so that most specimens of a given kind do not deviate from the natural order. None the less, it is useful to recognize that he includes two concepts in his concept of the usual—the concepts of the frequent and of the normal. It is not unreasonable (despite the powerful reasons leading Aristotle in the other direction) to acknowledge that most specimens of a species actually deviate from the natural and (to this extent) normal order. Indeed, Aristotle seems to acknowledge precisely this possibility in the case of species that are divided into male and female, since he believes that the female is a deformed male (GA 737a27-8). This belief certainly raises formidable difficulties for Aristotle's natural teleology; it is worth noticing here, however, since it shows how far his conception of the natural comes apart from his beliefs about what is most frequent.
108
T H. Irwin
The sacrifice of exactness resulting from an appeal to the usual is quite different for the two types of usual generalizations that we have distinguished. If we find only frequencies, we have no Aristotelian science. Aristotle distinguishes the scientific understanding of the explanation from the mere compilation of observed frequencies (Met. 981a12-30). If, however, the usual regularities that we find are normative regularities, we have the basis for a science." The teleological aspects of natural organisms are often expressed in usual regularities; nature does not invariably succeed in achieving the end that is naturally appropriate for the organism. Aristotle's recognition of this fact does not make him conclude that teleological regularities are unimportant for the study of living organisms. Since normal regularities, those that describe the natural situation, are more suitable than mere frequencies for scientific knowledge, it is reasonable to infer that when Aristotle allows scientific knowledge of the usual, he has normal and natural regularities primarily in mind.15 14 Sometimes, admittedly, he suggests that sciences require demonstration, and demonstration requires premisses that are necessary. Sometimes, however, he seems to be willing to allow usual premisses to be substituted for necessary premisses in a demonstration (APo 87b19-27, 96a8-19; see Barnes, Posterior Analytics, 191-3). Once he actually seems to allow that one type of necessary proposition is usual rather than invariable (Phys. (Physics) 198b5-6). Similarly, while some passages might suggest that Aristotle sometimes confines science strictly to what is necessary (EN 1140b31-3), his fuller statements allow scientific knowledge of what is usual. Whenever his statements about the objects of science mention both the usual and the necessary, he treats both as objects of scientific knowledge; and so we may be inclined to treat remarks confining science to the necessary as abbreviations rather than as deliberate restrictions. He recognizes that science includes both the invariable and the usual (Met. 1027a20-1); that is why the fact that chance events are neither necessary nor usual implies that there is no science of them. Scientific knowledge includes the grasp of an explanation. It cannot extend to the usual, if the usual consists simply of mere frequencies. If, however, the usual includes normal and natural regularities, it may be open to scientific knowledge; for our grasp of the right natural norms gives us an explanation. This role of usual regularities is easier to understand if we remember that Aristotle does not believe than an explanation must be a sufficient condition for what it explains. Barnes, Posterior Analytics, 193, points out (cautiously) that it is easier to see how usual statements can appear in demonstrations if they are taken to describe norms than if they are taken to describe mere frequencies. 15 The difference between frequencies and norms is important for understanding Aristotle's conception of social and political explanation. His generalizations about the city and its natural characteristics and functions are meant to help us to explain and understand the behaviour and the development of Athens, Sparta, Argos, and other actual Greek cities. It would be difficult, however, to show that the natural characteristics that Aristotle attributes to the city are more frequent than their opposites in actual cities (cf. Pol. (Politics), 1252a34, 1254a36—b2, 1255b3-4). In the MM, Aristotle rather awkwardly suggests that the naturally just condition is also the usual condition (MM 1194b30-1195a3). If he means by this that it is also the most frequent condition, he has given no reason for believing this. In the parallel discussion in the EN, claims about the natural condition remain, but the claim
Ethics as an Inexact Science
109
Ethics and Variation
When Aristotle claims that ethics provides usual generalizations, does he primarily have in mind mere frequencies or natural norms? Does he recognize ethical principles parallel to the normative regularities of teleological b iology? 1 6 Let us consider the example of wealth. Aristotle recognizes that wealth is not always, all things considered, good for everyone in every situation. But it is usually good, since it is good for those who use it correctly in normally favourable circumstances. Aristotle could hardly believe that wealth is usually good, however, if 'usually' were taken to imply 'more often than not. For it is not at all clear that most people are good enough, or that most circumstances are favourable enough, to ensure that wealth is more often beneficial, all things considered, than harmful to people who have it. This passage suggests that when Aristotle speaks of the usual, he really has in mind the normal, which may not be statistically the most frequent. If we are to fit his observation about the harms resulting from wealth into his general views about wealth, we must appeal to those claims that rely on assumptions about normality. A similar point holds for bravery. In favourable conditions, bravery helps us to preserve and enjoy external goods. Brave people who have reasonable luck do better in preserving their lives and cities than cowardly people do, but the relevant conjunction of moderately favourable circumstances cannot be guaranteed. Similarly, part of the reason for valuing justice is the fact that in a just community we are respected for our justice, justice leads to greater harmony between different groups in society, and so on; but we cannot count on these favourable conditions. The connections between virtue and external goods ensure that virtue secures happiness in normal conditions; but the example that Aristotle gives here refutes the unrestricted claim that virtue ensures happiness. that the natural condition is the usual one is absent. (Natural justice is discussed by F. D. Miller, Nature, Justice, and Rights in Aristotle's Politics (Oxford: Clarendon Press, 1995), 74-9.) 16 We can already see that some question about nature is in Aristotle's mind, if we look back at his remark about fine and just things. He remarks that some people argue from variation to conventionalism; they claim that nothing is just by nature, because they see that principles about justice have exceptions. We have already considered the discussion in Book V where Aristotle rejects this line of argument. He argues, as he argues in the biological works, that even if there are exceptions to general rules in nature, there is still a natural and normal condition; in fact he illustrates his point with a biological example (righthandedness, 1134b33-5; cf. PA 666b35-667a5).
110
T. H. Irwin
If the claim that virtue usually secures happiness refers to frequency, it implies that favourable circumstances and outcomes are more often encountered than unfavourable ones. Aristotle has no reason to believe this, and it is not especially relevant to his theory. If favourable circumstances are in our power to achieve, the fact that they cannot be taken for granted in most cases does not matter. Though admittedly Aristotle's theory cannot work if it requires too many things to happen contrary to the way they most often do, the specific claims we are concerned with do not need to be true more often than not. In present circumstances, virtue may not secure happiness more often than not. Still, we have good reason to choose virtue if in normal circumstances it achieves happiness, and normal circumstances are within our power to produce. In all three examples (about justice, wealth, and bravery), the usual generalizations that Aristotle alludes to are most plausibly taken to describe the normal situation rather than the most frequent situation. To see the connexion between these claims about ethics and Aristotle's claims about the usual in nature, we must remember the connection between the usual and the natural. Aristotle believes that external goods are both good without qualification and good by nature (EE 1248b26-37). His conviction that they are good by nature does not depend on the conviction that most actual people in most actual circumstances actually benefit from them. It depends on his general theory about what is good for a human being.' 7 In ethics, as in the study of animals, the usual regularities are important because they describe natural norms, not because they describe frequencies. If, then, we examine these specific examples, we can see what kinds of usual regularities Aristotle primarily has in mind in his remarks about the inexactness of ethics. He is primarily thinking of norms rather than frequencies. He does not explicitly distinguish these two types of usual regularities in his different claims about the usual. But we must notice the distinction; for we should attend especially to the consequences of treating ethical principles as statements of natural norms.
Usual Principles and Ethical Science
Why does Aristotle think it matters that ethical generalizations are usual? He might give either of two answers: (1) He emphasizes this to warn us that 17 Admittedly, this general theory itself is not completely separated from information about what happens to most people in different circumstances; but it does not follow that the particular claims about what is the case by nature are claims about relative frequency.
Ethics as an Inexact Science
111
we ought not to take ethical generalizations too seriously. They cannot claim the status of scientific principles, but should simply be regarded as summaries of experience. (2) He emphasizes this to warn us that we ought not to be deterred from taking ethical generalizations seriously; for they are still scientific principles, even though they have exceptions. To decide between these answers, we ought to remember, first of all, that Aristotle cannot take all ethical generalizations to be merely usual.18 It is easy to compile a list of generalizations that do not seem to have exceptions. Happiness is everyone's ultimate good; everyone's happiness consists in activity of the soul in accordance with complete virtue in a complete life; it is always better to aim at the mean than to aim at either the excess or the deficiency; it is always better to be brave than to be cowardly; it is right to care more about fine action for its own sake than about honour. These claims may seem too schematic to be practically useful; but Aristotle is also committed to less schematic generalizations with no exceptions. He believes, for instance, that one ought always to be willing to face great danger if some important cause is at stake, and one ought never to be willing to face it for some trivial reason. He believes it is always bad to cultivate the fearless attitudes of people who do not stop to think about dangers, or of people who care so little about living that they do not care about dying (1115b24-9, 1117b17-20, 1124b6-9). He believes that we ought never to make fun of people simply to raise a laugh, without any regard for what is fine or expedient (1126b36-1127a6, 1128a4-7). These principles have some practical content, since they clearly prohibit the attitudes of some readily recognizable types of people. If Aristotle is committed to many practically significant unqualified generalizations, he cannot consistently take his claims about the inexactness of ethics to rule them out. His claim that ethics states usual principles cannot mean that ethics states only usual principles; it must mean that ethics includes not only unqualified principles, but also usual principles. If Aristotle believes that usual principles are not scientific, he must say that ethics includes both unqualified generalizations that are suitable for science, and usual generalizations that are unsuitable for science. If, however, he believes that some usual generalizations are scientific, he claims that the scientific part of ethics includes both unqualified and usual generalizations. When he emphasizes that the science of nature includes usual principles, he means to affirm, not to deny, the scientific status of the study 18
18.
This point is emphasized by S. W. Broadie, Ethics with Aristotle (Oxford: OUP, 1993),
112
T. H. Irwin
of nature. Why should we not say the same about ethics? We would have reason to hesitate if we believed that the usual generalizations available in ethics do not embody normal and natural regularities; but we have seen that these are precisely the regularities that Aristotle has in mind in the specific examples he gives. I have emphasized these points because it is easy to suppose that Aristotle's remarks about ethical generalizations imply a sharp division between the epistemic status of ethical principles and that of the principles of theoretical sciences. This supposition is quite mistaken. The fact that ethics relies on usual regularities does not distinguish ethics from natural science, as Aristotle conceives it; and so we have no reason to suppose that he takes ethical generalizations any less seriously than he takes physical generalizations. We can strengthen this conclusion once we recall that Aristotle normally takes the principles of a discipline to be 'better known by nature' (gnôrimôteron phusei) than the initial beliefs that provide the starting points for inquiry; these initial beliefs are better known 'to us, but not better known by nature. The relevant aspect of this contrast, for our present purposes, is Aristotle's claim that what is better known by nature is also more universal, prior, and more explanatory (APo (Posterior Analytics), 71b29-72a7). If Aristotle in the Ethics is looking for principles that are better known by nature than the starting points, he is looking for principles that are more universal than, prior to, and explanatory of the starting points. The Ethics makes it clear from the beginning that Aristotle is looking for principles. He remarks that the road towards the principles is different from the road from the principles, and he insists that when we are on the road towards the principles, we must start from things that are better known to us. Aristotle contrasts, as he often does, what is known to us with what is known 'without qualification' (1095a36—b4), and hence known by nature.19 He implies that we are on the road towards principles (1095a30—b4). Later, he claims to have found a principle by giving an account of happiness. He accompanies this claim with a warning not to demand the same degree of exactness in all disciplines (1098a20—b8). This warning is not intended to qualify the claim to have found a first principle. On the con19 Both this passage and the next one I discuss raise some difficulties of detail. I have discussed them a little in 'First Principles in Aristotle's Ethics', Midwest Studies in Philosophy, 3 (1978), 268, n. 18. See also M. F. Burnyeat, 'Aristotle on Learning to be Good' in Essays on Aristotle's Ethics, ed. A. 0. Rorty (Berkeley: Univ. of California Press, 1980), 88-9, nn. 2-5.
Ethics as an Inexact Science
113
trary, it is intended to forestall an inappropriate objection to that claim, by showing that an inexact statement of a first principle is appropriate for the discipline whose principles we are seeking. The beginning of the Ethics, then, gives us good reason to believe that Aristotle takes some ethical generalizations, including some usual ones, to be principles, and therefore to be prior in the order of explanation, justification, and knowledge to the ethical beliefs that they explain. This conclusion counts against the view that Aristotle's belief in the usual character of some ethical generalizations reflects any commitment to particularism.
Why are Ethical Principles Usual?
Aristotle does not suggest that the presence of usual principles in ethics simply reflects our ignorance or the incompleteness of our theory. He suggests they are ineliminable, just as he assumes that they are ineliminable in natural science. Why is this? In natural science we have a choice between two explanations: (1) These usual regularities cannot be eliminated because the behaviour of matter is essentially indeterminate, so that exceptions to teleological generalizations cannot be exhaustively specified. (2) Aristotle's preference for usual regularities indicates his belief in the importance of teleological regularities and the unimportance (for these particular purposes, though not necessarily for all purposes) of exceptions to them, even if we can specify all the exceptions. The first explanation asserts that Aristotle affirms merely usual principles only because he believes that it is impossible to specify all the exceptions to teleological generalizations. The second explanation, by contrast, neither affirms nor denies that he holds this belief. It says that his holding it is not necessary for his affirmation of usual principles. The two parallel explanations in ethics are these: (1) We cannot, even in theory, find all the qualifications that would be needed to formulate the appropriately qualified principles. (2) It is unwise, for practical purposes, to try to build all the qualifications into our principles, even if it is possible to build them in. In natural science, the second answer is preferable. When Aristotle draws attention to teleological regularities in nature, it might simply be distracting to try to list all the qualifications needed to make the generalizations accurate. His aims are better achieved if he sets out the natural norms and teleological regularities, recognizing that they have exceptions, but not specifying the exceptions in detail.
114
T. H. Irwin
To see whether the second answer is also preferable in ethical theory, we must consider whether the purposes of ethical theory make it reasonable to stick to usual principles. Aristotle insists that ethical theory—the discipline practised in his ethical treatises—essentially has a practical aim. The appropriate age for studying ethics is to be decided by considering the age at which this study is practically useful, 'since the end is not knowledge, but action' (1095a5-6). This practical purpose of ethics explains why we ought to be satisfied with principles that are stated only roughly. To demand more would be as misguided as a carpenter's seeking geometrical precision about right angles (1098a29-31). Nothing about wood and right angles prevents carpenters from finding out how the right angles they try to produce in wood fall short of being true right angles; but it would be pointless to occupy themselves with these questions. Similarly, the practical purpose of ethics implies that we should not try to build in all the qualifications that we would need to add to find exactly correct principles. Aristotle's examples of usual principles suggest why we ought not to try to spell out all the exceptions. Virtuous people take the right attitude to wealth and to other external goods. They recognize that wealth is good without qualification, but not good for everyone, and they learn that virtue results in happiness in appropriate conditions, though not in all conditions without exception. It is more important to grasp these points about virtue, external goods, and fortune than to learn the more complex generalizations that would incorporate all the relevant exceptions to the unqualified generalizations. Aristotle argues, then, that we need not list all the exceptions and qualifications that would replace our usual rules with exact generalizations. Nor should we take the usual rules any less seriously simply because they are usual. These usual principles will be useless to people who cannot guide their desires by reason, but useful to those who can guide desires by reason (1095a4-11). If I learn that brave action is always better than cowardly action, but that it only usually results in happiness, then I will not believe it is ever in my overall interest to prefer the cowardly action. My confidence in preferring the brave action will not be shaken if I recognize that in this particular case the brave action will not result in happiness for me. The combination of the universal and the usual generalizations about bravery and brave action will strengthen my confidence in acting virtuously; the same will be true for every other virtue. Our recognition of the usual character of some rules helps us to take them seriously; for we will not be disconcerted to find that they have exceptions.
Ethics as an Inexact Science
115
If this is the point of recognizing usual principles, it does not imply that any special doubt or difficulty arises in deciding whether or not to do the brave action in this or that particular situation, or that something more than the theoretical principles is needed if we are to find the right thing to do in particular situations. Aristotle does not suggest, for instance, that the merely usual status of the generalization that bravery usually results in happiness ever gives us a good reason not to act bravely, or that we need any special exercise of perception to see that we must act bravely in this particular case where brave action will involve significant harm. These remarks on the inexactness of ethical principles do not support particularism.
Particulars and Inexactness
In Book II Aristotle tells us more about the practical aims of ethics, and adds a new claim about inexactness. Before presenting his general account of virtue, he reminds us that ethical inquiry aims at practice rather than theory, and that we must be satisfied with inexact accounts (1103b261104a5). He adds that further inexactness results when we try to deal with particulars: And when our general account is so inexact, the account about particular cases is all the more inexact. For these fall under no craft or profession, and the agents themselves must consider in each case what the opportune action is, as doctors and navigators do. The account we offer, then, in our present inquiry is of this inexact sort; still, we must try to offer help. (1104a5-11)
The parallel with medicine and navigation shows how we must understand the claim that 'every account of the actions we must do 20° has to be stated in outline, not exactly'. It does not exclude unrestricted generalizations from ethics. Aristotle suggests only that medicine and navigation must sometimes resort to inexact generalizations, to be applied by trained judgement in particular cases, because they seek to give practical advice. The same is true of ethics. This warning about the inexactness of ethical generalizations comes between Aristotle's claim that virtuous action is in accordance with right reason and his claim that virtue is a mean. Indeed, he begins his argument for the latter claim immediately after saying that the moral philosopher must 'try to help' despite the inexactness of ethics (1104a10-11). The doctrine of the mean certainly does not provide a precise quantitative guide to 20
I read prakteôn rather than praktôn.
116
T. H. Irwin
finding the virtues, since we cannot eliminate 'at the right time, to the right people, and so on, from accounts of virtue as an intermediate state. The doctrine is not at all empty, however; it conflicts with the view that virtue is simply continence, and with the view that virtue requires the elimination of all non-rational desires. It is difficult to see how such a doctrine could be represented as a summary of what virtuous people have found in particular cases; and Aristotle does not represent it in this way. Having stated the doctrine of the mean, Aristotle considers how it might be specified in suitable detail. 21 In order to give more detail, he specifies the mean with reference to different kinds of feelings. The different means are specific virtues of character, of which he promises a more 'exact' account later (1107b14-16). Aristotle does not suggest that generalizations are useless; for here he introduces the accounts of the specific virtues of character, which evidently include many generalizations about the characteristics of different virtuous and vicious people. He is coming closer to 'particulars' by providing more specific generalizations. Instead of simply saying that virtue consists in a certain kind of mean, he adds that bravery is this kind of mean, temperance is that kind, and so on. When Aristotle tells us to attend to particulars, he does not abandon generalizations; he describes the generalizations we should look for. Nothing in these remarks comes close to acceptance of particularism. Some of Aristotle's directions for approaching the mean give general advice about compensating for the extreme that we are more prone to, though he qualifies the advice with a warning about inexactness (1109a30—b23). We need to recognize inexactness not because we regard generalizations simply as summaries, but because we recognize them as giving normative guidance. Once we have accepted the doctrine of the mean, and seen the desirability of trying to achieve the mean, we see that we need to attend to particular cases; but we do not know what particular cases we ought to attend to until we have accepted the doctrine of the mean. The distinction between the statement of the doctrine of the mean and its application to particular cases corresponds to Aristotle's distinction between the 'general account' and 'the account about particular cases. The claim 'Virtue results in happiness' is a generalization with exceptions, but it does not try to give us advice about any specific circumstances we may 21 'However, we must not only state this general account but also apply it to the particular cases. For among accounts concerning actions, though the general ones are common to more cases [read koinoteroi rather than kenôteroi], the specific ones are truer, since actions are about particular cases, and our account must accord with these.' (1107a28-32)
Ethics as an Inexact Science
117
face. The closer we come to giving advice about very specific circumstances, the more inexact our advice must be. This second sort of inexactness cannot result simply from the fact that we mention a particular situation. However particular it might be, we can still advise agents to do the finest action open to them, or to do what is required, all things considered, by the virtues. This advice may, for all Aristotle says, be entirely correct in all circumstances, but it is not of much practical use in helping us to decide precisely what to do. To say that months in the year have fewer than thirty-five days is an unrestrictedly true generalization, but it is not much practical use if we want to pay our bills on the last day of each month and we need to know which day that is. For this practical purpose the generalization that the months in the year usually have thirty-one days is better than the unrestricted generalization we began with. But it seems that it would be even more useful to formulate a new generalization that includes the appropriate qualifications, instead of merely suggesting, by the use of 'usually, that some qualifications are needed. The most useful generalization says that all the months have thirtyone days, except for April, June . . . and so on. Similarly, knowing that Latin prepositions usually take the accusative is fairly useful, but it is more useful to know that they take the accusative, except for ab, ex, clam, coram . . . etc., which take the ablative. Once we have built in all the exceptions, we have replaced our inexact usual generalization with an exact generalization. Why does Aristotle not advise us to do this in ethics? He suggests his answer when he discusses the cases in which deliberation is necessary. 22 He describes a series of increasing degrees of inexactness that require larger contributions from deliberation. His 'exact and self-sufficient' cases are most like my examples of the months and the Latin prepositions. In the other cases we have fewer and fewer simple rules that guide us effectively in particular cases. The last sentence quoted sums up the situations in which deliberation is needed. What does Aristotle mean by saying that deliberation is needed when things are 'undefined'? He need not rely on any belief in some real 22 Now there is no deliberation about the sciences that are exact and self-sufficient, as, for instance, about letters, since we are in no doubt about how to write them. Rather, we deliberate about what results through our agency, but not in the same way on every occasions, as, for instance, about questions of medicine and money-making; more about navigation than about gymnastics, to the extent that it is less exactly worked out, and similarly with other [crafts]; and more about beliefs [read doxas; rather than technas] than about sciences, since we are more in doubt about them. Deliberation is in things that are usual, but [or 'and'? or `i.e.'?] unclear how they will turn out, and in which the right course of action [subject of adihoriston unexpressed in Ms text] is undefined.' (1112a34—b9)
118
T. H. Irwin
indeterminacy in things. 23 The claim that something is undefined may simply refer back to his remark that we are in doubt about what to do. We need deliberation in cases where the considerations are various and complex enough to make it inappropriate to appeal to generalizations that are sufficient by themselves to tell us what to do. If I want to write the French word for a cat, and I know that this word is spelt `c-h-a-t', I need not deliberate further about what letters to write. When more considerations bear on the appropriateness of one or another action, it is less reasonable to appeal to generalizations that can be applied without further deliberation. These remarks about deliberation suggest that choices in particular situations involve inexactness because we cannot expect to reach the right choices by appeal to exact generalizations that are immediately applicable, without further reflexion, to particular cases. This does not mean that generalizations are inappropriate, but that they must be qualified in certain ways, and that their limitations must be recognized. Recognition of the limits that we have mentioned falls far short of particularism.
Particulars and Perception 24
Some of Aristotle's remarks about particulars also mention perception. These may seem a more promising basis for attributing particularism to him. After stating the doctrine of the mean, and listing the specific means constituting different virtues, Aristotle gives us the general advice to come as close as we can to the mean, by trying to avoid the extreme to which we are more inclined. He warns that we cannot expect this general advice to give precise guidance in every case. The difficulty in determining questions about particulars seems to result from two things: (a) the differences between different people, and the consequent differences in the training that each person needs to approach the mean; and (b) the different particular conditions that each person faces and the consequent difficulty in covering all the circumstances relevant to a decision about what is needed to approach the mean (1109b12-23). Since we cannot readily define these 23 I leave aside the question of whether Aristotle actually believes in some real indeterminacy in things. 24 Some discussions of perception: R. A. Shiner, `Aisthesis, Phronesis, and Nous', Philosophical Studies, 36 (1979), 377-87; 'Ethical Perception', Apeiron, 13 (1979), 79-85; M. J. Woods, 'Intuition and Perception in Aristotle's Ethics', Oxford Studies in Ancient Philosophy, 4 (1986), 145-66.
Ethics as an Inexact Science
119
points in a general account, Aristotle concludes that 'the judgement about them depends on perception'. Aristotle suggests a similar role for perception in his remarks about the mean constituted by gentleness. When we try to distinguish the gentle (praos) person both from the people who are too prone to anger and from the people who are not angry when they should be, the boundaries are difficult to mark in general terms. Perception is needed to supply the deficiency of general accounts. We make the relevant judgement in particular cases by using perception (1126a31—b4). 25 The connection between particulars and perception is marked again when Aristotle denies that particulars are objects of deliberation. Perception is introduced here both for non-evaluative judgments (`this is a loaf') and for evaluative judgments (`it is cooked the right amount') (1112b34-1113a2). These remarks do not imply acceptance of particularism, for two independent reasons: (1) Aristotle does not rely on a claim that fully qualified generalizations are in principle impossible. (2) Even if he did believe they are impossible, he would not be committed to particularism, because he does not attribute to perception of particulars the sort of priority that is necessary for particularism. To explain the first reason, I refer back to Aristotle's views about the role of perception in relation to deliberation. If we are not content with usual rules, but we try to qualify them so much that they provide definite advice for every particular case that we meet, we will have to provide numerous qualifications. If these qualifications are extremely numerous, it may be better to equip the learner with some other means of finding the right answer. If generalizations become at all complicated, with many qualifications, the different qualifications will refer to different aspects of a situation, and the agent applying the generalization will have to recognize these different aspects. If agents equipped with unqualified generalizations and the capacity to recognize ethically relevant aspects of particular cases can reach the right answer, it is better not to burden them with extremely complicated qualified generalizations. One of Aristotle's examples may help to illustrate this point. The advice to bake bread until it looks done relies on some trained capacity to estimate when it looks done. Conceivably we could avoid reliance on any such capacity by providing the baker with a thermometer to insert carefully in 25 In 1126b4 I read kai (rather than kan) tê(i) aisthêsei hê krisis. The passage is discussed by N. Sherman, The Fabric of Character (Oxford: Clarendon Press, 1989), 35.
120
T H. Irwin
the bread, a chart to find the appropriate temperature for different kinds of bread, and a colour chart to determine the colour that it ought to be when it should be removed from the oven. Using all this apparatus is a nuisance, however; baking is normally simpler and more efficient if we simply learn to recognize when bread looks done. Similarly, we can see why an agent needs some capacity to recognize several ethically relevant aspects of a situation, if we consider some of Aristotle's own contributions to casuistry. In Book ix Aristotle refers back to his remark in Book II that it is difficult to give definite answers about particular cases. Still, he makes some suggestions (1164b25-1165a12). He urges, for instance, that usually we ought to pay our debts, but this general rule has exceptions, when something especially fine or necessary is involved. Aristotle has good reasons for not spelling out all the relevant exceptions and for not even trying to incorporate them into modified rules. Suppose that we could formulate general rules, with the appropriate qualifications built into them, to cover all these cases. In order to apply these rules, agents would have to recognize that a situation involves (say) an obligation to one's father, an obligation to someone who has already paid one's own ransom, an obligation to pay a debt, a request to lend money in return for having been given a loan, and a permission to refrain from lending in cases where one will probably lose one's money. If I can recognize these different aspects of situations, and I also have some idea of which of these considerations are more important, I will probably do quite well at getting the answers that would be given by the fully qualified generalizations. I need not be able, however, to formulate rules that express my views about the relevant considerations. Aristotle does not say why the practical purposes of ethics are better served if we learn usual generalizations than they would be if we set out to learn fully qualified generalizations. But his view is reasonable. Fully qualified generalizations would be so complicated that they would be difficult to learn and difficult to apply; moreover, the skills that we would need in order to apply them, once we had learnt them, will lead us to the right answers without having learnt the fully qualified generalizations. Instead of trying to learn fully qualified generalizations, it is better to try to learn to recognize and to compare the considerations that ought to guide us in assessing the different claims of different usual generalizations. These points support my first reason for rejecting a particularist interpretation. Aristotle's emphasis on perception is intelligible whether or not he believes that fully qualified generalizations are possible.
Ethics as an Inexact Science
121
Even if this first reason is mistaken, the second reason still stands. Aristotle's casuistical discussions do not suggest that we can resolve the difficulties raised by these situations, unless we understand some generalizations. Generally we must follow the rule of reciprocity, because reciprocity takes priority over disposing of our resources as we feel inclined; payment of a debt, therefore, takes priority over a generous impulse to give a present to a friend (1164b30-3). But the priority of reciprocity over generous impulses does not imply the priority of reciprocity over everything. If we also recognize that special obligations to our parents are prior to the general obligation of reciprocity, we will acknowledge the proper place for the rule of reciprocity. To acknowledge this, we must already have understood the moral basis for the presumption in favour of reciprocity. We are not really violating the rule of reciprocity; for the rule holds only on a presumption that (in this case) is not satisfied. This use of the basis of a rule to explain exceptions, or apparent exceptions, is especially clear in Aristotle's discussion of equity. He suggests that the provisions of written law need to be violated in some cases, but that violations do not violate the point of the law (1137b11-32). Once we see what the legislator aimed at in formulating the general rule, we can see that this very same aim requires violation of the rule in some cases. Ethical rules are different from the laws that need to be violated in order to fulfil their aim; for, in contrast to laws, ethical rules acknowledge their limitations by being expressed as usual generalizations. If we fail to do what they tell us to do usually, we do not violate them, if the principles underlying them justify our claim that this is not one of the usual cases. 26 Consideration of Aristotle's examples (friendship and equity) do not suggest that exceptions to general rules are irreducibly particular. We might argue, for instance, that a general rule about the priority of obligation to one's parents over the obligation of reciprocity explains exceptions to the rule of reciprocity. If Aristotle were a particularist, he would have to reject this explanation of the exceptions. Since he does not express any doubts about it, his treatment of these examples does not suggest that he is a particularist. In these cases, then, even though general rules do not prescribe precisely what we ought to do in a particular situation, their normative guidance contributes essentially to a correct decision about what to do. Aristotle's conception of their role, therefore, conflicts with particularism, which 26 Equity is usefully discussed by Sherman, Fabric, 13-22; R. A. Shiner, 'Aristotle's Theory of Equity', in Justice, Law, and Method, ed. S. Panagiotou (Edmonton: Academic Printing, 1987), ch. 12.
122
T H. Irwin
asserts the priority of perception over general rules. The only way to defend a particularist interpretation of Aristotle's view in these cases is to argue that the generalizations that are brought to bear on particular cases are themselves simply summaries of what we have found by perception in particular cases. But we have seen no reason to believe that Aristotle accepts this account of ethical generalizations.
Perception and Prudence
So far I have suggested that Aristotle thinks of perception primarily as a means of applying general rules to particular cases, so that we recognize all the relevant aspects of a particular situation to which our generalizations apply. Particularist interpreters, however, believe that this claim about perception cannot be all that Aristotle intends, since it fails to attribute to perception the normative priority that Aristotle claims for it. McDowell takes Aristotle to be a particularist in so far as having the right conception of eudaimonia involves having a number of 'motivational susceptibilities'. These cannot be properly described as the acceptance of any general principles—they are simply tendencies to focus on certain features of situations and to react to them in certain ways. These tendencies manifest a special perceptual capacity. 27 To explain what he means in speaking of a perceptual capacity, McDowell refers to the ability to see the relevant features in a particular situation and to concentrate on them. Here he follows Wiggins, who calls this capacity 'situational appreciation'. 28 For agents who have the appropriate sort of situational appreciation, the morally significant features of the situation are salient; these agents deliberate in the light of the fact that this situation provides an opportunity for (say) spending money for the public good rather than for displaying one's own wealth. This is the sort of situa27 This is the view that Dancy expresses in suggesting (on McDowell's behalf) that 'there is nothing that one brings to the new situation other than a contentless ability to discern what matters where it matters'. See fn. 5 above. 28 Wiggins describes situational appreciation as follows: 'A man usually asks himself "What shall I do?" ... in response to a particular context. This will make particular and contingent demands on his moral or practical perception, but the relevant features of the situation may not all jump to the eye. To see what they are, to prompt the imagination to play upon the question and let it activate in reflection and thought-experiment whatever concerns and passions it should activate, may require a high order of situational appreciation, or, as Aristotle would say, perception (aisthêsis): (`Deliberation and Practical Reason' in Essays on Aristotle's Ethics, ed. Rorty, 232-3.) McDowell cites Wiggins to explain his references to perception and salience (Mind, Value, and Reality, 68).
Ethics as an Inexact Science
123
tional appreciation we expect from an agent with the virtue of magnificence rather than the vice of ostentation. Interpreters who emphasize situational appreciation believe that, as Wiggins implies, Aristotle has it in mind in the various passages in Book vi where he connects prudence with perception. I will assume, for the moment, that these interpreters are right. According to this view, virtuous people tend to see things differently from the way other people see them, because they tend to focus on features of situations that other people tend to miss. Neither Wiggins nor Aristotle suggests, however, that situational appreciation is normatively prior to general rules. On the contrary, we might suppose that virtuous people's acceptance of the right generalizations explains their ability to identify morally relevant features of situations. They have learnt, for instance, that it is wrong to make jokes that humiliate their innocent victims but amuse the audience; this is why they train themselves to notice the tendency of a joke to humiliate an innocent victim. This role for situational appreciation does not support particularism. We will justifiably attribute particularism to Aristotle if we find that he treats generalizations as mere summaries of particular exercises of ethical perception or situational appreciation. A general principle, according to a particularist, constitutes 'a reminder of the sort of importance that a property can have in suitable circumstances'. 29 Such generalizations may be practically useful. A team planning a game, or a general planning a campaign, may reasonably plan in accordance with strategic principles and tactical maxims that summarize the moves that have succeeded and seem likely to succeed in specific circumstances. These simply tell us the importance that a property can have, and not the importance it actually has, because we make no error, relative to the aim of winning the game or the campaign, if we win the game or the campaign by violating the strategic principles. Some of the generalizations involved in moral deliberation may be understood as strategic principles of this sort. But they cannot all be understood this way. A team captain who treated the rules of the game, or the aim of helping his own side to win, as merely strategic principles, to be violated if he 'saw' that in this particular case it would be better to kick the opposing players or to help the other side to win, would be open to criticism. Similarly a general who 'saw' that it would be better in this case to kill thousands of innocent civilians as a means to victory, or who 'saw' that it would 29
Dancy, Moral Reasons, 67.
124
T H. Irwin
be better to change sides because the other side seemed more likely to win, would not deserve congratulation for winning by unorthodox tactics or strategy. For this reason, the examples of games or military campaigns do not help to illustrate the plausibility of particularism as an account of moral deliberation. For they do not provide cases in which we can treat all generalizations as mere reminders of the sort of importance that a property can have. Indeed, some of the relevant generalizations seem to have a prescriptive role that they would not have if they were simply reminders. We do not show that Aristotle is a particularist about moral generalizations if we merely show that he treats some generalizations as mere reminders; we must show that he treats all non-trivial generalizations as reminders that we are morally permitted to disregard if our perception of particular situations prompts us to act differently. Does anything commit Aristotle to this position? He claims that prudence is about particulars, and therefore requires perception (1142a23-30). This claim does not support particularism; for it says nothing about priority. He also claims that universals are derived from particulars of which we have the special kind of perception that is a kind of understanding (nous) (1143a32—b5). Aristotle is a particularist if he believes that the general principles accepted by prudent people are summaries of what they have noticed in particular cases. If they have been correctly brought up, they have been trained to notice that here and now they have an opportunity to repay a friend; they form the habit of focusing on that feature of a situation rather than on the fact that (say) they have some spare money that they could spend on their own amusement. From noticing this fact about the reactions of virtuous people, the ethical theorist can form general principles about the obligations of friendship; but prudent people do not react to these situations as they do because they have themselves accepted the general principles. We might interpret Aristotle in this way, if we focused simply on the claim that universals are derived from particulars. But such an interpretation faces at least two objections: (1) Even if this passage is taken out of context, as a general claim about the acquisition of universals, it does not support particularism. (2) It ought not to be taken out of context, and if it is taken in its context, it does not make a general claim about the acquisition of universals. To understand the first objection, we must notice that not every sort of derivation of universals from particulars implies that generalizations are merely summaries of particular perceptual judgements, or that perception
Ethics as an Inexact Science
125
of particulars has normative priority. Aristotle believes that, in some sense that is difficult to explain precisely, all universal judgements, not just those in ethics, are derived from particular judgements. This claim about learning gives particular judgements some sort of priority 'to us. None the less, as we have seen, universal judgements are prior 'by nature'; if we understand a scientific theory, we must recognize this priority of universals. If Aristotle affirms about ethical universals what he affirms about universals generally, he does not affirm particularism; for particularism does not assert simply the claim that particular judgements are 'prior to us' in learning, but the stronger claim that they are recognized as prior in the outlook of the virtuous person. Aristotle does not endorse this stronger claim. To understand the second objection, we must notice that Aristotle ascribes situational appreciation to the prudent person, who is presumed to have deliberated well about what promotes the ultimate end. If this deliberation includes the application of generalizations with normative force, situational appreciation may depend on acceptance of these generalizations. Such dependence does not conflict with Aristotle's claim that universals are derived from particulars. Aristotle might simply mean that prudent people revise their general principles in the light of their situational appreciation.30 ° If we apply several of our generalizations at once, we may see that they lead to unacceptable results; to see this, we may need to appeal to generalizations. If this is the point Aristotle refers to when he says that universals are derived from particulars, his claim makes good sense in its context, and falls far short of a particularist view about the role of theoretical generalizations. What is the role of perception and situational appreciation in this process of revision? Perhaps we can see more clearly what Aristotle might and might not be committed to if we connect this discussion with the frequent appeals to 'salience' in treatments of moral perception. We might think of a feature of a situation as being salient in two different ways: (1) Some features rather than others are salient in so far as they provide 3" This process is described by Wiggins: 'No theory . . . can treat the concerns which an agent brings to any situation as forming a closed, complete, consistent, system. For it is of the essence of these concerns to make competing and inconsistent claims ... The weight of the claims represented by these concerns is not necessarily fixed in advance. Nor need the concerns be hierarchically ordered. Indeed, a man's reflection on a new situation that confronts him may disrupt such order and fixity as had previously existed, and bring a change in his evolving conception of the point ( to hou heneka), or the several or many points, of living or acting' (p. 233). On at least one point Wiggins's suggestions go beyond what we can reasonably attribute to Aristotle. Wiggins suggest that our concerns essentially make competing and inconsistent claims, but I am not sure where Aristotle commits himself to a claim quite as strong as this.
126
T H. Irwin
morally relevant considerations. If I am to deliberate properly, I must be able to ignore irrelevant and confusing considerations, and focus on relevant ones; I must, for instance, keep in mind the fact that I have made a promise to A and that B needs my help urgently now, and I must ignore the fact that it would be slightly inconvenient for me either to keep my promise to A or to help B. I need some skill in recognizing salient features if I am to focus on what is relevant to my deliberation. 31' In these cases, then, salience is relevance. (2) One feature rather than other features may be thought to be decisive; it may be more important, for instance, that B needs my help urgently than that I have made a promise to A, so that I ought to help B even if I must also break my promise to A. In this case we might say that B's need rather than the promise to A ought to be salient to me; the relevant sort of salience is decisiveness. Both relevant and decisive considerations might be described as 'salient, but an appeal to perception is not equally appropriate in both cases. In the first case, several features will often be relevant in a particular situation, and it is reasonable to suggest that perception is needed, since some selective awareness of relevant features is presupposed by deliberation in a particular case; we might say that correct deliberation requires us to see some things and not others in a situation. In the second case, however, only one feature is decisive, since the decisive feature is the one that matters most. In this case, in contrast to the first case, some of the `non-salient' (i.e., nondecisive) features are morally relevant. We cannot say that we see only the `salient' (i.e., decisive) feature; for it would show a failure of moral awareness if we did not recognize the moral relevance of considerations that we take to be relevant but not decisive. Moreover, it is more difficult to attribute the awareness of decisiveness to perception; the relevant awareness follows, and does not precede, deliberation. This division between two types of salience may help us to see how Wiggins's position falls short of particularism. He does not suggest that the same perception that is aware of relevance is also aware of decisiveness; hence he does not suggest that situational appreciation is the ability that we need to resolve conflicts or to settle our priorities in these situations. Indeed, it is not obvious why the same capacity that enables us to pick out relevant features of situations should also enable us to settle on their morally decisive features. Even if we were to claim that awareness of relevance is normatively prior to acceptance of generalizations, we could not infer that awareness of decisiveness is normatively prior to acceptance of 31
In putting the point about salience in this way I have been influenced by Karen Jones.
Ethics as an Inexact Science
127
generalizations. We would still be free to claim that we can resolve conflicts and settle our priorities correctly only if we have grasped the appropriate generalizations and the ethical grounds for them. I use Wiggins's views to clarify the issue about Aristotle, because Wiggins clearly believes that situational appreciation is an indispensable component of virtue. Even if Aristotle agrees with Wiggins about the importance of situational appreciation, he need not be a particularist. A Different Role for Perception I have tried to make the issue about particularism clear by conceding for the sake of argument that Aristotle's remarks on perception are about situational appreciation. In conceding this point, I am making things easy for particularist interpreters, since this is how they interpret the remarks about perception. Having argued against a particularist interpretation on this assumption, I would like to express some doubts about whether Aristotle really has situational appreciation in mind. I do not deny that situational appreciation is important; I simply wonder whether Aristotle refers to it in his remarks on perception. A different task for perception is suggested by Aristotle's examples of its role in deliberation (1112b34). 32 He mentions 'it's a loaf' and 'it's baked enough' as judgements we reach from perception. He adds that if we have to deliberate about everything, we will face an infinite regress. We must be able to make some judgements without deliberation, if our deliberation is to be a feasible task. Some of these are what we might call ordinary perceptual judgements—in this case, something that anyone can notice, whether or not they know how to bake bread. Others are judgements that have to be perceptual, but refer to features that we have to be trained to notice—in this case, the signs that a loaf is baked. This requires trained perception, because it cannot be replaced by an effective guide that can be applied by someone with ordinary perception. How do these features show how universals might come out of these perceptual judgements? Perceptual judgements can lead to the formulation of generalizations that still cannot be applied without the relevant training and experience. We might, for instance, learn how to cook an omelette. We notice that it is better not to allow it to cook completely in the pan. We form the generalization 'Take it out when it's still a little runny.' This, 32
Quoted at p. 119 above.
128
T H. Irwin
generalization, however, still cannot be applied by means of ordinary perception alone; we need experience and trained perception to estimate how runny is a little runny. The moral parallel is to be found, according to Aristotle, in generalizations that include 'too much, 'too far; and other qualifications that signal something not completely spelt out in all the relevant details. We can see how this might work in familiar cases. Perhaps, for instance, we should be concerned about our friends so that we neither seem indifferent nor seem too intrusive; we should be helpful without being interfering; teachers should make clear the faults in a student's work without being discouraging, and should be encouraging without concealing important faults. These generalizations are useful, but they do not provide an effective procedure for just anyone, irrespective of their experience of such situations, to identify the actions that conform to them. This would be an intelligible role for perception that fits Aristotle's examples, and it allows the formulation of generalizations. It may be a better account of the passages in Book vi on perception than the account that appeals to situational appreciation. Some argument for this view of perception, and against the view that relies on situational appreciation, may be found in Aristotle's casuistical discussions (ix 2). I have already described them briefly. We can now look back to them to see where situational appreciation might be applicable, and whether Aristotle seems to have it in mind. It is clearly relevant, since casuistry requires recognition of both relevance and decisiveness; Aristotle mentions both relevant and decisive features of situations, and he recognizes that they require the modification of generalizations that we might have thoughtlessly accepted had we not considered the situations he mentions. At this point, we might take him to agree with Ross, for instance, who actually quotes Aristotle's claims about perception to explain how we can identify decisive features. 33 Aristotle, however, does not say what we might expect a particularist to say. Specifically, he does not suggest that perception has any special role in these cases; it seems especially significant that he does not suggest it has any special role in identifying either the relevant or the decisive features. 33 See W. D. Ross, The Right and the Good (Oxford: Clarendon Press, 1930), 41 f: 'For the estimation of the comparative stringency of these prima facie obligations no general rules can, so far as I can see, be laid down. . . . For the rest, en tê(i) aisthêsei hê krisis. This sense of our particular duty in particular circumstances, preceded and informed by the fullest reflection we can bestow on the act in all its bearings, is highly fallible, but it is the only guide we have to our duty.' In 'preceded and informed' Ross may be going beyond particularism.
Ethics as an Inexact Science
129
Probably, then, he does not believe that the functions he ascribes to perception involve situational appreciation. If, however, he has in mind the role for perception that I have just illustrated from his own examples, it is intelligible that he does not mention perception in the discussion of casuistry. For the questions that he raises do not involve the sorts of discriminations in particular cases that Aristotle attributes to perception. I do not suggest that this argument shows decisively that Aristotle's remarks about perception are not about situational appreciation. I mention it to show that the case for particularism may be even weaker than I suggested it was when I conceded that the remarks are about situational appreciation. Conclusion I have tried to show that the passages on prudence and perception do not require a particularist interpretation that assigns normative priority to perception. We might be tempted to accept a particularist interpretation of these passages, if we were already convinced that the passages on inexactness and on perception require a particularist interpretation. Equally, we might be tempted to accept a particularist interpretation of the passages on exactness and on perception, if we already believed that this is the only possible interpretation of the passages on prudence. The fact that several passages speak of particulars and of the necessity of perception reasonably leads us to interpret one group of passages in the light of the other group. I have argued that none of the three groups supports a particularist interpretation, so that none of them creates a presumption in favour of a particularist interpretation of the other groups. For these reasons, the evidence I have discussed does not support a particularist view. In ethics as in natural science, Aristotle believes he can find theoretically significant generalizations. He also believes that in ethics his generalizations are significant for the primarily practical purpose of ethical inquiry.
6
The Particularist's Progress Jonathan Dancy
Many people seem willing to call themselves feminists with little idea of what they are committing themselves to. The same, in my experience, is true of particularism in the theory of moral reasons. There is a common suggestion that to be a particularist is, at the outset, only to admit that circumstances can make a difference. But if that were all that particularism amounted to, it would be uncontentious. In this chapter I lay out what I think one commits oneself to if one accepts the general claim that reasons are sensitive to context—a claim sometimes called holism in the theory of reasons, and of which moral particularism is merely one expression. Of course holism here, as elsewhere, does come in degrees. The strongest form of context-sensitivity would be the claim that every reason is somehow altered with every change of context. The weakest form is the claim that some reasons are on occasions capable of being altered by a change in context. The form of holism that I recommend is pretty weak on this scale, so far as the modality goes, but strong on the extent of the domain. I maintain that all reasons are capable of being altered by changes in context—that there are none whose nature as reasons is necessarily immune to changes elsewhere. When I talk of altering a reason, I mean to suggest not that the consideration which is a reason is altered, but that its nature as a reason changes. Instead of being a reason in favour of some course of action, it ceases to be a reason for action at all, or even becomes a reason against. One could express this by saying that the practical relevance of the consideration at issue is sensitive to changes in context, and the practical relevance of the I am grateful to Eve Garrard and David McNaughton for many discussions of the issues discussed in this paper; also to Roger Crisp, Brad Hooker, Derek Parfit and to all those audiences around the world that have let me try to persuade them of the merits of particularism. I owe special thanks to All Souls College, Oxford, where as a Visiting Fellow I wrote an early draft of this paper.
The Particularist's Progress
131
consideration includes its polarity. A consideration reverses its polarity when, having been a reason in favour of action, it becomes a reason against, or vice versa. My holism holds that every consideration is capable of having its practical polarity reversed by changes in context. It is hard to be sure quite how extreme a claim this is, partly because of the awkward modality in its characterization. But I shall not be discussing that matter much here. Perhaps I will have to admit that not all reasons are sensitive to context in this way—that there are a privileged few, including probably the intentional inflicting of undeserved pain, which necessarily constitute the same sort of reason wherever they occur. If so, I will have lost a battle but won the war. For the main aim of my particularist position is to break the stranglehold of a certain conception of how moral reasons function—the generalist conception under which what is a moral reason in one situation is necessarily the same reason wherever it occurs. Generalism need not be false of every moral reason in order to be largely false, and hence false as a general account of moral reasons and the way they work. And if it is false as a general account of such reasons, rational constraints on moral thought and action—in particular, accounts of what consistency requires in these areas—must not themselves be based on generalist assumptions. It may be that my train of thought here is vitiated by being overconcerned with one specific opposing account of how reasons function— Ross's theory of prima facie reasons. But I might as well admit that I do have this theory constantly in mind, since it seems to capture so well the outlines of the position I am trying to dislodge. Omitting Ross's epistemology for the moment, the theory of prima facie reasons holds: 1. What is a reason in one case is the same reason in all.' 2. Judgement is the attempt to determine the balance of reasons, so conceived.
Holism in the Theory of Reasons
In this section I argue in favour of particularism in ethics. In the past I tended to argue largely from example. 2 This persuades some people but not 1 This is not fair to Ross: see D. McNaughton's 'An Unconnected Heap of Duties?' Philosophical Quarterly, 46 (1996), 433-47. 2 See my Moral Reasons (Oxford: Blackwell, 1993), ch. 4.
132
Jonathan Dancy
others. Here my argument will be more theoretically grounded—though there will still be considerable use of examples as well. As I said in the preamble, I see ethical particularism as merely one expression of an overall holism in the theory of normative reasons—that is, in the theory that discusses the reasons that favour one thing (action, belief) over another. Such an overall holism can be expressed as follows: 1. What is a reason in one situation may alter or lose its polarity in another. 2. The way in which the reasons here present combine with each other is not necessarily determinable in any simply additive way. There are theoretical reasons and practical reasons, reasons for belief and reasons for action. My holism is intended to hold on both sides of that distinction. I start by trying to establish that theoretical reasons are holistic. We will quickly find that theoretical reasons are perfectly capable of changing their polarity according to context, without anyone making the slightest fuss about the matter. For instance, suppose that it currently seems to me that something before me is red. Normally, one might say, that is a reason (some reason, that is, not necessarily sufficient reason) for me to believe that there is something red before me. But in a case where I also believe that I have recently taken a drug that makes blue things look red and red things look blue, the appearance of a red-looking thing before me is reason for me to believe that there is a blue, not a red, thing before me. It is not as if it is some reason for me to believe that there is something red before me, but that as such a reason it is overwhelmed by contrary reasons. It is no longer any reason at all to believe that there is something red before me; indeed it is a reason for believing the opposite. As I say, it seems to me that nobody ever thought of denying what I am claiming here. I know of nobody who has nailed themselves to an atomistic (i.e. non-holistic) conception of how theoretical reasons function. If generalism is taken to be the view that all reasons are general reasons, i.e. that if a feature is a reason in one case, it is the same reason in any other case, generalism is uncontentiously false of theoretical reasons. Let us now turn to ordinary practical reasons. We will find just the same thing there. There are plenty of examples to persuade us that such reasons are holistic (or non-generalist, if you like). For instance, that there will be nobody much else there is sometimes a good reason for going there, and sometimes a very good reason for staying away. That one of the candidates wants the job very much indeed is sometimes a reason for giving it to her
The Particularist's Progress
133
and sometimes a reason for doing the opposite. And so on. Now examples would be of little use if there were some theoretical obstacle to taking them at face value. But again we should remind ourselves that nobody has ever really debated the question whether ordinary practical reasons are holistic or not. There should be no parti pris on this issue; so the examples, which are legion, should be allowed to carry the day without resistance. Perhaps this is too quick. There is a theory-based reason for doubting my claim that practical reasons are holistic, one that derives from the common thought that practical reasons are grounded in desires of the agent in a way that theoretical reasons are not. What one wants should not affect what one judges to be the case, on pain of charges of bias or prejudice. But what one wants can perfectly well affect what one has reason to do. Indeed, many find it hard to conceive of our having any practical reasons at all if we had no desires. My own view on this matter, however, is that desires do not give us or ground our reasons. Reasons stem from the prospect of some good. If we have no other reason to do a certain action, wanting to do it will give us no reason at all; nor can wanting to do a silly action make it marginally less silly. (These are only the first moves in a long debate. 3 I mention them here only to show the sort of way in which I find myself denying the possibility of grounding practical reasons in desires of the agent.) This view of mine is, of course, an independent input in the present debate. I mention it only to show that a certain motive for doubting the analogy I have been drawing between theoretical and practical reason is itself contentious. It may be that here we come across the real motivation for generalism in the theory of practical reason—an adherence to the view that reasons for action are partly grounded in desires. For if we accept that view, and if we then think of desires as giving the desirer the same reason wherever the desire occurs, we will at least get the sort of generalism I discussed above. The right response to this, however, is to claim that even if all practical reasons are grounded in desires, the same desire need not always function as the same reason. Consider first the third-person case. That he wants power and she does not may be a reason to give the power to her rather than to him, as I have already said. (It may at the same time be a reason to give it to him, since according to me one feature can be a reason on both sides at once; but remember that here it is a reason not to give it to him, and that it need not always be such a reason.) Now consider the first-person case. Suppose that I am trying to train myself into indifference towards a girl. I want very much to spend time with her. But I also want not to have this 3 For the remainder of the debate, see ch. 2 of my Practical Reality (Oxford: Clarendon Press, 2000).
134
Jonathan Dancy
want, since she is permanently indifferent to me. It is better for me not to think of her at all. If I spend time with her, this will make things worse for me rather than better—so long as I have not yet succeeded in training myself into indifference towards her. Once I am indifferent towards her, I can spend time with her without loss. In this situation, it seems, my desire to spend time with her may be a reason for me not to do so. Before carrying on to consider moral reasons, which have been claimed to be non-holistic, I want to step aside for a moment to ask whether I have not already made a mistake. There is a distinction between epistemic and what one might call constitutive reasons. An epistemic reason is a reason for believing something or other; a constitutive reason is a reason why something or other is the case. That the butler's fingerprints are on the murder weapon is a reason for believing that he did the deed, but no part of what makes it the case that he did it or of why it is true that he did it. That the hedgehogs are hibernating early is a reason for believing that we will have a severe winter, but not any part of what makes it the case that the winter will be severe. And so on. Now holism in the theory of reasons should concern itself with constitutive reasons rather than with epistemic ones. But I appear to have argued only that epistemic reasons are holistic, for my first example, or fulcrum, concerned reasons to believe that there was a red thing before me. It is, therefore, technically irrelevant. 4 This is true, and I apologize for it. But matters can be redeemed. We should not suppose that all that I have shown is that epistemic reasons are holistic, it being left entirely open whether constitutive reasons are or are not. For many, possibly most, epistemic reasons are also constitutive. For instance, that an action involves the gratuitous inflicting of pain is held by many to make it wrong, but equally clearly functions as a reason to believe that the action is wrong. It is both an epistemic and a constitutive reason. Some epistemic reasons are not constitutive, and perhaps some constitutive ones are not epistemic; this is all that can be said. Now could it be the case that the epistemic ones are holistic but the constitutive ones are not? I think that this is inconceivable. The mere fact of the overlap between reasons of the two sorts should give us pause. But more importantly, can we suppose that the very logic of epistemic reasons is capable of differing at a very deep level from that of constitutive reasons? This supposition entirely undermines the sort of connection there needs to be between reasons why things are so and reasons for taking them to be so.
4
Thanks to Nick Zangwill for pointing this out to me.
The Particularist's Progress
135
I return, therefore, to the onward or outward spread of holism. So far we have it that theoretical reasons (constitutive ones) are holistic, and so are ordinary practical ones. Now could it be the case that moral reasons are quite different from others in this respect, being the only atomistic ones? This is what many have supposed, in supposing that moral rationality is based on the existence of a range of moral principles. Moral reasons, they have held, necessarily behave in regular (or rule-bound) ways, though other reasons see no need to behave in that way at all. About this I want to say that straight off it just seems incredible that the very logic of moral reasons should be so different from that of others in this sort of way. Consider here the sad fact that nobody knows how to distinguish moral from other reasons; every attempt has failed. How does that fit the suggestion that there is this deep difference between them? Not very well at all. Then of course there are examples to be considered, examples of apparently moral reasons functioning in a holistic way. I forbear to bore you with these. It just seems inevitable that moral reasons should function holistically in the way that other reasons do. This certainly makes it hard to hold, as many do, that the very possibility of moral distinctions, of moral thought and judgement, is predicated on the existence of a range of moral principles. Moral principles, however we conceive of them, seem all to be in the business of specifying features as general reasons. The principle that it is wrong to lie, for instance, presumably claims that mendacity is always a wrong-making feature wherever it occurs (pro tanto, of course, not necessarily absolutely). It cannot be merely a generalization, a claim that lies are mostly the worse for being lies, for if all moral principles were of this sort the argument that moral thought and judgement depends on the possibility of moral principles would simply be the argument that such thought is impossible unless there is a considerable preponderance of normal cases over abnormal ones. I have never seen this argument made, and I doubt, what is more, whether it would be persuasive if restricted to ethics. If moral reasons, like others, function holistically, it cannot be the case that the possibility of such reasons rests on the existence of principles that specify morally relevant features as functioning atomistically. A principlebased approach to ethics is inconsistent with the holism of reasons. All the same, it might be argued, we have to admit that there are some invariant reasons—some features whose practical relevance is invariant. And surely I should allow this, because holism, as I expressed it, concerns only what may happen, not what must. It could be true that every reason may alter or lose its polarity from case to case, even though there are some
136
Jonathan Dancy
reasons that do not do this. If they don't do it, this will be because of the particular reasons they are. Invariant reasons, should there be any, will be invariant not because they are reasons but because of their specific content. And this is something that the particularist, it seems, should admit. It is like the claim that a man can run a mile in four minutes, that Sam Smith is a man, and that Sam Smith cannot run a mile in four minutes. These claims are compatible, and so are the claims that reasons are variable qua reasons though some reasons are (necessarily, given their content) invariant. The invariance, where it occurs, derives not from the fact that we are dealing here with a reason but from the particular content of that reason. So can the particularist admit the existence of some invariant reasons? The obvious examples are things like the causing of gratuitous pain on unwilling victims. Surely, it is commonly urged, this is always for the worse, even if over all we might in some case be morally forced to do it. Well, the first thing to say is that admitting the possibility of some invariant reasons is a far cry from admitting that the very possibility of moral thought and judgement is dependent on our being able to find some such reasons. To support any such suggestion, we would somehow need to be able to locate a sufficient range of invariant reasons, ones that together somehow covered the moral ground entirely and themselves explained the nature and role of the variant reasons. This is quite a different matter from simply trying to refute particularism (which is merely an application of holism in the general theory of reasons to the moral case) by producing one counterexample of an invariant reason, which is normally what is going on. Further, we should remember that the question whether reasons are atomistic or holistic is a very basic question about the nature of rationality, of how reasons function from case to case. It is, I suppose, conceivable that though the vast bulk of reasons function holistically, there are a few that function atomistically. But if this were true we would have a hybrid conception of rationality. There would just be two sorts of reasons, each with their own logic, and moral thought would be the uncomfortable attempt to rub such reasons together. It is much more attractive, if at all possible, to think of our reasons as sharing a basic logic, so that all are atomistic, or all holistic. Let us consider, then, how the supposed invariant reasons function as reasons in the particular case. Take the well-known example of the fat man stuck in the only outlet from a cave that is rapidly filling with water from below. We and our families are caught in between the fat man and the rising water. But we have some dynamite. We could blow the fat man up and get out to safety. But the fat man is unwilling to be blown up (he, at least,
The Particularist's Progress
137
is safe from drowning); and, let us immediately admit, he is blameless in being where he is, and in being fatter than the rest of us. So what we propose to do involves the destruction of an unwilling and blameless victim. As such, we might say, this is some reason against lighting the fuse and standing back. The question I want to raise is whether this feature (that we are causing the death of an unwilling and blameless victim) is functioning as the reason it here is, in any way that is to be explained by appeal to the (supposed) fact that it functions in the same way in every case in which it occurs at all. It seems to me that this feature is the reason it is here quite independently of how it functions elsewhere. Of course if the feature is genuinely an invariant reason, this fact, should we discern it, will be of use to us in any case where we might be in doubt as to the contribution it is making. We can say 'This is an invariant reason, it makes such-and-such a difference there, and so it must be making that difference here.' But suppose that we were to treat one of these supposedly invariant reasons as potentially variant, so as to deny ourselves the use of that inference. What sort of mistake would we have made? Would it be a failure of rationality to treat an invariant reason as potentially variant, or just a mistake of fact? I suggest that the invariance of the reason is an epistemic matter rather than a constitutive one. That the reason functions invariantly is a clue to how it is functioning here, but in no way constitutes the sort of contribution it makes to the store of reasons here present. In that sense, the invariance of its contributions is not a matter of the logic of such a reason, and failure to treat the reason as functioning invariantly is not a failure to understand how it functions as a reason. It is a perfectly good reason case by case without our worrying about how it operates elsewhere. I conclude, then, that particularism should accept the possibility of invariant reasons, so long as the invariance is not a matter of the logic of such reasons, but more the rather peculiar fact that some reasons happen to contribute in ways that are not affected by other features. We can admit this without adopting a hybrid theory of rationality, so long as we treat the invariance of any invariant reason as an epistemic matter rather than as a constitutive one. Holism in the Theory of Value The next question concerns whether our holism in the theory of reasons spills over to generate a holism about value. This new holism, value holism, can come in various forms, just like the holism of reasons. In broadest
138
Jonathan Dancy
outline, in my hands it will amount to the claim that for any x that has value in some context, x may have a different value or none at all in other contexts, and if there is disvalue as well as value, x may have value in some contexts and disvalue in others. Is there any prospect of accepting a holism of reasons and denying value holism? This is not a matter on which there is a long history of debate. Presumably we should approach the issue by thinking about the relation between reasons and values. Thomas Scanlon has recently revived the view that value is to be understood in terms of reasons. 5 He defines a valuable object as one that has features that give us reasons to protect, promote, admire, respect, approve of it (etc.: this list is open-ended). On this view, it seems inevitable that a holism of reasons will generate a holism of value. There are, however, other views to be considered. One might allow that wherever there is value, there are reasons, but leave it open whether there can be reasons that are not directly connected with value. 6 In this way one would make room for certain forms of deontology, without moving very far from the Scanlon position. Michael Slote and Roger Crisp suggest that there are some reasons that do not derive from values.' One might also ask whether there are agent-relative values as well as agent-relative reasons; perhaps agent-relative reasons do not 'stem from' values at all. There is, then, a spectrum of views to be considered. On the most trenchant views, a holism of reasons must be matched by a holism of values. But even on the less trenchant views, it still remains possible that the holism of reasons must match that of values. Suppose we agree for a moment that most reasons are linked to values. (linked to' is the vaguest phrase I can think of.) And suppose that values are atomistic, i.e. insensitive to context. Immediately we have the problem of explaining how it can be that the reasons linked to those atomistic values are able to be holistic at all. Why are they not atomistic too? Of course, if there are some reasons that are not linked to values, and those too are holistic, we might think this some justification for supposing that the explanation of their holism will not make reference to any holism of values. And we could then argue that, by parity of reasoning, the holism of other reasons is to be explained in a similar 5 See Scanlon, What We Owe to Each Other (Cambridge, Mass.: Harvard University Press, 1999), 95-7). Similar views were considered by W. D. Ross and adopted by A. C. Ewing. See Ross's Foundations of Ethics (Oxford: Clarendon Press, 1939), 278-83, and Ewing's The Definition of Good (London: Macmillan, 1947), 148-9. 6 I consider this possibility in my 'Should we Pass the Buck?', in A. O'Hear (ed.), The Good, the True and the Beautiful (Cambridge: CUP, 2000). See M. Slote, Goods and Virtues (Oxford: Clarendon Press, 1983), ch. 5, and R. Crisp `The Dualism of Practical Reason', Proceedings of the Aristotelian Society 96 (1996), 53-73.
The Particularist's Progress
139
non-value-related way. One trouble with this procedure, however, is that the preponderance of value-linked reasons seems to be enormous, at the least, and we are therefore in danger of letting the tail wag the dog. Another trouble with it is that the connection between reasons-holism and valueholism seems so plausible. It is far easier to explain the holism of the few non-value-linked reasons in terms of their relation to the holism of the many value-linked ones than to cast off the only obvious prop we have. I take it, then, that reasons-holism does not entail value-holism, since it is at least possible that reasons-holism is to be explained in other ways. By far the most plausible picture is that, just as most reasons are linked to values, so their holism is linked to a holism of values. This leaves us with a strong incentive to be particularists in the theory of value. Of course this incentive needs to be supported by examples of values varying according to context. But these are not too hard to produce. The disvalue of pain may be affected by the question whether it is part of a merited punishment. This does not mean that the pain hurts less, but that the punishment is not as much the worse for involving the infliction of the pain as we might have been led to think by considering the disvalue of other pains that hurt as much. If value-holism shadows reasons-holism, the two views have to be structurally similar. Now as far as reasons go, the holism that appeals to me holds: 1. What is a reason in one situation may not be the same reason in another; it may even change its polarity. 2. The way in which the reasons here present combine with each other is not necessarily determinable in any simply additive way. By analogy, then, our value-holism should look like this: 1. A feature or part may have one value in one context and a different or opposite value in another. 2. The value of a complex or whole is not necessarily identical with the sum of the values of its elements or parts. And this is therefore the form of value-holism that I adopt. In this I differ from G. E. Moore, despite that fact that we could both be called `organicists' in the theory of value. Moore believed (2) above but not (1). He held that any feature or element necessarily retained the same value as it moved from context to context, but that it could contribute to a complex of which
140
Jonathan Dancy
it was a part a value other than the one that it had there. The whole, that is, could be more valuable because of the presence of a certain part than could be explained by the value of that part; a part can contribute more, or less, value than it actually has. 8 I don't believe any of this. What explains the difference between Moore and me? This difference needs explaining, since all the examples that impressed Moore are just the sort of examples that impress me. How then have we come to such different conclusions? The answer lies in the fact that Moore accepted without question a certain doctrine of supervenience. He believed that the intrinsic value of something supervenes upon its other intrinsic qualities, and so that where an element does not change its intrinsic qualities on moving from one complex or whole to another, it must retain the same intrinsic value. The examples of organic wholes that impressed him required him therefore to say that elements can contribute to a complex more, or less, value than they have got themselves there. I, however, do not accept Moore's doctrine of supervenience. I accept (or used to accept, and as far as the present debate goes continue to accept) a slightly but crucially different supervenience doctrine, that intrinsic value supervenes upon other qualities, but not that the intrinsic value of one object supervenes upon its other qualities. For me, it supervenes upon other qualities, including those of other objects. 9 This may seem perverse. But there is a ready explanation of it, once one remembers my commitment to particularism. First, I distinguish between those features from which some value results (the good-making features, as we might put it), and other features whose presence or absence would have made a difference. The latter features are obviously relevant to the value, but they are not playing the same role as that played by the good-making features; they are not themselves part of what we might call the `resultance base. Given this distinction between roles, I can announce that intrinsic value is value that results from intrinsic properties of the object concerned, but also allow that that value can vary because of changes elsewhere, that is, in those properties whose presence or absence can make a difference to the ability of the intrinsic properties to generate the value that they do. The notion of supervenience draws less fine distinctions than that of resultance, See G. E. Moore, Principia Ethica (Cambridge: CUP, 1903), 30. For an early proponent of this view of supervenience, see A. C. Ewing, The Morality of Punishment (London: Kegan Paul, 1929), 166: 'It does follow from the conception of goodness or value that the value of something cannot be different except as the result of some other difference, but this difference need not necessarily lie in the thing itself, it may lie in something else. We cannot therefore say that the intrinsic value of any quality will always be the same under all circumstances.' 8
9
The Particularist's Progress
141
and all we can say is that intrinsic value supervenes upon other properties, not particularly upon intrinsic properties. And once we have said that, there is no reason to stop short of allowing that the 'other properties' can be properties of other objects, or relations between them and the first one, or whatever. My doctrine, then, is a sort of global supervenience, since the supervenience base is cast so wide, while Moore's is a sort of local supervenience. Who is right? I do not see that there is any way of determining the answer to this question before we get down to arguing about particular cases. It is not, that is, going to be a logical question which of us is right. Nor is it going to be decided quickly by appeal to the notion of a reason, or to naturalism in metaphysics, or anything like that. So the overall situation, as far as value-holism goes, is that if values are to track reasons, and if the structure of value-holism is therefore to be the same as that of reasons-holism, we have to abandon one traditional formulation of the doctrine of supervenience in favour of something less familiar. But to appeal to the traditional doctrine to defeat my form of value-holism would be to beg the question.1°
Holism in the Theory of Choice
Let us suppose now that I am right on both counts: my reasons-holism is the truth, and my value-holism is the truth as well. What does this tell us about the possibility of a full ordering, in which everything has its place, and where for each A and each B, A is either better than B, worse than B, or roughly as good as B? Well, so far as what we have so far seen will take us, there might still be a full ordering of that sort, in which everything has its place in the table of values, from best to worst. For all the considerations we have so far adduced concern the way in which the value of a complex is determined from the values of its parts, and the way in which parts may change their values as they move from one complex (context) to another. Once the value of the whole is determined, however, it is not going to vary, and we can enter the whole in its proper place in the great ordering. This would mean, for instance, that transitivity is not threatened by particularism. For with everything in its own place, we are never going to get a situation in which A is above B, B above C, and C above A in the Great Order. 10 It seems to me that Susan Hurley does make this appeal in her Natural Reasons (Oxford: OUP, 1989), 235-7.
142
Jonathan Dancy
This pleasing picture is however not as secure as it initially appears, sad to say. Let us remember that some of the things that have value are actions, and actions are chosen out of a set of alternatives available to the agent at the time. Now to adopt the picture I have just described is to suppose that each alternative action has its place in the great ordering of values, a place that is not affected by the place or nature of other alternatives. And this seems to require that the value of an action is never affected by the question what the available alternatives are. Now this is a very attractive doctrine indeed, partly because it enables us to retain a plausible principle of rational choice which we might call the principle of the indifference of independent alternatives (IIA): IIA: If in one situation I prefer action A to action B, it can never be rational for me to prefer B to A in other situations which differ from the first merely in the fact that further alternatives are available. In simpler terms, if I choose A where my choice is between A and B, I cannot rationally choose B where my choice is between A, B and C. The availability of C may indeed alter my overall choice, but it cannot affect the relative ranking order of A and B that has already been established. Though this principle is very attractive, I am not convinced that stubborn adherence to it is fully compatible with the broadening particularist perspective that I have been developing. For it is not obvious to me how one can prevent available alternatives from counting as part of the context within which an action is placed. And if one cannot do this, then the general particularist claim that context can make a difference to value as well as to reasons seems likely to take us to the view that the value of an action or choice can be affected by the alternatives that are available at the time. There is a reply to this, however. The argument of the previous paragraph might have been merely that every alternative is an object, though not all objects are alternatives (to each other). Since every object may have its value affected by others, every alternative may have its value affected by other objects, including some that are alternatives to it. There could be no bar against this happening—no bar that ensured that only those objects that are not alternatives to this one are capable of affecting its value. But this, though true, does nothing to establish the controversial doctrine that is really what we are after. That doctrine is that when one object becomes an alternative to another, that change may make a further difference to the value of the second—a difference beyond that
The Particularist's Progress
143
made by the existence and nature of the first object. And this doctrine does seem very peculiar." But not as peculiar as all that, perhaps. There may be examples of this phenomenon—of the arrival of a new alternative making a difference to the relative values of two existing alternatives, in ways that are not explained merely by the existence or possibility of the thing that becomes an alternative, but rather by its new status as an alternative. Suppose that we have two alternatives A and B, and that we prefer B to A. Our original question was whether we might be rationally led to prefer A to B if there appears a further alternative, C.' 2 Any successful example of this must meet certain criteria. It must not be one in which we simply change our minds about our initial ranking of A and B, perhaps for the reason that the appearance of C as an alternative draws our attention to something that we had previously missed. If nothing changes, however, it is hard to see how the ranking of A and B can be reversed by the arrival of C. The question to bear in mind is whether the examples offered contain the right or the wrong sort of change. Suppose that I have to buy a house in Reading, and have a choice between a smaller house within walking distance of the university, and a larger and more expensive one that requires a bus ride. I prefer the larger one despite the bus ride. Then a third house, even larger but also further away than the second, comes onto the market. I realize that if I buy the second house, I will always regret not having bought the third. With this in mind, I buy the first house. Is this rational? I suggest that it can be. There is of course a change of information here, but it is not the sort of change that I tried to rule out earlier—a change that leads to a change in my initial ranking. If there were no further house available, I would still have chosen the second. I have not changed my mind about that. So we are 11 I have expressed this doctrine in terms of a change. But that need not be the point. The question could equally well be phrased in terms of the difference between the case where C is not an alternative and the case where it is; can the difference between C's being an alternative and its not being one make a difference to the relative values of A and B? Here there is no talk of change. There is, of course, nothing wrong with examples that do involve change. It is just that I should be careful to avoid supposing that change is essential to the point. 2 It is not, of course, strictly necessary for us to find an example in which the order of the initial choice is reversed. It would be enough if we found a case in which the relative values are altered, so that the one that we originally preferred we still prefer, but not by so much—or by more, perhaps. Then we could argue that this sort of change in relative preference is bound to lead, on occasion, to a change in ordering. But it is more striking to produce an example in which the ordering is reversed.
144
Jonathan Dancy
not dealing here with a simple revision of the initial choice in the light of new information. Have I cheated in the description of the example? The obvious mistake would be a slide in the nature of the 'objects' of choice. Let us be sure that these 'objects' are my buying house A, my buying house B and my buying house C. House C was there all the time, but it was not on the market. The mere possibility of a further house C that is even larger than B, though still affordable, is not enough to cause me to prefer the nearer house. There has actually to be such a house C available to me before my continual regret at not having bought it can turn into a reason for me to choose the nearer one. The simplest example I know, and perhaps for that reason the best, is found in Wilfred Thesiger's Arabian Sands.13 He wrote, about his travels across the desert on a camel: 'I would not myself have wished to cross the Empty Quarter in a car. Luckily this was impossible when I did my journeys, for to have done the journey on a camel when I could have done it in a car would have turned the venture into a stunt.' So it does look as if there may be examples in which independent alternatives are not indifferent.14 Maybe, then, even if particularism does commit us to the existence of such examples, this is not a disaster. But I raised a question earlier that I have not yet answered. Does particularism itself constitute a reason for rejecting IIA? My first attempt to show that it does was a failure. I argued that particularists should not be surprised to see the nature of one thing making a difference to the value of another. But this was irrelevant. The real question was whether the rather special feature of `being an available alternative' can make a difference to the relative values of two things. Particularism does not show that this must happen. Particularism is everywhere permissive rather than prescriptive; or perhaps we should say that it forbids some things and prescribes nothing but suspicion. In the present case it says that we should be open to the possibility London: Longmans (1959), 260. I am grateful to Lars Bergstrom for very helpful discussion of this and other potential examples. He referred me to M. Resnik, Choices (University of Minnesota Press, 1993), 40, for another example with the same structure. Here is a further one, which I owe to Eve Garrard. I have to choose between two men, Joe and Sam. Joe is dull but reliable, Sam is unreliable but exciting. I prefer Sam, because if I chose Joe I would always be missing the excitement that Sam would have given me. But then Sebastian comes along, who is even more exciting but yet more unreliable. I realize that if I chose Sam, I would always be missing Sebastian's excitements, even though Sebastian's unreliability is so terminal that he is ineligible. And this tells me that my reason for preferring Sam to Joe is no reason, in the new situation, since choosing Sam will not lead to my having no lost excitement to regret. So I choose Joe. 13
14
The Particularist's Progress
145
of such a thing and not make a fuss if some crop up. The only reason for supposing that there cannot be any examples would be the generalist claim that since the feature of 'being an available alternative' often makes no difference, it makes no difference anywhere.' 5 There remains a difficulty. There appears to be an argument that there could be no counter-example. If so, the situation is unstable. We would have an example, our holistic position that there could be examples, and an argument that there can't be. Here is the argument. Sadly, it is one to which there can be no particularist objection as such; it is not a switching argument, for instance. Let us start with a supposed overall ranking, the Great Ordering. Everything has its place in the order. We can compare the values of different objects, and compare different objects in order to establish their relative placing, and these activities cannot themselves alter the values of the objects compared or their relative places; otherwise the very notion of relative value would be incoherent. One can compare the values of merely possible objects, e.g. possible courses of action or possible states of affairs. Could there be any difference between the values of merely possible objects and the values of those objects should they become real? No; for otherwise the activity of establishing the relative values of different possibilities would be incoherent. And this would make deliberation before action incoherent, if deliberation is the establishing of relative values of possibilities so as to decide which to make actual. Suppose then that I ask you to rank ownership of each of ten paintings. What you are ranking is a set of possibilities. Suppose then that I give you all the paintings, and ask you to rank the ten actual ownings. There can be no conceivable reason for a change in your ranking order (unless of course you have changed your mind). Now: could there be a difference between an order of preference and a ranking order for 15 An interestingly different avenue of approach, which I will not pursue here, starts from something that Derek Parfit is apparently happy to admit, namely an analogous claim concerning not value but 'ought'. Parfit's view seems to be that it is possible that one ought to do A if B is the only alternative, but that if C is also available one ought to do B. He denies, however, the claim that the values of A, B and C can be related in a structurally similar way. See his Reasons and Persons (Oxford: OUP, 1984), 429: 'Whether I ought to act in one of two ways may depend on whether it would be possible for me to act in some third way. . . . I then ask whether, compared with A + , A would have been better. The relative goodness of these two outcomes cannot depend on whether a third outcome, that will never happen, might have happened.' So in the evaluative realm Parfit is what one might call a choiceatomist but in the deontic realm he is a choice-holist. My own view is that this position is unstable. And my reason is fairly predictable: that if one admits that there are examples of choice-holism in the deontic realm, structurally similar examples will emerge in the evaluative one.
146
Jonathan Dancy
choice, where what one is dealing with is alternatives? No: there is no possible relevant difference between a preference order and a choice order. Suppose that instead of giving you all the paintings, I give you the money to buy one. You should buy the one that came top in your preference ranking. So the feature of 'being an alternative' cannot make a difference. Matters are more complicated than this. The complications do not make a difference, but they are relevant to what happens later. There is an obvious difference between buying just one painting and ranking them all. The difference comes out when we consider a case where I give you the money to buy one painting, then enough to buy another, and so on until you have all ten (though you never knew in advance that I would give you the money for the next). The order in which you buy the paintings need not be the same as your original order of preference. To see this point, it is important to distinguish between two quite different preference orders. The first has ten slots, in each of which one is asked to put one item of the form 'I own picture n. The second has ten slots, the first of the form 'I own picture n, the second of the form 'I own pictures n and m, the third of the form 'I own pictures n, m and p, and so on. There is no reason whatever why either of these two orders should be extractable from the other. The point is that if you already have, say, six of the paintings, you might rationally choose to add to those six a painting other than the one that came seventh on the list. To get a true analogue of the original ordering, when it comes to choice, we have to suppose that I give you enough money to buy one, but that just as you try to buy it someone else gets in first; so you should go for the second on your original list—but the same happens again, and so on down to the tenth. The order of choice should be the order of preference. What we have, then, is an explanation of why a certain feature, 'being an alternative' cannot make a difference, and therefore of why particularism is compatible with a full ordering. In one sense (epistemically, perhaps) it is possible that 'being an alternative' can make a difference. But there is an argument (which is not a switching argument) that no instance of this could be found. So the situation seems to be that we have on the one side an example in which the feature does make a difference, and a weak general reason derived from our holism to expect this sort of thing to crop up, and an argument on the other side to show that it is impossible. Now this is not one of those situations in which there can be reasons on one side and reasons on the other, and we can just decide where the balance of probability lies, leaving the defeated reasons in place. If we go with the example, we have to show what is wrong with the argument on the other side.
The Particularist's Progress
147
Luckily this can be done. The property of being an alternative is incapable of making a difference to a ranking order already established because there is no relevant difference between overall preference and overall choice. And just as the ranking preference order may be affected by the list of things to be ranked, so that if we take something off the list, the rankings of the rest may change, so with the ranking of alternatives. But all that this shows is that preference is like choice, and like choice in the crucial respect that it deals with alternatives. Being alternatives is the same as being mutually exclusive. Not all preference rankings are rankings of objects conceived as mutually exclusive, as we have seen. But some are. And the same is true whether we are ranking existing objects or possible ones. So the explanation of why the feature of being an alternative cannot be the cause of a difference between a preference ranking and a choice ranking is that this feature is present on both sides.16 Our conclusion should be that being an alternative can make a difference to all three rankings: of possible objects, of actual ones, and of objects of choice. The existence of persuasive examples should then move us without further resistance from value-holism to a sort of choice-holism, which holds that: 1. The value of one alternative can be affected by the nature of other available alternatives. 2. Assessing the relative merits of different alternatives is not the same as assessing the various alternatives one by one and then comparing the results. I want to end this section by comparing what I have said here with something I wrote in Moral Reasons: My daughter trod on a sea-urchin on holiday a few years ago, and we caused her considerable pain (not entirely with her consent) in extracting the spines from her heel. Was the pain we caused her something which made our actions worse than they would otherwise have been? Here is a switching argument which says that it was. Had there been available a painless method of getting the spines out, we would and should have adopted it. We would have been wrong to continue digging in her heel with a needle, because of the pain. Surely this shows that as things were our actions were the worse for the pain they caused? 16 Thanks to Eve Garrard and David McNaughton here. It would be wrong to say, in reply to this argument, that the feature we were originally discussing was that of being an available alternative, not that of being an alternative. The notion of availability merely takes us from possible choices (preferences) between mutually exclusive options to actual choices.
Jonathan Dancy
148
I don't think it does show this. What we should say about cases like these is that a feature which would have made this sort of difference had there been any alternative choice need not necessarily make it if there is no alternative. It seems to me quite consistent to say that as things stood our action was not the worse for the pain it caused, though that pain should have led us to choose another method had one been available.' 7
The idea, expressed in terms of reasons rather than, as above, in terms of values, was that the pain is not a reason against the action if there was no alternative, pain-free course of action available. It is not just that it is not sufficient reason; it is not any reason at all. I presented this thought as an application of a style of switching argument, whose general form is: if this action were less F, it would be better; so its being F must detract from its overall value. But it can be seen immediately that the example I gave goes further than is required for that purpose. My use of an example that hangs on a point about alternatives was more of a distraction than a help, since the general point I was trying to make was nothing to do with alternatives. Talk about available alternatives was intended more as an explanation of the supposed fact that, in the example given, the action was not the worse for the pain caused, even if it would have been better with less pain. Since it was not possible to do it with less pain, the pain caused does not make the action worse than it would otherwise have been. In possible world terms: even though, in the nearest world in which there was an available pain-free alternative, the action we did was wrong, wrong because of the pain it caused, and the worse for that pain, the actual action is not the worse for that pain. This is just an application of the holistic thought that a feature can make a difference in one situation that it does not make in another. Where this occurs, particularists admit that there must be an explanation of it; the explanation is that in the actual world, there was no alternative. Perhaps, then, the situation is like this. Holism takes away from us one of our two main reasons for sticking to principles like IIA. Hone is a holist, it is going to be hard to think that the question whether something was a real alternative cannot make a difference. If IIA expresses a form of generalism, holists don't have that reason to believe it. They might have the other reason, which is that if we lose principles like IIA (and all the rest), we lose what is really the only detailed account of the 'logic' of choice. The loss of IIA seems to be another nail in the coffin of the idea that there is such a logic. 17
Dancy, Moral Reasons,
65-6.
The Particularist's Progress
149
One might suggest, however, that the loss of IIA is far worse for generalists than it is for particularists. For generalists, the pillars of practical rationality would really be tottering. Choice-holism and the Possibility of a Full Ordering of Values Suppose then that we accept a general choice-holism to go with our reasons-holism and our value-holism. Have we totally lost any possibility of a full ordering? There are two ways in which we could hope to retain anything like a full ordering. The first is to say that what we have established is only that the context of real choice (i.e. the actually available alternatives) can affect the value of an option. This result, we might say, is clearly disturbing. But it does not altogether disturb our full ordering. Choice-holism concerns itself with real situations, in which the question what alternatives are available is a serious practical one. As such, it is to be distinguished from any thoughts about the effects of merely comparing one option with another. The value of an option will not vary according to what we imagine as the possible alternatives to it; it will only be affected by what actually are the alternatives. And mere comparison is relevantly similar to imaginary choosing, we might say, so long as the purpose of the comparison is to establish relative value. So an object's value is not affected by the mere act of mental comparison with another; it is only able to be affected if the two objects become real alternatives for some agent. Even this, of course, will do something to upset our full ordering, if we suppose that the very same thing can occur in more than one actual choice situation. But we might deny that possibility, supposing instead that objects of choice are incapable of recurrence. We cannot have the same action again, that is, and we cannot have any other choosable object again either. For the objects of choice are not repeatables. If I offer you a chocolate bar today and you refuse it in favour of a pint of beer, and I offer the very same bar to you tomorrow, the fact that it is the same bar does not show that you have the same choice again. You only have a similar one, and holists allow that objects that are intrinsically similar may yet differ in value because of their context. Be that as it may, the position that this move is trying to defend is surely another unstable one. It holds that mere comparison of two objects A and B, which we can do at any time at will, is incapable of affecting the values of A and of B; but should they become actual alternatives, their values may
150
Jonathan Dancy
be affected. So I may compare A and B and prefer B, and yet when I have to choose between the two choose A without irrationality. This distinction between actual and imagined choice, or between the effects of choice and those of comparison, is surely unsustainable. And this means that the dream of a full ordering collapses entirely. For if I cannot compare two objects without being in danger of affecting their relative values by doing so, there is surely no sense left in which objects have their own place in the ranking order. The ranking order must mean that objects have their place on it whether one actually compares them or not; indeed, to compare them is just to establish their relative placings in the order. If one could affect those placings by the act of comparison, the notion of an order would be destroyed. So much for the first way of defending the possibility of a full ordering. The second way involves us in redescribing each option in terms of the available alternatives to it. 18 Instead of thinking of ourselves as having the three options of buying house A, buying house B, and buying house C, our three options are: 1. Buying house A when we could have bought houses B or C. 2. Buying house B when we could have bought houses A or C. 3. Buying house C when we could have bought houses A or B. Now, we might say, these three options have an unvarying value, and occupy a fixed place in the full ordering. For if one of the three houses is taken off the market, or a fourth house enters the equation, we no longer have any of these three options, but either two or four new ones. The difficulty that I see in this approach is that if it is to avoid the difficulty we have already exposed, that comparison of the value of two objects is relevantly similar to choice, we will have to relativize every item on the full ordering to all other items, first severally and then in pairs and so on up until each is relativized to all others at once. And there will be no way of predicting, from the value of an option that is relativized to degree n, what its value will be relativized to other degrees. Given this, the use of the full ordering will be limited indeed. Transitivity, for instance, must fail. For if we rank 'A when we could have had B' above 13 when we could have had A, and 'B when we could have had C' above `C when we could have had B, it in no way follows that we should rank 'A when we could have had C' above 18 I have borrowed this manoeuvre from John Broome, Weighing Goods (Oxford: Blackwell, 1991), ch. 5, and 'Can a Humean be Moderate?' in R. G. Frey and C. Morris (eds.), Value, Welfare and Morality (Cambridge: CUP, 1993), 51-73.
The Particularist's Progress
151
`C when we could have had A'.19 It will be perfectly true, that is, that every relativized option occupies one and only one place in the ordering, without this doing anything to preserve the conception of rational choice that the ideal of a full ordering was designed to promote. There is a further problem. Suppose that we have a full ordering of all relativized options. This locates each option with respect to every other. Suppose now that I ask of item 32 in the list how it compares in value with item 33. It need not be the case that my answer is that item 32 is more valuable than item 33. The option '33 when I could have had 32' is a different option from the simple option '33, no matter how internally complex option 33 may be—and the same goes for option 32. But if my ranking order does not even commit me to claims about the relative values of the items ranked, it is pointless.
Holism and Explanation
There remains one further matter that I think it worth bringing out. It seems to me that particularism commits one to a highly debatable doctrine in the theory of explanation. In Moral Reasons I was not so clear about this. I start by considering the relations between two doctrines. One is the now familiar holism in the theory of reasons. The other is a doctrine in the theory of explanation, which has no agreed name that I know of. Here they are: Holism: the ability of a consideration to stand as a reason for action can be affected by the context in which it occurs. Non-guaranteeing explanations: an explanation can be perfectly good without being 'complete, where a complete explanation is one that is inconsistent with the non-occurrence of the explanandum. Where E and 0 both occur, the occurrence of E can explain that of 0 without guaranteeing it. Perfect explanations can be 'non-complete'. The idea that non-complete explanations are enthymematic is a mistake. 19 Of course, relative to one and the same three-way comparison, transitivity must be preserved—or at least nothing that I have said gives us any reason to dispute that. If we do dispute it, we will probably do so for quite different reasons, i.e. those stemming from comparisons in which many different criteria are operating at once. For a recent rehearsal of such considerations, see Larry Temkin, 'Rethinking the Good' in J. Dancy (ed.), Reading Parfit (Oxford: Blackwell, 1997), 290-345.
152
Jonathan Dancy
The question that I want to start with is whether I was right or wrong to think, as I once did, that my doctrine of reasons-holism was effectively the same as this doctrine in the theory of explanation. If it is, we have uncovered another awkward consequence of particularism. One preliminary first. Some philosophers maintain that though guaranteeing explanations are not required when what we are explaining is an action, they are required when we are explaining events. A scientific explanation, then, will have to be a guaranteeing one. 20° My own view, however, is that we would need a lot of persuasion to say any such thing. The only reason for adopting it, I think, is a sense that we would do well not to demand something of action-explanations that we have no prospect of achieving; in science, by contrast, where prospects are better, we should not be satisfied with anything less than perfection. The proper riposte to this is that action-explanations are as good as any explanation needs to be, and that, so far as the purposes of explanation are concerned, there is no reason to think there is either need or room for anything better. We should therefore accept the possibility of non-guaranteeing explanations on both sides or on neither. 2 The concept of a non-guaranteeing explanation requires that of an enabling condition. If there is a non-guaranteeing explanation F of an event E, there must be an event 0 such that: 1. The occurrence of 0 is not part of the explanation of E, and 2. If 0 were to fail, we would have a situation (F, not-E). In such a situation, 0 would be an enabling condition for the explanation that F gives us of E. The reason why there must be such things as enabling conditions if there are to be non-guaranteeing explanations is that if every candidate for the role of enabling conditions were to turn out to be part of the explanation of the event-type E, all explanations would be guaranteeing ones (when `complete'). If we think that there are any nonguaranteeing explanations, then, irrespective of whether we think that no explanations are guaranteeers, we must be able to make sense of the notion of an enabling condition—a condition whose satisfaction is required for the explanation, but which is not itself a part of that explanation. And this sets us something of a challenge. 20 For a recent example of this view, see Philippa Foot, 'Does Moral Subjectivism Rest on a Mistake?', Oxford Journal of Legal Studies, 15, 1 (1995), 1-14. 21 My views on the nature of explanation in general bear interesting similarities to those of Nancy Cartwright. See her How the Laws of Physics Lie (Oxford: Clarendon Press, 1983).
The Particularist's Progress
153
But we face the same challenge in the theory of reasons for action, once we adopt holism there. This is not perhaps too surprising antecedently, since presumably the reasons favouring an action are reasons that explain that action's rightness (or whatever moral status it has). If we start from one case where there are reasons R1—Rn making the action right, but allow that changes elsewhere may affect the ability of those reasons, not merely to make the action right, but even to be reasons at all, we will again say that features over and above the reasons must be functioning as enabling conditions. They enable the features that are reasons to be the reasons they are in this case, without themselves being among the reasons why the action is right. To give a very simple example: suppose that ought implies can. Then if I cannot do the action, the features which, were I capable of doing it, would be reasons why I should do it, are incapable of playing that role. But we should not conclude that my ability to do it is one of the reasons why I should do it. It is a condition that enables the reasons why I should to be the reasons they are, but is not itself among those reasons. Allow this, and you will probably allow the next: that I have the opportunity to help is something without which the reasons why I should help would not be those reasons. Her need is not a reason for me to help her if I have no possible opportunity of doing so. It is only a reason for me to seek an opportunity, which is different. But that I have an opportunity to help is not itself among the reasons for doing so. Another similar example: if I were not alive, the reasons that there are for me to help the needy would not be able to be the reasons they are. But this does little to show that among the reasons why I should help the needy is the fact that I am alive. So we see the same structure both times. Both holism and the claim that explanations need not guarantee are committed to making sense of the notion of an enabling condition. One might think 22 that there are forms of holism that don't have this effect. We might try to adopt a sort of weak holism without adopting any distinctive doctrine in the theory of explanation. Weak holism: the ability of a consideration to stand as a reason can be affected by what other features are present as reasons (but not by other things).
With this in hand, we might hope to avoid the need to talk about enabling conditions at all, and avoid the need to allow non-guaranteeing 22
And I did think so, until Eve Garrard showed me that I shouldn't.
154
Jonathan Dancy
explanations. But we can only do this if we can add to our weak holism a sort of: Holistic generalism: taking all the reasons here present together, they guarantee the rightness of the action; i.e. where present together elsewhere, they will always have the same effect (non-causally, of course).
This holistic doctrine is certainly compatible with the denial of a different form of generalism: Atomistic generalism: each reason has the same tendency as a reason, no matter what the context in which it is found.
But in fact weak holism cannot be coherently combined with holistic generalism. If the ability of a consideration to stand as a reason can be affected by what other features are present as reasons, why is it that the presence of further reasons in a second case, in addition to all those present in the first, is incapable of making any difference to the original reasons? Surely the official statement of weak holism says that a new reason can upset others whether we take them one by one or all together. So the combination of weak holism and holistic generalism is not a sustainable position. Moving now to the theory of explanation, we find the same thing. There might seem to be a form of holistic generalism available there, that is compatible with the demand that all explanations be guaranteeing explanations. We might say, that is, that though no individual feature has its own explanatory potency, which it carries with it from case to case, regardless of changes in other explanatorily potent features, still the entire complex of features that together explain the event constitutes a guaranteeing explanation. It is a guaranteeing explanation because that complex of features could not have occurred unless the event to be explained had occurred (or been going to occur). But this attempt to make room for guaranteeing explanations within a holistic picture suffers from the same incoherence that we found in the theory of reasons. No explanation is given of why the arrival on the scene of a new element is capable of affecting the explanatory contributions of individual features, but not of a set of such features. Why is it that a whole set of explanatory features is necessarily invulnerable to the sort of difference that a new feature can make to individual elements in that set? There is just no answer to this question. So there is no way for a holist to avoid making
The Particularist's Progress
155
room for the concept of enabling conditions, and denying the possibility of guaranteeing explanations. 23 One useful consequence of these thoughts is that the incoherence of the combination of weak holism and holistic generalism gives us an answer to one leading reply to particularism. 24 This reply is that my examples only concern what we might call contributing reasons. It is true that contributing reasons do vary in their polarity from one context to another. But it is not true of complete reasons. These remain the same, as reasons, regardless of changes elsewhere. There are two ways in which I can respond to this. The first is, as above, to point out that there is no obvious explanation of why a new reason in a new case is supposedly able to change the behaviour of one reason but not of all of them at once. Why is it that the whole pack of them is immune to change, when no individual one is? Note that this question is asked within the constraints of weak holism; it would be a different matter to maintain that considerations that are not reasons in the new case are able to make changes in this way. The second way in which I can respond to this attack is to challenge its notion of a complete reason. This cannot be identical with all the reasons present in the particular case, since we have already seen that there is no justification for the view that the pack of all such reasons is invulnerable to changes brought about by the presence of a new reason in a new case. It must therefore be something greater than that; it will presumably contain all the enabling conditions, as well as the absence of disabling conditions (specified one by one). But, first, this collection is ceasing to look like a reason at all, and, second, there is beginning to be a prospect that a complete reason will expand indefinitely. 25 23 The same may be true about causal statements: a sufficient cause need not be a guaranteeing cause. But to try to argue this would take us too far away from present concerns. 24 I mentioned this reply earlier, in discussing holism in theoretical reason. A recent example is to be found in Jonathan Bennett, The Act Itself (Oxford: Clarendon Press, 1995), 80. 25 Jonathan Lowe pointed out to me that a valid deductive argument is something we could reasonably call a complete reason, since it is monotonic, i.e. no addition of premises will affect the validity of the argument. In this, deductive reasoning differs from probabilistic reasoning. So there is one model of a complete reason that I cannot undermine. This model is, however, not applicable to moral reasoning, unless there are absolute moral principles; reasoning that runs in terms of prima facie principles is non-monotonic. There is certainly available a notion of deductive reasoning in ethics. The only question is whether the reasoning that takes us from premises specifying the features that make an action right to the conclusion that it is right is ever deductively valid. (I mean by this form of words to exclude such premises as 'all actions of this sort are right', 'this feature is the only relevant one' and 'this feature is a pro tanto reason'.) I think not.
156
Jonathan Dancy
What I really want to say, of course, is just that my interests are in the ordinary notion of a contributing reason, not in the concocted notion of a complete reason, which in my view is designed merely to save a dubious philosophical theory from refutation. But this may seem to be a definitional sulk; so I support it by attacking the notion of a complete reason that is designed to save generalism from counter-example.
7
Ethical Particularism in Context David Bakhurst
Introduction In recent ethical writing of a realist bent there is an increasing appreciation of the `situatedness' of moral cognition. This manifests itself, first, in the contextualist idea that the nature and influence of moral reasons cannot be understood unless agents are seen as situated in traditions of moral thought and practice, traditions that contribute to the character of moral reality and that empower agents to discern its nature, and second, in the particularistview that the structure of moral reality is not best captured by systems of moral principles; moral judgement involves a sensitivity to context which outruns anything moral rules can establish. This chapter examines the role such ideas play in the work of two thinkers. One is Alasdair MacIntyre, whose masterly After Virtue represents a powerful statement of a contextualist position.' The other is Jonathan Dancy, who has given articulate expression to a particularist ethics where moral decisions must be made case by case, 'without the comforting support or awkward demands' of moral principles. 2 Particularist and contextualist ideas, though distinct, suit one another. They are naturally An earlier version of this chapter was given at a colloquium on ethical particularism at the University of Reading in July 1996. I am grateful to Jonathan Dancy, David McNaughton, Piers Rawling, Hanjo Glock, and the other participants at that session for their remarks, and also to the Advisory Research Committee of Queen's University for funds that supported my visit to Reading. That version appeared as `Particolarismo etico in contesto', Studi Perugini, 2 (1997), 155-76. Thanks also to Allison Dawe, Angela Fernandez, Cheryl Misak, and David Wiggins for their astute criticisms of an earlier draft. Alasdair MacIntyre, After Virtue: A Study in Moral Theory, 2nd edn. (London: Duckworth, 1981; (revd.) 1984). 2 Jonathan Dancy, 'Ethical Particularism and Morally Relevant Properties', Mind, 92 (1983), 530.
158
David Bakhurst
combined, for example, in the view that a tradition is a repository of skills, or `know-how', the subtleties of which resist codification into principles. My interest here is whether MacIntyre's and Dancy's views can be woven into a coherent and illuminating whole. In After Virtue, MacIntyre urges a return to the Aristotelian moral tradition. Yet in the sequels to that work, Whose Justice? Which Rationality? and Three Rival Versions of Moral Inquiry, it is Aquinas, rather than Aristotle, who takes centre stage as MacIntyre endorses a Thomistic synthesis of Aristotelian and Augustinian ideas. 3 Though this development is anticipated in After Virtue, its prominence in the later books has repelled many readers who find much to admire in the earlier volume. I want to consider whether the kind of particularistic realism advanced by Dancy, which also has roots in Aristotle, might become an alternative, secular complement to the contextualism of After Virtue. After outlining MacIntyre's contextualism and Dancy's particularism, I shall defend them from an objection that argues that neither has an adequate account of the justification of moral belief, and tries to set the two views at odds with one another. I hope to show that, notwithstanding their differences, the two positions indeed complement each other, so long as Dancy's particularism is strengthened by incorporating various MacI ntyrean insights.
Exposition
MacIntyre
In After Virtue, MacIntyre argues that our moral tradition has lost its bearings. The seemingly interminable moral disputes in our culture are not, as non-cognitivists suppose, evidence of the fundamentally subjective character of ethical attitude. Rather, they arise from the fact that our culture has only a dim appreciation of the true character of moral inquiry. MacIntyre argues that our 'moral decline' has the following explanation. The terms of ethical discourse have their origin in a teleological framework. Classical and medieval thinkers employed a contrast between human beings as they contingently are, and as they would be, were they to realize their telos. Ethics was seen as enabling and empowering humanity to accomplish this 3 Alasdair MacIntyre, Whose Justice? Which Rationality? (London: Duckworth, 1988); Three Rival Versions of Moral Inquiry: Encyclopaedia, Genealogy, Tradition (London:
Duckworth, 1990).
Ethical Particularism in Context
159
end, to express humanity's essential nature. This framework thus allowed for the assessment of moral judgements in terms of truth and falsity. Consequently, when modernity undermined teleological explanation, morality faced a crisis of justification. The thinkers of the Enlightenment responded by attempting to derive conclusions about the authority of moral precepts from premises concerning human—or rational—nature, but these attempts could not but fail without the teleology that gave moral discourse its sense. Though our moral thought and talk seem to state facts and to be open to rational evaluation, morality has become for us merely a means for the expression of attitude. Thus emotivism, though false as a timeless philosophical theory, is true as a sociological characterization of the role of moral discourse in a culture like ours. MacIntyre argues our choice is either to follow Nietzsche and portray morality as an empty system of rationalizations masking the fundamentally non-rational character of the will, 4 or to return to the Aristotelian tradition and the idea of virtue. He favours the latter. In this, he does not seek (at least in After Virtue) to endorse some pre-modern view of where virtue lies, but to recommend a certain picture of the character of moral judgement and its justification. MacIntyre invokes the notion of a practice: 'a socially established cooperative mode of human activity' with 'internal goods. A good is internal to some practice if it is not intelligible independently of the practice. The good in question can neither be characterized without describing the practice, nor identified and appreciated by those who lack experience of the practice. Moreover, internal goods evolve in harmony with the practices that pursue them. Musical and literary appreciation, education, scholarship, and romance are all practices in this sense. MacIntyre claims that practices provide 'the arena in which the virtues are exhibited'. 5 This is true in two senses. First, virtues are qualities, the exercise of which helps sustain our practices and achieve their internal goods. Practices are cooperative endeavours, and virtues like justice, courage, honesty, patience, modesty, etc. are necessary for their successful realization. Second, MacIntyre urges that we replace the dominant individualist view of the self, as a self-authenticating atom unencumbered by social attachment, with a much richer picture. He argues that we are socially situated 'narrative selves, who live out a story that runs from birth to death. A life well lived is one about which the right kind of story can be told. We thus aspire to see the practices in which we are engaged as a complex unity, the principle of unification of which is some vision of the good. By arguing 4
MacIntyre, After Virtue, 117.
5 MacIntyre, After Virtue, 187.
160
David Bakhurst
that each human life can be seen as a quest for the good, MacIntyre reintroduces a teleological dimension into moral explanation. Accordingly, the virtues are portrayed as qualities that sustain us in this quest. 6 There are certainly elements of particularism in MacIntyre's contextualism. In his critique of the 'Enlightenment project, he argues that attempts to justify morality as a system of rules fail—Kant, Ross, and Hare are duly criticized—and where he first moots a return to the concept of virtue it is in direct response to this failure.' MacIntyre does not jettison moral rules, however, for he assumes that a community's moral knowledge will necessarily find expression in precepts and rules, of which law is the most formal articulation. Yet such moral rules are parasitic upon a prior understanding of the virtues which inform them; no one could understand and apply them properly unless they possessed the relevant sensibilities. 8 Dancy
The origins of Dancy's position lie in the work of John McDowell. In several influential articles, McDowell claims that moral requirements are genuine constituents of the world, there to be discovered. 9 This is nevertheless MacIntyre, After Virtue, 219. 7 MacIntyre, After Virtue, 119,232-3. We should note that, with time, MacIntyre has increasingly stressed the importance of rules. In Whose Justice? Which Rationality?, MacIntyre laments that he is often read as advancing a morality of virtue in contrast to a morality of rules, and he insists that even in After Virtue 'any adequate morality of the virtues was said to require as its counterpart a morality of laws' (p. ix). Nevertheless, although Whose Justice? emphasizes the importance of rules in moral education (31, 114) and the law, MacIntyre continues to maintain that the exercise of the practical intelligence (phronesis) is not, and cannot be, ultimately governed by rules (95, 115-17, 119-20, 194-6). Thus even where our moral view is represented in a system of principles or laws, questions about how the principles or laws are to be understood and applied cannot ultimately be settled by appeal to further rules: judgement must be exercised. In this, MacIntyre is at one with Dancy. More recently, however, in a review of Charles Taylor's Philosophical Arguments, MacIntyre claims that to submit our practices to rational evaluation, we are required to represent them as governed by rules. He chides Taylor for advancing a view of rule-following where 'what is involved is primarily a usually inarticulate and partly inarticulable embodied understanding of how to implement rules in varieties of complex and often ambiguous situations' (p. 96) and contrasts this with the idea, which MacIntyre prefers, that our practices are governed by inexplicit rules that can be articulated through rational reflection. Here MacIntyre's position seems remote from particularism. It is revealing that he portrays his disagreement with Taylor as one about the nature of rule-following, without considering the idea that our practices, or some of them, are not best seen as a matter of following rules at all. 9 These include 'Are Moral Requirements Hypothetical Imperatives?', Proceedings of the Aristotelian Society, 52 (supp. vol) (1978), 13-29; 'Virtue and Reason', The Monist, 62 (1979), 331-50; Non-Cognitivism and Rule-Following', in S. Holtzman and C. Leich 6
8
Ethical Particularism in Context
161
compatible, McDowell argues, with the recognition that moral properties are anthropocentric in character, and that the requirements they generate weigh only with beings able to have certain concerns, or rather, beings for whom certain concerns are fitting. The space of moral reasons is real, though it can be occupied only by beings of a certain kind. Drawing on Aristotle, McDowell argues that the ability to discern moral reasons involves the exercise of a form of practical wisdom (phronesis). McDowell's position is contextualist in that it sees practical wisdom as a capacity acquired in enculturation, a dimension of our 'second nature. Its acquisition represents our coming to occupy the moral point of view, from within which alone moral demands can be rendered fully perspicuous. The position is also particularist in that practical wisdom is viewed as akin to a perceptual capacity (to discern the good) which is non-codifiable in character: the deliverances of mature moral judgement cannot be captured in a set of moral rules. Both these points rest on a distinctive view of the relation of the moral and the non-moral. Moral properties cannot be reduced to, or otherwise linked with, non-moral properties in such a way that a being who did not share the moral point of view could grasp how moral concepts operate by discerning how the use of moral terms is associated with the regular occurrence of certain non-moral properties. One has to be within morality to see how natural features of the world are morally significant. Events that warrant description with the same moral predicate will not necessarily exhibit a pattern of similarity perspicuous in non-moral terms. This view challenges a certain non-cognitivist account of moral principles. On a non-cognitivist view, since reality does not contain moral value, agents cannot be criticized for failing to perceive what is objectively required of them. At best, we may hope to show that a wrongdoer's attitudes are inconsistent. It is natural for the non-cognitivist to argue that our ethical practice is consistent just in case we follow rules that pair sets of non-moral properties with their associated moral values, rules which (eds.), Wittgenstein: To Follow a Rule (London: Routledge & Kegan Paul, 1981), 141-62; `Aesthetic Value, Objectivity, and the Fabric of the World', in E. Shaper (ed.), Pleasure, Preference, and Value (Cambridge: CUP, 1983), 1-16; 'Values and Secondary Qualities', in T. Honderich (ed.), Morality and Objectivity: A Tribute to J. L. Mackie (London: Routledge & Kegan Paul), 110-29; 'Might There be External Reasons?', in J. Altham and R. Harrison (eds.), World, Mind, and Ethics: Essays on the Ethical Philosophy of Bernard Williams (Cambridge: CUP), 68-85; 'Two Sorts of Naturalism', in R. Hursthouse, G. Lawrence, and W. Quinn (eds.), Virtues and Reasons: Essays in Honour of Philippa Foot (Oxford: Clarendon Press, 1995). These papers are all reprinted in McDowell's Mind, Value, and Reality (Cambridge, Mass.: Harvard University Press, 1998).
162
David Bakhurst
ensure that we respond to relevantly similar sets of non-moral properties (the only genuine features of the world there to respond to) with the same moral judgement. McDowell's view that, from beyond the moral perspective, natural features are morally 'shapeless' undermines non-cognitivist appeals to principles of this kind.10 Dancy builds on the particularist elements in McDowell by attacking head on two leading accounts of moral principles. The first is R. M. Hare's doctrine of universalizability." Hare's is a non-cognitivist account of the kind described—one designed to show how parcels of non-moral properties may be tied to our moral responses in a way that generates principles that determine the rightness or wrongness of further actions. Hare argues that when we judge a particular action, A, 'wrong, we do so in virtue of some of its non-moral properties, say xyz. Consistency then demands that we find any action wrong that shares just those non-moral properties. Thus our judgement in the particular case is 'universalized' into a moral rule: any action which is xyz is wrong. This version of universalizability is too crude, for though A may be right in virtue of xyz, action B, which is also xyz, may yet be wrong in virtue of further properties. If I approve of an action because it is kind and generous, I am not thereby committed to approve of the gift of stolen property however kindly and generously meant. The universalizability theorist must therefore move to the view that if we call A wrong in virtue of xyz, we must call wrong any action similar 'in relevant respects, where a property is relevant to A's goodness if its presence or absence would affect the value of A. The particularist counters that, since the list of 'relevant respects' will be long, the resulting principles will be cumbersome. Further, Dancy argues that even if we can identify the properties in virtue of which we call action A good, and those which, were A to have possessed them, would have
10 Much in McDowell's view stands in marked contrast to Kant's moral philosophy. Not only does McDowell play down the significance of moral rules, he also argues that when we fail to exercise practical intelligence well, this is not always, or even often, a failure of rationality; it is rather our humanity that is wanting. There is, however, an important Kantian theme in McDowell's work, for he sees our ability to inhabit the space of reasons in general, and of moral reasons in particular, as exhibiting our distinctive character as autonomous, thinking beings. And this, he argues, is crucial to a satisfying general understanding of the relation of mind and world, see McDowell's Mind and World (Cambridge, Mass.: Harvard University Press, 1994), esp. lecture iv. 11 First expounded in R. M. Hare, The Language of Morals (Oxford: Clarendon Press, 1952), and revised and refined in Freedom and Reason (Oxford: Clarendon Press, 1963); Moral Thinking (Oxford: Clarendon Press, 1981), and numerous articles.
Ethical Particularism in Context
163
robbed A of its goodness, A will still have a large (infinite?) number of further features irrelevant to its moral worth. Yet though each of these further properties may be morally irrelevant in itself, the presence or absence of some combination of them may be morally relevant. But can we really expect our principles to anticipate the influence of such combinations? 12 Ofcourse,wldavth'nrespcuifd,bthen resulting principles will be vague. The particularist concludes that universalizability cannot deliver clear, workable principles. Dancy also attacks Ross's analysis of moral principles in terms of prima facie duties.' 3 Any absolute principle—say 'Tell the truth!'—is bound sometimes to conflict with other requirements. If we respond to such conflicts by qualifying our principles (e.g. 'Tell the truth, except where doing so causes harm of such and such a kind'), they again threaten to become unmanageable. Ross therefore proposes we treat principles as specifying only prima facie or conditional duties. The prima facie principle (P) 'Acts which produce pleasure are right' asserts that an act which produces pleasure is right, just so long as it has no further characteristics which make it wrong overall. (P) states a conditional duty in the sense that if the only morally relevant property of an action was that it caused pleasure, that action would be right. Ross's position embodies the idea that each prima facie principle identifies a property that is 'generally morally relevant, that is, morally relevant in the same way wherever it occurs. So properties like being a cause of pleasure, being truthful, being law-abiding are 'right-making characteristics' of actions: actions that bear these properties are always the better for them, even though they may turn out to be wrong overall in virtue of further properties. 12 This argument is made in Dancy's 'On Moral Properties', Mind, 90 (1981), 367-85, but it is not supported by example. A case like the following may be what Dancy has in mind. Suppose a man, M, gives a gift to a young woman friend, W. We judge that this action was good in virtue of being (p) kind and (q) generous. We also recognize that the properties of not being a bribe (—r), and of the gift's not being stolen (—s) are relevant to the moral value of the act. So we can list p, q, (—r), and (—s) among the morally relevant characteristics of the act. But note that while the facts that M is not sexually interested in W (—t), and that M is not a Catholic priest (—u), are each morally irrelevant to the value of the act (and so are not specified in the list of relevant respects), the combination of (t) and (u) would be morally relevant. It is absurd, however, to think that combinations of absent properties like (—(t+u)) should be specified among the morally relevant features of the act. Interestingly, Dancy does not pursue this argument in Moral Reasons (Oxford: Blackwell, 1993), moving straight to the point that the defender of universalizability, in an effort to incorporate all features possibly relevant to the judgement in question, is forced to broaden the `universalizability base' to the degree that it becomes a trivial truth that any relevantly similar action deserves the same moral evaluation. 13 W. D. Ross, The Right and the Good (Oxford: Clarendon Press, 1930), ch. 2.
164
David Bakhurst
Dancy attacks this generalist account of moral relevance. He does so by example. He points out that some actions are the worse for causing pleasure (cruel ones, for example), and that the truthfulness of an action is sometimes irrelevant to its moral value. If I tell a colleague before his graduate students that his latest work is a feeble contribution to an empty debate, is my action the better for being truthful? (That might make it worse.) So features which are sometimes, perhaps usually, right-making characteristics are elsewhere morally irrelevant or even wrong-making characteristics. As a result, cases can be constructed where an action deemed prima facie right for having some property turns out to be wrong overall in virtue of that same property. Take the prima facie obligation to obey the law. Someone who objects to enforced military service may do so precisely because the law requires it (that person might not object to voluntary military service). This generates the paradox that being commanded by law would be both a right-making characteristic of conscription, and the reason why it is wrong all things considered! Incoherence looms. Particularism is sometimes read as a purely negative doctrine, 'a natural default position to take up once one has found that no suggested moral principle seems to be fully plausible'.1 4 This characterization misses the fact that Dancy's criticisms of Hare and Ross are informed by a positive metaphysics of moral reasons. Dancy is a pluralist and a holist about moral relevance; that is, he argues that there is an irreducible plurality of features potentially relevant to the moral assessment of actions and to the constitution of moral reasons, and that which properties are morally relevant in some case, and how they are so, is determined holistically by the interplay of the various features of the case. What is significant in one case—perhaps so significant as to constitute a sufficient reason for action—may in other cases be irrelevant. There is some uncertainty about the character of Dancy's holism. It is unclear whether he wishes to hold the weak position that the moral relevance of some property in a particular case can be affected by the influence of other properties; or the strong thesis that no property has moral relevance independently of its relation to other properties.15 While the weak view is compatible with some account of general moral relevance, the strong view undermines the idea that we can assign to properties degrees of moral relevance antecedent to their appearance in particular cases. It is 4 Torbjorn Tännsjo, 'In Defence of Theory in Ethics', Canadian Journal of Philosophy, 25, 4 (1995), 573. 5 Dancy is non-committal at Moral Reasons, 103.
Ethical Particularism in Context
165
also unclear whether Dancy holds the narrow thesis that holism applies to the moral relevance of non-moral properties; or the wide thesis that holism also applies to those properties picked out by 'thick' ethical concepts, such as cruelty or kindness. Here I shall assume that Dancy's preference is for holism strong and wide. In any case, Dancy follows McDowell and argues that moral competence should be understood as a kind of perceptual capacity to respond to the specific configuration of morally relevant properties each case presents. It is as if the properties of a situation have a particular 'profile' or 'shape, within which some are especially salient. To grasp which features are salient in a case is to understand how one is required to act. Objection Despite differences of method and substance, MacIntyre's and Dancy's positions have significant similarities. Both stress that practical intelligence cannot be exhaustively defined by rules. Both are hostile to the legacy of the early-modern period and the Enlightenment, particularly the idea that we must deploy a 'disenchanted' view of objective reality—that the world, as it really is, contains no features characterizable only from a distinctively human point of view. 16 Accordingly both offer an `internalist' view of moral justification, where such justifications can be compelling only to those situated within a moral tradition. With diversity, difference and particularity fashionable in our intellectual culture, is the time ripe to herald a contextualist–particularist turn in moral realism? There would be two reasons for caution: if all the talk of situatedness were wrong-headed, or if the two views under scrutiny were somehow in tension. Consider the following objection, by Torbjorn Tännsjo." Tännsjo argues that any viable moral realism must provide an account of the justification of moral belief. Ideally, we want to know not just what a justification would look like, but which basic moral beliefs are justified. That way, moral philosophy can help direct us towards the good. Yet particularists have nothing interesting to say about justification. McDowell and Dancy portray moral judgement as a matter 'seeing the situation before one aright, discerning the correct 'shape' of the facts, etc. But they have no account of what makes a certain 'way of seeing' things correct. Dancy, for 16
Dancy, Moral Reasons, 30-2.
' 7
Tännsjo, 'In Defence of Theory in Ethics'.
166
David Bakhurst
instance, maintains that moral justification has a narrative structure. To display one's reasons is to present the salient features of the particular case in the appropriate narrative order.' 8 He does not explain, however, how we know when a narrative is true to what it describes. Tännsjo argues that advocates of phronesis, like MacIntyre and McDowell, typically gesture at a supposed body of shared moral understanding with reference to which we are said to assess the deliverances of practical wisdom in particular cases. Tännsjo doubts such shared understanding exists. But even if it does, we must acknowledge that its character is contingent on historical and cultural factors. So much we learn from MacIntyre, who stresses the plurality of moral traditions, the cultural diversity, even incommensurability, of visions of the good. So either we admit that no moral beliefs are justified or we make their justification relative to culturally specific moral traditions. At best relativism, at worst scepticism. What's a realist to do?' 9 Tännsjo's complaint is not that our thinkers cannot provide an 'external' justification for morality, one that would convince an amoralist to adopt the moral point of view. The charge is that they cannot give sufficient content to the structure of moral justification as it proceeds within the terms of morality. Tännsjo argues that, in this post-foundationalist climate, we are forced to embrace a coherence theory of justification. But this is impossible for those with particularist leanings, because the best coherence theory makes essential use of principles. This is Rawls's method of reflective equilibrium, where we arrive at justified moral beliefs by showing that our judgements about particular cases cohere with our general moral beliefs, stated in the form of principles that explain the moral properties of particular cases. Moral inquiry involves adjusting our particular judgements and
Dancy, Moral Reasons, 13. The charge of relativism has often been brought against Aft er Virtue, and in later works MacIntyre attempts to show, first, that the Aristotelian—Thomistic tradition offers a view of justification according to which the merits of rival traditions, or of different historical stages of one tradition, may be assessed by genuine standards of rationality. On this view, a tradition is held superior to a rival tradition (or to an earlier stage of itself) if it transcends the limitations and failures of its rival (or earlier stage), limitations and failures which, though recognized by the rival's (earlier stage's) own standards of rationality, cannot be overcome within that framework (see MacIntyre, Three Rival Versions, 119-20, 180-1; Whose Justice?, ch. xviii—ixx). As observed in fn. 8 above, recent remarks of MacIntyre's suggest that he believes that rational assessment requires us to represent our practices as rule-governed. For an historicist account of objectivity in justification that is consistently particularist see Dancy's Moral Reasons, ch. 9. 18 19
Ethical Particularism in Context
167
the principles that explain them until all our beliefs cohere. When we reach that happy state of 'reflective equilibrium, our beliefs are justified. Tännsjo maintains that the particularist, in contrast, is stuck with the worst kind of Prichardian intuitionism. The agent confronts a moral reality mysteriously containing reasons for moral belief and action that must somehow be intuited. No general structure can be discerned in moral reality, by appeal to which we might justify present judgements and determine how we must judge in future. Judgement proceeds case by case, atomistically. Moral reality seems capricious and moral claims groundless. At this point in the argument, someone might attempt to supplement Tännsjo's objection by turning MacIntyre against Dancy. MacIntyre sees the intuitionism of Moore, Prichard, and Ross as a graphic illustration of the poverty of ethical thinking after the collapse of the Enlightenment project. The intuitionist responds to our failure to theorize moral judgement by declaring moral judgement largely untheorizable. In this, he simply adopts the problem as his position. MacIntyre might well argue that Dancy's particularism inherits all the ineptitude of 'classical' intuitionism. Of course, MacIntyre might add, the sociology of particularism is different from that of intuitionism. Moore, Prichard, and Ross were secure in their moral universe, confident that the exercise of intuition by enlightened individuals would converge upon a set of agreed moral truths. In contrast, modern-day particularism occupies a more unstable moral and intellectual culture. There is widespread lack of conviction in, and even disdain for, Enlightenment ideals of truth and objectivity. How often we hear of the collapse of foundationalism, the arbitrariness of rationality, and the demise of 'grand meta-narratives, such as Christianity and Marxism. In this climate, particularism looks like the uneasy position of someone who has been persuaded that there is no hope for general moral truths and no place for moral theory but, unable to embrace the nihilism of academic postmodernism, clings to a residual faith in moral objectivity. As such, it is just another symptom of the postmodern condition. Dancy would no doubt reject this characterization. He might argue that particularism seeks a path between the ideals of the Enlightenment and the nihilistic protestations of postmodernism, and that MacIntyre's own attempt to carve such a path has been unsuccessful precisely to the extent that he has diminished the particularist dimensions of his position. The Thomistic Aristotelianism, to which MacIntyre is increasingly committed, operates with a monistic ideal of a single supreme good for human beings to which all other goods are subordinate. 20 Accordingly, MacIntyre argues 20
See MacIntyre, Whose Justice?, 165-6.
David Bakhurst
168
that the structure of practical reasoning involves argument from fundamental 'first principles' that articulate the nature of the telos. 21 Tännsjo would no doubt view this development in MacIntyre's position as a step in the right direction. But, the particularist counters, monism about value is grossly implausible. In the modern world it is simply impossible to believe that there is, in any substantive sense, a single supreme good for human beings. This view becomes no more attractive when imbued with the scent of divinity, the idea that the ethical first principles at issue are precepts of natural law, and so on. MacIntyre correctly perceives that our best chance for a coherent set of moral principles is to assume a monistic system of value. But, the particularist concludes, we are better off abandoning the chimera of a single supreme good, and moral principles along with it.
Response
Instead of continuing to envisage how Dancy and MacIntyre might squabble, I propose to consider how particularism might be defended and developed so that it can form a common front with contextualism against Tännsjo's criticisms. What Particularism Needs
In my view, if the objections above are to be silenced, the particularist must complement her holistic metaphysics of moral properties with a moral psychology that achieves two ends. First, the particularist must explain how, although our decisions are made 'case by case, they are nonetheless informed by general moral knowledge of some kind. This would lessen the impression that moral judgement is atomistic in a deleterious sense, and reveal how specific attempts to justify moral belief are made in light of perspectives that have continuity and integrity. Second, the particularist must show how her view of moral judgement is itself morally illuminating. Dancy suggests that particularism is an empowering doctrine, arguing that the self-conscious adoption of a particularist metaethics would bring us closer to moral reality. To show this is true would undermine the attempt to turn MacIntyre against Dancy by dispelling the impression that particularism is a view bred of moral inarticulateness.
21
MacIntyre, Whose Justice?, 92-3.
Ethical Particularism in Context
169
I shall suggest that it is precisely by incorporating Maclntyrean insights that Dancy's particularism can best develop a plausible moral psychology. Much of my response, however, draws on ideas present in Dancy's work, but Dancy makes them hard to see, in part because his style is terse, and in part because there are prominent passages, which Tännsjo seizes upon, which suggest that moral psychology is a closed book for the particularist. For example: [0] ur account of the person on whom we can rely to make sound moral judgements is not very long. Such a person is someone who gets it right case by case. To be so consistently successful, we need to have a broad range of sensitivities, so that no relevant feature escapes us, and we do not mistake its relevance either. But that is all there is to say on the matter. To have the relevant sensitivities just is to be able to get things right case by case. The only remaining question is how we might get into this enviable state. And the answer is that for us it is probably too late. As Aristotle held, moral education is the key; for those who are past educating, there is no real remedy. 22
This is an unsatisfactory passage. It is one thing to argue that metaethical theorizing will not invest the theorizer with the practical skills necessary to be a good person (a truth clear to anyone who knows a fair sample of moral philosophers), quite another to imply that the good person's view of things cannot be analysed and illuminated by philosophical speculation. Indeed, on the same page Dancy shows how we might say more. Generalism, he argues, encourages 'a sort of looking away' from the specifics of new cases. It breeds rule-fetishism, where agents refuse to see what the particular case `obviously call [s] for' because they are anxious to adhere to rules they have invoked in the past. Conversely, we can infer that particularism encourages sensitivity to the fine details of particular cases and willingness to inquire of each property exactly how it may be morally relevant. The particularist must be attuned to complexity, and prepared to articulate reasons by highlighting features of the present case, rather than solving matters by falling back on some general vision. This requires subtlety of mind and the courage not to use past judgements as an excuse for quick solutions to present problems. The generalist will likely reply that nothing about generalism rules out sensitivity to the details of particular cases. For example, the method of reflective equilibrium demands that we constantly consider whether our principles are sufficiently nuanced to explain the features of particular cases. Generalism does not encourage either the blind adherence to moral 22
Dancy, Moral Reasons, 64.
David Bakhurst
170
rules or their hamfisted application. In fact, Tännsjo might argue, it is the particularist who represents us as acting blindly. For if, as Dancy claims, we are unable to predict from the moral relevance of some property in this case its relevance elsewhere, then what we decide here is no help next time. With each case we begin anew. So, while the generalist can counsel us to 'treat each case as it comes, considering exactly how our principles bear on it, all the particularist can say is that each new leap in the dark must be made on its own merits. This generalist rejoinder is an overreaction, albeit a natural one, to Dancy's remarks about prediction. What Dancy claims is that we cannot predict exactly what difference some morally relevant property will make in future cases from the difference it makes in this one. This claim, however, is compatible with an agent's possessing a wealth of general knowledge about the kind of contribution moral properties often make. McDowell and Dancy sometimes treat moral judgement as a skill akin to an aesthetic ability. Just as an artist can see that it is appropriate to complete her painting just so, so the moral person discerns how it is appropriate to act in this case. Both judgements involve the recognition of a certain kind of necessity, and both require an essential background of aesthetic or moral experience. We have no problem admitting that the aesthetic judgement is made in light of considerable general knowledge about the practice in question that does not take the form of principles. A concert pianist may be able to discourse about the kind of considerations relevant to the interpretation of a Beethoven sonata. Although this knowledge is general in form, there are no strict rules that determine that this note at this particular point must be played just so. That judgement can be made only in the appreciation of the moment. Such is true of many skills exercised in the course of practices, where agents typically make informed and reasoned choices in light of general knowledge that is not formulated as a set of principles (it may not be formulated at all). We should see ethical judgement in the same light. This is in keeping with Dancy's claim that particularists need not abandon moral principles altogether, but can view them as reminders of about `the sort of importance a property can have in suitable circumstances'. 23 Suchprinlesdotbaflchekistongamhter; they can embody a rich and general appreciation of the way certain properties (kindness, cruelty, forms of pleasure, etc.) feature in our lives. Generalists may concede that, life being what it is, there can be no rules for how to live. Nevertheless, there may be rules about how not to live. 23
Dancy, Moral Reasons, 67.
Ethical Particularism in Context
171
Although there are no recipes for interpreting Beethoven sonatas, there are principles that rule out certain things as possible interpretations. If you pour paint into the piano, you are not interpreting Beethoven. Of course, it is a feature of many practices that they include, perhaps essentially, critical reflection upon the limits of the practice itself. This is particularly evident in art and philosophy. Much moral thinking concerns the limits of moral behaviour, and some views of morality see the setting of such limits as the primary task of moral theory. Ethics can define standards of behaviour, failure to adhere to which puts an agent beyond the pale of morality. What, asks the generalist, could such constraints be except rules or principles? And how could any moral psychology be adequate if it failed to countenance commitments to principles of this kind? One might expect Dancy to dismiss the idea that the limits of morality can be defined by deontological constraints. But he does not. Consider the following candidate constraints: A. Do not kill innocents. B. Do not cause unnecessary suffering. C. Do not exploit your fellow human beings. Dancy grants that such constraints can be true. He suggests, however, that the form of their expression somehow guarantees their truth. 24 To describe someone as 'innocent' is to say that they have done nothing that would justify their being killed. Thus (A) comes out true. Similar moves can be made for the other principles: 'unnecessary suffering' is suffering the cause of which cannot be justified and 'exploitation' is just the illegitimate use of people to serve the ends of others. The constraints are true, but only attention to particular cases will determine whether some act of killing, or causing of suffering etc., is justified. Dancy concedes that the constraints fix a `default' position about killing, causing suffering, etc.—such acts are wrong in the absence of suitable justification—and he admits this recalls Ross's prima facie duties. The worry is, of course, that the particularist's manoeuvre makes the constraints trivial. Turning Dancy's argument against Ross against Dancy, the generalist will complain that all we are told here is, for example., that killing is wrong in those cases where it cannot be shown to be right. This is almost empty. The purpose of constraints is to convey that there is something about killing, suffering, exploitation, etc., that makes them wrong, 24
Dancy, Moral Reasons, 229.
172
David Bakhurst
abominable, intolerable, etc., and which explains why one needs a special justification in cases where one is required to infringe them. It seems impossible to capture this without recourse to ideas of the general, enduring moral relevance of certain properties and general moral truths, such as `Kantian' ideas about respecting other people as 'ends-in-themselves'. No moral psychology, the generalist continues, will be adequate unless it can capture how such constraints can be internalized as abiding dispositions of agents. Surely anyone recognizable as a virtuous person appreciates the force of such constraints, a force which derives from the character of the acts they proscribe, rather than from 'structural' features of their articulation. Here we arrive at what is indeed a genuine problem for the particularist. The particularist need have no difficulty with the idea that acts which cause suffering, or involve killing or exploitation, are typically abhorrent and in those cases the grounds for abhorrence derive from the character of the acts in question and from that alone. The difficulty, however, is to describe how a good person might set herself against actions of certain kinds, and do so in a profound and meaningful way, while retaining a particularist appreciation of the complexities of predicting moral relevance and acknowledging that features which somewhere make actions abhorrent may elsewhere be morally inconclusive. The particularist needs an account of moral commitment, a way of making sense of how our moral lives can be guided by an allegiance to certain substantive visions of the good. This problem seems acute where particularists are strong, wide holists, holding that no property, not even those picked out by 'thick' moral concepts, has moral relevance antecedent to its interaction with other properties in particular cases. Particularism, Commitment, and Justification
To address the problem defined above, I propose to adjust a feature of Dancy's position. Dancy speaks of properties 'mattering, or 'making a moral difference, or 'having moral weight. He uses these various expressions interchangeably, for behind them lies a single conception of moral relevance: a property's moral relevance in some case, C, is a function of its contribution to the overall moral value of C, which in turn determines the agent's reasons for action in C. This notion of moral relevance is unduly narrow. I think we should admit that, say, suffering retains moral significance even in cases where the presence of suffering is not a critical factor in determining the agent's reasons for action (i.e. where it is not relevant in
Ethical Particularism in Context
173
Dancy's sense). Suffering is enduringly significant because it is something a morally sensitive agent must countenance. It is not something that such an agent could leave out of a moral description of an act where it is present, even in cases where its presence does not determine the structure of the agent's reasons or otherwise contribute to the value or disvalue of the act. The enduring significance of suffering derives from the fact that it is an appropriate object of moral concern, a source of moral commitment, and such concerns and commitments are critical to our understanding of ourselves as moral agents. This broader notion of moral significance is not, I think, at odds with particularism, because an agent could see some property as significant in this sense, and have concerns and commitments generated by it, while recognizing that its presence sometimes has no bearing on what he should do. We are forced, however, to modify Dancy's account of the structure of moral reasons. As we have seen, Dancy argues that, in some particular case, certain morally relevant features will be 'salient' (`they stick out or obtrude, and should catch our attention if we are alert' 25). These salient features jointly constitute the 'shape' of the situation and to grasp this shape is to see what one has reason to do, since 'to see a feature as salient is to see it as making a difference to what one should do in the case before one'. 26 The broader notion of significance requires us to admit that features may be 'salient, in the sense that their presence is something no morally sensitive agent could ignore, even where they have no bearing on the agent's reasons. This is where it is inviting to enrich Dancy's position with pivotal ideas drawn from MacIntyre. Dancy sometimes writes as if deciding what to do is a matter of staring at the situation, rather as if it were a 'magic-eye' 3D picture, until its 'shape' jumps out at you. As such, he focuses on the context of the decision without putting the decision into context. Moral agents enter situations with dispositions to treat certain properties as important and these dispositions are partly constitutive of their understanding of themselves as moral agents. Hence, what is salient in any moral situation is not just determined by the world. Cruelty, joy, pain, respect, courtesy, etc. are not features that I can fail to see as morally significant when I encounter them. I would change dramatically as a moral agent if I were no longer antecedently attuned to their presence. The properties thought to generate deontological constraints possess enduring moral significance in just the
25
Dancy, Moral Reasons, 112.
26 Dancy, Moral Reasons.
174
David Bakhurst
sense that a special regard for these properties is fundamental to our sense of moral identity. The crucial point, as MacIntyre brings out so well, is that the context of each decision includes the fact that it is made by an agent struggling to determine not just what to do, but what kind of person to be. The latter quest requires general moral competence of a richer kind than simply an appreciation of the importance certain properties can have, or typically do have, in certain circumstances. It involves acts of selfdefinition made in part by the articulation of an attitude to those properties as they typically manifest themselves. Such considerations must influence our understanding of moral justification. Deciding what to do involves more than just opening oneself to the facts; one has to make sense of the facts from a moral point of view. As we saw, Dancy holds that displaying one's reasons for action involves telling the right kind of story about the morally relevant features of the situation, a story which reconstructs the 'shape' of the facts. The idea that moral justification has a narrative form provides a cognitive counterpart to the concept of 'shape. Narrative allows us to reproduce the shape of the facts and to convey the compellingness of the reasons that shape constitutes. To take this appeal to narrative seriously, however, we must recognize, again with MacIntyre, that moral description is not a matter of telling isolated stories concerned only with fidelity to the contours of the events described. Rather, how I describe events in this case has to ring true in light of the stories I have told about previous events. Moreover, my integrity as a moral agent depends on the extent to which all these particular stories make sense as stories of which I am the author. MacIntyre's view of the self seems particularly pertinent, since it draws our attention to the fact that the descriptions I give of particular moral situations are events in a life, the unity of which is also a matter of narrative coherence. They are stories within wider stories—the story of my life conceived as a quest for the good, which is in turn set in the context of the communities and traditions in which I participate. MacIntyre's insight allows the particularist to countenance the full context of particular moral judgement, which includes the agent's attempt to live a certain kind of life, without compromising Dancy's holistic vision of the constitution of moral reasons in particular cases. The particularist–contextualist is now well placed to stare down Tännsjo's objection, for she has ample resources to illuminate the structure of moral justifications and the criteria for assessing them. Their structure is narrative in kind, and as narratives they must ring true in light of what we have said, and may want to say, about other cases (similar and different), about the coherence of the agent's life from a moral point of view, and
Ethical Particularism in Context
175
about the shape of a worthwhile life per se.27 There is, of course, no Archimedean point from which to assess these justifications, our terms of art remain contestable, and there will likely always be substantive disagreement. Yet though disagreement is a feature of our moral communities, the terms of controversy are, for the most part, shared by the participants. We know what we are arguing about and where and why our arguments break down. This is all the shared understanding we need if the dialectic of moral inquiry is to proceed. On this view, justification is certainly concerned with the ideal of coherence, with which Tännsjo is enamoured. But this is not coherence between a principled theory of the good and judgements in particular cases. It is rather a matter of the (internal) coherence of descriptions of particular cases, and the coherence of those descriptions with each other and with a developing vision of what constitutes a life well lived, a life the character of which is manifest in and through apt judgements in particular cases, though much else besides. The test of coherence here is not formal; nor is it a matter of harmonious evidential relations. It is a matter of what kind of lives can coherently be lived. Questions of coherence are thus intimately related to issues about the meaning and purpose of human lives that are matters of substantive moral debate. It remains to ask whether the psychology we have developed on the particularist's behalf is morally illuminating, or whether, as the objection above suggested, it reflects an impoverished moral perspective. The particularist enjoins agents to be as sensitive as possible to particular cases and not to embrace principles that will dictate answers to moral problems, serving as an excuse for due attention to the facts. Agents must be prepared to acknowledge that the factors relevant to our moral decisions can vary in their influence from case to case, sometimes dramatically. Properties which are in some cases sufficient to constitute moral reasons are elsewhere irrelevant no matter how important they seemed. I have tried to show that an agent can accept all this, while possessing considerable general moral knowledge and while fostering cares, concerns and commitments directed towards certain ways of living and dispositions to view certain properties as of enduring moral significance (a significance so profound as sometimes to find expression in deontological constraints). This is hardly a picture of a moral agent at a loss before the mercurial dance of moral properties. 27 Particularism need not, of course, be an individualist doctrine. It can be informed by a communitarianism, as recommended by MacIntyre, which stresses that the quest for the good is one pursued jointly with others. Much depends on exactly how the particularist employs the key notion of a tradition of moral inquiry.
176
David Bakhurst
Conclusion A number of prominent moral philosophers, MacIntyre among them, have complained that moral philosophy has focused unduly on questions of what it is right to do at the expense of questions of how it is good to live. This charge can be levelled at Dancy's particularism. As it stands, his brand of realism is too world-centred, focused too heavily on ethical requirements dictated by the world. In this chapter I have sought to add to the particularist picture a view of moral personhood, or at least to amplify the view already present, in order to lend particularism an account of moral commitment, without which no plausible theory of moral identity—of our quest to determine what kind of people to be—will be possible. MacIntyre's insights, I argued, are especially helpful in this regard. It might be argued that this shift of emphasis from right action to good character is unnecessary. After all, ethical truth must surely be a matter of what the world asks of one, and the coherence of a moral life is important only in so far as that life is lived in accordance with objective moral reasons. People's lives are composed of deeds, and their ethical integrity depends on whether they do what they know they should. Thus a 'person-centred' view is subsidiary to an object-centred approach. Such a response is mistaken. Our choice ought not to be between 'worldcentred' and 'person-centred' approaches to ethics. Persons are in the world and lives are open to moral assessment, just as actions are. We know someone by their deeds, but we understand the character of deeds, at least in part, by reflecting on the kind of lives lived by those who do them. So the question 'What should I do here now?' is essentially related to the question `What kind of person should I be?' Neither should have priority. This would seem in harmony with an Aristotelian view of ethical judgement as `situational appreciation, which, as David Wiggins puts it, seeks to articulate 'the reciprocal relations of an agent's concerns and his perception of how things objectively are in the world'. 28 I do not believe, then, that moral philosophy faces a choice between act- or virtue-centred approaches. They are not, or should not be, mutually exclusive. I hope I have shown that the two views under scrutiny can be powerful allies. Dancy's particularism has much to learn from MacIntyre's contextualism, and not just on the issue of moral identity. Particularism is crying out for an infusion of the kind of historical selfconsciousness which 28 David Wiggins, 'Deliberation and Practical Reason', in his Needs, Values, Truth (Oxford: Blackwell, 1987), 237.
Ethical Particularism in Context
177
informs MacIntyre's approach. At the same time, when suitably developed, particularism offers a secular framework in which to advance moral philosophy beyond the insights of After Virtue. Dancy's pluralism reminds those who, like MacIntyre, are prone to speak of the quest for the good, that we should not see this as the search for some final, general vision, but rather as an activity unbounded by the prospect of final success, but undiminished nevertheless by its infinite openendedness. Just like, we might add, the search for truth.
8 Particularity and Principle: The Structure of Moral Knowledge Jay L. Garfield
`When God created Adam, he whispered in his ear, "In all contexts of action you will recognize rules, if only the rule to grope for rules to recognize. When you cease to recognize rules, you will walk on four feet". '1
This remark is surely as true of our moral lives as it is of the more narrowly cognitive domain of language use and ratiocination in the context of which most recent discussion of rule-following has been conducted. Moral universalists might take comfort in such an insight, inferring that since the moral domain is rule governed, our moral knowledge consists in the knowledge of universal generalizations. But (as Sellars would have been the first to note), they would be wrong. For everything hangs on whether what it is to follow a rule and what it is to know a rule are understood as the grasp of a universal generalization from which knowledge of particular instances is derived, or as the knowledge of how to respond to paradigm instances, with an appropriate but perhaps inarticulate ability to generalize. Putting the point this way raises yet another issue central to debates about moral parI thank Kathryn Addelson, Anna Alomes, Marguerite LaCaze, Gwen Nettlefold, Leila Shotton, and Tim Sprod for helpful comments on an earlier draft as well as two anonymous commentators on a subsequent draft. Special thanks to Jane Braaten, Jonathan Dancy, Susan Levin, Barry Smith, and Elizabeth V. Spelman for extensive critical comments that forced me to see things differently. Thanks also to audiences at the Australasian association of Philosophy meetings in 1998, at the Australian National University Research School of Social Sciences, at the University of Canterbury, and at the University of Otago; particularly to Frank Jackson, Philip Pettit, and Michael Smith, as well as Derek Browne, Steve Gardiner, Dorothy Grover, Dianne Proudfoot, Charles Pigden, and Kurt Baier for hard questions and useful suggestions. 1 W. Sellars, 'Language, Rules and Behavior', in J. Sicha, (ed.), Pure Pragmatics and Possible Worlds (Ridgeview: Ridgeview Press, 1980), 138.
Particularity and Principle
179
ticularism: Is the debate between particularists and universalists one about the ontology of the moral domain or a debate about moral epistemology? In arguing for a particularist understanding of what constitutes a rulegoverned ethical domain, I will be locating the debate firmly in the domain of moral epistemology; in defending a particularist moral epistemology I will examine the role that concrete, particular experiences plays in shaping the inarticulate background against which we enact our distinctively human rule-governed lives; against which we represent our lives and the world in which we live them; and against which we formulate even the most explicit rules by which we live. These issues arise in particularly sharp relief in a recent exchange between Onora O'Neill and John McDowell, and I will explore them through attention to that exchange. In her fine book Towards Justice and Virtue Onora O'Neill defends the compatibility of a commitment to a broadly Kantian moral theory that places justice and a reliance on universalizable moral principles at the heart of morality with a commitment to the virtues as central to moral psychology. 2 She also argues that particularist accounts of morality—accounts that deny the foundational character of universal moral principles in favour of the view that moral judgement is ineliminably bound to particular contexts in which matters of perception, judgement, individual relations, etc. play roles that cannot be captured by general principles—can neither explain moral judgements, guide moral action and reasoning, nor enable moral criticism or discussion. She argues that the particularist critiques of those like John McDowell 3 and others of rule-based universalist moral theory underestimate the ability of general moral principles to apply to particular cases in the requisite fashion. Universal principles, O'Neill argues convincingly, are not thereby uniform. A consistent principle will require different actions by different agents in different circumstances. While O'Neill's defence of these three theses (Kantian universalism; the compatibility of virtue theory and deontology; the failure of particularism as a suitably general account of moral knowledge, action, discourse, and motivation) is both plausible and persuasive, I do not think that it ultimately succeeds. I will argue in this chapter that in fact universalist deontology and particularist virtue theory are rival accounts of moral 2 O. O'Neill, Towards Justice and Virtue: A Reconstructive Account of Practical Reasoning (Cambridge: CUP, 1996). 3 J. McDowell, 'Virtue and Reason', The Monist, 62 (1979), 331-50; Non-Cognitivism and Rule-Following', in S. Holtzmann and C. Leach (eds.), Wittgenstein: To Follow A Rule (London: Routledge & Kegan Paul, 1981).
180
Jay Garfield
knowledge and motivation, and that particularist virtue theory in fact provides a better analysis of moral knowledge, action, discourse, and motivation than does O'Neill's version of Kantian moral theory. In doing so I will defend a version of McDowell's Aristotelian account of moral knowledge, as well as the Wittgensteinian epistemology underlying it whose significance I will argue O'Neill underestimates.
O'Neill's Understanding of the Particularist Attack and Her Defence Against it O'Neill correctly locates the focus of the particularist attack on universalism: Particularists of any stripe—Humean, Aristotelian, or otherwise— emphasize what Williams 4 has called the 'thickness' of the morally relevant descriptions of actions: their saturation with cultural and social meanings which render them non-transportable from context to context. 5 A 'thick' description cannot be cashed out in culture- or context-neutral terms, but rather implicates a rich set of values and commitments, which inform, guide, and motivate action. Particularists argue that universal descriptions —those that abstract from this particularity and from the specificity of the situation of individual moral subjects or actors—necessarily fail to be morally relevant or action-guiding precisely because they abstract from the very semantic connections that render moral descriptions relevant to action and to criticism. O'Neill rehearses this view as follows: Particularists object strongly to the abstract character of act descriptions which universalists identify as the proper content of universal principles. They insist that we do not and cannot grasp, guide, or judge action by using abstract descriptions, or principles which incorporate them. Action must rather be grasped and interpreted in terms of 'thick', culturally and socially specific descriptions, which render it intelligible and accessible to the particular, restricted audiences who are B. Williams, Ethics and the Limits of Philosophy (Cambridge: CUP, 1985). This non-transportability is of course both weak and matter-of-degreeish. Moral descriptions are of course applicable in multiple contexts and typically in multiple cultures. The point is rather that such transportability is limited precisely by the similarity of those cultures and contexts, and is not guaranteed by any universal principles underlying those descriptions. As many universalists are quick and correct to point out, appeal to thick properties and to particular relationships can and often do function to mask bigotry, nepotism, and a host of other vicious attitudes and practices by shielding them from the scrutiny that comes from asking what morally relevant features of particular agents, situations, or patients justify particular actions, and whether other agents, situations, or patients might share those features—the Kantian demand for universalizability. Particularists must always be wary of this danger. But universalism is not the only way to avoid it. 4
5
Particularity and Principle
181
familiar with those descriptions, but equally may make it unintelligible and inaccessible to those familiar with the different thick descriptions. 6
Not all particularists share this conception of moral descriptions, and not all hold that the 'thick'/'thin ' distinction is even relevant to the particularist/universalist debate. Jonathan Dancy points out that he, for one, does not.' On the other hand, Dancy's position differs from McDowell's and mine in at least two other respects connected to this point: First, Dancy's defence of particularism is ontological in character, rather than epistemological. He argues not from claims about the structure of our knowledge to a thesis about the structure of the moral domain, but rather in the reverse direction. This point deserves some emphasis. For there are two versions of the particularist/universalist debate, reflecting two very different conceptions of the subject matter of that debate: Some, like Jackson, Pettit, and Smith (Chapter 4 in this volume) and Dancy (Chapter 6 in this volume) take the debate to be about moral ontology—about the character of moral facts. Others, like McDowell, O'Neill, and myself, take it to be about moral knowledge. While these debates are, to be sure, related, they are distinct. To my mind, since ethical decisions presuppose moral knowledge, the most important questions are questions about moral epistemology— questions about what we know when we grasp moral truths or properties, and about how we deploy that knowledge in the context of action. Secondly, and more importantly, Dancy's particularism is more radical than either McDowell's or mine: He chides us for conceding that there are any moral rules, or that moral claims are even universalizable. For reasons that will emerge below, neither of us could accept this stronger particularist position. We defend precisely a particularist understanding of moral rules and of our knowledge of them, not a rule-nihilism. This also deserves emphasis. Ontological particularists like Dancy, as well as ontological universalists like Jackson, Pettit, and Smith would infer from the truth of particularism that there are no rules and in this would agree with O'Neill and other epistemological universalists. They disagree only on whether to apply modus ponens or modus tollens. We epistemological particularists on the other hand argue that moral knowledge does consist in the grasp of rules, but that that grasp must be understood in particularist terms. This is the position I will defend at length below. While to some extent this might suggest that the debate between McDowell and me and O'Neill is orthogonal O'Neill, Towards Justice, 67-8. J. Dancy, 'Can a Particularist Learn the Difference Between Right and Wrong?' ( World Congress of Philosophy, Boston, 1998). 6
Jay Garfield
182
to more ontological debates, I think the outcome suggests that we are actually debating on the right ground, and that the ontological debates are rather beside the point. O'Neill offers two independent and plausible defences against this particularist attack on universalism as action-guiding. First, she argues that, as she puts it, 'intelligibility is not threatened by abstraction's —that is, that despite the fact that many moral descriptions are 'thick,' this is no bar to moral reasoning that abstracts from those thick descriptions, or the intelligibility of such reasoning to those from communities for whom those thick descriptions are foreign. Second, she argues, virtue theorists—the particularists most prevalent in the history of western ethics and those she takes most seriously—fail to provide an account of the moral relevance of the dispositions regarded as virtues, and that an account of that moral relevance must advert to general properties of actions. Let us consider these replies in turn. O'Neill defends the intelligibility of the abstract as follows: Reasoning that abstracts from culturally specific, locally intelligible act descriptions can remain intelligible to those from whose daily 'thick' descriptions it abstracts. Particularists are simply wrong to think that intelligibility is inevitably threatened by abstraction. All act descriptions, whether thicker or thinner, are abstract to a degree, and the more abstract are not necessarily the less intelligible. 9
This argument is as elegant as it is simple: All descriptions—whether thick or thin—O'Neill notes, are abstract in that in capturing action (or any other descriptum for that matter) under a verbal sortal we abstract from all of the properties not shared by all instances of the kind that sortal term denotes. Since intelligibility itself presupposes and exploits description and classification—after all, we come to understand actions and moral evaluations even on the particularist understanding precisely because we describe and classify them in particular ways—on O'Neill's understanding, intelligibility, rather than being threatened by abstraction, demands it. [ S] ome particularists insist that universalist writing ... does not take the demands of intelligibility seriously enough ... because it focuses too little on the sources and embodiments of action. Many particularists concentrate variously on the particular dispositions, states of character, sensitivities and commitments from which action flows, and the particular practices and commitments and traditions in which it is embodied, preserved, articulated and developed and from which further action flows. They favour 'agent-centred' or 'practice-centred' over 'actcentred' ethics; they insist that abstract principles may lack intelligibility and in
8
O'Neill, Towards Justice, 68.
9 O'Neill, ibid.
Particularity and Principle
183
any case will fail to capture the subtlety and complexity of a virtuous state of character or of good practices.10
The second part of O'Neill's defence against the particularist attack hinges on the alleged impossibility of dispensing with action as a focal category in moral assessment. Without taking actions and their moral properties seriously as the ground level phenomena of moral theory, she argues, the particularist cannot even provide an account of why the traits of character s/he picks out as virtues are morally desirable in the first place. Hence, she argues, the very particularist project of understanding the moral properties of action and our cognitive access to those properties in terms of the virtues and our possession of them cannot get off the ground. Virtues and vices are in fact usually construed . . . as dispositions or as traits of character of individuals, or as the practices and traditions of communities .. . [1] he category of action is neither dispensable nor dependent. Dispositions are merely tendencies to do acts of this or that sort. If virtues and vices were merely dispositions, action, intelligibly described, must be the basic focus for thinking about virtue or vice. If virtues and vices are . . . traits of character, which are not invariably, but only more or less reliably, expressed in action, once again the category of action must be the basic focus for an account of virtue or vice."
That is, O'Neill argues, on any plausible account of virtue, what makes a trait of character a virtue in the first place is its tendency to give rise to morally praiseworthy action. That means that any analysis of a virtue must begin with an analysis of action, and so the central part of the particularist programme in fact presupposes the universalism she defends. We now turn to O'Neill's positive account of and defence of the need for a moral theory expressed as a set of general moral principles. For O'Neill a central moral fact is moral disagreement, and a principal motivation for and task of moral theory is the principled resolution of serious moral disagreement. She faults particularism for its alleged impotence in resolving such disagreement, and hence for its failure to address the central task of moral theory—a task for which she argues that universalism is admirably suited: Particularists . . . have difficulty in dealing with ethical disagreement. . . . They stress the perception of particular cases rather than procedures for dealing with types of cases, and there is little they can say when disagreement appears. Yet ethical difference and disagreement are pervasive in contemporary life and the prospect of ignoring them or dealing with them by convergence of perception is 10
O'Neill, Towards Justice, 71.
11 O'Neill, Towards Justice, 72.
184
Jay Garfield
vanishingly slender . . . For reasons that any Wittgensteinian would recognise, we cannot just point and hope that the light will dawn for others; or even that we can point to anything determinate. When we deal with disagreement we need to contrast and debate, to compare and infer; in doing so we formulate and we follow varied rules of varying scope . . . Rules and principles may be embedded deep in individual characters or social traditions, whose practices and habits, virtues or vices they constitute. In such cases those who act may have little conscious awareness of rules, and perhaps no clear sense of the degree to which they leave matters open, or about their status or revisability. Even so, their use of rules and principles to guide action will be readily apparent.' 2
O'Neill makes two important points here. First, even if a particularist takes himself to be successfully indicating moral facts or morally significant instances by ostension, the facts or instances so ostended must have whatever suasive force they do under some description or other, and that description, if it is to be morally relevant must capture some general principle. So even the use of particular examples in moral discourse favoured by the particularist, O'Neill argues, presupposes the truth of universalism. Second, she suggests, even though the particularist might correctly deny that a moral agent or even a culture can articulate the rules that underlie moral judgements regarding or reactions to particular cases, this would not count against the importance of such principles as explaining those judgements or reactions.' 3 Appeal to introspection or even to the anthropology of moral discourse itself would therefore not count against her universalism. Finally, O'Neill argues, particularists either overestimate the power of moral communities to yield genuine ethical insight or—more problematically—use them as a cloak for morally objectionable discrimination, finding morally relevant differences that no disinterested party could find between relevantly similar particular instances: [I)] articularist strategies for fixing who or what has ethical standing are [not] convincing. Claims to discern who has ethical standing by inspecting the scope of 'our' particular community or nation or relationships are often decently clad in the vocabulary of commitment, care, and attachment, but neglect the hard cases for these approaches. These arise where none of these wholesome bonds is to be found: those who fix the scope of ethical consideration by reference to the limits of 'our' community or 'our' attachments and relationships must seemingly O'Neill, Towards Justice, 86-7. We might compare this to the way that grammatical principles opaque to speakers of natural languages none the less explain their judgements regarding our reactions to sentences of their native languages according to Chomskyan linguistic theory. 12
13
Particularity and Principle
185
exclude from ethical consideration those from whom 'we' care not a whit or whom `we' see as outside 'our' community, as well as those to whom we are bound by hatred or rivalry rather than by friendlier ties.' 4
So far we have three distinct objections to particularism. Let us recall them together for future reference, as we will wish to reply to each: (UD) All appeal to particulars in particularist accounts of moral knowledge of discourse presupposes universalist descriptions under which those particulars are subsumed. (TK) The moral principles that guide and explain the use of particulars in moral reasoning and discourse may be only tacitly known to those who so deploy them. (MA) The unprincipled use of particulars in moral reasoning leads to morally arbitrary moral distinctions being drawn. Before replying, however, we will need to tour some neighbouring conceptual terrain.
The Wittgensteinian Problem and Her Reply Thereto
McDowell's own particularism is grounded in his Wittgensteinian epistemology and O'Neill is especially concerned to refute that defence of the particularist moral epistemology. Her discussion of the Wittgensteinian account deserves special note: Some radical particularists offer further reasons for thinking that principles are empty formulae that cannot guide action. One of their principal inspirations lies in Wittgenstein's considerations on rule-following. They take it that the view of rules and actions required by any act-centred practical reasoning has been undercut by Wittgenstein's claims that rules cannot supply objective standards that determine us how to go on in one way rather than another. The notion of 'doing the same thing' is not fixed by rules or principles; hence universal principles, rather than demanding uniformity, determine nothing at all.' 5 If one attempted to reduce one's conception of what virtue requires to a set of rules, then, however subtle and thoughtful one was in drawing up the code, cases would inevitably turn up in which a mechanical application of the rules would strike one as wrong—and not necessarily because one had changed one's mind;
4
O'Neill, Towards Justice, 96.
15 O'Neill, Towards Justice, 78.
186
Jay Garfield
rather one's mind on the matter was not susceptible of capture in any universal formula.' 6
O'Neill argues that McDowell erroneously claims that rules, if they are to determine action must 'bear on action as external causal mechanisms', causing action according to deterministic psychological laws.' 7 Moreover, she argues, he subscribes to the more radical doctrine that what seems like rule-following is no such thing; rules yield not incomplete guidance, but no guidance'.18 O'Neill reads McDowell as concluding from the fact that rules cannot serve as a foundation for regular activity that they play no role whatsoever in such activity, reading McDowell as following Wittgenstein in this view. She responds as follows: Rules can be indispensable and yet indeterminate; they can be indeterminate and yet action-guiding. Agents can use rules to shape action, because rules do not function as mechanisms and in spite of the fact that they provide no algorithms for action. In using rules we shape our lives, we make judgements—often nuanced judgements—both about the situations we face and about the lines of action we will pursue. In short, the rule-following considerations provide no grounds for thinking that rules or principles, provided they are not conceived of in the misleading Platonic way, are impossible, or dispensable, or corrupting, or even dispensable in articulating the nuanced agreement, discussion and complex articulation of action to which those who are sceptical about rules point. Rules are not the enemy but the matrix of judgement. 19
So, O'Neill concludes, McDowell, following Wittgenstein, erroneously argues that rules play no role at all in our moral lives because they are incapable of determining action. Instead, she insists, rules, while not bearing mechanically on action, do in fact determine action, are used to shape action, and form the indispensable context which constitutes our moral knowledge and in which moral life is lived. But O'Neill is wrong, both in her interpretation of McDowell and in her understanding of the import of Wittgenstein's analysis for ethics. McDowell suggests not that moral descriptions are irreducibly particular—that in describing or evaluating actions we do not appeal to morally relevant properties satisfiable by other particular actions or states of affairs. Nor is he suggesting that there are no moral rules. Rather he argues that no conception of virtue could be reduced to any such set of rules, and that moral knowledge cannot consist in the mechanical application of a set of 16 18
McDowell, Virtue and Reason, 336. O'Neill, Towards Justice, 83.
17 19
O'Neill, Towards Justice, 80-83, passim. O'Neill, Towards Justice, 85.
Particularity and Principle
187
criterial rules.20 This is a much weaker claim than that O'Neill takes McDowell to make (but for all that not an uninteresting position). I will argue that in fact it survives O'Neill's critique precisely because she confuses it with the stronger no-rules/no-universals position and attacks the latter. The confusion, we will see, is double-edged: Not only does O'Neill fail to reply to McDowell's actual position because she mistakes it for a stronger and less plausible position, but she also fails to appreciate the insights McDowell's actual position offers because of that confusion. And these insights in fact provide a more satisfying context for her own account of the role of moral rules in moral knowledge. Let us begin by getting the contrast clear between McDowell's position as he articulates it and as it is represented by O'Neill. To do that, I would like to draw an analogy from a domain at least as important as the ethical: Australian rules football. Consider the difference between the way that two very important rules function in the game—the rule concerning what counts as a goal and that concerning what counts as a mark. The first really does determine what a goal umpire should say (even if it fails adequately to motivate with disturbing regularity). There are clear, unambiguous criteria for something counting a goal, and we can reduce the property of being a goal to that of the ball's being last touched by the foot (including the leg below the knee) of an attacking player and passing strictly between the big sticks (or their vertical extensions). To put the point starkly, by borrowing a bit of technology from fencing and some from air traffic control, we could make AFL goal umpires redundant. But field umpires' jobs are safe, not because of the poverty of technology, but because their calls require judgement. (For a player to be credited with a mark the umpire must rule that he controlled the ball after having caught it on the full at least ten metres from the point at which it was kicked, or that he would have controlled the ball were he not tackled or otherwise interfered with in the process of taking the mark.) What counts as 'controlling the ball' is a matter of jointly satisfying to an appropriate degree a number of soft constraints, some counterfactual, which might vary appropriately from game to game, depending on conditions, or from league to 20 F. Jackson, P. Pettit, and M. Smith, 'Ethical Particularism and Patterns', in the present volume (Chapter 4), argue that the supervenience of the moral on the descriptive 'tells us that there are necessarily true conditionals that take us from descriptive ways things might be to moral ways things might be' but note that 'it is a separate question whether there are moral principles in the sense of patterned connections between descriptive ways things might be and moral ways things might be.' This is precisely the point. Supervenience does not entail reducibility. See also Garfield, 'Propositional Attitudes and the Ontology of the Mental', Cognition and Brain Theory, 6 (1982), 319-31.
188
Jay Garfield
league. This is of course essential to the beauty of the game—its transcendence of expressibility in the aesthetic sense is partly grounded in its transcendence of expressibility in the technical sense. For all that, though, it would be ludicrous to say that there is no rule in the code concerning marks, or that field umpires are not guided by the rule in making their rulings. The rule simply is learned and guides action differently from the way that the rule concerning goals is learned and guides action. Let us note some salient features of that difference, as well as some important similarities. First, the similarities: Both rules guide the action both of umpires and of players. Both rules are clear enough for the game to proceed without endless argument about each case (excluding that among the fans). Both rules can be mastered to expert level by ordinary persons in a finite period of time. For each rule there is a clear difference between fair and arbitrary application. Finally, each rule is constitutive in its domain. Now, the differences: The goal rule can be applied mechanically; the mark rule cannot. One can get indefinitely better at applying the mark rule, capable of increasingly subtle discriminations and adjustments of one's judgement;21 goal umpiring has a natural ceiling on ability. These are uncontroversial, and O'Neill and McDowell would agree that in each of these respects moral rules are like the mark rule and unlike the goal rule. But the second two differences are crucial, and indicate the difference between O'Neill and McDowell both regarding moral knowledge and regarding rule following in general: (1) The goal rule could be learned completely without any experience of its instances; the mark rule could not. Clear instances of marks, of non-marks, and of borderline cases, together with a commentary on each explaining why each counts as a clear case either way or a borderline case would be necessary in teaching that rule, and what is learned is a discrimination based on similarity relations to paradigm cases. 22 (2) While there is a single descriptive property characteriz21 Tim Sprod (personal communication) points out that there is another possibility: With respect to field umpiring there is an asymptote of expertise at which even experts will disagree in particular cases. This might be a better way of expressing the truth. I'm not sure. But I note that either analysis serves to mark the distinction in the way requisite for the present argument. Probably only the metaphysics of football hangs on this point—a deep topic beyond the scope of this paper. 22 Tim Sprod disagrees here as well: He argues that we can learn the mark rule just fine simply by reading the rulebook; it is the application we can't learn without instances. But this isn't right. (First I note that even were it correct, all of the argument goes through mutatis mutandis with 'application of the rule' substituted for 'rule' throughout.) To learn the rule in the sense relevant for umpiring a football game (like learning moral rules in the sense relevant to making moral judgements) is not to learn a verbal formulation, but is
Particularity and Principle
189
able independently of the language of the rule universal to all instances of goals in virtue of which they are goals, there is no such property universal to all marks. One could not therefore, learn a formula and thereby learn what a mark is. 23 It should be clear where I am going. To assimilate moral rules to the goal rule and moral knowledge to that demanded of goal umpires is to align oneself in one's analysis both of moral principles and of moral knowledge with Kant. To assimilate moral rules to the mark rule and moral knowledge to that demanded of field umpires is to align oneself in these domains with Aristotle (epistemologically with Wittgenstein as well) and with McDowell. To assimilate the two kinds of rules is falsely to regard the second position as rule-nihilism and to align oneself with O'Neill. 24 That, I have argued, is
precisely to learn how to apply and how to deploy a rule. Someone who could recite the rulebook but who couldn't tell a mark from a hit-out could not be properly said to know the rule. Similarly, someone who knows the words 'Act only by that maxim which you could at the same time will be a universal law' but who, despite professing the Kantian creed, consistently judges acts of unbridled egoism to be obligatory doesn't know that rule in the relevant sense. 23 This is independent of O'Neill's avowal that rules can be 'indeterminate, 'provide no algorithm', and form the 'matrix of judgement'. Here I am focusing on the epistemic priority of instances over rules in cases such as this, as opposed to the priority of rules over instances in cases like the goal rule. It is this distinction that O'Neill misses and which is crucial to McDowell's account. Dancy, 'Can a Particularist Learn the Difference Between Right and Wrong?', suggests that connectionism might give us the right account of how we learn such open-textured properties. Maybe. On the other hand, non-linear dynamic cognitive theory might explain it, in which case any claim that there must be a finite or even recursively enumerable description of the descriptive properties on which the moral supervenes collapses immediately. Frank Jackson and Philip Pettit (personal communication), on the other hand, argue that there is no difference in kind between these cases, but only a difference in degree of vagueness that issues in greater agreement in goal cases than in mark cases. They point to the vagueness inherent in predicates like 'below the knee', etc. But this also misses the point. Vagueness may indeed be endemic in our language, and hence in all rules. But the mark rule is not simply vague—it is counterfactual in character; it is contextual in a strong sense; and its understanding is inextricably bound up with particular instances and with similarity relations. In all of these ways it is fundamentally different in cognitive character from the goal rule. 24 Again, it is worth emphasizing that this is not a dispute about whether there are moral principles, or about whether we use those principles in moral reasoning. Rather it is a dispute about the relationship between such principles and such phenomena as particular instances, ways of seeing, traits of character, and so forth. It is worth comparing the confusion with which I am charging O'Neill with readings of the Philosophical Investigations according to which Wittgenstein denies the existence of rules or that anyone ever follows rules in using language.
190
Jay Garfield
to fail to recognize important features of our knowledge and application of rules in open-textured domains and to fail to draw a real distinction. 25 We can now come back to O'Neill's three objections: (UD) All appeal to particulars in particularist accounts of moral knowledge of discourse presupposes universalist descriptions under which those particulars are subsumed. (TK) The moral principles that guide and explain the use of particulars in moral reasoning and discourse may be only tacitly known to those who so deploy them. (MA) The unprincipled use of particulars in moral reasoning leads to morally arbitrary moral distinctions being drawn. The replies are all straightforward at this point: (UD) involves a conflation of two senses of 'universalist description If we understand that phrase just to mean that, for example, all courageous acts can be described as courageous, or that all cruel acts can be described as cruel, the claim is true but trivial and harmless. Everything, we reply, depends on whether the account of the description and of our knowledge thereof is particularist or universalist. If, on the other hand the phrase really means that there must be a reduction of the predicate in question to a more fundamental descriptive vocabulary determining application in every case, we have shown (UD) to be false: Moral rules, like many other rules, display the open texture that entails that the predicates they employ will evade a precise noncircular specification of satisfaction conditions. 26 25 Here I note that Jackson, Pettit, and Smith's 'Ethical Particularism and Patterns' 83, simply get particularism (and Wittgenstein) wrong in an important way when they say, `The doctrine is that there is not even a highly disjunctive commonality or pattern that unites the right acts when described in descriptive terms. It is not, for example, like Wittgenstein's famous example of a game and, more generally, of family resemblances. In these cases, it can be difficult to spot or state the pattern, but the fact that, given a large enough diet of examples, we can say of some new case whether or not it is, say, a game .. . shows that there is a pattern that we can latch onto; our ability to project shows that we have discerned the complex commonality that constitutes the pattern.' Wittgenstein never argues for the existence of 'highly disjunctive commonalities'. He argues that no finite disjunction could do (and infinite disjunctions are neither well-formed in human natural languages nor, qua disjunctions, learnable). And to say that the fact that we learn such family resemblance concepts shows that there is a 'complex commonality' among the cases is not only unfaithful to the argument of the Philosophical Investigations, but also questionbegging in the present context. The particularist argues with Wittgenstein that we grasp such patterns despite the fact that they are not statable as commonalities, disjunctive or otherwise. 26 And for all of her talk about the need for rules to be applied flexibly and with 'nuanced judgement', if O'Neill has a complaint against McDowell on this score it is this complaint:
Particularity and Principle
191
(TK) can be granted, but in light of the foregoing considerations, no longer stands as an objection. Whether the knowledge in question is tacit or explicit, we have argued that it can be particularist in character and not universalist. Though moral principles are known by the virtuous, guide virtuous action, are non-arbitrary and are constitutive of the moral good, it does not follow that they can be finitely stated in terms of criteria of application. Moreover, learning them requires that one grasp their instantiation in paradigm cases, which cases are central to our grasp of the rules themselves, and that one grasp relevant, but not statable similarity dimensions determining correct and incorrect application of the rules. 27 Finally, we have shown, there is no reason to take (MA) seriously: Moral knowledge, moral reasoning and the application of moral rules understood as the particularist or the field umpire understands it is no more likely to lead to arbitrariness than is universalist knowledge, reasoning and application. In fact, ironically, one of the most important morals of the Private Language Argument (as Saul Kripke correctly emphasizes 28) is that while it might appear that the charge of arbitrariness attaches to the conventionalist, it is in fact the proponent of the determination of action by rules who is seen on analysis to posit an arbitrary move at each application. This apparent arbitrariness is only eliminated by appeal to the background of a form of life of the kind to which McDowell, following Wittgenstein, appeals. She insists on a general description under which an action satisfies the ethical predicates the particularist employs. Elizabeth Spelman (personal communication) points out that there is a confusion at work here between universal rules and universalizable rules. When we commit ourselves to the claim that if an act that satisfies a description is, in virtue of satisfying that description, a morally praiseworthy or blameworthy act then any act that satisfies that same description is as well, we commit ourselves to the universalizability of our moral judgements. But that is not to say that they are thereby universal. For there may be no general description of the act in question—none that abstracts in any interesting ways from its particular features—in virtue of which such a universalized principle can be stated. 27 It might be argued that similarity judgements require us to use principles in order to determine relevant dimensions and degrees of similarity—that without such principles to guide us any similarity judgement would be arbitrary. There are two ways of seeing how wrong this would be. In the first place, it flies in the face of much developmental psychological literature on concept formation (see esp. E. Rosch, and E. Lloyd (eds.), Concepts and Categorization (Hillsdale: Lawrence Erlbaum, 1978) ); and of connectionist work on concept representation in that spirit. Second, recall the important discussion of rule-following in Philosophical Investigations: To apply a rule itself requires that we be able to recognize what it is to go on in the same way, and to be able to recognize appropriate application and compliance conditions. These perceptual and behavioural dispositions can't be reduced to further instances of rule-following on pain of regress. 28 S. Kripke, Wittgenstein on Rules and Private Language (Cambridge: Harvard University Press, 1982).
192
Jay Garfield
To sum up the morals of this section: McDowell's particularism, which O'Neill paints as a nihilism with respect to moral rules, is not a rejection of the reality of moral rules, but an insistence on a Wittgensteinian, as opposed to a Kantian, understanding of the way moral rules operate in human life. O'Neill, on the other hand, while she explicitly grants the distinction I have been urging, and while she would even agree with McDowell that moral rules are more like the mark than the goal rule, argues that to adopt the particularism McDowell urges—to place the epistemic weight he does on particular cases and to urge that the content of moral concepts is more akin to similarity relationships than like description-satisfaction—is to reject any role for moral rules at all. And in this she is wrong.
Why General Principles of O'Neill's Kind Can't Guide
We could stop here, and we would have got somewhere regarding this dispute in moral epistemology. But this diagnosis enables a few more insights to be gleaned from McDowell's intriguing blend of Wittgensteinian and Aristotelian approaches in this domain. First, it is interesting to note that despite O'Neill's arguments to the contrary, it is McDowell's analysis that enables us to understand how moral principles can in fact guide action. If O'Neill were right, on the other hand, it is paradoxically difficult to see how moral rules could guide action. And here we can rely directly on the Wittgensteinian considerations motivated by McDowell and disparaged by O'Neill to make the point. O'Neill grants the open texture of the moral domain. That is, she grants that moral rules are 'indeterminate' and that, as she puts it 'rules do not function as mechanisms and ... that they provide no algorithms for action ... in using rules we shape our lives, we make judgements—often nuanced judgements—both about the situations we face and about the lines of action we will pursue.' Now, despite O'Neill's misleading construal of Wittgenstein as denying that we ever apply rules in such open-textured domains, Wittgenstein's subtle analysis in the Philosophical Investigations is in fact an account of just what it is to follow a rule in such domains. This is not the place to rehearse or debate the subtleties of interpretation of the `Private Language Argument'. Suffice it to say for present purposes that, as McDowell recognizes, Wittgenstein argues that in such domains—of which both much natural language use and morality are instances—the application of rules requires us to make use of similarity judgements, atten-
Particularity and Principle
193
tion to paradigm instances, and in the end to the learning of sets of social practices and ways of seeing that we learn only from example and from correction (§§ 143 ff). Rules, he argues, in such domains, are like sign posts: Their interpretation is fixed if, and only if, we share a common background of practices for their use (§§ 84-8). Knowing what a game is, he argues, consists in knowing instances, and extrapolating similarities. Without sharing the form of life that makes salient these practices of interpretation and application, together with relevant dimensions of similarity, even the rule for generating the series of even numbers cannot guide action satisfactorily. By emphasizing that moral knowledge shares this structure—that it is founded in our shared judgements about particular cases, and about shared practices of interpretation, application and similarity judgement— McDowell explains both, following Aristotle, how virtue and moral knowledge are acquired and, following Wittgenstein, how it is that the principles we learn in acquiring virtue can non-mechanically guide our action. In doing so, he calls attention—as Wittgenstein repeatedly does not only in the Philosophical Investigations but in On Certainty and Zettel as well—to how much shared inarticulate background is presupposed for such moral knowledge to be possible, and for it to have this action-guiding role. It is this shared background and its centrality to ethics and to ethical life that O'Neill ignores, and in virtue of ignoring it, has no way either of responding to or exploiting this deep Wittgensteinian insight, and so no way of explaining just how it is that the grasp of a rule determines anything at all. 29
29 Jackson, Pettit, and Smith, (`Ethical Particularism and Patterns', 89-90), argue that 'it is hard to see why we could not capture [the descriptive pattern on which some moral property supervenes] in words . . . Surely, if the pattern that connects matters described in descriptive terms with matters described in moral terms can be discovered, then, after we have made the discovery, we can tell people about it.' No. Anyone who understands the import of the Wittgensteinian argument sees that in countless cases we grasp patterns in inarticulable ways. Connectionist and dynamic learning theory provide empirical support for an analysis of this phenomenon. The following remark of Jackson et al. is apposite: `Before we leave the question as to whether there is a descriptive pattern in the right acts over and above the pattern given in the response-dependent role they play, we should emphasise that this question is separate from the question as to whether we can see the interest or point of the pattern independently of the role played' (96). This is very much to the point. In fact, it is the heart of the matter. What divides particularists and universalists is not a thesis about moral ontology, but about moral epistemology. Jackson et al. simply argue on the wrong ground, and talk past their particularist target. O'Neill, on the other hand, joins the battle where it ought to be joined—on questions about moral epistemology.
194
Jay Garfield Why O'Neill's Rules Can't Motivate
Moral knowledge must not only guide action; it must—as the Humean tradition in ethics correctly emphasizes—motivate action as well. Here, too, McDowell's account has an advantage over O'Neill's. Like any Kantian, O'Neill has the problem of explaining just how her account of the nature of virtue gives anyone a compelling reason to be virtuous. Even if we grant (as we will more explicitly in a moment) that some set of principles may be descriptively adequate with respect to the good, to the extent that they require us to set aside narrow self-interest in order to satisfy them we will need an answer to the question, 'Why be good?' and that is a question notoriously difficult to answer if it asked in earnest—that is, by someone not antecedently motivated to be good, as opposed to a philosopher examining the structure of moral psychology. 30 Part of the problem for a Kantian like O'Neill is that abstract moral principles are not self-motivating. They always, as Hume properly emphasizes, leave the springs of action untouched. Those springs lie in the moral imagination and in its power to arouse moral sentiments and passions sufficient to motivate action. And it is a fact about human moral psychology, remarked not only by Hume but by such contemporary moral psychologists as Annette Baier, 31 Owen Flanagan, 32 and Peter Unger 33 and exploited by those in the business of arousing them—all along the moral spectrum from Save the Children to the most xenophobic political parties and racist organizations—that our moral sentiments and passions are best aroused by the actions or plights of particulars (though of course this is not to say that our moral passions—raw or educated—provide an infallible compass for action). For any aretaic moral theorist—and especially for one who is also a Wittgensteinian epistemologist—such brute facts about human moral psychology are important if we are to understand the structure of genuine moral knowledge and virtue. For it is our moral knowledge and our virtue 30 Of course, universalists such as Kant and Habermas have provided candidate answers to this question. It is beyond the scope of this paper to develop a critique of those answers. But as the subsequent discussion will make clear, I find the moral psychology that underlies those replies implausible, even if the normative theory were to be granted. 31 Baier, 'Doing Without Moral Theory', and 'Morality and Reflective Practices', in Postures of the Mind (Minneapolis: University of Minnesota Press, 1985). 32 Flanagan, Varieties of Moral Personality (Cambridge: Harvard University Press, 1991). 33 P. Unger, Living High and Letting Die (New York: OUP, 1997).
Particularity and Principle
195
that motivates our moral actions, and in teaching a child to be morally good, we do this not by lecturing on an axiomatized moral theory set out in a set of moral principles, but by arousing compassion for the suffering, admiration for the kind, etc. And we do this by offering instances of moral goodness and evil for consideration and by praising and condemning instances of the child's behaviour. We do this in part because the passions and sentiments are so fundamentally attached to particulars, and so easily generalize from them. If these passions and sentiments were indeed in the first instance attached to moral principles, The Foundations of the Metaphysics of Morals might replace Aesop, Bible Stories, Jataka tales and their like in the world's cultures. But a world in which that were the case, while imaginable, would not be ours. Having said this, I must confess that this overstates the case. We do often teach children moral principles (as Pritchard points out34 ) but we do so as summaries, codifications of and as stimuli to moral responses we expect and engender, rather than as moral verities in their own right. We aim to make our children good, not merely dutiful.35 A further fact about moral and cognitive development poses difficulties for O'Neill but is readily explained by the particularist: As Aristotle notes in Nicomachean Ethics (1142a 10 20), while there are mathematical prodigies there are no moral prodigies. This, he explains, is because moral knowledge requires extended experience of many particular cases, and not knowledge of a universal, which can be acquired quickly. If universalism is right, we should expect moral prodigies. Particularism correctly predicts their non-existence. 36 -
34 Pritchard, Reasonable Children: Moral Education and Moral Learning (Lawrence: University Press of Kansas, 1996). 35 It would be unfair and inaccurate to suggest either that Kant himself suggested that the goal of moral education is the inculcation of duty as the primary moral motivation and, that the development of virtue is unimportant, or even that he neglected the importance of concrete examples in moral education. Marcia Baron, Kantian Ethics Almost Without Apology (Ithaca: Cornell University Press, 1995) correctly emphasizes that Kant enjoins the cultivation of virtue, sympathy, and generosity in children and that we provide children with an appropriate diet of examples, avoiding both the tendency towards sentimentality and that towards cold-heartedness. So here I am in fact in agreement with Kant on the details and methods of moral education. But this does not amount to agreement on the central questions concerning moral epistemology. We differ, that is, regarding the role these examples are to play, and the primary object of moral knowledge. 36 What has been said is confirmed by the fact that while young men become geometricians and mathematicians and wise in matters like these, it is thought that a young man of practical wisdom cannot be found. The cause is that such wisdom is concerned not only with universals but with particulars, which become familiar from experience, for it is length of time that gives experience, but a young man has no experience; indeed one might ask this
196
Jay Garfield
We are indeed capable of generating passions for justice, for fairness, against racism, and so forth. And general principles regarding these phenomena indeed—as O'Neill correctly emphasizes—play central roles in our fully formed moral reasoning and moral life. But these motivating passions derive in the first instance from our consideration of instances of justice, of fairness, of injustice, or racism, and so forth, and not the other way around. The roots of our moral sentiments, that is, are in the particular, and while the final flower of moral consciousness may appear to be universal, all of its motivational qualities are grounded in those particularist roots. For this reason, moral knowledge and virtue understood in McDowell's sense is straightforwardly not only action-guiding but also action-motivating, while if they are understood in O'Neill's sense remain in need of a credible theory of moral motivation, one whose relation to the moral theory itself one feels would be uncomfortably external and ad hoc.37
Why Moral Principles do Describe our Moral Life
There is a well-worn and useful distinction between two roles behavioural rules may play. 38 On the one hand, rules may accurately describe human behaviour much as natural laws describe the behaviour of inanimate objects, or even of infralingual and cognitively simple animals. Such rules are merely codifications of regularities. 39 But importantly, they need not be question, too, why a boy may become a mathematician, but not a philosopher . . . Is it because the objects of mathematics exist by abstraction, while the first principles of [ethics] come from experience?' I thank Susan Levin for drawing my attention to these remarks. 37 One might also argue that it is a mistake in the first place to ask of morality that it motivate. The content of morality, one might argue, is cognitive, and specifies the good. It is one thing to know the good and another to be motivated to act in accordance with that knowledge. Barry Smith (personal communication) argues that many criminals know perfectly well what morality requires, but just don't care. Moral knowledge hence appears to be independent of moral character. It is certainly correct that moral knowledge may fail to motivate. But this does not undermine the claim defended above that it is part of its function to do so, just as the failure of a heart to pump blood on some occasions is no argument against the claim that its function is to do so, nor is the argument that you need more than a heart in order for blood to circulate. My claim here is that a particularist understanding of moral knowledge, unlike a universalist understanding, explains how it can perform this function, not the stronger, implausible claim that such an understanding entails that it always does so. 38 W. Sellars, 'Mind, Meaning and Behavior', Philosophical Studies, 3 (1952), 83-95; L. Wittgenstein, Philosophical Investigations (New York: MacMillan, 1953), 147ff. 39 Depending on one's view of natural laws, and of the applicability of natural laws to the Geisteswissenschaften, these rules may be modalized.
Particularity and Principle
197
explicitly represented, known, or even dreamed about or formulable by those whose behaviour they describe. Indeed they may be the elusive quarry of social scientists. When discovered, they are not held to have been even tacitly known by those they cover—only to predict correctly and to describe their behaviour. The rules that predict our gaze patterns may be unknown to many of us, but may be very useful predictors for advertisers. On the other hand, rules may prescribe. In this guise, for rules to have any grip on behaviour they must be known—at the very least implicitly, but in order to facilitate reflective discourse, explicitly—by those whose conduct they govern. If we want 'Stop on red' to govern our behaviour, we must ensure that people come to know it, unlike those rules governing gaze patterns. Prescriptive rules that become widely known and widely followed can often then do apparent double duty as descriptive rules. We can accurately subsume and predict the behaviour of drivers, ceteris paribus, under the traffic laws. But this apparent double duty is deceptive. The real descriptive force is achieved by conjoining the rule with the statement that the rule is known, and that those who know it indeed obey it. No such epistemic or compliance condition is needed in the case of purely descriptive rules. We might sometimes come to believe that a rule is functioning prescriptively for a person or a group of persons when in fact the rule is merely descriptive. We might come to such a conclusion in view of quite naturally ascribing a motivational structure and epistemology, or a cognitive psychology, to those whose behaviour is governed by the rules we observe according to which it is overwhelmingly likely that they satisfy those rules in virtue of representing them to themselves (perhaps tacitly) and then obeying them. And in so hypothesizing, we might on occasion be wrong. Some argue that Chomskyans are wrong in exactly this sense. A man might always buy the same brand of soap; it might be the cheapest brand. An observer might ascribe to him adherence to the prescriptive rule 'Buy the cheapest soap.' But he may be oblivious to price, and even to the uniformity in his behaviour. He might just grab the rightmost soap on the shelf at eye level, as psychologists and advertisers would predict, which soap, in the supermarket he frequents, happens to be the cheapest. Now this duality draws our attention to the fact that to say that rules, including moral principles, play an important role in our moral life or in our ethical theory is multiply ambiguous. We could be asserting that explicitly articulated moral principles undergird and guide our moral perception and reasoning, like 'stop on red' undergirds and guides the behaviour of competent, law-abiding drivers. We could on the other hand be
198
Jay Garfield
saying that they merely describe moral perception and reasoning that is determined by quite other psychological mechanisms, like 'buy the cheapest brand. Or, we could be suggesting that the rules are known only tacitly, but yet guide our behaviour and moral thinking, like the rules of universal grammar. We might think that the rules we posit operate 'in real time' as we act and judge; we might think that we use them only heuristically or critically, on reflection. This range of possibilities suggests that the fact that universal moral principles appear in some way in our moral lives and moral theories in no way entail the truth of universalism. As the analogy of the field umpire makes clear, and as should be clear in any case from a positive reading of Wittgenstein and of McDowell's appropriation of his ideas in the service of Aristotelian moral theory, much of our moral reasoning, criticism and discourse employs moral principles, formulated in universalist language. This is natural and appropriate, and it would be surprising if it turned out not to be so. O'Neill gets this right. So, if an aretaic theorist like Aristotle, Hume or McDowell were to deny that we use such principles, or if it were to turn out to be a consequence either of virtue theory or of Wittgensteinian epistemology that we never use such principles, this would be a straightforward reductio on the position. But of course neither do any of these theorists deny that we do, nor is it a consequence of any of their views that we do not use moral universal moral principles, just as it is not a consequence of Wittgenstein's views about meaning that Peano's postulates are false. McDowell's framework, instead of denying that there is a role for universal moral principles, explains precisely what their role is, and how it is possible for them to play it. Universal moral principles, on this understanding, are ineliminably open textured summaries of our moral knowledge, and as such can be used in explicit moral discussion, including discussion aimed at transforming the moral intuitions and behaviour of others, and in moral training. In the same way, the mark rule is a true, universalizable, but ineliminably open textured summary of a field umpire's football knowledge, which can be used profitably in the discussion both of play and of umpiring practice, including the training of umpires and the teaching of the game. All that is denied is the foundational status of the rules with respect to the knowledge they express. 40 4" Note that this is different from saying with Ross (The Right and the Good, Indianapolis: Hackett, 1988) that moral principles and obligations are always prima facie principles and obligations. We do not, for instance, say that the mark rule is a prima facie rule. Particularism of this kind is not the claim that there may always be defeaters in the
Particularity and Principle
199
This account of the positive role of universal principles in moral life and discourse makes it clear that O'Neill's retreat to the claim that unstatable moral principles may be tacitly as opposed to explicitly known is beside the point. Universal principles of the relevant kind may indeed be perfectly explicitly known, despite being defeasible or standing in need of judicious interpretation. In fact, we might say, contrary to the role O'Neill would assign to them as the tacit underpinning of explicit practice, it is more accurate to say that the only role that such rules have is as the explicit but always incomplete articulation of practices which, because in the end never completely specifiable, are themselves tacit collective understandings.'" Pedagogical and Critical Roles for Rules: Apodeictic vs. Deictic Discourse
We formulate and use moral rules of universal form for at least two good reasons: moral pedagogy and moral criticism. Let us turn first to their use background for any principle, but that principles are the wrong objects of moral knowledge in the first place. Nor is this the adoption of a two level theory of the kind advocated by R. Hare (Moral Thinking: Its Methods, Levels and Point, Oxford: Clarendon Press, 1982), as has been suggested to me both by Barry Smith and Kurt Baier in oral comments on this chapter. On such a two-stage moral theory the objects of moral knowledge are rules which approximately codify a set of elusive moral facts to which only our moral intuitions give us access. On my view moral principles are not the objects of moral knowledge at all, but are at best dialectically useful summaries of that which we know. Charles Pigden in oral comments argues that principles have another and more ineliminable role in moral life—that when we don't know how to handle a hard moral case we appeal to principle for guidance. This role, he argues, demonstrates their foundational role. I deny this, though. To be sure, we often do resort to moral principles in our reasoning and disputation concerning difficult moral decisions. But this is only because we use principles in order to draw our attention to relevantly similar cases, and to particular dimensions of similarity, not because we believe that those principles have force of their own independent of examples. That is, we use principles in such cases as a bulwark against arbitrariness not by using them as universals under which to subsume the new case, but rather as guides to other cases with which to pair it in our consideration. I am hence urging a casuistical and not a Kantian account of the role played by moral principles. 41 I should note that the account of moral knowledge and moral reasoning I proffer is very close to the account of casuistry developed and defended in great detail by Jonsen and Toulmin in their excellent volume, The Abuse of Casuistry (Berkeley: University of California Press, 1988). Exploring casuist moral reasoning and the relation of my position to it in detail would take us far afield, and indeed Jonsen and Toulmin do an admirable job already. Here I take myself not so much to be defending a distinct moral theory, but rather providing an epistemological framework within which to understand moral casuistry and to be defending this approach to ethics and to moral epistemology against a new universalist attack. My project, I emphasize, is principally one of moral epistemology, and not primarily an investigation of theoretical metaethics.
200
Jay Garfield
in moral pedagogy. As O'Neill correctly notes, even on a Wittgensteinian account of the use of paradigm cases and their ostension in teaching, these cases must be picked out and characterized as satisfying relevant moral descriptions. And if they are to function as instances of morally relevant types, we must be committed to generalizations ascribing similar moral properties to relevantly similar acts or circumstances. To describe a particular saving of a drowning child from shark-infested waters as courageous is to commit oneself to the claim that all relevantly similar savings are courageous. Any suggestion to the contrary would make the use of such instances and particularism in general incoherent, and would fall prey to O'Neill's charge of moral arbitrariness in an especially acute way. Moreover, in explaining the moral domain to children or to those who come from a very different moral culture, we exploit the power of universal generalization, suitably festooned with ceteris paribus clauses, to describe the contours of our moral landscape, and to explain why particular instances are treated as they are. These explanations advert to their similarity on relevant dimensions to other such instances, which similarities can be approximately and conveniently captured by universal generalizations like 'stealing is wrong' and 'saving drowning children at risk to oneself is courageous'. But we know that these generalizations are not exceptionless, and that they derive their content from the more basic particular facts and similarity relations they capture. That is, while any moral judgements we make may be universalizable, they are typically not universal. 42 In this sense, we can compare these universal descriptions to what Sellars has called 'distributive singular terms' such as 'the lion' in sentences such as ' The lion is dangerous.' 43 Well, not all lions. But the typical lion. As we diverge from typicality our reliance on the generalization becomes weaker. Consider, for example, The dog makes a good housepet.' Depends on the dog (attack-trained Rottweiler or reliable border collie), the house (inner city flat or suburban palace), the cat, etc. This does not undermine the truth of the apparently universal claim, but rather draws attention to its less than universal pretensions. Still, such a claim would be useful in teaching someone about dogs, pets, and their role in Australian culture. Mutatis mutandis for ethical principles. 42 As Skurzynski's Honest Andrew (New York: Harcourt, Brace, Jovanovich, 1980) discovers (see Pritchard, Reasonable Children for a nice discussion) the generalization An otter ought to be honest' is indeed important for a young otter to learn, but it is equally important for a morally serious otter to learn what situations call for what degree of honesty. I thank Tim Sprod for calling this nice example to my attention. 43 W. Sellars, 'Abstract Entities', Review of Metaphysics, 17 (1963), 627-71.
Particularity and Principle
201
We also use moral generalizations in moral criticism. We might, for instance, argue that someone who believes that all human beings are deserving of equal respect, and who sincerely strives to live up to this moral maxim but also believes that it is morally acceptable for a landlord to refuse to rent to Muslims ought to abandon this latter view. And in convincing him of the immorality of his religious bigotry we might appeal to the moral principle we share regarding universal respect—and not simply as an ad hominem device. O'Neill gets this just right. But using moral principles in this way in no way contradicts, and in fact can best be explained by, particularism. For generalizations in this context call our attention to relevant similarities. If they did more than this—if they pretended to true universality—they would fail to have rhetorical force simply in virtue of their falsity. But by showing our once bigoted friend that Muslims are relevantly like himself, like atheists, Christians, and others he deems worthy of leaseholding, we convince him to go on to that particular case on the basis of earlier cases in the same way we do. We teach him to see that case differently; to participate in our moral form of life. That we use a principle to do this in no way undercuts the fact that it is at bottom the concrete moral situation that grounds those generalizations and to which we return in moral consciousness and in our everyday as well as our philosophical moral argument. Heidegger44 usefully distinguishes deictic from apodeictic discourse. 45 A discourse in which we adduce arguments, proffer generalizations, and employ empirical or conceptual investigation is an apodeictic discourse, where truth claims are held hostage to the availability of logical argument or empirical evidence. Such discourse, including appropriate use of moral principles in ethics—such as those we have just been examining—is essential in much of our lives. Its salience, Heidegger argues, often distracts us from the more fundamental discourse which is its ontological and existential precondition—deictic discourse. In deictic discourse—of which for Heidegger poetry and religious discourse are central examples, but which Borgmann points out comprises as well most of our aesthetic discourse and deep discourse about nature and 44 M. H eidegger, Being and Time (San Francisco: Harper, 1962); and On the Way to Language, trs. J. Stambaugh (New York: Harper & Row, 1971). See also A. Borgmann, Technology and the Character of Contemporary Life (Chicago: University of Chicago Press,
1987) for an excellent discussion of this distinction and its importance in value theory. 45 The terms `deictic' and `apodeictic' are Borgmann's ( Technology, 1987). They capture Heidegger's distinction between discourses that point to fundamental values and to horizons and those discourses that characterize phenomena against those horizons, presupposing those backgrounds and shared values.
202
Jay Garfield
fundamental values46—we communicate by ostension values which lie too deep and which are too fundamental for further exposition. If you were to ask me why I so love the Himalayan region where I now write these words, no universal generalization about the capacity of snow mountains, eagles, terraced fields, and broad vistas to stir the heart would do so well as bringing you here and showing you, or as second best, describing in appropriately vivid prose, or showing you pictures. In none of these cases would we dispute the issue of the beauty of this place by offering arguments, or by adducing evidence. The difference in ontological–existential level at which these discourses operate is amply in evidence in environmental disputes. In defending the Franklin River, Mother Cummings Peak, or the Tarkine Wilderness from aesthetically destructive economic exploitation (and here I use 'aesthetic' in a very broad sense to include fundamental values attaching to wilderness) we confront apodeictic arguments in favour of exploitation, defended by appropriate cost/risk benefit analyses balancing tourist dollar benefit streams against logging and pulping benefit streams, etc. We respond to these—sometimes successfully, sometimes unsuccessfully—not with contrary cost/risk benefit analyses, but rather with often compelling and poignant deixes—insistences that we look at what is threatened. Such deixes presuppose that those we address will be thereby moved. The fact that such a discourse is so fundamental to our axiological lives is why Peter Dombrovskis' photographs could have been so influential in the Franklin River debate, and why Ansell Adams' photography saved the Yosemite Valley from a similar potential catastrophe. This Heideggerian distinction provides us with another way of seeing the relation between particularist and universalist understandings of moral knowledge. The particularist emphasizes the deictic side of moral discourse, while the universalist emphasizes its apodeictic side. To be sure, moral discourse, like all discourse, can be carried out at both levels. But like all fundamental discourses about value, the deictic level provides the background presupposed by the apodeictic. Without sharing fundamental values, there are no explicit values to feed into the apodeictic reasoning which takes them as objects. 47 Borgmann, Technology, 1987. This is not to say that deictic discourse is epistemologically foundational, or that the phenomena ostended by such discourse are given independently of the conceptual framework articulated in apodeictic discourse. These two levels and the concepts they mobilize can only be understood as interdependent. Specifying this interdependence takes us far beyond the scope of this chapter. That is the project of Being and Time (Heidegger, 46 47
Particularity and Principle
203
But if apodeictic discourse about values presupposes the possibility of successful deictic discourse (deictic discourse in which our interlocutors respond at least to a considerable degree to our deixes as we intend them to), it is equally clear that the possibility of agreement in deictic discourse presupposes in turn that we share a broad range of values and of ways of seeing—ways of taking up with the world—in virtue of which objects become salient, similarity dimensions are commonly noted, aesthetic and emotional responses are in broad agreement, and so on. That is, we must, in the existential sense, inhabit the same world; we must, at a fundamental level, care about the same things. 48 As Wittgenstein and Heidegger argue, this commonality of form of life, or of life-world, underlies all possibility of agreement, and hence of discourse itself. 49 In coming to share such a form of life we come to share ways of seeing situations, ways of recognizing which similarities are important, ways of responding to our fellows and their circumstances, to temptations, to achievements and to failures, to fear, to art, to suffering, and to happiness. But the cultivation of these traits of character, including, inter alia, the virtues as classically understood, is achieved not through instruction in a set of universal moral principles, but rather, as Aristotle also mentioned, through training—through exposure to particular situations, and through 1962). The important point to emphasize here is simply that moral discourse draws its power and motivational force from the deictic, and not the apodeictic side. 48 Or at least about human beings to the same degree, and this, as Jane Braaten (personal communication) points out, may be the universalist's real concern. 4-9 This shared background has both biological and social roots. Evolution has certainly wired us with certain dispositions to respond to particular stimuli and situations in particular morally relevant ways, and as Santideva, Hume, and more recently Baier (`Doing Without Moral Theory', 1985, and 'Morality and Reflective Practices', 1985), Matthews, The Philosophy of Childhood (Cambridge: Harvard University Press, 1994), and Flanagan, Varieties of Moral Personality, 1991, have emphasized, these biologically grounded dispositions are crucial to our moral psychology and our moral development. All of these theorists join Aristotle and McDowell in emphasizing the crucial role of socialization both in shaping and extending these innate emotional and moral dispositions, but also in cultivating an overlay of entirely socially constructed moral, aesthetic and emotional responses which enable a person to be recognizably one of us—to participate in our moral community and form of life. As Aristotle so aptly puts it, 'That is why how we raise our children makes a great difference—or rather, all the difference' (Nicomachean Ethics, 1103b25). It is undoubtedly the fact that we share so much biology and psychology that accounts for the great overlap in our cultures, despite their vast differences, and undoubtedly this overlap, together with our shared biologically and psychologically grounded sentiments and ways of seeing that accounts for the great agreement—again, despite considerable disagreement as well—in moral judgements, and our even greater capacity to come to agree if we talk long enough. This, I think, is what the commonalism of those like M. Smith's ('Rights, Right Actions and the Cultural Context of Morality', Social Theory and Practice, 5 (1978), 409-15, and 'The Best Intuitionistic Theory Yet', (1999) ), gets right.
204
Jay Garfield
approval, correction, and description of particular actions. In the end, we come to learn our culture's morality, including the canons of its apodeictic moral discourse just as we come to know what a game is, and how to speak our mother tongue without recourse to principles, and only later to articulate its grammar. We easily fall from awareness of the importance and depth of our shared backgrounds of concerns and ways of engagement with the world, and treat all of our knowledge as articulate, principled, declarative knowledge, thereby concealing the fundamental deictic dimension behind the derivative apodeictic. In the same way, we too easily forget how moral principles represent but the summary and the entry into the domain of critical discourse of a complex network of habits, dispositions, and ways of seeing, always rooted in the particular, and cultivated by training. We thereby are led to invert in normative theory the structure of moral knowledge, giving the universal the foundational role and occluding the fact that the particular is what is most essential to human moral life.
9
Against Deriving Particularity Lawrence Blum
Since the mid-1980s, the distance in substantive ethical views between impartialists or adherents of principle-based ethics (generally, Kantians and consequentialists) and partialists or particularists has greatly diminished. There is general agreement that compassion, friendship, generosity, and familial devotion are good things that should play a part in any morally adequate life. Both sides agree that virtue is a central moral idea, and that an adequate account of the moral life can not jettison it. At the same time, both sides seem to agree that concern for the welfare of strangers, commitment to some general principles, and a concern for the general good must play their part as well. Disagreement between the two general camps (which I will refer to for the time being as 'impartialists' and `particularists, bracketing for the moment differences within each rubric) has tended to shift to a theoretical level. Virtuists and adherents of care ethics represent two general approaches to ethics that require particularity and partiality to pose a I would like to thank audiences at a National Endowment for the Humanities Summer Institute on 'Principles and Practices' in 1992, a conference on Virtue Ethics at the University of Santa Clara in 1994, the 2nd Annual Philosophy Colloquium at the University of Utah in 1996, the Philosophy Department of the University of Maine in 1997, John Doris of the University of California at Santa Cruz, and the editors of this volume, for valuable feedback on earlier versions of this article. 1 'Care ethics' has an irreducibly particularist dimension, focused on response to the particularities of particular other persons' situations. (However, there can be both partialist and non-partialist forms of care ethics—i.e. those that accord distinct moral standing to particular personal relationships (the more common form) and those that do not. ) 'Virtue ethics' is theoretically agnostic on the particularism issue—there could be a non-particularist form of virtue ethics—but in general virtue theorists recognize that some virtues (compassion and generosity, for example, as I will argue below) have an irreducibly particularist dimension. Similarly, most virtue theories wish to make room for partialist virtues such as familial devotion and friendship, but it would be possible to imagine one that did not.
206
Lawrence Blum
significant challenge to the theoretical commitments of impartialism. 1 Impartils,ncowhretaipmcyforlitandgeeral principles. Some particularists, such as Jonathan Dancy, accord full theoretical priority to particularism. 2 In this chapter I defend a more moderate position, one that is pluralistic in character. At the psychologically deepest level of the moral life, and of its philosophical expression, lie notions of virtue, particularistic moral response, and partialist moral concern. These notions provide irreducible constraints on the construction of any theories or accounts of the character of morality. Morally, the intuition that we have special moral ties to members of our families which we do not have to others is as secure a moral conviction as that every human being should be treated with dignity. The belief that generosity and compassion, which fall outside a purely principle-based approach and are irreducibly particularist (an argument for this is given on p. 212), are ethical virtues is as secure as the impartialist view that in moral thought everyone should count equally. At the same time particularity is, on my view, no more fundamental than impartiality. There is no unitary 'moral point of view' that can be defined in terms of a single moral notion or procedure.
Three Impartialist Approaches to Deflating Particularity
I am concerned with ways various impartialist accounts of morality have, or could, respond to the challenge posed by particularity (which, from this point on, I will use to mean 'particularity or partiality, except when the context makes it clear that I mean one of them in particular), by muting the theoretical significance of particularity. They attempt to 'deflate' the claims of particularity to theoretical significance at the deepest level of a philosophy of morality. 3 I distinguish three impartialist approaches (which I will sometimes call `projects') to deflating the theoretical significance of particularity. Often Jonathan Dancy, Moral Reasons (Oxford, 1993). Impartialism has also been criticized for failing to accord legitimacy to 'the personal point of view' and, relatedly, for undermining personal integrity or generating alienation, a concern initiated by Bernard Williams' seminal articles 'Person, Character, and Morality', in Moral Luck (Cambridge: CUP, 1981); and 'A Critique of Utilitarianism', in Williams and Smart, Utilitarianism: For and Against (Cambridge: CUP, 1973). My concern is not to validate the personal point of view as such—for example, to defend personal relationships as a form of personal good—but to argue for the moral (or ethical, or quasi-moral) value of devotion to friends and family, compassion, and other particularistic phenomena. 2 3
Against Deriving Particularity
207
particular impartialists adopt more than one of these approaches without drawing clear lines between them. 4 The dismissive approach presupposes a sharp distinction between moral and non-moral value, and then denies moral value to particularist phenomena entirely, though perhaps according them some other kind of value. For example, acts of love, compassion, generosity, or familial loyalty may be seen as good, but not morally good. This approach reflects Kant's famous claim in the Foundations of the Metaphysics of Morals that actions from sympathy have no moral worth. Though it deserves 'praise and encouragement' it is a mere inclination and does not partake of the 'far higher worth' that comes from acting from duty. 5 The derivation approach, by contrast, acknowledges the moral worth of particularistic phenomena, but sees them and their worth as directly derivable from, and thus ultimately reflective of, impartialist notions. While on the surface particularity may seem other than and inconsistent with impartiality, if we look deeper we will find that particularity is really a form of impartiality, rightly understood. The validation approach agrees with the derivation approach in according moral or ethical worth to particularistic phenomena, but, in contrast to it, does not see these phenomena as derivable from impartialist notions. The psychological source of particularist phenomena does not lie in impartialist principle. More significant, the moral value of particularistic phenomena is not derived from—or not fully derived from—(the moral value of) impartiality. What makes the validation approach deflationary is that it claims that the moral or ethical worth of particularity requires legitimating, validating, or authorizing by impartiality. According to validation, compassion and generosity may be virtues, and ones not of a principle-based or impartialist-based character; however, in order to be accorded ethical worth, they (or particular instances of them) must pass the test of permissibility, a test whose character is determined by universal principle or impartiality. Though their full moral character is not exhaustively accounted for by impartiality, still impartiality must give these virtues its 4 To one degree or another, many contemporary theorists partake of the deflationary project. The one most straightforwardly critical of my own work Friendship, Altruism, and Morality (London: Routledge & Kegan Paul, 1980) and 'Gilligan and Kohlberg: Implications for Moral Theory' (Ethics, 98 (1988), 472-91), on these particular grounds is Jonathan Adler, 'Particularity, Gilligan, and the Two-Levels View' (Ethics, 100, October 1989, 149-56). I am indebted to Adler for providing the impetus to write this chapter, as a defence of my views against his criticisms. 5 I. Kant, Foundations of the Metaphysics of Morals (1785), trs. L. W. Beck, (Indianapolis: 1959), 14-15.
208
Lawrence Blum
imprimatur. Commitment to impartialist notions constitutes a necessary condition for the moral legitimacy of particularity.
What is Being Deflated?
Different deflationary projects aim at deflating distinct entities. Their targets may be non-universality, partiality, or particularity. My argument aspires to defend any of these against this attempted deflation. A 'universal' principle is one that is both general and also makes no reference to the agent's previously existing personal relationships. So the principle, `Devote oneself to the welfare of one's child' is general but non-universal; 6 whiletprnc,'Pomhegdfal(inptoherlationship to oneself)' is universal. Some anti-particularists regard a deflationary project as adequate if it retains the former (generality but non-universality) as its fundamental theoretical commitment in terms of which particularity or partiality is deflated. Others would regard this stance as insufficiently impartialist, and would press to eliminate the essential reference to the agent's relationships at the level of moral fundamentals. For example, Alan Gewirth argues that such partialist, generalist principles as 'Devote oneself to the welfare of one's child' are antithetical to impartialism, and tries to show that they can be derived from purely impartialist principles.' `Partiality' is generally understood to mean that it is morally right to give a higher priority in one's actions to the good of those to whom we stand in certain sorts of relationship (friend, teacher, parent) than to those to whom we stand in no relationship, everything else being equal. 'Particularity' is sometimes used to mean partiality (as Gewirth does) but can be given other meanings as well. For example, some use it to mean that the agent's particular moral identity may permissibly be taken into account in deciding on right action, as a reason for departing from a purely universal, or perhaps even general, approach to a situation (i.e. what anyone should do). 6 This use of 'universal' is, I trust, sufficiently familiar. Yet in a sense it is arbitrary. One could regard the principle 'Devote oneself to the welfare of one's child' as universal in the sense that it applies to any agent. Yet it is not universal in an equally legitimate sense of privileging particular relationships over an equal concern for all independent of such pre-existing relationships. I am marking this distinction by calling the principle in question 'general' (applying to anyone who occupies the relevant situation) but 'non-universal' (making essential reference to pre-existing relationships of the agent). In Alan Gewirth, 'Ethical Universalism and Particularity', in Journal of Philosophy, 85 (June 1988), 283-302.
Against Deriving Particularity
209
I use it with a meaning it is often given in care and virtue ethics, and that is that the agent responds to the particulars of a situation without drawing (explicitly or implicitly) on any general (much less universal) principles of an act-determining nature. Thus a parent might in fact be devoted to the welfare of her child without acting from any principle to that effect. She may take no stand on whether others should be similarly devoted to their children, or whether they should perform some particular act of devotion that she performs. But her devotion to her child instantiates the virtue of `devotion to the welfare of one's child. This meaning of 'particularity' will be further elaborated below, but it is in this sense that I aim to defend particularity, partiality, and nonuniversality against whichever of these the impartialist chooses to target. While this may seem to render my argument incomplete, since defences of particularity do not necessarily serve to defend partiality (and the reverse as well), by considering major forms of attempted deflation I hope to make the case that deflation is unlikely to be successful with regard to any of its usual targets.
The Dismissive Approach
The dismissive approach suffers from a fatal arbitrariness. Accepting for the sake of argument the sharp line it draws between the moral and the non-moral, it would be reasonable to inquire what sort of non-moral value acts of compassion, friendship, and generosity are to be assigned. Nonmoral value may be articulated through the categories we possess under that rubric—aesthetic, prudential, intellectual, economic, and the like. The value of these actions is not aesthetic, or at least not primarily so, nor more so than actions reflective of impartiality (for example, just actions, or actions promoting maximal agent-neutral value). Nor is the value involved in these acts merely prudential, aiming at the agent's interest. Nor do the virtues in question fall into the category of intellectual virtues—such as cleverness, ingenuity, or intellectual depth. As long as the line between moral and non-moral is claimed to be sharp, as the dismissive approach requires, it is difficult to see why action from virtues such as friendship, compassion, and generosity should be entirely excluded from the 'moral' category, absent a prior commitment to a conception of morality that requires impartialism (in some particular form, as in Kant's view cited above). Until an argument independent of the dismisser's prior conception of morality is provided to warrant consigning
210
Lawrence Blum
acts of compassion, generosity, and family devotion to a category of nonmoral value, to do so remains arbitrary. Actions or motives informing friendship or familial devotion may not seem paradigmatically moral. Nevertheless, unselfish concern for the friend's or spouse's welfare seems pretheoretically more like a form of moral concern than a definitively non-moral concern. My argument here is not meant positively to affirm the moral significance of the particularistic phenomena. The argument is rather that a view that is itself invested in a sharp and exhaustive moral/non-moral division fails to make a plausible case for putting the particularistic virtues on the non-moral rather than the moral side. 8 In fact the 'dismissive' approach—entirely denying moral significance to particularistic phenomena—has not been much followed by contemporary adherents of the impartialist traditions. By and large their approach has been rather to acknowledge the moral worth and significance of particularity—of virtues of personal relationships, caring and compassion, and the like—but to claim that these phenomena fail to pose a significant challenge to the structure and commitments of impartialist theories. In addition, sometimes the particularistic phenomena are seen as having either less, or a derivative form of, moral value than impartiality and universal principle. That is, impartialists tend to engage in either the 'derivation' or the 'validation' approach, each of which acknowledges the moral or ethical value of particularity but rejects the claim of particularity to constitute a theoretically significant element in morality. I will focus in this chapter on the derivation project, much the more robustly deflationary of the two. Because the validation project concedes that impartiality cannot account for the ethical value of at least some phenomena, it leaves open the possibility that a good deal of what is of moral 8 I myself would follow Bernard Williams' suggestion, in Ethics and the Limits of Philosophy (Cambridge, Mass.: Harvard U. Press, 1985), that in this context we employ a notion of 'ethical' that does not require a sharp separation between ethical and non-ethical,
and that leaves room in 'the ethical' for a plurality of kinds of consideration. Such a broadly Aristotelian conception allows, for example, that honesty, compassion, generosity, and 1 oyalty might fall clearly within the category of 'ethical' (bound up most centrally with important facets of the way we lead our lives in the context of other persons and their lives). It allows that virtues such as tactfulness and resourcefulness may or may not count as 'ethical'; yet even if they do not, they can still be acknowledged as bearing some relationship to the ethical. So the line between ethical and non-ethical is not a sharp one. Williams' conception does not require that each ethical virtue be directly concerned with others' interests, as do the altruistic or benevolent virtues. It leaves room, for example, for virtues that can be put entirely to self-regarding uses, such as courage or temperance. But we will not be able to understand these as virtues without reference to some regard for the lives of other persons.
Against Deriving Particularity
211
significance in our lives stems from sources other than impartiality. This calls into question its claim that impartiality is in any overall sense more fundamental than the ethical notions underlying these other phenomena (whether virtue, particularity, partiality, or whatever), and leaves it with at best a tepid form of deflation. 9
The Derivation Approach
The derivation approach aspires to show that particularity is in some way derivable from, and is thus in its inner character ultimately reflective of, impartiality. To count as a candidate for a derivation, a view must satisfy the following two conditions: (1) Genuine particularity—not merely an appearance of it—must be acknowledged on the level of lived morality, that is, of moral agents' actual motivation, dispositions, and understandings of their own actions. This condition guarantees that the candidate for particularity-deflating acknowledges a genuine and not merely apparent or illusory particularity. (2) The view must attempt to explain the lived particularity in terms of impartialist notion(s). The impartialist concept must be proffered as explaining the character and value of particularity. Failing this condition, the view will not have attempted a derivation of particularity from impartiality. An instance of a view which fails to satisfy these two conditions is one that claims that agents who appear to be acting from particularistic motives and virtues are in fact, whether they consciously recognize this or not, operating from universal moral principles. If pressed to scrutinize their motives more deeply, these agents would be compelled to acknowledge that, indeed, they were drawing on some universal principle. An example illustrating this claim is the following generous act: Sarah, upon discovering that a friend is particularly fond of a kind of shirt that she herself owns (but the friend does not know this), and of which she is fond but not deeply attached, gives the shirt to the friend simply because she wants her to have it. Sarah takes herself to be motivated simply by her friend's fondness for the shirt, and (rightly) does not regard herself as obliged to do so. This motivation coupled with various other conditions 9 The argument that the 'validation project' does not really deflate particularity, is made in my 'Gilligan and Kohlberg: Implications for Moral Theory', (see fn. 4). That article has been rewritten for my Moral Perception and Particularity(New York: CUP, 1994).
212
Lawrence Blum
(e.g. that Sarah in fact has no obligation to provide the shirt, that the friend does not have a dire need for the shirt) renders her act a generous one.'0 The purported deflationary view being considered here would claim that despite what Sarah takes her motivation to be (a desire to satisfy her friend's wish for the shirt) she is actually operating from some general and universal principle—perhaps to the effect that whenever one possesses something of a certain degree of value and learns of a friend's desire for that object (a desire indicating that the friend would get more pleasure out of it than one would oneself), then one should give the item to the friend. The view goes on to claim that whenever a moral agent apparently acts from a motive of an apparently particularistic nature, such as non-principle-based compassion for a particular person or concern for a family member, in fact that person is drawing on a universal principle which specifies and prescribes that very same act. Even if we grant the implausible claim that a general principle could be located that prescribed (for example) every generous act, its existence would not be sufficient to show that the agent is actually acting from that principle when she engages in her generous act. While not all moral principles from which an agent operates need be explicitly consulted on each occasion—principles may be so deeply ingrained that their motivational force becomes habitual--there remains a difference between an agent's actually possessing the principle as part of her implicit and explicit motivational structure, and such a principle's being formulable but not playing any role in a given agent's psychology. (That psychology contains not only direct motives but conditions under which motives operate.) Only a purely theoretically driven commitment to the ubiquity of moral principles provides grounds for asserting affirmatively that on every occasion on which an agent does not take herself to be acting from such principles (but simply to be acting from an inclination to give her friend pleasure in this particular situation), she is, nevertheless, doing so." So the view in question is not an instance of a deflationary derivation project at all. It does not attempt to deflate particularity at the lived level by 10 I draw this account of generosity from James Wallace in Virtues and Vices (Ithaca, NY: Cornell U. Press, 1978), 131-36. Wallace calls this form of generosity, in which goods and services are given to the other 'economic generosity', to distinguish it from other forms, such as being generous in evaluating someone. 11 The principles in question must be sufficiently contentful to count. For example, the injunction 'be generous' does not provide sufficient guidance to action to count as the sort of principle a principle-based account needs to constitute a genuine alternative to a virtue or particularity-based view. If it did, every virtue-based action would be automatically principle-based, since it would instantiate the principle 'be virtuous'.
Against Deriving Particularity
213
deriving it from something else, but rather to deny it altogether. It says there are no genuinely particularistic motives; all apparent ones are really instances of general principles. Thus, this approach does not retain particularity at the lived level. It violates condition (1).' 2 If the purported deflator retreats to the mere existence of a universal principle prescribing the generous act (leaving aside the implausibility of that postulation), she violates condition (2). This view fails even to attempt to explain lived particularity by reference to an impartialist concept. It does not attempt to deflate particularity but only to show that alternative ethical approaches would generate the same actions. Note that even if, as both these views postulate, the action prompted by the particularistic motive were in some meaningful behavioural sense the same as that prompted by a universal principle, it would not follow that the ethical description of the action—hence the value we attach to it—would be the same in the two cases. To take our earlier example, if Sarah gives her friend a shirt in the generous manner described earlier, this action is valued as a 'generous act. The same behaviour of giving the friend a shirt is not necessarily to be accounted a generous act if Sarah acts on the basis of a universal principle prescribing the act. The latter action could be valued, as, say, a conscientious act or a dutiful act; but neither of these is the same as generosity. That which explains the value and character of a dutiful or conscientious act would not thereby explain the value and character of a generous act.
Two-level Views
I want now to consider a family of views familiar in moral theory that (in contrast to those above) do satisfy conditions (1) and (2), hence are genuine candidates for derivation projects. These approaches involve a 'twolevel' structure, perhaps most famously articulated by R. M. Hare but applicable to many other theories. While not all two-level views are of one type, nor do all serve the same purposes within moral theory, many of them are explicitly or implicitly deflationary of particularity. A feature 2 The position taken by the purported deflation is actually a version of a position not catalogued in my original description of options for impartialist theories. It differs from the dismissive view in that the latter acknowledges particularity as a distinct psychic motivational phenomenon, denying only that it has moral value; the view being considered here denies particularity as a distinct psychic phenomenon itself. This move can be found in Jonathan Adler, 'Particularity, Gilligan'.
214
Lawrence Blum
common to most of them is a distinction between a level of lived, intuitive, or commonsense morality, and another, deeper, more normatively or philosophically substantial level taken as defining the moral point of view and expressed in terms of the concepts the theory takes as characterizing morality in its fundamental nature.13 Two-level deflationary theorists argue that moral notions at the commonsense or lived level need not on their face reflect the notions characterizing the normatively sounder level; in fact the point of the two-level theory is that the two levels have distinct psychological and apparent ethical characters. Nevertheless, the ethical legitimacy of the intuitive or lived notions stems ultimately from their being derived from, and thus reflecting, notions of a different character at the normatively sounder level. Regarding particularity, such two-level views acknowledge that we sometimes appropriately act from motives of a particularist or partialist nature, such as (to take a frequently mentioned example) the preference accorded members of our own families when we are in a position to benefit them but are equally in a position to benefit others. A further elaboration of this example postulates a sentiment—such as love for one's family—that underlies that actional preference. The two-level theorist approves of the sentiment and the preference, yet sees both as apparently inconsistent with maximization of agent-neutral value or whatever impartialist notion the theorist takes as characterizing the normatively sounder level. For a direct impartialism would allow no special concern for the good of one's family over the good of strangers. Thus, the two-level theorist in question acknowledges an apparent conflict between these two commitments (in this case, partialist and impar13 Some examples of forms of two-level-ness that do not, or cannot, coherently aspire to deflating particularity are the following (not all of these permit the two levels to be accurately characterized as 'lived' and `deeper/normatively sounder'): (1) The lived level is spontaneous and unreflective action; the deeper level involves action based on reflection. (This is not deflationary because particularity can be either reflective or unreflective.) (2) The lower level comprises the kind of thinking required for decision-making when one must act fairly quickly; the other level comprises moral reflection that need not be tied to specific decisions, and that is not constrained by time. (Particularity is no more centrally tied to quick decision making than is impartiality.) (3) The lower level consists in simple, coarsegrained moral precepts; the higher level involves more refined and accurate principles, required for situations of conflict between lower-level principles. (Particularity is not generally principle-based, and so may fit into neither of these categories.) (4) The lived level consists in accurate principles; the deeper level is a fuller account of why principle-based action is right than that supplied merely by principle. (Again, particularity does not generally involve principles, nor need an account of the value of particularity deflate it.) (Options (3) and (4) are insightfully discussed in T. M. Scanlon, 'Levels of Moral Thinking', in D. Seanor and N. Fotion, Hare and Critics: Essays on Moral Thinking (OUP, 1990).)
Against Deriving Particularity
215
tialist). She attempts a reconciliation between the two by claiming that the maximization of agent-neutral value would best be promoted precisely by dispositions of this partialist character. Although on some particular occasions acting from such dispositions involves acting contrary to the promotion of this end, in the long run these dispositions best realize the deeper-level impartialist commitments of the theory. The deflationary character of these two-level theories may not be immediately apparent. They are often framed as attempts to preserve space and legitimacy for particularity and partiality that would be excluded by interpretations of impartialism requiring the agent to act directly from explicit impartialist principles on all occasions. Nevertheless, two-level theories regard the lived particularity as ethically acceptable and legitimate only in so far as it can be derived from the deeper level impartiality. Accommodation at the lived level is acquired at the price of deflation at the philosophical/theoretical level. This deflationary aspect is in fact part of the implicit point of most (though not all) of the most widely and currently influential forms of two-level theories. It is to ensure that particularity (in its various manifestations) not threaten the theoretical primacy reserved for the impartialist notions taken as defining morality. There are several distinct types of two-level deflationary views. Perhaps the most familiar is consequentialism, in which the deeper level is characterized in terms of promoting good consequences overall and with indifference to particular persons, each of whom is regarded as equally worthy of having her good promoted. I will consider this view in detail in a moment, but want to mention, for the sake of contrast and to indicate the breadth of two-level particularity-deflating views, three other types. One is T. M. Scanlon's contractualism, in which the deeper level is characterized in terms of the standpoint of principles that reasonable persons could not reasonably reject." A second is dialogical, as in Jurgen Habermas's view, in which the deeper level is characterized in terms of a discussion, subject to certain constraints, among persons seeking to resolve differences in interest and moral outlook.15 A third, represented by Alan Gewirth's 'Ethical Universalism and Particularism', 1 6 casts the deeper level as a specific moral principle—the right to freedom and wellbeing—and 4
T. M. Scanlon, 'Contractualism and Utilitarianism', in A. Sen and B. Williams (eds.),
Utilitarianism and Beyond (Cambridge: CUP, 1982), and What We Owe to Each Other
(Cambridge: Harvard University Press, 1998). 5 See, e.g., J. Habermas, Moral Consciousness and Communicative Action (Cambridge, Mass.: Harvard University Press, 1991). 6 Gewirth, 'Ethical Universalism and Particularity'.
216
Lawrence Blum
claims that particularistic principles can be derived from that nonparticularist principle.
Conditions for a Successfully Deflationary Derivation Approach
To be successfully deflationary, a derivation approach must satisfy two further conditions beyond the two considerations earlier: (1) acknowledgment of genuine particularity on the lived level; and (2), attempting to account for the character and value of the particularity on the deeper or normatively/philosophically sounder level. (3) The characterization of the normatively sounder level must be distinctly impartialist, expressed in terms of concepts genuinely incompatible with particularity at that same level. While the relevant two-level views aspire to meet this condition, not everything that conforms to the specifications for two-level views does so; for not every concept characterizing the deeper, basic level of moral concepts is non-particularist. Take the view that virtue (rather than impartiality, maximizing agent-neutral value, or universal principle) is the fundamental concept of ethics, and that ethical phenomena at the lived level—for example, specific virtuous acts—have ethical value precisely and only in so far as they instantiate virtue. Such a theory would not be deflationary of particularity, since virtue allows for particularity. Yet its theoretical scaffolding would still be in the form of a `two-level' view. More generally, the mere fact that a view proffers an account, or theory, of morality, does not render that account necessarily impartialist or non-particularist.' 7 Furthermore, not every two-level view need characterize the deeper level in terms of a unitary conception of the moral point of view such as characterizes all the two-level views so far mentioned. Rather the deepest level could consist of a plurality of mutually irreducible ethical notions all taken to be basic, but none more basic than the others. There might, for exam17 By speaking of 'account or theory', I mean to avoid the current controversy about the place of theory in ethics. I am accepting the idea that philosophy needs to be able to provide an account of why the moral phenomenon in question on the lived level (compassion, honesty, justice, filial loyalty) possesses the kind of value it does. Sometimes this account is spoken of as 'justifying' the lived level phenomenon. I am trying to avoid using this term as I fear that it implicitly tilts toward impartialist conceptions of the philosophical account. What are taken as requiring justification are particularist or partialist phenomena rather than impartialist ones. If 'justification' can shed this connotation, I do not object to it. I do not want to rest content with purely particular intuitionist judgements of the moral worth of specific actions, but seek a more general level that accounts for this worth.
Against Deriving Particularity
217
ple, be both impartialist and virtuist or particularist notions at the deeper level. Views that, at the deeper level, contain concepts themselves reflecting particularity—either by themselves or as one of a set of mutually irreducible basic ethical notions—do not satisfy condition (3), and would not be successful as deflation projects. The deflating of particularity can be accomplished only if the impartialist (or other particularity-inconsistent) conceptions are the sole occupants of the philosophically sounder level, thus excluding particularistic notions from being fundamental to the philosophical account of morality. Only then will they have undercut the claims of particularity to characterize morality at the deepest and most fundamental level. (4) The final condition is that the impartiality, universality, and the like, on the deeper level must succeed in fully accounting for the particularity on the lived level. That is, it must account for the character of the particularity; and it must account for the type of value that the particularistic phenomena is pretheoretically taken to possess. (Remember that the derivation approach does not challenge the pretheoretic moral worth of the particularistic phenomena as does the dismissive approach.) The impartiality must provide a form of explanation of the particularity that fully reveals its (form of) value.18 The deeper level explains why the lived level phenomenon is good. For example, if the derivation accounts for an act of generosity or familial love as morally permissible, but accords and accounts for no value beyond this, it will have failed condition (4); for generosity is pretheoretically valued (ceteris paribus) as a positive moral good, not merely morally permissible, such as is, say, drinking tea. Condition (4) distinguishes the derivation project from the validation project, which does not aspire to give a full account of particularity's form of value but only to ensure the phenomenon's consistency with impartiality. For example, for validation to be successful on its own terms, it would be sufficient that an act of generosity be consistent with certain limiting conditions set by universalizable principles of permissibility, not that this consistency account for the (full) moral value of the generosity. Hare's Indirect Consequentialism as a Derivation Approach
Let me now turn to the best-known version of a particularity-deflating two-level theory: consequentialism. Its deflationary aspect is more an ' s
I am grateful to Bernard Williams for help in formulating this condition.
218
Lawrence Blum
upshot than an explicit aim of consequentialists' discussions of particularity, which are set in a framework of defending consequentialism against certain objections that involve particularistic phenomena. For example, R. M. Hare responds to the charge that because of their non-impartialist nature the moral validity of loyalties to members of one's family constitutes an objection to utilitarianism. 19 Hare counters that such loyalties are not contrary to utilitarianism, but the upshot of his argument is that family loyalties are morally valuable only in so far as they are derivable from utilitarianism. They have no moral standing independent of their serving an impartial concern for the welfare of all. So the legitimacy of partiality is acknowledged only on the lived level; its aspirations to a fundamental place in ethics are rebuffed. Hare's version of consequentialism (in his case, utilitarianism) exemplifies a deflationary approach in its pure form. Some earlier utilitarians (such as William Morris) thought that acting from familial love or loyalty was contrary to the demands of utilitarianism, since doing so often requires promoting the good of a less needy person (the member of one's family) rather than a more needy stranger. This view also seems an implication of act-utilitarianism in its classic form—that a moral agent should, in her every action, aim at the maximal agent-neutral good (or happiness), that is, the good of persons independent of their relationship to the agent. In Moral Thinking, Hare, joining the trend among contemporary utilitarians and consequentialists, decisively rejects this understanding of utilitarianism. He says it is a good thing that people (in general) have emotions that lead them to give preferential attention and aid to those closest to them emotionally, such as friends and family members. He crafts a version of consequentialism that validates these partialist phenomena. At the same time, Hare equally decisively retains maximization of agentneutral good as the ultimate moral criterion for assessing the moral worth of behaviour. In doing so, he puts forth a version of what has come to be called 'indirect consequentialism'. This is the view that individual actions are not the only, or even the primary, subject of consequentialist assessment; for example, general dispositions to action are to be assessed consequentialistically. It follows that moral agents ought not necessarily directly aim at maximizing agent-neutral value in each distinct action. Using the example of a parent's preferential concern for her children, Hare reasons as follows: If a parent possessed a sentiment of caring for all children equally, this would have the effect of so diluting the sense of 19
R. M. Hare, Moral Thinking: Its Levels, Method, and Point (Oxford: OUP, 1981), 135 f.
Against Deriving Particularity
219
responsibility of parents toward children that all children would be cared for less well than if each parent possessed a preference for her own children.20 Hence, Hare appears to reason, in the long run the currently familiar forms of agent-relative preference for one's own children serves maximal agent-neutral good better than aiming at that goal directly, even if in some particular case it would be consequentialistically better to attend to the needs of children other than one's own. 2 1 Other indirect consequentialists see a problem in Hare's view that leads them to seek a different form of that doctrine. It is that for Hare familial preference, love, and loyalty do not possess value in their own right but only in so far as they serve the external, agent-neutral end of (for example) maximizing all children's welfare. Their value is purely instrumental. For Hare it is a deficiency of human nature that we require ourselves to be inculcated with familial preferences in order to stave off a weakening of responsibility toward children. Familial love plugs up this responsibility gap, as a means to overall consequentialist good. It is not a human good in its own right. The form of indirect consequentialism I wish to consider retains our ordinary intuition that familial and philial relationships, and the sentiments and dispositions they require, have some worth in their own right independent of their instrumental value, and that Hare's view is deficient in denying this.
Railton's Indirect Consequentialism as a Derivation Project
In 'Alienation, Consequentialism, and the Demands of Morality' Peter Railton develops a form of indirect consequentialism (which he calls `sophisticated consequentialism') that affirms the intrinsic worth of personal relationships yet aspires to deflate particularistic sentiments and motives.22 Railton says 'We must recognize that loving relationships, friendships, group loyalties, and spontaneous actions are among the most Hare, Moral Thinking, 137. Consequentialists disagree whether to retain an element of act-consequentialism in the view that we should continue to call 'right' those actions which in fact promote the best consequences, as Sidgwick proposed; or whether to call 'right' those actions which would be performed by an agent whose set of dispositions was good-maximizing in the long run, even if in some particular case that act was not good-maximizing. This dispute does not bear on the aspects of consequentialism relevant to the deflation of particularity. 22 P. Railton, 'Alienation, Consequentialism, and the Demands of Morality', in Samuel Scheffler (ed.), Consequentialism and Its Critics (Oxford: 1988), 93-133. 20 21
Lawrence Blum
220
important contributors to whatever it is that makes life worthwhile; any moral theory deserving serious consideration must itself give them serious attention: 23 While Hare sees the dispositions involved in personal relationships as validated by utilitarianism because their presence is instrumental to agent-neutral good, Railton sees these agent-relative goods as themselves components of the good that is to be maximized. That overall good is understood pluralistically, to include a range of intrinsic and mutually irreducible non-moral goods. (Knowledge, solidarity, autonomy are instances beyond those enumerated above.) Railton takes his view to be deflationary in that the value of particularity—his dominant example is of personal love, especially for one's spouse—must be affirmed from some standpoint 'more general' than that of an individual agent simply conferring value on his loved one, or on his love for him or her. Railton sees Bernard Williams as having denied this point, and he may be correct about this (at least prior to Williams's Ethics and the Limits of Philosophy). 'She's my wife, Railton says, is not a sufficient warrant for conferring preferential benefit on her. Railton's view is only successfully deflationary, however, if the more general standpoint that confers value on particularistic phenomena is itself anti-particularistic in character. Railton appears to assume that this is necessarily the case—that any general standpoint must be impartialist (and, in particular, consequentialist). As noted earlier, however, not all general standpoints are either impartialist or exclusively impartialist. For example, a `virtuist' standpoint is general yet non-impartialist. Of course, for Railton as a consequentialist, the more general point of view—the deeper level articulating the fundamental character of morality—is consequentialist, hence impartialist. Examining consequentialism as a deflationary project involves looking at consequentialism in a different light than that familiar in current debates. The usual disputes focus on whether consequentialism is consistent with certain moral intuitions that at first glance seem inconsistent with it, regarding dispositions to engage in certain acts; or whether, if it cannot be made consistent with those intuitions, it provides sufficient reason for abandoning them. An oft-debated question, for example, is whether the disposition to show preference regarding beneficence toward family members or friends as opposed to perhaps needier strangers is consistent with consequentialism as an impartialist theory. Showing that maximizing agent-neutral value prescribes certain partialist acts or dispositions does not, however, suffice for deflating particularity. 23
Railton, 'Alienation', 98.
Against Deriving Particularity
221
This is so for two reasons. First, particularistic phenomena such as generosity, compassion, unselfish love of one's friend, and familial devotion involve more than actions and dispositions. The specific sort of value they have depends on further emotional reactions, forms of attentiveness and personal vulnerability, as well as other conditions inherent in these virtues and sentiments; that value is not reducible to the value of the acts and dispositions that may also be a component of these phenomena. For example, consequentialism will not have accounted for generosity by showing that a disposition to benefit others in certain circumstances is generated by consequentialism. For generosity is more than such a disposition; it requires that the item that is being given to the other person be of value to the agent (and that she recognize this), and that the action lies outside what is morally prescribed (and that the agent recognize this). 24 It requires that the agent be motivated purely or dominantly by a direct concern for the other, or a desire to give her pleasure. All these features contribute to the specific value that we pretheoretically confer on generosity, a value not reducible to the mere disposition to benefit. We value generosity the way we do because it lies outside (and often beyond) the structure of generally accepted obligation; and also because (e.g. in contrast to a kind of thoughtfulness that sees that an acquaintance would like to have an item one possesses that is no longer of any value to oneself) it involves giving up something valuable to oneself. Unless these other morally significant features are built into what it is that consequentialism is taken to prescribe, it will not have prescribed generosity. An entirely different point—which would survive finding some way for consequentialism to prescribe actual generosity (and other particularist virtues in their complex psychic and moral structure)—is that showing that consequentialism can generate, prescribe, or approve of certain virtues does not yet offer—at least not explicitly—a view of the source of value of those dispositions. To meet condition (4), a deflationary project must aspire to, and succeed in, accounting for the source of value of the particularistic phenomena being explained. If the view merely says, 'These particularistic phenomena are valuable, and consequentialism approves of them' then it remains in the domain of a validation rather than a derivation project. Later I will consider the significance of this possibility. First, however, I want to explore the resources Railton's view provides in the way of a deflationary project—that is, that it proffer an account of the pretheoretic value of goods of a particularistic character. I will focus 24
See my earlier argument about generosity, p. 211 f.
222
Lawrence Blum
particularly on two examples. One is generosity, as discussed above. The other is familial devotion, as portrayed in the following example: Manuel is an exceptionally devoted parent, one manifestation of this being that he is acutely attentive to his children. He makes exceptional efforts to understand them and their needs. He gives his children's welfare an important place in his life priorities, and is frequently willing to jettison his own personal plans of the moment to attend to their needs. In addition, let us assume, Manuel gives no more, though no less, than average attention to the ways in which he might be able to help other people's children. What does Railton's view provide as an account of why generosity and familial devotion are good? One candidate might be that these are simply among the non-morally good things in life, and are thus to be maximized in action. Railton implies this in the quote earlier where he lists some of the things (including 'loving relationships') that make life worthwhile. This interpretation is clearly unsatisfactory as a derivation of particularity, hence inadequate as a deflationary project. For it gives no account of the character or value of these goods in impartialist terms; it simply says that they are goods, leaving the question of why entirely open. This option therefore violates condition (2) that it proffer an explanation of particularity in terms of impartiality—hence (4) as well—that the explanation be satisfactory. That these goods are intrinsic is perfectly consistent with their having an irreducibly particularistic character, rather than being ultimately, or at a deeper level, impartialist. (Indeed, this is what Railton sometimes implies that he believes.) Beyond this difficulty, the interpretation in question requires Railton to construe family love, friendship, and group loyalty as non-moral goods, in contrast with the good of right action, or of the production of maximal agent-neutral value, which is a moral good. Consequentialism requires this particular structure—what is moral maximally generates non-moral good. Morality, then, is a value applicable to agency, while other non-moral values attach to states of affairs and other non-agentic phenomena. However, given this particular division between moral and non-moral goods, the goods of generosity and familial devotion cannot be consigned to the category of the non-moral. This is particularly obvious in the case of generosity, which is generally pretheoretically regarded as a moral virtue. It may seem less so for familial devotion. The good of family relationships could be seen as a personal good to the parties to it, but not a moral good in one of those parties. I think it is both. The good of family relationships can be realized only as the good Railton intends in his placing it on the list of one of most —
Against Deriving Particularity
223
worthwhile things in life if the family members possess certain virtues that are partially, but essentially, constitutive of those goods. One of those virtues is, as Manuel exemplifies, (parental) familial devotion (which, in any case, is the particularist phenomenon in question, rather than familial relationships). Without Manuel's attentiveness, understanding of his children, and willingness to give their welfare a high priority in his life plan, the full good of familial relationships will not be realized by the members of his family (including, at least to some extent, himself). So familial devotion is a moral or moral-like virtue. In any case it can not be consigned to the category of purely non-moral good. 25 It is also true that, as a personal good, the good of family relationships can be treated as a state of affairs to be sought in action. However, part of what it will take to accomplish this goal will be the realization of moral or moral-like virtues on the family member's part. Unless the family member comes to love the members of his family for their own sake, and not for the sake of securing a good for himself, he will not secure that good for himself. (Railton is perfectly aware of this complexity.) 26 So the personal good to be sought requires a moral good, contrary to the supposition of this interpretation of Railton's view as deflationary. A second interpretation of a Railtonian account of the value of particularistic acts and virtues shifts the locus of value from the to-be-realized non-moral good to the agentic process of realizing that value. On this account, what makes Manuel's actions admirable is that they help to realize the good of personal or familial relationships. Here the form of value is a value of agency, and that value is construed as lying in the production of the personal goods discussed a moment ago. But this account does not in fact capture the kind of value we actually attach to Manuel's devotion to his family, or to generosity. We value generosity not because the agent strives to bring about the good of generosity in the world but because generosity reflects a certain kind of moral relatedness to and concern for other persons. It is not even clear that an act motivated by a desire to bring generosity into existence would be a generous act. 25 One reason family relationships may be thought not to be ethical goods is that they are conceived of solely as goods of good fortune. One is fortunate if one has the goods of family love in one's life, but this is a matter of luck rather than ethical effort or attention. This view is only a partial truth. It may be a matter of some degree of luck whether one ends up forming a family with people one is able to love and be loved by in ways necessary for the good of familial love to be fully realized. However, actually loving members of one's family in the way exemplified by Manuel is a matter of ethical attention and effort. 26 Railton, 'Alienation', 110.
224
Lawrence Blum
The type of value in the familial case differs from that of a person striving to bring about the good of familial relationships, in her own life or the life of others. For we admire Manuel, not because he seeks the goal of realizing the good of familial relationships, but because he seeks the welfare of his particular children in a way that instantiates various virtues (even though possessing those virtues is in fact one of the best ways to realize those goods). If Manuel's dominant motivation in attending to his children were the desire to bring about a family life that is deeply satisfying to himself, we would probably admire him at least somewhat—and perhaps a good deal—less than if he seeks to realize his children's welfare simply out of a concern for them for their own sake. It might be replied that the value of the promotion of the good of familial relationships should not be seen as lying in something the agent engages in. It should rather be taken as providing a reason for a consequentialist to promote in others, and perhaps in herself as well, the psychological structures constitutive of familial devotion. This interpretation, however, compromises the sense of agency attributable to Manuel. Railton may not think this troubling, if his consequentialism is of the type that severs the production of maximal good from the individual human agent. However, independent of this concern, this move prescinds from even proffering an account of the source and character of the value of Manuel's parental devotion. It would thus preclude Railton's view (on this interpretation) from being a candidate for a deflationary project. There is a third possible interpretation of Railton's view of the good of particularity. That is that what makes Manuel's action good—like what makes anything connected with agency good or worthy of approval—is that it aims to realize, or contributes to realizing, the overall good that reflects the consequentialist commitment to impartial concern for everyone's interests. This contrasts with the account immediately prior in which the good the production of which is held to be the value in question is the specific category of good (in our case, loving family relationships) that is most intimately tied to the particularistic virtue in question (family devotion). In this third interpretation, the agent's goal must be overall good, not merely one specific type of good. Here the agent must in some way be aware of the role of the specific good in the panoply of all goods constituting overall good, and must be promoting this specific good because and in so far as it occupies the place it does in that overall good. This view does no better in accounting for the form of valuing we actually see in Manuel's actions, and in generosity. The argument for this could equally well be given against Hare, despite Railton's deliberate departure
Against Deriving Particularity
225
from Hare in according intrinsic value to friendship and familial relationships, while Hare accords them only instrumental value. Manuel's actions are good not because they contribute to overall good. They are, rather, good because they manifest certain virtues involved in familial devotion. In a sense this third interpretation of Railton takes the inadequacy of the second one step even further in the wrong direction—away from the focus on the particular persons whose welfare is served by the virtue of family devotion, toward an even greater level of abstraction and distance from those persons. Similarly with generosity. Generosity is not the specific moral virtue that it is, with its particular kind of value, because it involves striving to maximize agent-neutral value. It does not typically have such an aim, and would often be inconsistent with it. At the same time this third interpretation of Railton at least retains the virtue of bringing the consequentialist character of the theory into closer contact with a conception of moral agency.
Does Consequentialism Aspire to Derive and Deflate Particularity?
Thus, in sum, Railton's theory fails to provide the resources for a satisfactory account of the form of value involved in particularistic virtues and actions. Hence it violates condition (4)—providing an adequate account of the value of the particularity—and so fails as a derivation project. As suggested earlier (p. 218), my argument is perhaps not appropriately construed as a criticism of Railton. Perhaps he should be taken not as attempting to account for the value of moral phenomena on the lived level, but simply as showing how the consequentialist moral notions at the deeper level can prescribe these phenomena (e.g. generosity, or family devotion), or at least prescribe the conative dispositions involved in them. Impartialist consequentialism would, then, be declaring these phenomena good in some way, if not in the way that aspires to capture their pretheoretic value. On this interpretation, Railton's consequentialism would have backed away from a goal it often appears to aspire to—to give an overall philosophical account of the character of the moral or ethical life. It would instead be aspiring to an account of only one part of that life, leaving other sources of ethical value uncommented upon. Such a view would perhaps usefully be seen as a validation project, rather than a derivation one. Like the neo-Kantian version of the latter it would have provided a moral standard for the approval and disapproval of purportedly moral phenomena. In at least one version of the Kantian case,
226
Lawrence Blum
the standard is a limiting condition grounded in the idea of universalizable principle. In the consequentialist case it would be compatibility with the production of maximum agent-neutral value. In neither case would the view aspire to account for the full ethical value of the phenomena approved of. There would be an important distinction between the two cases, however, which lends a greater credibility to the neo-Kantian validation project. The neo-Kantian project presents itself as supplying limiting conditions on phenomena (or maxims) whose source lies entirely outside neo-Kantianism itself. We get the proposed maxims from our inclinations, desires, and other ethical and non-ethical sources. But before carrying them out, we check them against the limiting condition. If they do not pass, we refrain from acting on them. Consequentialism has much greater difficulty in allowing sources of ethically acceptable disposition that lie entirely outside of its own edifice. For if a disposition is to be consistent with promoting maximal agent-neutral value, then it must actually promote agent-neutral value. In this sense, consequentialism raises the bar quite a good distance beyond where the neoKantian puts it. It is, ultimately, difficult to see how consequentialism can stop short of saying that the approved of dispositions are in an important sense actually generated by the fundamental principle of consequentialism itself. This will be the only way of ensuring that a disposition actually meets that standard. In conclusion, then, I have argued that some influential consequentialist forms of deflation of particularity fail. I hope to have suggested that particularity and partiality, as manifested in various virtues, are no less fundamental features of the moral life than are impartiality and universal principle—and that whatever theory or philosophical account of morality we finally rest with must reflect this truth. I have suggested that this account will be pluralistic, with several irreducible ethical notions playing fundamental roles. The philosophical account thus provides for recognizing multiple moral concerns, including those of a non-particularist and non-partialist nature. It therefore bids a moral agent reflect on her aims and dispositions. It does not promote the conception of particularity as mindlessly following one's emotions and inclinations and intuitions of the moment. At the same time it recognizes, as none of the impartialist projects discussed here do, that various sorts of particularistic and partialist motives, sentiments, and perceptions are a source of moral or ethical value distinct from that provided by impartialist theories, and are no less fundamental to the moral life and to theories thereof.
10 Why Practice needs Ethical Theory: Particularism, Principle, and Bad Behaviour Martha Nussbaum
Innocence is indeed a glorious thing; but, unfortunately, it does not keep very well and is easily led astray. Consequently, even wisdom— which consists more in doing and not doing than in knowing—needs science, not in order to learn from it, but in order that wisdom's precepts may gain acceptance and permanence . . . Thus is ordinary human reason forced to go outside its sphere and take a step into the field of practical philosophy, not by any need for speculation (which never befalls such reason so long as it is content to be mere sound reason) but on practical grounds themselves ... Thus when ordinary practical reason cultivates itself, there imperceptibly arises in it a dialectic which compels it to seek help in philosophy. Immanuel Kant'
Enemies of Ethical Theory
Ethical theory is under attack. In itself that is nothing new. Attacks on ethical theory began, in the western tradition, 2 with the subject itself, which I am very grateful to Steve Burton for the occasion to develop these ideas, to Scott Brewer, Ken Cress, Dan Kahan, David Luban, and Catherine Wells for insightful comments at the Holmes conference, and to Eric Brown, Tracey Meares, Richard Posner, Eric Schliesser, and Cass Sunstein for valuable comments on an earlier draft. I am especially grateful to Bernard Williams for lengthy comments, to many of which I have tried to do justice in revising. I am sure that I have not answered all of his questions. I am citing from the Ellington translation, and he translates the title as Grounding, but for reasons of familiarity I shall use the more conventional rendering throughout, since both seem acceptable. Translations from Latin are mine throughout. (Kant, Ethical Philosophy (1785), trs. James Ellington (Indianapolis: Hackett, 1983).) 2 I don't mean that these things didn't happen in other traditions: Confucius, for example, was also an embattled figure. But I will be focusing on the western tradition here.
228
Martha Nussbaum
alarmed people who saw advantages in the unexamined life. No sooner did Socrates gain a following than he was indicted, convicted, and killed. Aristotle, fleeing a second time into exile, said that he did not want the Athenians 'to sin twice against philosophy'. 3 The emperor Nero knew Stoic moral theory too well to be content with its defence of liberty. After dispatching his mother, he turned, in AD 69 to his philosophical mentor Seneca, who patterned his mandatory suicide closely on the death of Socrates. 'Even in his last moments his eloquence did not fail him. He called scribes in and dictated a good deal, which, since it is published in his collected works, I shall not bother to adapt. '4 Marcus Aurelius philosophized with impunity until death, but then he was the Emperor. Other thinkers under the Empire were less fortunate. The fifth century saw the death of the eminent neoplatonist philosopher Hypatia at the hands of a Christian mob in Alexandria, incited by a local bishop, who said it was unchaste for women to argue in public. They dragged her from her litter and beat her to death with sticks; the bishop became St Cyril. Some time thereafter, again under Christian influence, the schools of philosophy at Athens were closed entirely. The Middle Ages greeted philosophical theorizing with much scepticism; it was not until Thomas Aquinas that the subject established itself in Church-dominated universities. Renaissance humanists revived Greek ethical theorizing at great personal and political risk. In the seventeenth century Grotius advanced his theory of the just war from exile, being smuggled out of Holland in a trunk by his wife and family. 5 Nor did danger fail to greet his Enlightenment successors. The Scottish Enlightenment was relatively gentle in its restrictions: Hume's alleged atheism did not jeopardize his life or stop the publication of his works. But it did cause him to be denied the chair of philosophy in Edinburgh, a judgement approved by one contemporary opponent of ethical theory, on the grounds that ethical theory should derive its first principles from religious authority.° On the continent, philosophers faced sharper opposition. In 3 See I. During, Aristotle in the Ancient Biographical Tradition (Goteborg: Almqvist & Wiksell,1957), Vita Marciana sect. 41, 105. Like all the material in the ancient lives, this story is of dubious authenticity, but it shows what many people believed about the events. 4 See Tacitus, Annales, ed. Furneaux (Oxford: Clarendon Press, 1907), xv.63. Seneca was implicated in the conspiracy of Piso, which, like other anti-Imperial conspiracies, was inspired in part by Stoic ideals of libertas. 5 Grotius' trial and imprisonment were for Arminianism, thus for a religious rather than a philosophical view; and yet his Christian and his Stoic/philosophical views are so mixed together that it seems impossible to isolate a purely religious aspect of his career. 6 Alasdair MacIntyre, in Whose Justice? Which Rationality? (Notre Dame: University of Notre Dame Press, 1987).
Why Practice needs Ethical Theory
229
France, many works of ethical theory, viewed as anti-clerical, had to be clandestinely circulated along with works of sexual pornography. In Germany the book trade was more decorous, but Kant, no friend of the pornographers, still had to fight for philosophy's freedom of speech. In 1795, writing on the conditions of a lasting peace among nations, he cited as 'Secret Article of a Perpetual Peace' the freedom of speech of the moral philosophers, without whose aid, he argued, governments cannot succeed in making a productive plan for the containment of aggression. 'Kings or sovereign peoples,' he wrote, 'should not ... force the class of philosophers to disappear or to remain silent . . . This is essential to both in order that light may be thrown on their affairs.'7 Nor, in our own century, has moral philosophy failed to threaten and be threatened. We have seen the political persecution of moral philosophers such as John Dewey and Bertrand Russell (imprisoned twice for his arguments against war and tossed out of a job at the City University of New York for allegedly obscene writings on marriage, later praised in his Nobel Prize citations); the stocking of Eastern European philosophy departments with drones and sycophants, the virtual impossibility of doing moral philosophy at all in Cuba, China, and many other parts of the world, even when other parts of the subject, such as logic and the philosophy of science, are permitted to proceed more or less as usual. 9 7 I. Kant, 'Perpetual Peace', in Kant: Political Writings, ed. H. Reiss (Cambridge: CUP, 1991), 115. 8 Russell was in prison in 1918 and 1961 (when he was 86); Marriage and Morals was condemned as obscene by a New York court, upon which Russell lost his job at CUNY. 9 Williams here makes two valuable points: first, that the attackers may not have been enemies of ethical theory as such, but only of a particular kind of (secular anti-clerical) theory; they may well have been fond of some religious ethical theories. Second, that his own position does not entail the rejection of philosophical activity, and thus the attack on philosophy as such is importantly distinct from his more limited assault on ethical theory, which retains a large role for the critical scrutiny of tradition. The second point is true and important, and I try to deal with it in the last section of the paper; I do believe that the attackers would have been far less worried about Kant, Rousseau, etc. if they had had no systematic views that contradicted received views, and so I continue to believe that the theoretical aspect of the theorizers' work was salient in motivating the attacks. (This was true even of Socrates, who was understood by his attackers, rightly or wrongly—I think rightly—to have a definite set of subversive views.) About the first point, I have some doubt. We have to be careful here how we draw the line between a religious and an ethical theory, and some ethical theories (such as that of Aquinas) have been housed within religious systems of thought. My own view would be, however, that they cease to be ethical theories of a philosophical kind when they submit to temporal religious authority for their interpretation and further development. I doubt that the attackers were really friendly to any philosophical theory, even that of Aquinas, that they could not rigidify and codify in their own way.
230
Martha Nussbaum
It is nothing new, then, that ethical theory should be assailed from outside, by religion, politics, and custom, by power and anti-reason and sheer bad behaviour. What is new, however, is that these days it is also under attack from within. During the past decade a number of prominent moral philosophers, including Bernard Williams, Annette Baier, Alasdair MacIntyre, and Cora Diamond, have assailed ethical theorizing, especially in its Enlightenment forms, as both useless and pernicious, as distorting practice and contributing nothing that could not be gained through more informal types of ethical reflection. It might strike one that the external assault gives some evidence against the 'uselessness' part of the internal charge. Any type of intellectual activity that is so vigilantly opposed by power is unlikely to be utterly without practical value. Nor do Williams and Baier, at any rate, want to yield the scene of social and personal decision-making to the conservative and/or authoritarian forces who, in these various cases, opposed the philosophers. 10 One might, then, suppose that the persecution of philosophers would give these thinkers, too, at least some reasons to defend the influence of ethical theory as not altogether pernicious, holding not only that it has a practical impact but also that this impact has done some good. That they do not support the enterprise is, however, clear. Baier declares: `I want to attack the whole idea of a moral "theory" which systematizes and extends a body of moral judgments,' and inveighs against 'that arrogance of solitary intellect which has condemned much moral theory to sustained self-delusions concerning its subject matter, its methods, and its authority."' For Williams, the major modern moral theories are 'not well adjusted to the modern world, and 'governed by a dream of a community of reason that is too far removed . . . from social and historical reality and from any concrete sense of a particular ethical life—farther removed from those
10
Annette Baier, 'Doing Without Moral Theory?' in S. Clarke and E. Simpson (eds.),
Anti-Theory in Ethics and Moral Conservatism, (Albany: SUNY Press, 1989), claims that she
does not share Alasdair MacIntyre's 'very gloomy estimate of the prospects of a secular culture'; Williams goes to considerable lengths to show that ethical reflection should criticize experience that is disordered: see his response to critics in J. E. J. Altham and R. Harrison (eds.), World, Mind, and Ethics (Cambridge: CUP, 1995), 218-19; see however his comment below, judging that in some respects moral philosophy is further removed from social and ethical reality than religion is. Diamond's positions on political and religious authority have not yet been articulated, so far as I can see. MacIntyre's position is highly complex, since he recommends a large role for philosophical argument, but within bounds set by first principles handed down by religious authority. I shall not discuss his views further in what follows. 11 Baier, 'Doing Without Moral Theory?', 29-48, at 33,46.
Why Practice needs Ethical Theory
231
things, in some ways, than the religion it replaced'.1 2 Diamond's position is more elusive, since she doubts that we can succeed in finding any widely agreed account of what the enterprise of ethical theory is; but in so far as she does propose such an account, she concludes that 'we should not take those rules seriously, or the conception of moral philosophy which they determine'.13 There are complexities in all these thinkers' positions; it is not clear that any of them opposes all the prominent ethical theories, or that the grounds of their opposition cannot be met by something that most of us would agree in calling a theory. I shall be pressing these questions in what follows. It is evident, however, that they all take themselves to be showing that on at least some widely shared understandings of what ethical theory involves, it is both unimportant and, so far as it does affect things, mostly damaging, a squeezing and deforming of particular experience that may actually prevent us from making the more valuable types of criticism of our daily lives.14 My purpose in this essay will be to state these objections and to contest them. Having enumerated some of the central criteria of ethical theory, as it has been defined in debates both ancient and modern, I shall introduce a distinction familiar in ancient Greek and Roman Stoicism, but largely absent from the modern debate. The Stoics recognized not two categories, theories and concrete judgements, but three categories: theories, rules, and concrete judgements. I shall argue, with Seneca, that the distinction between theories and rules is an extremely important one, which enables us to avoid a number of confusions. I shall lay out some Stoics' arguments for thinking that there is a natural alliance between theory and particular judgement, in that theory enables us to understand the limitations of general rules in ways we could not otherwise, therefore to correct the deficiencies inherent in any system of rules. Thus criticism of systems of rules need not entail criticism of ethical theory, and can in fact give us reasons for turning to an ethical theory. 12 Bernard Williams, Ethics and the Limits of Philosophy (hereafter ELP) (Cambridge, Mass.: Harvard University Press, 1985), 197. 13 Cora Diamond, 'Having a Rough Story about What Moral Philosophy Is New Literary History, 15 (1983), 155-69, at 168. Diamond is here referring to an essay of mine on The Golden Bowl, in which I try to show that the novel makes a valuable contribution to moral philosophy, in the process offering what I call 'a rough story about what moral philosophy is'. Diamond argues that on one common idea of what moral philosophy is (not, however, as she recognizes, the one I use in my article), a project like mine (which she defends) would not be possible. 14 See Williams, in Altham and Harrison ( World, Mind, and Ethics, 218-19): theory 'can be an effectively articulated expression of these distortions . . . blocking radical reconsideration of our morality'.
232
Martha Nussbaum
With all this in place, I shall then identify the most prominent recent objections to ethical theory and argue that there is none that cannot be met by something still recognizably theory—although some of the more telling objections will lead us to reject theories that (unlike all ancient and most interesting modern theories) identify theory with a system of rules. Finally I shall argue that we urgently need theory for the reasons given by Kant and Seneca. In a world in which moral perception is corrupt and judgement likely to be thrown off the track by temptations of all sorts, we need all the explicitness and articulateness we can muster if we are to elicit the best from ourselves, to identify defects in our social world, and to devise appropriate institutional and educational remedies.
What is an Ethical Theory? 15
Before we can begin, we need some account of the item in dispute. This is tricky, since all sorts of different items have turned up in definitions of ethical theory, some so contentious that their acceptance would entail directly that ethical theory is in grave difficulty. (Thus, if it were granted that something is an ethical theory just in case it states that there are no moral dilemmas, or that there are no exceptions to generally binding rules, many people would immediately concur in the rejection of theory. But it seems implausible to make this stipulation, since something that most of us would agree to call theory can easily reject these contentions, as quite a few well-respected theories have.) Nor have the antagonists always been forthcoming with their criteria. Baier mentions explicitness, universality, systematicity, and hierarchical ordering, but offers no general definition of theory that would show which of these items she views as necessary and/or sufficient for it. Moreover, some of her concrete claims, for example the claim that neither Aristotle nor Hume has an ethical theory, cast doubt on some of the criteria enumerated, in so far as they characterize the work of those thinkers as well; her historical remarks are more puzzling than helpful, in the absence of further evidence about the way in which she is interpreting the authors in question. Diamond, as I have said, rejects the whole enterprise of giving even a 'rough story' 1 6. about what ethical theory is; it is 5 I shall make no distinction between 'moral theory' and 'ethical theory'. There seems to be no generally agreed distinction between the two, and it seems best to proceed by using a single term and defining its use carefully. 16 My own term, in the article she is criticizing.
Why Practice needs Ethical Theory
233
therefore extremely difficult to tell what her antitheoretical remarks are directed against. Williams is much more direct, defining theory as follows: 'An ethical theory is a theoretical account of what ethical thought and practice are, which account either implies a general test for the correctness of basic ethical beliefs and principles or else implies that there cannot be such a test.'" But we need to ask what this statement means: for Williams grants that Rawls' theory is an ethical theory, and yet Rawls' theory, by his own description, holds that the process of justification is holistic and that there is no single `test' for beliefs, the criterion of rightness being given by overall fit in the system as a whole. Sometimes, moreover, Rawls holds, the concrete beliefs test theoretical claims. So Williams' general definition must be intended to be elastic enough to include this more holistic sort of 'testing') 8 Elsewhere, Williams mentions other criteria for theory: theory involves the attempt to `systematize' (p. 116); it 'looks characteristically for considerations that are very general and have as little distinctive content as possible'; it is a 'structure of propositions' that 'in part provides a framework for our beliefs, in part criticizes or revises them' (p. 93). But it remains not fully clear which of these, if any, is intended as a necessary condition of theory. I shall proceed in two ways: first, by mentioning examples of ethical theory that seem to me so central that any account of theory that doesn't cover them is peculiar; second, by mentioning some of the features of theory that I take to be most pertinent to the contemporary debate—though here I shall not include some features that simply don't fit a significant number of the examples. The examples: Aristotle, the Greek and Roman Stoics, Aquinas, Hume, Kant, Adam Smith, Bentham, Mill, Sidgwick, Rawls. (I select these in part for their obviousness, in part for the fact that they are central cases of theory for the attackers. I myself would be happy to include Spinoza, Hobbes, Rousseau, and quite a few others; I shall later include Grotius, as one example of the Stoic tradition; but this list is sufficient for our purposes.) A good way of thinking about what an ethical theory is, is that it is a set of reasons and interconnected arguments, explicitly and systematically articulated, with some degree of abstractness and generality, which gives
Williams, ELP, 72. Williams has helpfully clarified this point: his phrase 'implies a general test for the correctness of was not meant to mean 'implies that there is one single test which excludes holistic . . . argument'. 17 18
234
Martha Nussbaum
directions for ethical practice. The criteria of theory on which I shall focus are six.' 9 1. Gives recommendations about practical problems. An ethical theory gives direction for practice: it shows us how to make progress on ethical problems. There is no reason to think that this progress need be easy or straightforward: one way of making progress is to identify complexities and difficulties. Nor is there any reason to suppose that this progress need involve direct application of theory to unreformed practice—as opposed, say, to instructions for the reform of institutions and of moral education. 2. Shows how to test correctness of beliefs, rules, and principles. I revise Williams' definition to make room for his own central case, Rawls' account of the search for reflective equilibrium, in which principles and judgements inform one another and none is held absolutely fixed. Some theories (for example, those of Socrates 20 and, to some degree, Cicero) prefer to hold concrete judgement fixed and to use those as a test for principles; others prefer to use principles as test for judgements; Rawls, like Aristotle and Sidgwick, allows illumination to travel in both directions. All of these seem acceptable ways for a theory to proceed; what is essential to its being a theory is that it give us some account of how to proceed in doing this; this account may allow a good deal of room for further interpretation and judgement, as it does in the case of Aristotle and Rawls. 3. Systematizes and extends beliefs. One of the major purposes of having an ethical theory is to bring the material of ethical experience into a perspicuous ordering, rendering the incoherent coherent (by suitable revision and discarding), and showing how one thing relates to another. This also makes it possible to extend the application of principles to previously unconsidered material, or to see how one concrete judgement can be extended to similar cases. 4. Has some degree of abstractness and generality. An ethical theory will be only as abstract as its insights deem relevant: we should not take it as a necessary condition for an ethical theory that it regards all concrete situations 19 Compare the very helpful list of twelve characteristics in R. Louden, 'Virtue Ethics and Anti-Theory', Philosophia, 20 (1990), 93-113; and in his Morality and Moral Theory. My list is closely related to his, though shorter. 20 Of course, any claim about Socrates is bound to remain controversial. Some would deny that he had a theory at all. Moreover, my claim is complicated by the fact that on some occasions Socrates appears to hold fixed not only concrete judgements about action-types, but also some very general judgements, such as that a good person cannot be harmed. For my own current view of Socratic inquiry, together with a discussion of Vlastos' important account, see my review of Vlastos: Socratic Studies, in The Journal of Philosophy, 94 (1997), 27-45.
Why Practice needs Ethical Theory
235
and relationships as ethically irrelevant, or we should omit virtually all of the theories on our list of examples. (For Kant, particular relationships are highly relevant to duties of beneficence; even Utilitarians can admit them as relevant to the strategies by which agents maximize universal happiness.) On the other hand, a theory cannot consist of simply a collection of reports on concrete particular judgements in concrete contexts. It will be a theory only if it can give guidance for the future (see 1.); but this means that it must abstract to at least some degree. There are, obviously, both good and bad ways to do this, and this will concern me in what follows. 21 5. Is universalizable. An ethical theory should be applicable to all agents as such, rather than to agents simply in virtue of being members of a given community or religious group. This does not mean that an ethical theory cannot recommend latitude for people to pursue attachments to their communities or religious groups. It also does not mean (see 4.) that an ethical theory cannot allow its principles to be applicable only to people whose ethical situation is relevantly similar; any sensible theory will have an account of relevant similarities and differences. These may or may not include family relationship, friendship, the character and propensities of the people in question, whether the agent is oneself or not, and so forth. Again, to say that a theory may not take these into account would be to make theory a virtually empty category. (In general, any categorization of theory according to which Kant does not have a theory should be condemned as empty rhetoric.) Universals may be, in this sense, highly concrete, and probably they must be to give appropriate guidance. 22 6. Is explicit. A theory is, if it is anything, a set of explicit guidelines for practice: it is written down or otherwise promulgated, and is available to be consulted by all. This does not mean that it may not take account of features of the inner world of people that cannot be easily made explicit, such as their loves and their religious attachments and convictions. Nor does it mean that it cannot say that in certain circumstances there is no explicit guidance it can give; most theories in fact contain an account of their own limitations. (Again, it is empty rhetoric to caricature all ethical theory as demanding explicit rules about everything, as if theories as such could not allow that human life has areas of mystery. But it's also a good idea to remember that to point to a mystery shouldn't automatically be taken to immunize the area in question from the critical scrutiny of theory: thus 21 See the illuminating discussion in Onora O'Neill, Towards Justice and Virtue: A Constructive Account of Practical Reason (Cambridge: CUP, 1996), ch. 2. 22 On the concrete universal, see R. M. Hare, Moral Thinking (Oxford: Clarendon Press,
1981).
236
Martha Nussbaum
Mill was correct that romantic views of love and sex can all too easily function to seal off that part of life from criticism, and that this move should be resisted for the sake of equality and justice.) These six criteria, as described, will be treated as both necessary and (jointly) sufficient for ethical theory. I do this for the sake of clarity in argument, and not because I really believe the tradition has such a unified shape; in reality, there are some characteristics that are regarded as more central in some eras and others in others. But these six do seem to turn up in some form in all the major examples, and they also appear to be among the characteristics most contested by theory's opponents.
Three Things, Not Two: Theories, Rules, Concrete Ethical Practice
No ethical theory is a system of rules. None of the examples on my list simply enumerates the rules governing conduct, for example 'don't lie, 'don't kill, 'don't steal, and then organizes these into a system. Indeed, it is obvious that we associate this way of proceeding with religion and custom, far more than with ethical theory. We might without much exaggeration say that it was systems of rules that ethical theory came on the scene to displace. Why? What has ethical theory traditionally taken as its task that leads it to have reservations about the usefulness of rules and rule-governed conduct? I shall now draw on Seneca's Moral Epistle 95, as he attempts to answer this question. 23 1. Point and purpose. First of all, the maxims by which people usually guide their conduct are obtuse: without something more, they do not show their own point and purpose. 'Don't lie' is all very well, if the person is docile; but if a person is inclined to want to lie for some particular purpose, she will need to understand something that the rule doesn't tell her—why she should not lie. Conventional systems of rules say 'because this is the way we do things'. Some religious systems of rules say, 'Because God commanded you not to lie.' Philosophical theories proceed in opposition to both of these answers. As Sidgwick puts it well, they recognize that con23 See the related threefold distinction in Sidgwick, The Methods of Ethics, Book I (London: Macmillan, 1907), ch. viii, where Sidgwick describes the transition from particular intuitions to rules of conduct, and from both of these to philosophical explanation. (Sidgwick, however, conceives of the more abstract as always being more firm than the concrete, not seeing the Stoic point that theoretical principles can help us identify exceptions to generally valid rules.)
Why Practice needs Ethical Theory
237
ventional moral codes, even when they give generally good guidance, are only 'an accidental aggregate of precepts, which stands in need of some rational synthesis. In short, without being disposed to deny that conduct commonly judged to be right is so, we may yet require some deeper explanation why it is so.'24 Theories acknowledge the reasonableness of the agent's question, and they attempt to give her a good answer, giving reasons for the value of the rule in question. In so doing, they treat people as independent thinking adults, rather than as children. The form such answers take varies, of course, with the style of ethical theory being proposed. Some theories give the point of rules in terms of a single highly general end that all good conduct is alleged to promote, and give agents reasons for thinking this end the best one. (Utilitarianism is the most obvious example of such a theory.) Others give the point in terms of a plurality of (usually interconnected) ends, each of which is taken to have independent value; they then set about showing agents that this list of ends is a reasonable one to adopt, and, if rules are the issue, that the rules in question promote those ends. (Aristotle's theory is an obvious example of such a theory.) Others give the point in terms of some more general imperatives, and then try to show the point of those imperatives by convincing agents of their overwhelming importance. (Kant's is an obvious example of a theory of this type, although Kant is also concerned with the relation of duties to the happiness of agents, and in that sense has a mixed theory.) What is important is that the reason of the agent is addressed with persuasive considerations that illuminate rules of conduct, giving an intelligent being something to go on in deciding whether she wants to adopt the rule in question.25 Opponents of theory need to consider more than they typically do what the alternative to this way of proceeding usually is. 2. Motives and character. One of the most common complaints made by the ethical theory tradition against systems of rules is that they prescribe conduct without saying anything about the actor's state of mind and emotion. But—given, once again, that we are respecting people as adults rather than treating them as children to be ordered about—we feel that people's conduct is right or wrong depending not just on the bodily movements they perform, but also on how their thoughts and emotions are working. If a person acts in a childlike way, fulfilling the rule because an authority figure says so, all major ethical theories will agree that this is not a case of right Ibid., Book I, ch. VIII, sect. 4. See, e.g., 95.12, where Seneca insists that theory is addressed to a rational being as such, and that rules by themselves are 'rootless' if they don't attach to an overview of human life. (Seneca, Epistulae Morales, ed. C. D. Reynolds (Oxford: Clarendon Press, 1965).) 24 25
238
Martha Nussbaum
conduct.26 If the person fulfills the demands of the rule with great struggle and reluctance, many ethical theories will also deny that this is a case of right conduct—those that hold that emotions and desires can and should be cultivated to love the good and right. Ethical theories undertake to specify the state of mind and emotion in which a suitably performed action counts as right and virtuous. Thus they need to answer questions about what the passions are, how and to what extent they can be cultivated, and to what extent we can expect them to agree with judgement. These issues are as much a preoccupation of Kant as they are of Aristotle, though he arrives at different (and, I believe, mistaken) conclusions regarding the malleability and intelligence of the passions. The tendency of the antitheorists to caricature Kant as preoccupied with rules to the exclusion of moral psychology is a most unfortunate distortion. It's perfectly obvious that from page 1 of the Groundwork on he is preoccupied with motive and intention, and that the entirety of the Doctrine of Virtue is an attempt to describe virtuous states of mind. 27 3. Exceptions to rules. Systems of rules seem obtuse in another way: they do not mention special circumstances in which the rule may not give good guidance. This is connected with the fact that they don't describe their own point and purpose—so we can't easily see from rules alone when that point is better served by a divergence from the rule. This is why the Stoics and Aristotle insist that there is a natural complementarity between theory and fine-tuned practice. If you have the illumination of theory, and you understand the point and function of the rules, then you will be able to see the new particular case more clearly, seeing, frequently, that this is a case where following the rule would not make sense. Cicero thinks it's obvious, for example, that if you promise to show up in court on a particular day to help a friend plead his case, and then when the day comes your child is very ill, you don't have to keep that promise, and your friend would have a false idea of morality if he complained. 28 Now Cicero thinks most people would act this way anyhow, but having a theoretical account of the importance of 26 See 95.40: (Seneca, Epistulae Morales), purpose and manner are all important, and this cannot be given in a system of rules. 27 For two convincing accounts, which show the close relationship between Kant's theory and Aristotle's, see Barbara Herman, The Practice of Moral Judgment (Cambridge, Mass.: Harvard University Press, 1993); and Nancy Sherman, Making a Necessity of Virtue: Aristotle and Kant on Virtue (New York: CUP, 1997). See also Onora O'Neill, Constructions of Reason (Cambridge: CUP, 1989); Christine Korsgaard, Creating the Kingdom of Ends (Cambridge: CUP, 1996); and S. Engstrom and J. Whiting (eds.), Aristotle, Kant, and the Stoics: Rethinking Happiness and Duty (Cambridge: CUP, 1996). 28 Cicero, De Officiis, 1.32, ed. M. Winterbottom (Oxford: Clarendon Press, 1994).
Why Practice needs Ethical Theory
239
care for one's family and its relationship to one's other duties will prevent someone from getting confused about the reach of the rule. In other more controversial cases, theory does more work. In killing Julius Caesar, Brutus and the other conspirators were killing a human being; they were also killing a friend. Cicero thinks, plausibly enough, that conventional systems of rules, and also conventional moral practice, would condemn their action. He believes, also, I think, plausibly, that his complex Stoic-based ethical theory29 can show the point and purpose of the rules against killing and in favour of friendship, and that, when it has done so, we will see that the assassination was morally justified. This is so because the same theory shows the crucial importance of republican political institutions in giving people the liberty within which to cultivate their humanity and to pursue friendship. If, then, we agree with Cicero that Caesar's ascendancy represents a dire threat to republican institutions and that the assassination gives a reasonable hope for preserving them, we will agree that the assassination represents a rare exception to the usual rules. 30 The tyrannicides might have seen all this without theory (though in fact they didn't: Brutus was a moral theorist); what is more important to Cicero is that theory can help the general populace to see the overwhelming importance of republican liberty, at a time when it is threatened with extinction and Antony's rhetoric is confusing the issue. In short, theory is not obtuse in the way that systems of laws can frequently be obtuse: by turning to theory, which gives us the point and purpose of rules, we learn when we may diverge from them. And we can go further: divergence from rule without point and purpose is itself obtuse. Theorists who license exceptions to generally good rules are legitimately anxious lest their permissions encourage bad behaviour. They therefore plausibly insist that the exception should be taken only when an argument can be given that convincingly links the conduct in question to the overall purpose and point of human conduct. By imposing a stringent and public intellectual burden on the exception-taker, these theorists discourage selfserving exception-taking. 4. Gives arguments. Theories, unlike systems of rules, address their recipients as reasoning beings. This means that, unlike systems of rules, 29 It is an interesting question whether Stoicism by itself could justify this conclusion, or whether Cicero is here, as often, relying on a hybrid version of Stoic and Academic ideas. 3(1 See Cicero, De Officiis, 111.19; a similar example is at 111.40, where Brutus (not the same one!) deposes his colleague Collatinus from office right after the expulsion of the kings, on the grounds that having a member of the royal family in high office is a threat to the founding of the Republic.
240
Martha Nussbaum
they give arguments for what they conclude. 31 They begin by giving accounts or definitions of their key terms, and they proceed to lay out explicitly the course of their thought, showing how the general overview of human ends can be thought to dictate a particular course of conduct. This, theorists emphasize, is a way of respecting recipients. Nothing is hidden, everything is out in the open. If you don't like the definition, you are at liberty to object, or to propose one of your own. If you don't like the way the conclusion is reached, find fault with the inferences, or the premises. Or bring in some other consideration that points in the opposite direction. All these familiar philosophical manoeuvres are made possible by the reasonableness and explicitness of theory, which addresses the recipient as an equal. It cannot be emphasized strongly enough that this way of doing things is established in deliberate opposition to other ways of giving rules for conduct, those deriving from custom and religious authority. So Socrates hears the oracle, but he won't allow it to govern his conduct until he has worked out the argument on his own. He sees himself and others as separate beings, centres of reflection and choice, possessed of the right to work out the purposes and patterns of their own existence. A life without that he sees as slavish and base. That is the message of philosophical theory in ethics, and it is for precisely this reason that it has been under attack ever since its beginning. Mental liberty is not popular with world leaders. None of these points implies that ethical theories should dismiss rules as pointless. Typically they do not. Most ethical theories draw heavily on the wisdom embodied in rules and conventions; from Socrates onward, they regard it as in general a point in favour of a theoretical account if it can preserve at least those general judgements that we regard as especially sound. Thus the rule against killing holds an important place in Stoic moral theory, as one of the data of human moral experience on which theory goes to work, and which it would be surprising for a theory to reject wholesale. Indeed, a challenge to one well-entrenched rule is frequently given by showing that it conflicts with another that is regarded as more fundamental: thus, conventional rules mandating unequal education for girls and boys were opposed by the Stoics with arguments that held this rule to be incompatible with the rule that every human being deserves the maximal development of its powers toward virtue. Rules are regarded as helpful in other ways as well. 32 They summarize the decisions of wise judges, whom we have some reason to trust as possi31 32
See Seneca, Epistulae Morales, 95.55, 61-2. On this, see Seneca, Moral Epistle, 94.
Why Practice needs Ethical Theory
241
bly shrewder and more experienced than ourselves. They keep us on the track when we are likely to go astray through partiality to self or friends and the special pleading to which that can so easily give rise. 33 (This is an especially important function for rules in a system of law.) Again, when we are faced with a complex particular situation, rules refresh our memory, shape and inform our vision, and focus our attention on aspects of the situation that we might otherwise have missed. On this basis Seneca argues that even the person who is familiar with the most general purposes and ends of human life, as given in a moral theory, still needs to hold on to rules of conduct: for our minds are often undertrained in discerning the salient features of a situation, and rules help us see correctly. 34 (Again,thsueflro inasytemflw,ciuaons be described in indefinitely many ways, and rules give us a set of categories that are of at least prima facie relevance.) Finally, rules save time. A Henry James novel may take six hundred pages to give an account of the relevant features of a practical situation; a legal system cannot afford such leisured description, and frequently our lives can't either. We need to summarize, classify, subsume—even when we admit that this is not always the best way of doing justice to all features of the particular. This suggests that we will want to think hard about when this is rightly done and when it is most likely to prove inadequate—when, for example, the fuller narrative of a criminal defendant's life might usefully come into view, and when we may properly confine our vision to the time of the crime itself. So we have three items: our concrete ethical practice, rules of conduct of various types, and ethical theories. The ethical theorist claims that an ethical theory gives important guidance for ethical practice and a set of guidelines for the proper use of rules, by sorting out the material of conduct in a more explicit and perspicuous way, giving the point and purpose of maxims of various types, and providing an account of human psychology that will both direct programmes of moral education and show when basically appropriate conduct is or is not fully virtuous.
33 See, on this, my account of Bob and Fanny Assingham in Henry James' The Golden Bowl, in "'Finely Aware and Richly Responsible": Literature and the Moral Imagination', in my Love's Knowledge (New York: OUP, 1990). 34
See Seneca, 94.32-33.
242
Martha Nussbaum
Objecting to Ethical Theory What are the charges of the antitheory philosophers, and how far can theory answer them? The current debate contains six primary charges. 1. Theories neglect an agent's own particular projects and her special relation to them. This is among Bernard Williams' central complaints about ethical theories. Indeed, in an essay on Sidgwick, he has stated that theories demand neglect of our relationship to our projects, and that this fact 'presents an insoluble problem to ethical theory'. 35 By asking us to assess courses of action 'from no point of view at all, these theories make an impossible and also inappropriate demand. For this simply is not how people live their lives and endow them with meaning. To this we should reply, first, that although this may be an accurate characterization of Bentham's and Sidgwick's utilitarianism, it is far from being generally true of ethical theories. Even Kant's theory, while frequently asking us not to take account of our own particular situation and relationships, insists that we ought to do so when we are thinking of duties of beneficence. And there is always one special position, one's relationship to oneself, that occupies an absolutely fundamental structuring role in Kant's theory. The Stoics, Kant's mentors in many respects, also insisted that family ties, civic ties, and many other features of our situation should get special consideration, though not as much as many people give them. That Kantians typically deny the moral relevance of many other features of people's situations, for example their race and gender, might be thought to be a point in the theory's favour; I am sure Williams thinks it is. As for Aristotle's ethical theory, it is entirely built up around the project of an agent who attempts to build a complete life for herself; this is a reason why Williams is more sympathetic to it than to other theories—though, despite his sympathy, he advances special reasons for dismissing it that seem to me inadequate. 36
35 B. Williams, 'The Point of View of the Universe: Sidgwick and the Ambitions of Ethics in Making Sense of Humanity and Other Philosophical Papers, 1982-1993 (Cambridge: CUP, 1995) 153-71, at 170. The essay was first published in 1982, and thus predates the final version of Bernard Williams' ELP. 36 See our exchange in World, Mind, and Ethics (Williams in Altham and Harrison (eds.) ). Williams first made some of these criticisms in 'Moral Standards and the Distinguishing Mark of Man', in Morality (Cambridge: 1973); they are further developed in Williams, ELP.
Why Practice needs Ethical Theory
243
We might add that the demand to see our situation from a point of view external to our own is a demand that arises within the ordinary point of view on ethical matters. Often we feel that we are too self-focused; even small children soon acquire the idea of a fair division of good things, and criticize those who think only of their own goals and projects. The idea that a division should be impartial, made as if from the point of view of no particular individual, is more common on the playground (and at the family dinner table) than in politics. So it is not correct that in making the impartial perspective central the philosophers who do so are asking us to depart from ordinary life and practices. Whether they are right to make this one aspect of our practices central and to demote others is, as I have said, a disputed point within ethical theory itself. But Williams is surely shortchanging a part of ordinary life when he represents the non-theoretical agent as immured within a personal perspective on the world. 2. Theories ignore moral psychology and the importance of emotions, etc., in good action. This objection is suggested in various ways by Diamond, Williams and Baier 37—but it seems, frankly, just mistaken. There is no major ethical theory that considers only an agent's reasoning processes important in arriving at correct choices. All have a very deep interest in the passions, and all have accounts of how institutions and moral education can shape the passions so that they are more likely to support good action. This is as true for Kant as for Aristotle, though his account of the passions is different. The Stoics are among the most profound psychologists in the entire history of philosophy, and the eradication of anger and hatred in favour of general human sympathy is a major part of what they propose. Adam Smith's ethical theory is entirely focused on the passions; Hume and Mill have a good deal to say about them too. Most of the theories, moreover, ascribe at least some positive value to the passions in making good choices. Kant is no exception, as his ambivalent discussion of pity shows. The Stoics are indeed exceptions, urging a thorough extirpation of all anger, grief, pity, and the other items they call `passions'; but they have arguments for this conclusion, and they do leave agents other affective motives, such as the love of humanity, and even a type of erotic love, 38 to steer them in choosing. If to some degree the Stoic life strikes us as bloodless, we will need to grapple with their argument (which Spinoza develops 37 Diamond, in 'Anything but Argument?' Philosophical Investigations, 5 (1982), 23-41: Diamond focuses on a demand made by Onora O'Neill, in reviewing a book on animal rights by Stephen Clark, that Clark 'reach beyond assertion to argument'. 38 See M. Nussbaum, 'Erôs and the Wise: the Stoic Response to a Cultural Dilemma', Oxford Studies in Ancient Philosophy, 13 (1995).
Martha Nussbaum
244
further) that it is only in this way that society can be purified of hatred; we cannot simply dismiss them as people who don't see what passion can contribute to life. 3. Theories hold that there can be no moral dilemmas; they neglect the plurality of goods. There are actually two distinct points here: for a theory can accept the plurality of goods while giving us a decision-rule that will allow us to resolve every apparent moral dilemma. But at a deep level the two issues are linked: for the recognition that each of a plurality of diverse goods exerts its own sui generis pull on moral agency does seem incompatible with holding that this pull is altogether removed by a contingent conflict with another obligation. So a theory that recognized distinct sources of value sensibly should recognize contingent moral dilemmas. Utilitarianism, especially in the version propounded by Sidgwick, does set out to remove moral dilemmas (clashes between right and right in which any course an agent might choose seems to involve some wrongdoing). The preferred strategy does indeed involve the homogenization of values. Kant seems determined to claim, as well, that there are no genuine moral dilemmas, though without proposing a single common coin of value. But this feature of his theory is difficult to integrate with the rest of it. Given his recognition that each human being is an end, of intrinsic worth, and given the obvious fact that society often makes it impossible for us to satisfy all our responsibilities to all the people with whom we have dealings, it would have seemed natural for him to recognize that contingent social facts can confront agents with moral dilemmas. This would provide an incentive to redesign those institutions (as Hegel saw). 39 So we may have a theory of a basically Kantian type that does recognize moral dilemmas. Aristotle's theory doesn't give prominent recognition to moral dilemmas, but he probably does recognize them; his theory, recognizing plural sources of intrinsic value, is certainly compatible with that recognition, and there is some evidence that this is the right reading of his text. 40 Cicero tends to treat every prima facie conflict between obligations as resolvable by creating a suitable exception to one of the rules: but there is no necessity that he handle things in this way, and one might adopt a basically Ciceronian theory in ethics while recognizing that not all conflicts of value can be resolved in this way. Such a theory would make a lot more sense of the complex cases Cicero introduces. 39
See Ruth Barcan Marcus, 'Moral Dilemmas and Consistency', Journal of Philosophy,
77 (1980), 121-36. 49
See Michael Stocker, Plural and Conflicting Values (Oxford: Clarendon Press, 1993).
Why Practice needs Ethical Theory
245
In general, a moral theorist is not likely to defend a picture of the human good in which the distinct and plural goods recognized by the theory are intrinsically in conflict with one another; most moral theorists have been attached to consistency in that sense. But there is absolutely nothing to prevent the moral theorist from acknowledging that two goods that are in principle compatible, for example one's attachment to one's children and one's attachment to one's work, may come into conflict in particular contingent circumstances. Indeed, that is exactly what a moral theorist who recognizes plural values (especially in an imperfect world) ought to say. The frequency and gravity of such conflicts is often the product of social arrangements: for example, societies that provide for parental leave have fewer and less grave conflicts between family and work than societies that do not. So recognition of the conflict will often, and reasonably, provide the theorist with an occasion to propose institutional reform. But this surely does not mean that the theorist is attached to consistency in a perverse manner, or refuses recognition of the texture of value in real human lives. In fact, people usually prefer not to face moral dilemmas; and if we can arrange things so that both of the good things they pursue are more securely within their grasp, that's a victory. Theory can contribute to that victory, by articulating the features of the good in a perspicuous way and making the location of likely conflicts evident. We must now speak about another type of value-pluralism. Different cultures frequently endorse different lists of valuable things. Different groups within a given culture do so too. Theory, as I have defined it, is in its nature antirelativistic: it says, that is, that at the most general level there is an account of ultimate ends that applies to all people as such, and therefore implicitly denies that the norms of a given society or group are the court of last resort in ethical matters. This does not prevent theory from leaving spaces for local or personal specification of ends: and liberal theories of justice standardly do so, defending the freedom of conscience and other types of liberty. Nor does it prevent theory from articulating its goals at a high level of generality, in order to allow for multiple specification even of the goals it does lay out. The theory I favour does exactly this.'" With respect to the second type of conflict among ends, then, theory as I characterize it does constrain pluralism and repudiate relativism; but there is a lot of space inside it where legitimate types of pluralism can flourish. 4. Theories give crude guidance, fail to direct agents in handling the complex contexts of life. As I have tried to show, to some extent this criticism 41
See, e.g., 'Aristotelian Social Democracy', in R. B. Douglass, et al., Liberalism and the
Good (New York: Routledge, 1990).
246
Martha Nussbaum
relies on confusing theories with rules or systems of rules. Nor is there any major ethical theory that claims to have provided an algorithm that makes tough moral reflection about particular cases otiose. That, indeed, is why all these theories have been preoccupied with moral psychology, feeling that the best way to produce appropriate choice is to produce a certain type of agent. We may still feel that in some areas some of the major theories have been too crude to guide us well; that is why it remains important for theories to consult good practice. Here Diamond's and Williams' interest in literature has real force, since literature can frequently, as Diamond suggests, cultivate our moral sensibilities and refine our ability to interpret complex situations. But literature can play this role as a partner of ethical theory; 42 one need not jettison theory in order to accept some of the most attractive suggestions of the antitheorists. 5. Theories ask agents to deliberate about everything; but good ethical practice is frequently intuitive rather than reflective. Again, this criticism, pressed vigorously by MacIntyre, and implicit in the work of Williams and Baier, seems misguided. Ethical theories do demand critical reflection: on the whole, they demand that all agents reflect in some way at some time about the shape of their lives as a whole. 43 (That is not true of one version of utilitarianism, which requires reflection only for a utilitarian elite; but this inegalitarian and nontransparent character of deliberation seems to me, as it does to Williams, a grave defect in the view. 44) But no ethical theory requires explicit reflection before each ethical choice. Theories vary in the amount of latitude they give well-trained agents to trust their own dispositions and sentiments; this variation depends on the theory's account of the passions. Thus Kant, who believes that bad inclinations can be suppressed but not very much modified, requires more testing and reflecting of agents than does Aristotle, who thinks it reasonable to expect well-brought-up people to come to love the good that they pursue. But even Kant urges agents sometimes to trust to their sentiments: for example, by strengthening our responses of pity through visits to hospitals and sickrooms, we acquire responses that guide us well when attention to duty might prove insufficient.45 And Kant certainly thinks that people of good will need not 42 See M. Nussbaum, 'Introduction', Love's Knowledge: Essays on Philosophy and Literature (New York: Oxford University Press, 1990). 43 On this aspect of ancient ethical theories, see Julia Annas, The Morality of Happiness
(Oxford: Clarendon Press, 1994). 44 S ee Williams, 'Point of View of the Universe'. Williams notes that neither act-utilitarianism nor R. M. Hare's version of utilitarianism raises this problem, though he objects to Hare's view on other grounds. 45 See Kant, Doctrine of Virtue, sect. 34, in Kant, Ethical Philosophy.
Why Practice needs Ethical Theory
247
work through Kant's moral theory every time they act; awareness of the idea of treating humanity as an end should infuse moral education very early, so that later in life the theory will serve to impart strength to a character that is already basically good. 6. Theories will not persuade bad people, and thus will prove impotent in practice. It is unclear what this objection is supposed to show. The fact that a medicine will not cure everyone is usually not taken as a point against using it to cure those whom it will cure. Most ethical theories expect to improve practice not by winning universal adherence, but by improving the average of practice, so to speak, and, especially, by attracting the attention of people, such as legislators and judges, who are in a position to do something about the people who are not reached by reason. Kant did not expect to end war by convincing people to love one another; he expected to prevail on the governments of republican states to bind themselves to certain conventions of international law. Nor is there any doubt that his theories have greatly influenced the development of modern international law." Similarly, feminist theorists do not expect to win universal support from the male sex for their proposals regarding rape, sexual harassment, and domestic violence. But they do not need to: they need, instead, to work on two fronts, by raising the level of awareness in the population as a whole, and by focusing especially on the makers of laws. Again, there is every reason to think that they have succeeded in changing things on these two fronts, in areas such as sexual harassment, marital rape, and domestic violence. (I shall give examples below.) More generally: in the history of western politics, the philosophical theories of the Enlightenment have proven highly influential in shaping constitutions and laws, and at the same time in informing the moral education of people generally. 47 Of course they have not persuaded everyone; but they have had a decisive impact even on the development of the religions within modern liberal democracies." But it is insufficient to rebut the objections to theory. For the attackers frequently make a further point: that theory is not necessary for the type of critical ethical reflection we need in our personal lives and in modern liberal society. Why, then, do we want explicitness, universality, abstractness, 46 See Matthias Lutz-Bachmann and James Bohmann (eds.), Frieden durch Recht: Kants Friedensidee and das Problem einer neuen Weltordnung (Frankfurt: Suhrkamp, 1996), English version as Perpetual Peace (Cambridge, Mass.: MIT Press, 1997); with articles by
Jurgen Habermas, Richard Falk, Axel Honneth, Thomas McCarthy, and others. 47 See my criticisms of Williams' paper 'Saint Just's Illusion', part of Making Sense of Humanity, in Ethics (forthcoming). 48 See John Rawls, Political Liberalism (New York: Columbia University Press, 1996).
248
Martha Nussbaum
systematicity, and a general account of how we may test our concrete judgements? In answering these questions I shall focus on two problematic areas of our lives that have, I believe, been profitably addressed by theory: aggression and world peace, and the nature of sex relations.
The Need for Ethical Theory
Antitheorists claim that the exchange of ethical criticism in daily life is sufficient to uncover distortions in our practices. In so suggesting, they employ a tacit picture of everyday life that we should question. Williams conveys the strong impression of thinking that when we do away with theory we will be left with people like Bernard Williams: they will lack philosophical theory, but they will still be energetically critical and self-critical, not captive to any other theory either, and alive to the possibility of distortion and hierarchy in the experiences that are the basis for their judgements. Life might then be like an Oxford common room in one of the more liberal colleges; or a Henry James novel with liberal politics thrown in. 49 Diamond,slrygethainbscofrmlaguentsw will all be reading novels and expanding our moral sensibilities. The first thing we should say, then, is that 'ordinary life' is not (or not only) a place of cultivated sentiment, critical exchange, and refined imagining. That is a very naive picture that even the antitheorists don't seriously endorse. Certainly Williams, at least, does not endorse it. Thus, although he makes a valuable point when he stresses that good ethical practice has a lot to offer theory, and can frequently be a better guide than the cruder sort of theory, it is somewhat odd that he is so confident that doing away with theory would leave good self-critical practice in the ascendancy in our daily lives. 'Ordinary life' is in fact filled with theories about conduct, some sophisticated and some extremely crude. Some of these are religious theories, some theories based upon convention and habit, some involving magic, astrology, and New-Age views of the psyche. In the absence of philosophical theory, people live their lives, to a great extent, in accordance with 49 There is room for much uncertainty here, since several figures prominently admired by Williams—the Greek tragic poets, Nietzsche, D. H. Lawrence—do not suggest at all the sensibility I describe. (Nor do they suggest liberal politics—although Williams plainly, and prominently, espouses liberal views.) The difficulty here comes from the fact that the sensibility expressed in Williams's text is to some extent at odds with the sensibilities of figures he praises; my attempt to characterize the former has possibly shortchanged the role of the latter in his thinking.
Why Practice needs Ethical Theory
249
unphilosophical theories, some of them very ill-considered and crude, many of them impeding the sensitive perception of individuals. Consider relations between men and women. People who don't think about the Kantian/Rawlsian theory that each human being should be treated as an end, or the Millian theory criticizing women's subjection to men, do not therefore simply go out and relate to one another in accordance with refined particular perceptions. Their mutual interactions are governed by a variety of theories, some metaphysical and religious, some customary. Stereotypes deriving from these theories prevent people from seeing what is before their eyes: thus, as Mill pointed out, we think we know all kinds of things about differences between the sexes, but when we put that knowledge to the test we discover that it rests on a totally inadequate foundation, one that we'd never accept in many other domains. What a philosophical theory like Mill's does is to open our eyes to the defects of these other theories, bringing sharply into focus the empirical and logical inadequacy of a part of our daily life. A vivid example of how it takes good theory to drive out bad is found in the history of the law of marital rape. For many centuries the conception of marriage dominant in the United States and most of the western world made marital rape a conceptual impossibility. The theory, which entered the law from custom and religion, was that man and wife are a single person; in this compound person, the husband was the possessor of rights and the wife was in his keeping. 50 As Sir Matthew Hale expressed the dominant theoretical view, 'The husband cannot be guilty of a rape committed by himself upon his lawful wife, for by their mutual consent and contract the wife hath given up herself in this kind unto her husband, which she cannot retract.' This theory was deeply internalized by many males in their daily conduct, and was used to rationalize bad behaviour. 51 The critical reflection of individuals and groups proved insufficient to dislodge this theory, deeply entrenched in our legal systems. It was only when feminist theory rebutted its presumptions with arguments and with an overall picture of women's dignity and autonomy that the legal system began to take notice. Between the 1950s and the present, the increasing influence of feminist theory (applying, we should note, insights gained from the tradition of 5(1 For the relevant texts, see Rebecca M. Ryan, 'The Sex Right: A Legal History of the Marital Rape Exemption', Law and Social Inquiry, 20 (1995), 941-1001. 51 See Ryan's discussion of John Galsworthy's 1908 novel, The Man of Property, in which Soames Forsyte calms his conscience by reflecting on his legal rights. The night of behavior that at first disturbed him becomes, as he reflects on theory, the 'night on which Soames at last asserted his rights and acted like a man'.
250
Martha Nussbaum
Kantian moral philosophy) caused perceptions to shift and eventually routed the other theories of spousal duty and right. They could not have done this had unease not been present—had thoughts about women's dignity and rights not already been on the scene, conflicting with the theoretical judgement that women are merely chattel. But it took the reach and systematic power of a counter-theory to put things in order in a new way. This theory ultimately had to become a legal theory—but it was of crucial importance that it was the sort of legal theory that was securely grounded in a moral theory of the Kantian type. 52 Consider, again, our lives with respect to war and peace. When Cicero, Seneca, Grotius, and Kant theorize philosophically about limitations on aggression and the hope for lasting peace, they are not operating in a sphere governed, otherwise, by fine-tuned Jamesian perceptions. Their opponents are customary theories of manly honour, of the pride of nations, of the inferiority and rightlessness of aliens. To go against these theories they felt the need to produce something theory-like, something with an argument to answer each consideration on the other side, something that could connect people's deeply held intuitions about human dignity to determinate practical and political considerations about aggression, showing them that their own customary theories of manly honour are to some extent in conflict with their own views about human dignity. Perhaps this would not have been necessary had ordinary life not been distorted by such theories. Then intuitions embodying sound judgements could have come to the fore without opposition and connected themselves with action. But bad theories were blocking those connections. Theorizing about the just war would have had no impact if it didn't tap an already existing fund of thought about the worth of human beings and the badness of doing certain things to them. But to get those thoughts to have power required derailing the bad theories that overlay and to some extent silenced those thoughts. The first thing theory needs to do for practice is, then, to defeat bad theories that silence important thoughts. But it is not only bad theories that silence our thoughts. Often, as Kant saw, it is our self-interested desires and passions, which pose a danger to practice even in a culture that is relatively 52 See now my 'Rage and Reason', in Nussbaum, Sex and Social Justice (New York: OUP, 1999), 240-52; and a review of Andrea Dworkin's Life and Death: Unapologetic Essays in the Continuing War Against Women, The New Republic, August 1997, where I stress the Kantian ethical origins of Dworkin and MacKinnon's legal proposals. See also my 'Objectification', Philosophy and Public Affairs, 24 (1995), 249-91. In the latter, I discuss the fact that Kant drew perverse and to some extent damaging conclusions from his own theory, and argue that it was not the theory but Kant's own misapplication of it that was to blame.
Why Practice needs Ethical Theory
251
free of bad theory. His argument in the Groundwork in favour of ethical theory goes as follows. If we could live our lives on the basis of the sound thoughts that (let us suppose) a sound culture has taught us, we might not need philosophy. This much Kant grants, stating that 'ordinary understanding in this practical case may have just as good a hope of hitting the mark as that which any philosopher may promise himself'. 53 But that is unlikely to happen, Kant argues. That is so because we find within ourselves many counterweights to the moral demands that we recognize as legitimate. These counterweights come from our selfish inclinations, our aggressive feelings, and so forth. (Kant thinks of these passions as innate and irremediable; his Stoic forebears think of them as culturally transmitted but ubiquitous. This difference means that for the Stoics, and for me, since I accept their view, the passions themselves are a kind of bad cultural theory, and this point will be a further application of my previous point.) When we feel these counterweights—when, for example, our deeply habitual conceptions of honour and personal prestige cause us to resent another person or group and to contemplate aggressive action against them, we may tell ourselves that we have good thoughts about the worth and dignity of human beings, the badness of treating human beings as mere means to our ends, and so forth. But what then happens, according to Kant, is that our passions begin to quibble with these good thoughts, telling us that it is ridiculous to think of dignity, when someone has just insulted our own, or unnecessary to think of women as ends, given that they are parts of our very own household. These quibbles then start to eat away at the good thoughts: 'Thereby are such laws corrupted in their very foundations and their whole dignity is destroyed—something which even ordinary practical reason cannot in the end call good.' (Such quibbling with the idea of equal worth misled Kant himself, clearly enough, as he vacillates oddly between ideas of equal human dignity that have proven highly fertile for feminist thought 54 and other ideas about females that treat them as chattel and deny their equal worth.) Kant is right: quibbling does eat away at the good thoughts. For instance, in the area of sex and marriage, as Mill noted, men who lord it over their wives allow thoughts about the equal worth of persons and their liberty (which he plausibly thinks most of his fellow citizens have) to be silenced Kant, Groundwork, trs. Ellington, 404. See Susan Moller Okin, Justice, Gender, and the Family (New York: Basic, 1989); and Barbara Herman, 'Can it be Worth Thinking about Kant on Sex and Marriage?' in L. Antony and C. Witt (eds.), A Mind of One's Own: Feminist Essays on Reason and Objectivity (Boulder: Westview, 1993). 53 54
252
Martha Nussbaum
by aggressive desires that give rise to counter-thoughts and rationalizations: I have a right to this exercise of power, she is mine, and so forth. Such rationalizations even underwrite marital rape as right and proper, although, as Mill noted, such an act is flatly incompatible with other thoughts the man is likely to have. By the time such a rationalizer rapes his wife, he feels that it is right and proper for him to do so. 55 Again, men who harass women in the workplace aren't in general evil through and through. Instead, they are likely to be self-indulgent rationalizers, people who have absorbed from society some pretty good thoughts about women's equality and dignity; but when their desires and their interest in power gain the upper hand, those good thoughts are silenced. If we remain at the level of untheoretical critical discourse and practice, as the antitheorists recommend, we will always be left, Kant plausibly claims in the Groundwork, with some good thoughts, corrupted by selfishness, aggressiveness, and urges to dominate. Even if not utterly silenced, these thoughts will not steer practice in a consistent manner. For this reason, ordinary judgement reaches beyond itself to seek the help of philosophy, asking to have the good thoughts laid out perspicuously and systematically, so that it will be clear ahead of time exactly what they entail in different areas of life. We get something to cling to, to look to when we are tempted, so that the self-deception of quibbling is less likely to prevail. Theory, then, can help our good judgement by giving us additional opposition to the bad influence of corrupt desires, judgements, and passions. One device theory uses in pursuit of this project is that of estrangement or defamiliarization. 56 Our judgements frequently feel so natural to us that it is hard for us to doubt them. And of course these intuitions are one part of the data that good theory will take seriously. But by asking us to look at the logical form of our judgements, and by urging us to describe them in an unfamiliar theoretical language, theory offers us a perspective on them that can be very valuable as we ask to what extent we have been engaging in self-interested rationalization. Just as Brecht famously urged the theatrical spectator to suspend identification with the theatrical characters and their 55 See E. Laumann, et al., The Social Organization of Sexuality, on the discrepancy in numbers between women who report being forced sexually and men who report using force—the discrepancy being especially prominent within marriage. 56 I owe this point to Scott Brewer, who develops it extremely well, in connection with the role of logic in moral and legal reasoning, in his excellent 'Traversing Homes's Path toward a Jurisprudence of Logical Form', in Steven J. Burton (ed.), The Path of the Law and Its Influence (Cambridge: CUP, 2000), 94-132. He appeals to Russian literary theory for clarification of the process; the Brecht analogy is my own, and one that I develop, with comparison to Stoic moral theory, in 'Poetry and the Passions: Two Stoic Views in J. Brunschwig and M. Nussbaum (eds.) Passions and Perceptions, (Cambridge: CUP, 1993).
Why Practice needs Ethical Theory
253
lives, in order to scrutinize the represented situation from a critical practical perspective, so good philosophy often gets us to do this with ourselves and our own lives. We look at the overall form of our judgements in ways we frequently don't, and we use the unfamiliar language of 'the kingdom of ends' or 'the categorical imperative' to test reactions we usually don't even scrutinize. Often this can help us overcome our tendency to rationalize by getting us to see relationships that had eluded us in our daily thinking. Thus the very detachment and remoteness in theory that antitheorists find problematic can serve a valuable practical function. Does all this mean that theory is necessarily hostile to the passions? Not in the least. Even Kant's theory is not generally hostile, as I have said: it asks us to cultivate our responses of pity or compassion. But it is inclined in that direction because of Kant's view that the passions represent an unthinking animal side of our personality and can never themselves be cultivated and made integral parts of a good character. If we were to select a theory such as Aristotle's, or Adam Smith's, according to which appropriate passions can be cultivated as parts of good character, then we would find many occasions on which ordinary thought and judgement could trust to the passions for guidance. (Indeed, once a theory is accepted, it can itself inform the structure of the passions, through self-criticism and especially through moral education.) But, as Aristotle and Smith show, the guidance of passion is never altogether trustworthy—no more than are the social norms that the passions embed, or any other belief we might use in moral reasoning. Therefore we need critical scrutiny of our passions, especially where we feel that we are likely to be selfish or shaped by untrustworthy social forces. So even an Aristotelian can agree with Kant's perceptive point: we are driven to theory by our unwillingness to trust ourselves in many of the most important matters. Kant imagines theory improving the practice of an ordinary individual. But he also envisages other roles for theory. Centrally, good theory shapes laws and institutions. Kant does not think that world peace will come about because all people will ultimately concur in testing their maxims by the categorical imperative. It will come about, if it does, because enough people think this way to make good laws that will constrain the behaviour of other people. And this is exactly how international law has evolved. The theories of Cicero and Grotius, developed by Kant, have had a formative influence on the conduct of governments and international agencies, so that by now, if we aren't exactly progressing toward a state of 'perpetual peace, we at least have in place many mechanisms to deal with egregious offences against human dignity. In particular, the world community has become
254
Martha Nussbaum
much more sensitive to the need to protest violations of basic human rights within nations, and that idea, first tentatively broached by Grotius citing Cicero, is now increasingly seen as a legitimate domain of international law. Bad behaviour is still ubiquitous; good theory prompts us to create institutional mechanisms to get at it. Once again, the antitheory position looks naive. Innocence, as Kant says, is a glorious thing. But the world community is not innocent, nor is it especially reflective and self-critical. It takes an explicit theory to generate a crusade like the modern human rights movement. Had Mill not been willing to make systematic and abstract and explicit claims about sex relations, his work would have had little impact on the shape of societies; by being a theory, it has by now influenced the structure of sex relations in nations as distant as China. Had Catharine MacKinnon made a series of concrete critical judgements, rather than articulating a theory that offered a systematic, explicit, and abstract account of the structure of sex relations, the very concept of sexual harassment would not have been forged. Women would have gone on having experiences of it, but without an abstract and systematic conceptual structure we would not have been able fully to name what we were experiencing. We do not live in the innocent world of orderly ethical practice that Kant admiringly and Seneca nostalgically imagine. We live, as they well know, in a world full of bad crude theories, self-serving passions, and tainted judgements, where the good passions and judgements need all the help they can get to prevail and even survive. That is why we need theory. As Seneca says, when our enemies take up elaborate weapons, we need to make the weapons on the side of good more elaborate and systematic to keep up with them. There are many obstacles in human life to this good role for theory, and to its practical success. Some of these obstacles are so deeply entrenched that we probably cannot ever get rid of them. As Seneca says of anger and hatred, 'Slow is the resistance to evils that are continuous and prolific.' But there is one obstacle that we can resist: it is the one Mill identified near the opening of The Subjection of Women, comparing his nineteenth-century culture to the eighteenth century: it is the tendency to mistrust argument and to prefer passion and intuition. It is always difficult, Mill says, to argue against deeply entrenched prejudice. But it is doubly so in an age that mistrusts Reason on principle, that substitutes Instinct for Reason and calls by the name of instinct 'everything which we find in ourselves for which we can trace no rational foundation. Contemporary antitheorists are not identical with the Romantic antitheorists whom Mill attacks. Williams, at
Why Practice needs Ethical Theory
255
least, does not prefer 'instinct' to reason, nor does he deny that we need some kind of rational scrutiny of what we find in ourselves. And yet there are tendencies in his thought that point in a Romantic direction, and his attack upon Enlightenment theories leaves Reason in a very reduced position. In a world in which irrational forces and their associated theories are increasingly getting the upper hand, it seems better to strengthen reason's bargaining position by allowing it to use all the resources of which it is capable. Innocence is indeed a glorious thing; but the real world, being guilty, needs philosophy.
11 Unprincipled Ethics David McNaughton and Piers Rawling
The term 'ethical particularist' has sometimes been used, in a broad and loose way, as a label for anyone who expresses hostility to the view that a decision about what we ought to do in some particular case can be mechanically 'read off' from a general moral principle or principles. Rather, it is urged, a correct moral verdict can only be reached by paying close attention to the individual case—to what differentiates it from other cases as much as what it has in common with them. As well as an understanding of the correct moral principles, we need fine judgement, sensitivity and even something approaching a perceptual capacity to appreciate the saliences of the circumstances in which we find ourselves. Particularism in this broad sense, which claims that a grasp of moral principles is insufficient for the correct moral appreciation of the particular case, has won many adherents in recent years. We will call this view, with which we agree, moral verdict particularism. It is a position explicitly held by intuitionists' (and no doubt by some other moral theorists) as can be seen from Rawls' classic definition: Intuitionist theories, then, have two features: first they consist of a plurality of first principles which may conflict to give contrary directives in particular types of cases: and second, they include no explicit method, no priority rules, for weighing these principles against one another. 2 The authors would like to thank Roger Crisp, Chris Daly, Jonathan Dancy, Andre Gallois, Eve Garrard, Brad Hooker, Philip Stratton-Lake, Stephen Wilkinson, members of the Philosophy Department at Manchester University, and participants at the NEH Institute, Ethics: Principles or Practices? And Piers Rawling would like to thank the University of Missouri Research Board and the University of Missouri-St Louis Research Awards for financial support. 1 We are not using the term 'intuitionism' to cover all moral theories that take the epistemological view that intuition tells us what is right. 2 J. Rawls, A Theory of Justice (Cambridge, Mass.: Harvard University Press, 1971), 34.
7
Unprincipled Ethics
257
But the term particularism is also used to refer to more controversial claims, one of which is the radical claim that there are no moral principles. It is not that general principles are insufficient to guide us in our consideration of the particular case—they simply do not exist. It is this more radical view we consider here. It arose, in part, as a response to Hare, who coined the term `particularist' .3 We shall examine this response to Hare, and go on to argue that it does not give us reason to abandon all conceptions of a moral principle (as Dancy sometimes does). 4 We urge a more moderate position that acknowledges the truth and importance of certain weak moral principles. 5
I Two Conceptions of a Moral Principle
As Hare defines particularism, 6 it is, at least in its `extremest' form, the rejection of the supervenience of the moral upon the non-moral—that is, the rejection of the notion that if two circumstances are alike in all nonmoral respects, then they must be alike in all moral respects. As Dancy uses the term, however, it means something different. Since Dancy sets up his position in conscious opposition to classical intuitionism, of the kind championed by W. D. Ross, we need to briefly expound Ross' notion of a prima facie or (better) pro tanto duty.' Suppose the fact that an act would be just is always a reason in favour of doing it. Then justice is a universally right-making feature of acts. Or, as Ross puts it, there is a prima facie duty to be just. This is not to say that all just acts are right (there can be occasions when justice is outweighed by other considerations), but, rather, the fact that an act would be just always counts for that act, deontically speaking. Similarly, if the fact that an act would be a betrayal of an innocent friend always counts against it, then the betrayal of innocent friends is universally wrong-making (which, to reiterate, is not to say that it is universally wrong). To employ a partial analogy with the chemical notion of valence, we have the following pair of principles: justice has a universally positive valence; whereas the betrayal of innocent friends has a universally negative valence. (A valence of zero R. M. Hare, Freedom and Reason (Oxford: Clarendon Press, 1963), 18. J. Dancy, Moral Reasons (Oxford: Blackwell, 1993). 5 For a classification of particularisms, see pp. 36-41 of R. Audi, 'Moderate Intuitionism and the Epistemology of Moral Judgement', Ethical Theory and Moral Practice, 1 (1998), 15-44. 6 Hare (Freedom and Reason, 18-19). W. D. Ross, The Right and the Good (Oxford: Clarendon Press, 1930). 3
4
258
D. McNaughton and P. Rawling
corresponds to moral irrelevance in the circumstance). Such principles, of course, are not sufficient for determining the deontic status of an act, since they may conflict (we will refer to such principles as weak). Dancy's particularism (which is clearly consistent with supervenience) combines moral verdict particularism (the claim that there are no absolute principles for determining the overall deontic status of an act) with the more startling claim that there are no weak principles (that is, there are no properties, apart from the thin moral properties right, wrong, etc., that have universally and counterfactually invariant valence). We dub this latter claim, with which we disagree, moral valence particularism. It is important to note that these claims are independent of one another. Hare's 'universal prescriptivism' commits him, among other things, to the `thesis of universalizability'. 8 Where 'M' abbreviates any moral term (such as 'good, 'right, 'just, or `courageous'), we have: (U) If an act or circumstance is M, then any relevantly similar act or circumstance is M (where relevant similarity is a matter of sharing some set of descriptive (i.e., non-moral) properties). When we judge an act, say, morally proscribed, we do so because it possesses non-moral properties, say, P i—Pn. And we are then rationally committed to the following strong moral principle: (P) Any action with non-moral properties P i—Pn is morally proscribed. Hare thinks that all moral judgements rest on principles of this form. Thus his position is inconsistent with moral verdict particularism. But this does not commit him to the view that P i (say) has invariant moral valence. Dancy sometimes speaks 9 as if universalizability commits one to thinking well of arguments such as the following. Suppose one accepts that a person's lack of direct benefits from the local schools (she has no children) is irrelevant to the issue of whether or not she should contribute to them, then (we have heard it argued) one must accept that a person's lack of direct benefits from the departmental coffee fund (she does not drink coffee) is irrelevant to the issue of whether she should contribute to it. But universalizability entails no such commitment: it is quite consistent with (U) to maintain that a lack of direct benefit can have varying valence. 8
R. M. Hare, The Language of Morals (Oxford: Clarendon Press, 1952) and Freedom and
Reason. 9
Dancy, Moral Reasons 88-90.
Unprincipled Ethics
259
It is only the extremest of anti-particularists who would maintain that the valence of all properties is invariant; and we doubt that there are any such extremists. It is certainly not a view held by the intuitionists. Urmson, having quoted Rawls' definition with approval, gives a second definition in terms of reasons. Some fact will be a primary reason for acting in a certain way if that fact's obtaining is always a reason for acting in that way, though not necessarily a sufficient reason. Some fact will be a secondary reason for acting in a certain way if that fact's obtaining brings about some fact which is a primary reason for acting in that way.10
If the fact that an act is just, to revert to our earlier example, is a primary reason for performing it, then justice has universal positive valence, and we have the weak principle: (J) In so far as an act is just, this counts in its favour. But there are plenty of examples of what Urmson would call secondary reasons—considerations with varying valence. The fact that some act would be illegal is often a reason against doing it, but we need not assume that the fact that an act is illegal will always count against doing it. It may be in virtue of other things that are generally true of illegal acts that the fact that an act is against the law is a strike against it. There might be special cases where its being illegal is morally neutral or even counts in favour of an act, as perhaps was true in the case of the Nuremberg laws in Nazi Germany. Or, to take a different example, that someone is strong can sometimes be a morally acceptable reason for giving them a job, and sometimes not. Justice may well be thought to require that we discriminate between job applicants only where the properties in question are relevant to the job. But which properties are relevant to the job will vary from case to case. Endorsement of (U) is consistent with sharing these judgements. To put matters in a different way: relevant similarity in Hare's sense can be interpreted holistically. That is, the valence of non-moral features in a principle such as (P) can be sensitive to context—P 1 in the context of P2–P a mightaven lc,butimghaveposlncither context. However, unlike Dancy, Hare does think that there are some terms (other than the thin moral terms) that have constant valence: the 1 () J. 0. Urmson, 'A Defence of Intuitionism', Proceedings of The Aristotelian Society 75, (1975), 112. The notion of bringing about some fact is not clear. We try to say a little more in connection with Ross' theory in part iii.
260
D. McNaughton and P. Rawling
`secondarily evaluative' terms, such as 'courageous'. These are, roughly speaking, coincident with Williams'thick' moral terms." And, modulo certain wrinkles that will appear in section iii, we concur with Hare that thick moral terms are univalent. However, the significance of this claim is very different on our view from its significance on Hare's. When we say that justice has universally positive valence, we mean that it always counts in favour of an act that it would be just; and that just acts are commendable in so far as they are just. Justice is an important feature that must be weighed up with the other considerations present in determining what to do. Hare simply means that to declare an act just is, on pain of violating the current prescriptive meaning of the term 'just, to commend it for being so—it is part of his project to avoid commitment to properties such as commendability. Indeed, Hare's model of morality is not one based on the notion of weighing reasons against one another. The weighing of reasons is a crucial aspect of intuitionism and other theories that endorse moral verdict particularism, and comports with our ethical intuitions. Reasons for action are not overriding, of course: there will typically be reasons for and against any action. On our view, (rational) agents confronted with (difficult) choices seek out reasons for and against their various options and weigh them up.12 Hare's model is very different. We noted above that the valence of any non-moral feature can vary from context to context on Hare's view. But the view is also consistent with non-moral features having no moral valence at all. The model simply has it that the bearer of some set of non-moral features has some particular moral property; it is no essential part of the model that the features have (context dependent) moral valences taken one by one in such a way as to facilitate the evaluation of their individual contributions to the whole. 11 B. Williams, Ethics and the Limits of Philosophy (London: Fontana Press/Collins, 1985), 129, 140. 2 In the case of reasons, not only is there no weighing algorithm, but, of course, agents can fail to know the facts that constitute their reasons (or know the facts and fail to realize that they are reasons). The relation between an agent's reasons and her obligations is, we think, complex—and we shall not discuss it further here, beyond noting the failure of a naïve equation of what an agent ought to do with what she has most reason to do. This is brought out by the following adaptation of an example due to H. A. Prichard (in 'Duty and Ignorance of Fact' in Moral Obligation and Duty and Interest (Oxford: Oxford University Press: 1968, 29). Do you have a reason to slow down before entering a main road? Legal issues and example-setting aside, if there is no traffic coming, you have no reason to slow down. Yet we concur with Prichard in thinking that you should slow down, even if there is no traffic coming.
Unprincipled Ethics
261
Among other things, the contribution of thick properties is rendered opaque. On our view it is helpful, in deciding whether to perform an act, to determine whether it would be, say, just. Hare, on the other hand, seems concerned only to look to non-moral properties in making such a decision. On his view, the following principle: (Q) Any action with non-moral properties Q i—Qm is just can tell you that an act is just, but it is only some set of the act's non-moral properties that feature in the determination of its rightness (and, as far as we can tell, this set may or may not include various of properties QI-QM). 13
There is a certain irony here: for the particularist (such as Dancy) who denies even that any morally thick properties have invariant valence, a similar issue arises. On this view, having determined that an act is just, we then have to ask the further question: does justice count for or against here, or is it morally irrelevant? So far we have classified Dancy's position as being opposed to classical intuitionism, as defined by Rawls. If intuitionism is understood in a rather wider sense, however, as the view that there is an irreducible multiplicity of morally relevant considerations that have to be weighed to reach a moral verdict, then even a radical particularism such as Dancy's can be classified as an intuitionist theory. It will be simpler, for purposes of exposition, to think of the question of whether there are properties with invariant moral valence as a family squabble within intuitionism. All intuitionists are moral verdict particularists, but they differ in their attitude to moral valence. We shall distinguish three positions. The first variety of intuitionism claims that there are non-trivial cases of universally and counterfactually invariant valence, and these all involve thick moral properties. We shall call this thick intuitionism." (We speak of counterfactual invariance because the 13 We have our doubts as to whether Hare, in The Language of Morals, can maintain his imperative model of prescription here. (Q) does not entail, of course, that any action with non-moral properties Ql–Qm has any particular thin moral property. Rather, the prescriptive meaning of 'just' presumably entails merely: 'In so far as an act is just, commend it' as distinct from 'Commend all just acts'. So we have the possibility of commending an act in so far as it is just, but not commending it overall. The relevant imperative here would appear to be something along the lines of: 'In so far as an act is just, do it'. But it is unclear what it would be to follow a command so qualified. It is, of course, not the conditional command: 'If an act is just, do it'. Rather one might be commanded to do the act in so far as it is just, but not do it overall—which makes little sense. 14 We have not the space here to discuss the issue of whether all morally thick concepts have invariant valence. Blackburn, 'Through Thick and Thin', Proceedings of The
262
D. McNaughton and P. Rawling
claim is not merely that a feature happens to have invariant valence in all actual acts, but also that it would count in the same direction in any act possessing it.) The straightforward cases are along the lines of principle (J). But there are less straightforward cases that we shall discuss in section m. The second variety of intuitionism adds some (but not all) non-moral properties to thick intuitionism's list of features with invariant valence. To mark this abundance we shall call this fat intuitionism. And the third variety of intuitionism makes a sweeping subtraction: it claims that the only cases of invariant valence are trivial, such as the claim that murder is wrong, if murder is construed as wrongful killing. On this view, which we shall call thin intuitionism, there are no primary moral reasons in Urmson's sense. This is Dancy's moral valence particularism. Contemporary particularism arose, in large part, as a response to Hare. We turn now to the arguments of McDowell and Dancy against Hare's conception of morality, and examine their impact on the cases for the varieties of intuitionism.
II Particularist Arguments
Perhaps the first question that arises for Hare's conception is that of how to justify an initial moral judgement. As we saw above, having once proscribed an act because it has non-moral features P l—Pn, (U) dictates that we proscribe any future act with those features. But, as Dancy points out, nothing is said about how to justify the initial proscription.15 And there is a further difficulty. 16 How are we to rule out the possibility of an act instantiating Pl—P n to the required degree together with some countervailing property, Pn+1 (a `defeater'), the combination of which requires us to alter our attitude to one of approbation? The obvious response to such an instance would be to add the absence of the countervailing property to P l—Pn to yield
Aristotelian Society, suppl. vol. 66, (1992), 285-99, for example, denies that 'lewd' is always
used with disapprobation—someone might declare: 'What we need here is an injection of lewdness into the proceedings.' And lewdness might be considered morally thick. We agree that the term is culturally encrusted, but we do not see it as having the stature of justice. Furthermore, the relevant issue for our purposes is not that of the usage of the term for purposes of expressing disapproval or otherwise. The issue is whether or not it counts against an act that it is lewd. 5 Dancy, Moral Reasons, 82-3. 16 Dancy, Moral Reasons, 80-1.
Unprincipled Ethics
263
(P') Any action with non-moral properties [P I–Pn and not Pn+I ] is morally proscribed. But this is a trivializing manoeuvre (as Hare realizes 17), unless there is reason to declare some stopping place: if all non-moral properties are included, then no two numerically distinct acts share all the relevant properties. And it is precisely the burden of Dancy's plaint here that no such stopping place can be determined in advance. Even if there are principles such as (P) that stop short of listing the entire supervenience base, we have yet to find any, let alone any that are of practical use. This argument supports all three types of intuitionism, since it attacks a vision of morality that all intuitionists reject (they are all moral verdict particularists). However, it also places a burden upon the advocates of fat or thick intuitionism. The thin intuitionist is radically holistic about moral reasons: she claims that the pertinence and force (i.e., both the sign and the magnitude of the valence) of any reason contributing to a moral judgement in a circumstance are dependent in part upon the other reasons there present. Thus the addition of P n+1 above might reverse proscription by modifying the valence of other factors. The challenge to the fat or to the thick intuitionist is: what reason have we to suppose that the valence of any consideration (non-moral or thick) is invariant across all possible cases? Why cannot the positive valence of a consideration in a particular case be reversed (or equal zero) in a new case? We will address this question further in sections in and iv, where we continue our argument for thick intuitionism. But we note here that we are not claiming that (say) justice has the same degree of relevance in all cases—it can vary in import depending upon other features of the case. But its import never falls to zero or turns negative. We consider now a pair of arguments due to McDowell.' 8 McDowell sees strongly principled views such as Hare's as attempting to ground moral judgement and justification in the following of rules, and draws upon Wittgenstein to argue that this strategy is wrong-headed. To take the standard example of the rule for adding 2, we declare that 1002 is the result of adding 2 to 1000; but what justifies this? Adverting to a further rule is otiose: we would then be required to justify the claim that this further rule Hare, Freedom and Reason, 12. J. McDowell Non-cognitivism and Rule-following' in S. Holtzman and C. Leich (eds.) Wittgenstein: To Follow A Rule (London: Routledge & Kegan Paul, 1981). Repr. in McDowell Mind, Value, and Reality (Cambridge, Mass.: Harvard University Press), 198-218. 17 18
D. McNaughton and P. Rawling
264
is being followed—and if justification must be grounded in rules, we clearly have a vicious regress. So, if the motivation for adverting to moral principles is to ground moral justification thereby, such views are ill-motivated: there is no such grounding to be had. Consider the question: 'How do I know that this act is morally proscribed?' In the case where the act is 'relevantly similar' to some previous act that was morally proscribed, Hare's answer is that there is a known strong principle of the form of (P) in force. This advertence to principle, on McDowell's interpretation of Hare, supposedly grounds the present judgement in apodictic fashion. But to make this grounding claim is to misconceive rules as antecedently present 'rails' extending from old judgements to new. (No sequence can dictate further entries.) According to McDowell, Hare supposes that strong moral principles are required in order to ground our moral judgements. But this they cannot do. There is a response to this challenge, however. The rule-following considerations do no more than undercut a motivation for the appeal to strong principles in ethics. Mathematics is principled in a strong sense— (ungrounded) justification in mathematics is rule-bound. Provided we have mastered the relevant subject matter, we can, in mathematics, derive theorems from axioms in accord with derivation rules (for example). And, according to Hare, we can derive and justify judgements (provided, again, we have mastered the subject matter) from strong moral principles and current conditions. There is the obvious problem of how we arrive at the first judgements of each type. But supposing this overcome, (ungrounded) moral justification would require judging in accord with the strong moral principles. According to the opponent of strong moral principles, of course, morality is not rule-bound in this way. But since the Wittgensteinian rulefollowing considerations apply to mathematics, care must be taken not to assume that they tell the whole story against moral principles.19 They tell against the possibility of ultimate grounding in both the mathematical and moral spheres. Indeed, if the motivation for Urmson's view that primary moral reasons have unvarying valence is to ground some notion of justification thereby, then this motivation too falls prey to them. But suppose one sees the dispute between Urmson (or Ross) and Hare simply as a matter of debating what (ungrounded) justification is in the moral sphere. On this account, they can both acknowledge Wittgensteinian 'vertigo, and we must measure the merits of their positions by other yardsticks. 19
Thanks to Bert Dreyfus for alerting us to this point.
Unprincipled Ethics
265
We turn now to another, related argument of McDowell's. Implicit in Hare's view (recall (Q) ) is the notion that our moral classifications coincide neatly with our non-moral classifications: acts can be classified into groups on non-moral criteria, each group being such that the same moral judgement applies to all its members. Against this McDowell takes another Wittgensteinian line: supervenience [of the moral on the non-moral, in the present case] leaves open this possibility ... however long a list we give of items to which a supervening term applies, described in terms of the level supervened upon, there may be no way, expressible at the level supervened upon, of grouping just such items together. Hence there need be no possibility of mastering, in a way that would enable one to go on to new cases, a term that is to function at the level supervened upon, but is to group together exactly the items to which competent users would apply the supervening term. Understanding why just those things belong together may essentially require understanding the supervening term. 20°
Suppose Hare were correct and we could codify moral judgements in the manner of: (Q) Any action with non-moral properties QI—Qm is just, there would be a list comprising arbitrarily many principles of this form: (Q) Any action with non-moral properties QI—Qm is just, (Q') Any action with non-moral properties Q' I—Qp' is just, (Q") Any action with non-moral properties Q" I—Qq" is just, and so on. In order to apply the term 'just, we would need to see what all the groupings [Q I —Qm ], [Q'I—Qp'], [Q"I —Qq"] and so on, have in common with one another, and be able to proceed into the future. Why, for instance, is [Ql—Qm] grouped with [Q' I—Qp']? We have seen that the possibility of defeaters gives us reason to doubt that there are (useful) principles such as (Q), but even if there were, what reason have we to suppose that the grouping of acts as just is transparent or makes any sense at the non-moral level? Why would we have one grouping rather than another if that were so? The point of the groupings is left obscure on Hare's view. (Compare the case of tin openers: at the non-functional level, it makes no sense to group together such a physically diverse set of objects under a single rubric.) 20
McDowell, Mind, Value and Reality, 202.
266
D. McNaughton and P. Rawling
As Dancy puts it: although the moral supervenes upon the natural, 'the subvenient base may be naturally shapeless'. 21 The onus is placed upon Hare to explain in what sense justice can have a 'shape' on his view, according to which it is a non-moral construct with a prescription contingently attached. 22 The thick intuitionist does not bear this burden so heavily. She can acknowledge a property of justice visible, because of its invariant valence, at the moral level. The valence is not a merely contingent attachment, it is an essential feature of justice. And this helps explain why justice shows up in the moral perspective. At least Hare does embrace the univalence of justice. The complaint is that he generates it in the fashion of an afterthought (it is as if we could group things into tin-openers and others without looking to the function of opening tins). The thin intuitionist denies this univalence, and this might yield its own set of problems when it comes to appreciating the `shape' of justice. Justice is a moral concept, and we suggest that understanding it, qua supervening term, requires an apprehension of its essential connection to the right. But the thin intuitionist denies that there is such an essential connection. With its variable valence, justice is in this respect, for her, on a par with the non-moral properties. Why, then, is justice shapely from, and only from, the moral perspective? The fat intuitionist has non-moral properties with invariant valences. And such invariance could lend more 'shape' to moral practice from a nonmoral perspective than McDowell's argument might lead us to expect. III The Thick, the Thin, and the Fat In order to examine further the relative strengths of the three varieties of intuitionism, it is helpful to have a model of thick intuitionism before us. We take as our starting point Ross' ethical theory. As Urmson rightly points out23 Ross' famous list of prima facie duties is intended as a list of primary reasons. 24 Ross' provisional list (he tries to reduce it yet further a little later in the chapter) contains duties of reparation, gratitude, fidelity, justice, self-improvement, beneficence, and non-maleficence. These are all terms for thick moral concepts; it is thus tempting to regard Ross as a thick intu21 Moral Reasons, 79. This is not strictly accurate. It is actually the set of instances of a moral property that may be shapeless at the natural level. 22 Hare, Freedom and Reason, 187-9. 23 Urmson, 'A Defence of Intuitionism', 113. 24 Ross, The Right and the Good, 20-22.
Unprincipled Ethics
267
itionist. If so, as Dancy readily admits, Ross' generalism is immune to the shapelessness argument. 'There is no suggestion in Ross that those who lack the relevant moral concepts should still be able to see the natural shape of those concepts.'25 Ross' theory, so understood, appears able to take on board the other two arguments employed in favour of particularism. As we have seen, the rule-following argument only undermines the notion that rules can provide apodictic justificatory grounding. The argument does nothing to show that there cannot be rules or general principles. Moral particularists often support their view by appeal to holism in the theory of reasons. Holism claims that reasons are context-dependent; what counts as a reason in one context may not count as a reason in another. Particularists hope to establish holism by appeal to particular examples. 26 But what these examples establish is only that there are many considerations whose valence can and does change with context. This is unsurprising and should, as we have already seen, be uncontentious. It is hard to see, however, how appeal to a few examples can establish that there are no considerations with unvarying valence. If one holds to the distinction between primary and secondary reasons then any convincing example of a switch in valence will merely be taken to show that the consideration in question is not a primary reason. The strategy of supplying examples is thus powerless in itself to show that there are no primary reasons. It can only show that the considerations in the examples are not primary reasons. To present the choice as being between holding that every consideration that can function as a reason is univalent and holding that none are would be to leave out of account an obvious and attractive range of middle positions. One of us has argued elsewhere 27 that Ross is best read as offering a theory of primary and secondary moral reasons, in which secondary reasons have force only in so far as it is their presence in the particular context that ensures the presence of the primary reason. Take the case of lying. That an act is a lie is, for Ross, only a secondary reason against it. Lying is normally prima facie wrong. But, in Ross' view, this is due to the fact that in standard cases lying will be both harmful and in breach of an implicit undertaking between members of a society to tell the truth. But we can imagine cases where this isn't so, such as the one Dancy offers us where I play Contraband Dancy (Moral Reasons, 95). E.g. Dancy (Moral Reasons, 60-62); D. McNaughton Moral Vision: An Introduction to Ethics (Oxford: Blackwell, 1988), 192-4. 27 D. McNaughton, 'An Unconnected Heap of Duties?' Philosophical Quarterly, 46, (1996), 433-47. 25
26
268
D. McNaughton and P. Rawling
with my children. Since the whole point and fun of the game is to lie, no harm is done and the implicit contract is suspended. In such cases, that a statement would be a lie does not tell at all against making it; indeed, given the purpose of the game, it may well be a reason for making it. There is nothing in Ross' account that rules out the possibility that any property of an act may be secondarily morally relevant on some occasion. But many of these features do not, in themselves, seem to have any moral significance. The beauty of Ross' account is that it explains how some features, such as lying, are ones that do normally have moral significance—because in all standard contexts they matter morally—while allowing that there can be special circumstances in which their normal relevance is cancelled. We develop below an account of thick intuitionism according to which non-moral features can enter weak moral principles, but only with evaluative riders attached. The evaluative riders lend moral shape to more commonly important non-moral features. 28 Thin intuitionism, by contrast, seems unable to offer any satisfying explanation for the fact that there are some non-moral properties that are more 'central' than others, and yet can still switch valence on occasion. 29 It explains neither why they do commonly count, nor why they don't count when they don't. How could a Ross-style thick intuitionism be shown to be unsustainable? It is not sufficient to show that Ross' particular list is in error, for he may have made a mistake of mere detail. What would have to be shown is that no sensible amendments to Ross could yield a plausible thick intuitionism that fitted well with our moral intuitions. Since a thick intuitionism occupies a midpoint between the thin and the fat, Ross' theory is vulnerable on both sides. On the one, there might be arguments to show that there are plausible fat principles. That is, it might turn out that there is some feature, on Ross' list or some successor, that is a plausible candidate for a primary reason and that can be wholly spelled out in non-evaluative terms. On the other, it might turn out that no plausible examples of primary reasons can be found, even when couched in evaluative terms. It might be that there just are no features that only count one way. Clearly, fully assessing both Ross' list and possible amendments to it is beyond the scope of this chapter. But by discussing a few examples we intend to bring out certain structural features of any plausible thick intuitionism. Some of the features on Ross' list appear to fit the thick model very well. It can plausibly be held that the fact that an act is just, for 28 The strategy we employ is, of course, not the same as the one attributed to Ross in the previous paragraph, but develops Ross' approach in a significantly different way. 29 Dancy, Moral Reasons, 67.
Unprincipled Ethics
269
instance, always counts in its favour. And justice is a thick moral concept whose content, it seems reasonable to assume, cannot be spelled out in descriptive terms because it is naturally shapeless. Certainly Ross' account of justice makes unabashed use of the ineliminably evaluative notion of merit, for he takes justice to require distribution according to desert. Non-maleficence is less straightforward. First, can one be injured as a result of fair competition? If no, then the concept of injury involves that of fairness, and is thick. If yes, then an evaluative rider is required: it is prima facie wrong to injure others unless in the context of fair competition. In any case, there are other grounds for thinking injury a morally thick notion. It is linked to (among other things) the idea of thwarting another's needs, where needs include morally thick aspects of a life. 30 Other members of the list are trickier still. Take the duty to keep promises. It is hard to see how promise-keeping could fail to be on any deontologist's list of prima facie duties. Yet it might well be claimed that, while promise-keeping is a notion with moral significance, it is not itself an evaluative notion, but one that can be spelled out in non-moral vocabulary.3 1 So it looks as if what we have here is a fat rule (a univalent non-moral feature). However, there are occasions on which the fact that I have promised carries no moral weight, gives me no reason to keep it. Suppose that I have promised to do something deeply immoral, such as a contract killing. Do I have some moral reason to carry out the killing in virtue of the promise, a reason that is outweighed by the duty of non-maleficence? It seems more plausible to think that here the promise counts for nothing, morally speaking. Or take the case of a promise extracted under severe duress, such as torture or threats. The duress does not just weaken my obligation; it cancels it. We suggest that a moral principle may draw attention to a feature that is always relevant, and relevant in the same way, under certain implicit conditions. 32 In our example, provided that one's promise is not given under duress and is not an undertaking to do something immoral, then it supplies one with a moral reason to act so as to keep it. We don't normally spell out these conditions when articulating the principle, both to keep it simple and because the principle is, as it were, written with the standard case in view. But the list of conditions is not open-ended, and it is knowable in advance. It has a rationale. The conditions are inherent in a proper understanding of Thanks to Berys Gaut for this suggestion. J. Searle, 'How to derive "ought" from "is"', Philosophical Review, 73 (1964), 43-58. Repr. in P. Foot (ed.) Theories of Ethics (Oxford: Oxford University Press, 1967), 101-14. 32 We are grateful to Tom Baldwin for this suggestion. 3() 31
270
D. McNaughton and P. Rawling
the nature and role of promising in the fabric of our moral life. Since promising is an institution for placing oneself under a moral obligation to perform an act, there would clearly be something self-defeating in allowing that one could use the institution to place oneself under a moral obligation to do an immoral act—an act that one had a moral obligation not to do. Similarly, if one could be coerced into putting oneself under an obligation, then those who acted immorally in acting coercively could put their victims in the position of being under an obligation to them, which is clearly morally objectionable. Note that we are back with thick intuitionism, because the conditions cannot be spelled out in purely non-moral terms. The notion of coercion here, for example, is not the mere use of force, for promises that are forced from one can still have moral weight. Once Milosevic had agreed to withdraw from Kosovo it would have been a breach of faith if he had attacked NATO troops, even though his promise was extracted by the use of force and the threat of further force. And we condemn those Argentinian soldiers who, during the Falklands war, are alleged to have indicated that they were surrendering and subsequently opened fire, even though force was used to make them surrender. It is, rather, unjust coercion (the use of force where the person using it has no right to do so) that invalidates the normal force of a promise, as does the use of trickery—another counter-instance that can be treated along similar lines. In the case of promising, the primary moral reason is that one made a promise meeting the relevant conditions: it is not a promise to do something immoral and it is not extracted under unjust duress. That an action would be the keeping of such a promise always counts in its favour. We have a univalent feature. But it is not purely non-moral—it relies upon the notions of immorality and injustice. Hence it does not fit the fat mould. And the list of conditions is not open-ended in the way thin intuitionism suggests. Even if our account of the conditions is not complete and someone were to come up with an ingenious counter-example requiring supplementation or amendment to the principle, we can be confident that the needed amendments would be variations on the kind of theme we have already seen. It is plausible to suppose that the concept of fidelity, which Ross uses as a blanket term to cover obligations of this type, is governed by the evaluative conditions we have spelled out. Someone who ignores a promise extracted from her under unjust duress is not committing a breach of faith. The notion of fidelity is thus, as one would have expected, a thick moral concept that cannot be fully explicated in purely non-moral terms. Indeed,
Unprincipled Ethics
271
there is a further respect in which fine evaluative judgement is required in order to decide whether there are reasons of fidelity for acting in a certain way. Ross rightly stresses that not all agreements or commitments are made by explicit promise. In many cases one comes to have a commitment in virtue of undertaking a series of actions; it is often a matter of moral judgement whether the way one has so far acted has placed one under a binding commitment that it would be a breach of faith not to honour. Other prima facie duties on Ross' list may need similar qualification if we are to discover a feature that always counts one way. For example, it may be that one need owe no duty of gratitude to someone who has helped you, but who has acted immorally in procuring you the benefit that he is now conferring. We suspect that whether you are beholden to your benefactor in such a case will depend on the precise circumstances—in particular, on his intentions. But it looks as if we need a similar exclusion clause to one of those we had in the case of promising, and for roughly the same reason. It is worth mentioning one complication. We have suggested that there are weak moral principles with implicit conditions. These principles spell out complex evaluative features that always (morally) count one way. But our original exposition (following Urmson) was in terms of primary reasons. The original idea was that there were features of actions that were univalent reasons for acting. But the univalent feature that constitutes the reason here turns out not to be 'having made a promise' but 'having made a promise that was neither a promise to do something immoral nor extracted under unjust duress'. Dancy has urged that this is not the right way to think of reasons. 33 I have reason to return your book today, because I promised to do so. It is true that, had the promise been extracted under duress, or had my returning the book been an immoral thing to do, then I would not have had such a reason. But that does not mean that the absence of such defeaters is itself part of the reason why I should return it. Similar remarks apply if we construe prima facie duties in terms of right-making properties. What makes it (prima facie) right to return the book is that I promised. It is not part of what makes it right that my promise was not extracted under duress, though if it had been, the returning of it would not have been (prima facie) required of me. There are two ways of responding to this objection. The first is to tough it out. To say that I promised is to give the most important part, but only a part, of the full reason, which should include the absence of the exclusionary circumstances. Though it sounds odd to say that the fact that the 33
Dancy, Moral Reasons, 77.
272
D. McNaughton and P. Rawling
promise was not coerced is part of the reason why it is right to return the book, that oddity is to be explained in terms of the normal conventions of conversation and thought. It is not metaphysically odd or redundant. The other response is to admit the force of Dancy's objection, and to reformulate thick intuitionism accordingly. We might say, for example, that the fact that I have promised does not always provide a reason for acting, but always does so whenever certain specifiable conditions are met. This would be to give up on the idea that there are univalent reasons here, but this only commits us to the letter of thin intuitionism in this instance, not its spirit, which holds that there is no way of specifying in advance how the valence of promising will alter from one new context to another. IV In Defence of the Thick
So far we have tried to show that there are some reasons for thinking that there might be plausible weak moral principles involving thick concepts, and none involving non-moral concepts. Without a good deal more work we could not show that a fully articulated and plausible thick intuitionist moral theory can be constructed. What we hope to do in this final section is to give further reasons for preferring a thick to a thin theory. A thin intuitionist might claim that the flexibility of her view accords better with our moral intuitions. But it is not as though thick intuitionism is inflexible: whilst the sign of the valence of the thick concepts cannot reverse or fall to zero on this view, the magnitude of the valence can vary widely. In the case of promise-breaking (where the promise meets the relevant conditions) the valence is always negative, but the magnitude depends on such factors as the content of the promise, to whom it was made, and the circumstances under which it was broken. Suppose someone goes too far in the keeping of a promise, and does the wrong thing thereby—she should have broken the promise in order to save a life. Does the valence of promise-breaking reverse in such a case? No, it is simply that its magnitude is outweighed by another consideration. Hare points out that the term 'industrious' is 'normally used to commend; but we can say, without any hint of irony . . . "too industrious".'34 Is this a case of the valence of 'industrious' reversing, or is it a case in which industriousness has been carried too far in relation to other considerations? We suspect the
34
Hare, The Language of Morals, 121.
Unprincipled Ethics
273
latter. And we are even more convinced in the central cases, such as that of someone being 'too just'. Justice always counts in favour, but it must be weighed against other considerations. It might be possible for an act to be the worse for being just. But this is not a case of valence reversal; it simply means that the act was wrong, and the right act was unjust or less just. (It might also be a remark about the motivations of the agent, but that discussion would take us too far afield.) Thin intuitionism does not appear to do justice to the role of thick moral concepts in our moral thinking. Recall Dancy's challenge to the fat or thick intuitionist: what reason have we to suppose that the valence of any consideration (non-moral or thick) is invariant across all possible cases? Why cannot the valence of a consideration in a particular case be reversed (or equal zero) in a new case? Our view is that, when it comes to certain thick concepts, the onus is on the thin intuitionist to justify her claim that the valence can vary—it is a weakness of the position that its radical holism marks no distinction between the thick and the non-moral. There are a number of thick concepts, roughly those corresponding to the traditional virtues and vices, that seem to play a crucial role in moral justification. These moral concepts mediate between the non-moral and the thinly moral. Part of moral wisdom is the ability to locate an act in terms of all the thick properties under which it falls. Moral judgement is then aided by the fact that we know the valences of the thick properties—we know on which side of the scale to weigh them. We have already mentioned justice, fidelity and gratitude, but there are many other examples. That an act is cruel, mean, or dishonest counts against it; that it is kind, generous, or honest counts in its favour. On thin intuitionism, thick moral properties have no more intrinsic moral significance than non-moral properties. It will, presumably, turn out that these properties are 'commonly more important' than some others (although thin intuitionism owes us an account of why), but that not only understates their force, it seems to mislocate their centrality. It is not just that it is helpful to look at them first because they often count; their counting is central to their being thick moral concepts. The valences of thick properties are discovered ontogenetically as the concepts are acquired: learning about rightness and acquiring an understanding of the thick moral concepts occur in tandem, and are mutually reinforcing. Learning about justice, for example, deepens the understanding of rightness. And learning about rightness deepens the understanding of justice: we see how justice and its lack feed into rightness and wrongness. To fail to see that justice has invariant positive valence is to fail to understand fully the relevant concepts, where invariance is a modal notion
274
D. McNaughton and P. Rawling
because, as we have seen, it applies counterfactually—of any act, if it were just, that would count in its favour. Davidson, initially in response to Goodman, utilizes the notion of 'fit' between predicates. 35 For example, he sees a lack of fit between the mental and the physical. In somewhat similar fashion, we might recast McDowell's shapelessness claim as the claim that there is a lack of fit between the moral and the non-moral predicates we use. And the thick intuitionist claims that there is a degree of fit between thick and thin moral properties—the fit here being responsible for the modal fact of invariant valence. The thick intuitionist gives us the best account of moral modality. 36 Dancy wants to make room for an element of modality in the moral domain.37 He attempts to account for this as follows. 38 What we learn in a particular case, when we notice a feature making a moral difference, is that it is possible for the feature in question to make a difference. And in certain modal logics (e.g., S5) the following holds: if possibly P then necessarily possibly P. So we have necessity in the picture. However, we are not convinced that this is a satisfactory account of moral modality—it is too easily paralleled in low-level non-moral cases. My car does not start, and I learn that this is due to water in the distributor. So I learn that water in the distributor can make a difference to whether my car starts. Hence I learn that necessarily it is possible for water in the distributor to make a difference to whether my car starts. The fat intuitionist denies shapelessness and claims that there is a degree of fit between some non-moral features and the thin moral properties. And she also owes us an explanation. What other work do weak principles do for the thick intuitionist? Do they, for instance, help us see what is relevant in a new case? On Ross' view, I come to apprehend the prima facie moral principles by a process of intuitive induction; that is, I notice that one or more particular acts are prima facie right in being (say) just, and then come to realize that there is a selfevident necessary general truth here. Since, however, I could only come to recognize the general principle because I am capable of recognizing 35 D. Davidson, `Emeroses by other names', Journal of Philosophy, 63 (1966), 778-80. Repr. in Essays on Actions and Events (Oxford: Clarendon Press, 1980), 225-8. Davidson, `Mental Events', in L. Foster and J. W. Swanson (eds.), Experience and Theory (University of Massachusetts Press and Duckworth, 1970). Repr. in Essays on Actions and Events, 207-24. 36 A lot more needs to be said here of course, concerning such tricky matters as scientific `laws' and their support of counterfactuals but space does not permit. Dancy, Moral Reasons, 69-70. 38 Dancy, Moral Reasons, 70, and n. 11,72.
Unprincipled Ethics
275
straight off that an action is prima facie right in virtue of being just, I do not use this general moral principle to infer that another just act is prima facie right—I can see this straight off, on normal occasions, just as I could in the first cases. 39 Dancy concludes from this that 'moral principles play no epistemological role'. 40 However, first, if one admits, as Ross does, that there are exceptional circumstances in which one 'apprehends individual facts by deduction from general principles' 41 then it is an exaggeration to say that principles play no epistemological role. And, second, Dancy's argument seems to assume that the only epistemological role a principle could play is by acting as a major premise in a deductive inference. But it can inform what we are able to see `directly' (this is part of the point of the discussion of shape). Cases vary in complexity and obscurity. It may be that it is only when we have grasped the general principle that we can see in less straightforward cases both that an action is just (among other things, we know that this is something to look for) and that it is prima facie right in virtue of being so. Weak principles help us apprehend what to do; they help us capture relevant features and help us see how they are relevant. Weak principles also help us to justify our actions, even in tricky cases. It may be unclear whether the fact that I have promised does make a difference. Where there is genuine uncertainty, it is helpful to be able to appeal to the principle of fidelity and to reflect on the kinds of case that constitute the exceptions. Only conditions of a certain kind are allowed to disqualify the making of a promise from having its normal force. While we still have to judge whether this case falls under the general rubric, having a general rubric helps us to justify our action in addition to alerting us to look out for the relevant conditions. Whilst we favour thick intuitionism over its fat and thin rivals, we hope to have done justice also to their common underlying conception of morality. Moral judgements cannot be 'read off' strong principles. Rather, factors must be assessed in situ to determine their role, if any, as reasons. And these must then be weighed. 3-9 4()
W. D. Ross, Foundations of Ethics (Oxford: Clarendon Press, 1939) 168-71. Dancy, Moral Reasons, 95. 41 Ross, Foundations of Ethics, 171.
12 Moral Generalities Revisited Margaret Olivia Little
I
A great deal of discussion in ethical theory has recently centred on the particularity of moral situations. In some cases, the point of discussion has been to emphasize how important it is to attend to the nuances and contextual details of a case before applying any moral norms, rules, or principles to it. But in a number of discussions, the point has been to question the idea that there are any moral norms, rules, or principles capable of codifying the moral landscape. Theorists from a variety of camps, including neo-Aristotelians and a certain brand of British moral realist, have urged that moral answers resist capture in any general formulae.1 The situations we confront are often saturated with unique combinations of morally salient features; no principles, however subtle or complicated, provide an adequate guide or model of how we should navigate through them. Two questions naturally present themselves on hearing of such a position. First, why on earth would one hold such a view; second, just how radMany thanks for useful and enjoyable background discussions to Alisa Carse, Jonathan Dancy, David DeGrazia, Jon Faust, David McNaughton, Amelie Rorty, Nancy Sherman, and Michael Smith. I'm especially indebted to my colleague Mark Lance, who has influenced my views at a number of points, and the members of the graduate seminar on particularism I co-taught with him at Georgetown. Sarah Buss, Madison Powers, and Brad Hooker provided helpful comments on an early draft of this work. Nascent versions of these thoughts were presented at the Kennedy Institute of Ethics, the philosophy departments at Keele University and the University of North Carolina at Chapel Hill, to the University of Maryland-Georgetown and the Philamore reading groups, and to the British Society for Ethical Theory. My sincere thanks to all who joined the discussions. 1 The classics here are John McDowell, 'Virtue and Reason', The Monist, 62 (1979), 331-50; and Non-Cognitivism and Rule Following', in Steven Holtzman and Christopher Leich (eds.), Wittgenstein: To Follow a Rule (New York: Routledge, 1981), 141-62; Jonathan Dancy, Moral Reasons (Oxford: Blackwell, 1993); and David McNaughton, Moral Vision (Oxford: Blackwell, 1988).
Moral Generalities Revisited
277
ical is the doctrine's rejection of moral generality? These two questions, it turns out, are intimately related, for the implications of the doctrine flow, in large part, from the model or argument that underlies it. In this chapter, I want to explore just why one might be motivated to accept moral particularism; more importantly, I want to articulate what does and does not follow from the doctrine. I first set out why we have reason to doubt that there are any moral principles, even very complicated ones, capable of codifying the moral landscape. The claim is not that such generalities are impossible, but that we have reason not to expect any: any we might come across would be, as it were, philosophically serendipitous. Particularists, of course, are often accused of forgetting to provide any argument for their position, contenting themselves with simply announcing their pessimism about the existence of adequate principles or with constructing ever more imaginative counterexamples to proposed principles—a strategy which, clearly, does not arbitrate the contested ground between rejecting principles and acknowledging that they are highly complex. To understand the pessimism, though, we have to explore the model that backs it—a model which, in essence, rejects the principlist's equally stubborn optimism that there must be such principles. I then explore just what kind of doctrine follows from the argument so laid out. I want to urge that it is both more and less radical than various parties have tended to appreciate. Some have accepted particularism almost blithely, their confidence trading on the belief that the doctrine, while important, is limited in scope: some codifiable law-like generalities remain; they are simply fewer or less ambitious than previously thought. On this prevalent interpretation, particularism is simply an articulation of the important but by now well-worn lesson that there is no exhaustive set of hierarchically arranged, deductive principles capable of doing our work for us, as it were; there are, none the less, some codifiable principles, perhaps tempered by pro tanto or ceteris paribus clauses, that can still play some of the justificatory and theoretical roles traditionally ascribed to principles. In contrast to this modest interpretation, I argue that the reasoning behind particularism grounds a more radical metaphysical claim: we have reason to believe there are no codifiable law-like moral generalities whatsoever, even those tempered to seemingly innocuous forms. Others, in contrast, take an alarmist view of particularism's implications. They believe that, properly appreciated, the argument backing particularism means that moral epistemology would have to proceed without any reference to generality. In contrast, I argue that, the above claim notwithstanding, moral generalities play an indispensable role beyond admittedly
278
Margaret Olivia Little
important heuristic and pedagogic functions. 2 There are aspects of our moral epistemic lives that necessarily look beyond the particular to invoke moral patterns, indeed law-like patterns, to carry justificatory weight. But nothing in the argument for particularism requires us to renounce such reliance. Indeed, I argue, particularists who believe otherwise are still in the grip of a picture their own argument is designed to dispel.
II
A deeply influential theme in the last few centuries of moral philosophy holds that ethical inquiry is the search for the architecture of morality. It is a search for the pattern that lies behind specific cases; but more than that, it is the search for generalizations that hold, not as mere accidents (as when we realize that Dora is usually cruel to Jack), but with a level of necessity and systematization to reveal the very nature or structure of the moral landscape. Just as scientists try to parse out how the various forces in physics interact systematically, moral theorists try to capture how moral considerations so identified, such as the requirements of justice and beneficence, are ordered in relation to each other. And again, just as scientists work to unearth bridge laws linking, say, the property of temperature to the property of mean kinetic molecular energy, the job of moral theorists is to identify which natural properties make an action count as just or beneficent. Moral particularists have a very different picture of morality. Without questioning that there are moral answers, they question that those answers are, or need to be, backed by any architecture, that those answers constitute pieces that fit together to form a structure with recoverable law-like relations. To be sure, no one (sensibly) rejects principles that tell us to 'respect autonomy' or to be kind. But the particularist denies that we can unpack those very abstract principles into generalizations that are both accurate and contentful enough to be action-guiding. Particularism denies that we can usefully specify how the demands of kindness and those of autonomy weigh up, or which nonmoral features suffice to make an action kind in the first place. 3 2 Things need not be propositions, much less true propositions, to be pedagogic and heuristic tools. After all, a pedagogic tool is anything that helps to educate; a heuristic anything that helps one to discern or interpret something in a situation. Reading an inspirational novel can teach, and turning one's head sideways (not to mention drinking a beer) can help one see the duck in the duck–rabbit figure. 3 One can, of course, agree with the first piece of this without agreeing to the second. It is fairly common these days to find theorists arguing that relations amongst moral considerations, so identified, cannot be codified—that morality, unlike arithmetic, say, is not
Moral Generalities Revisited
279
That's the intuition, though, not an argument. Obviously, to defend particularism, it is not enough to keep offering counterexamples to proposed principles. Even if they are accepted, just what they suffice to show is precisely what is in question—those attuned to the richness of morality but loyal to the existence of principles will see counterexamples as evidence of complexity, not of irreducible complexity. (Besides, there's something not a little farcical about measuring dialectic success in terms of who can outlast whom—those who want to refine the principles or those who want to find exceptions.) But the particularist's doubt does not stem from philosophical obsession with counterexamples or lazy extrapolation from them; it is not brute pessimism floating free of any other philosophical commitment. The particularist doubt is born of reflection about the nature of the moral domain. Let's take a look. In denying that we can codify the connection between moral and nonmoral properties, the model backing particularism clearly belongs in the non-naturalist camp. It shares the intuition that moral properties are, to use Simon Blackburn's felicitous phrase, 'shapeless' with respect to the nonmoral. 4 Now one reason for advancing a doctrine of shapelessness is the belief that we cannot mark out the boundaries of moral concepts in purely nonmoral terms: the items grouped together under a moral classification such as 'cruel' do not form a kind recognizable as such at the natural level. The thought here, familiarly, is that, of the infinitely many ways of being cruel—kicking a dog, teasing a sensitive person, and forgetting to invite someone to a party might each qualify—there is no saying what they have in common (and why, say, the pain inflicted during a spinal tap is different) except by helping oneself to the moral concept of 'cruelty'. This is to believe non-naturalism because one denies there are any usefully, finitelyspecifiable conditionals of the form If M then N'. But the particularist, while she may well agree with this, comes to nonnaturalism from what is arguably a more radical—and more interesting— an internally codifiable system. W. D. Ross, The Right and the Good (New York: OUP, 1930) is the classic here. I'm reserving the 'particularist' label for those who deny codification at both levels—a position, as we shall see, that is much more radical. In this chapter, I concentrate on discussing anti-codification at the moral to nonmoral level. Though I do not pursue the point here, the model described below also helps to make sense of denying that moral considerations, so identified, are codifiable. For more on this connection, see Margaret Olivia Little, `Wittgensteinian Lessons on Particularism', in Carl Elliot (ed.), Wittgensteinian Bioethics (Duke University Press, forthcoming). 4 Simon Blackburn, 'Reply: Rule-Following and Moral Realism', in Steven Holtzman and Christopher Leich (eds.), Wittgenstein: To Follow a Rule (New York: Routledge, 1981), 163-87. For Blackburn's latest intriguing thoughts about shapelessness and its implications, see his Ruling Passions (Oxford: Clarendon Press, 1998), ch. 4.
280
Margaret Olivia Little
route: she denies that we have reason to hope for any usefully, finitelyspecifiable conditionals of the sort 'If N them M'. Whether the examples just cited themselves qualify as cruel depends irreducibly, she urges, on the contexts in which they are situated. A set of features that in one context makes an action cruel can in another carry no such import; the addition of another detail can change the meaning of the whole. Indeed, the very valence' of a feature's contribution is context-dependent. The fact that an action is fun, to give an example cited by Jonathan Dancy, often counts in its moral favour, but at other times may be precisely what makes it morally problematic: that the sadist enjoys inflicting pain seems precisely what is wrong with the action, and not the 'moral silver lining' of the situation. 5 The point is not to deny that natural features serve as 'good- and badmaking' properties. When classifying an action as cruel or just, we certainly regard the moral status as obtaining 'in virtue' of certain of its nonmoral features: those natural features are what make the action cruel, are the reason it is kind. The point, rather, is to deny that such considerations carry their reason-giving force atomistically. 6 Natural features do not always ground the same moral import, which then goes in the hopper to be weighed against whatever other independent factors happen to be present. The moral contribution they make on each occasion is holistically determined: it is itself dependent, in a way that escapes useful or finite articulation, on what other nonmoral features are present or absent. It isn't just that we haven't bothered to fill in the background conditions because they are so complex—holism is not complicated atomism. The claim, rather, is that there is no cashing out in finite or helpful propositional form the context on which the moral meaning depends.' Natural features carry their contribution to an action's moral status in the way that a given dab of paint on the canvas carries its contribution to the aesthetic status of a painting: the bold stroke of red that helps balance one painting would be the ruin of another; and there is no way to specify in non-aesthetic terms the conditions in which it will help and the conditions in which it will detract. 8 Just so, whether a given feature counts as any moral reason at all—and if so, with what valence—is itself irreducibly dependent on the background context. This is not to deny that the moral supervenes on the nonmoral. Two situations, it's agreed, cannot differ in some moral respect without differing Dancy, Moral Reasons, 61. (Dancy attributes the example to Roy Hattersley.) This is the central point of Dancy, Moral Reasons. 7 See section III for an explanation of the gloss here on 'finite or helpful'. 8 This nice example is David McNaughton's (from conversation). 6
Moral Generalities Revisited
281
in some nonmoral respect. But this can be so freely admitted because supervenience is so weak: nothing in the doctrine implies that a given moral difference (say, the difference between being just and unjust) need always be found in the same nonmoral difference. 9 That is, while the doctrine of supervenience entails that the presence of a moral difference must be accompanied by a nonmoral difference, it does not entail that there are any interesting or helpful patterns to the ways in which the two sorts of differences line up. (We'll be returning momentarily to residual concerns about supervenience.) While the moral properties of actions, then, are in some sense determined by their natural features, there is no pattern discernible outside the evaluative practice to how those individual determinations add up. Particularism thus argues that there is no one moral import a given set of nonmoral features must carry, for the import depends, in a way that cannot be cashed out, on the background context. This notion raises suspicions for some: it strikes them as an oddly radical feature of the moral domain, once again confirming morality's dubious character. But the idea of particularism grounded in holism is a familiar—indeed a downright homey—notion in post-positivist philosophy. It is present unproblematically in a number of domains. We've already seen its presence in aesthetics, and it seems an inescapable feature of assessing matters of value more broadly. When the presence of rain adds to or detracts from a day depends in a way that escapes articulation on the context: it can add to the day if one is in a contemplative mood in front of a roaring fire, detract if one is setting out on a picnic. Perhaps most telling for those who regard holism in morality as suspicious, though, such holism is a completely familiar feature of epistemic justification. A central lesson from Quine, inspired by Duhem and others, is that beliefs and experiences do not carry their justificatory import atomistically. The fact that one has a perceptual experience as of seeing a table can be excellent reason to conclude that there is such a table, but in other contexts it will count as excellent reason against drawing such a conclusion—as when you have just helped yourself to a hefty psychotropic drug that has induced hallucinations of tables in the past. There is no way to codify the conditions under which an experience as of seeing a table is evidence for there being a table. Evidential justification is an irreducibly contextual enterprise—a point still sometimes missed because, as Gil Harman puts it, we tend to confuse rules of inference with the rules of implication. 10 The 9 10
See McDowell, Non-Cognitivism and Rule Following'. Gilbert Harman, Change in View (Cambridge: MIT Press, 1986), 3-4.
282
Margaret Olivia Little
former must be holistic to accommodate the background of beliefs we bring to a given epistemic situation; the latter are context-independent rules of logical relations amongst propositions. This should ease concern, persistent in some corners, that particularism born of holism renders morality unacceptably subjective. If the particularist is right, someone who gazes upon the natural world will find that some things get clustered together by us as 'cruel' and others excluded, but according to no pattern that can be seen at that natural level. How then, one might worry, is there any measure of consistency; how could we ever come to understand what belongs to the category and what does not? These intuitions, though, reflect a cluster of biases. It is a falsely narrow notion of consistency that counts us as going on in the same way only if the measure of sameness can be found at the natural level." Such a notion privileges a subset of our commitments, namely those from science, as the judge of all others; it engages in the pejorative sense of 'metaphysics', in which a substantive preconception of what facts and objectivity look like constrains ahead of time our view of what the world contains. (It is one thing to say, if we do, that science is objective, and quite another to say that science is the exclusive arbiter of objectivity.) The fact that the category of cruelty has no shape at the natural level does not mean it has no objective shape: it has the shape, precisely, of cruelty. Likewise, the proper measure of consistency is that one calls 'cruel' those things, and only those things, that are cruel.' 2 Of course, the particularist claims not just that the conditions of something being cruel can't be spelled out in natural terms, but that they can't (except trivially) be spelled out at all. Against one traditional philosophy of mind, this will seem worrisome: how can we explain what leads us to count certain instances but not others as cruel—and in a way that points to anything shared—if we cannot articulate the criteria of application? But the model is just that—a model—and it's one the particularist, citing Wittgensteinian reasons, urges us to reject. One of Wittgenstein's most cen11 The point has been made—and still ignored—before. For clear discussions, see McDowell, 'Virtue and Reason'; McNaughton, Moral Vision, 60-2, 192-4; and Dancy, Moral Reasons, 82-6. 12 It is obviously possible that there is in fact no such property, and in this case it will turn out that there is no objective measure by which our categorizations count as consistent. The point, though, is that this must be settled by open-minded investigation, with terms appropriate to the subject-matter, not by appeal to some philosophically driven picture of what facts and proof must look like. Just so long as our commitments to a moral property survive reflective scrutiny, we have available a criterion of consistency in our moral categorizations.
Moral Generalities Revisited
283
tral points is that we can share things, such as understandings, skills, and practices, that outstrip finite sets of propositions. If we feel that such a picture leaves an unsettling lacuna in our psychological explanations, we are reminded that it is fantasy to regard the lacuna as filled where codification is available. (After all, even with mathematical rules such as 'add two, our previous behaviour is consistent with an infinite number of rules; the idea that we possess something, such as a Platonic concept, that guarantees for us the trajectory of moves is an illusion.) In the end, we can explain why we make the moves we appropriately do only by saying that we share a practice.13 If we start to wonder how someone could ever come to 'catch on' to a rule whose shape can't be cashed out, we should remind ourselves that this question is generic as well. 14 How, Wittgenstein asks, can we ever come to catch on to a given rule from a finite set of examples, given that the examples we are shown could logically have been the products of an infinite number of rules? His answer, of course, is not to retreat into scepticism, it is to emphasize that we can, as members in a 'whirl of organism', outstrip the conditions of learning. Thus with morality, it is certainly true that we will come to understand a moral concept such as fidelity by reference to certain paradigmatic examples, such as intentionally told falsehoods. But this is just to say that we learned to become competent with the concept under circumstances in which the most easily accessible breaches of fidelity happened to be actual or mythic cases of intentional falsehoods. Once we have come to 'catch on' to the concept, though, we are able to discern the very different shape fidelity and its breaches take in different contexts. To think we cannot is to confuse the conditions of learning with the content of what is learned. An important pattern emerges if we step back from the examples just adduced. The examples—from morals, aesthetics, practical reason, and epistemology—all involve evaluative or normative concepts. I suspect that holism is a unifying feature of the evaluative. I suspect, that is, that there is no way of cashing out propositionally the ways in which nonevaluative properties contribute to the evaluative natures of situations, actions, characters. (Those persuaded by Davidsonian considerations about 13 Helpful discussions of this point can be found in Robert Fogelin, Wittgenstein (New York: Routledge, 1987); McDowell, Non-Cognitivism and Rule Following'; and Hubert Dreyfus and Stuart Dreyfus, 'What is Moral Maturity? Towards a Phenomenology of Ethical Expertise', in James Ogilvy (ed.), Revisioning Philosophy (Albany, NY: SUNY Press, 1986). 14 It's misunderstanding of just this point, I would argue, that leads Jackson, Pettit, and Smith astray in their article from this volume (Ch. 4).
284
Margaret Olivia Little
interpretation will have theoretical backing to explain this unifying tie: whatever other connections between the evaluative and nonevaluative domains we might acknowledge, there is an essential anomological relation between them, where this is read as rejection of articulable laws, because each domain's concepts answer to distinct substantive interpretational constraints.' 5 ) This point helps to explain the ubiquity of the particularist's pessimism about moral principles—a ubiquity which might otherwise seem simply odd. The first reaction many have when they hear the particularist's banner cry is that it is surely exaggerated: surely some moral principles survive scrutiny. Seemingly clear-cut cases spring to mind: surely rape, for instance, is always cruel. But the case of rape is a good example of the point at issue here. If we really mean 'rape' to be a nonmoral concept (so that this principle is indeed a case of codifying relations at the tier between moral and nonmoral features), then it is something that is cruel or unjust only against a certain kind of context—one which, of course, obtains in our society all too widely. It is actually very difficult to cash out that kind of context—to define the act we mean to condemn—without helping ourselves to concepts that turn out, on reflection, to be irreducibly laden with the evaluative (think of trying to isolate the instantiation conditions of consent without the help of moral notions like fair, or of force at the level of physical mechanics).' 6 If reason-giving considerations function holistically in the moral realm, we simply shouldn't expect to find rules that mark out in nonmoral terms the sufficiency conditions for applying moral concepts. Notice, now, that such an argument for particularism simply obviates what is perhaps the most persistent traditional argument offered in defence of deductive principles. 17 It is often said that such principles must be lurking in the background if a moral conclusion is to count as the right one. After all, it is thought, if the reason for the conclusion is adequate—if it really operates as a reason—it must be an instance of a generality that holds through all circumstances: an exception elsewhere shows that one must refine one's claim here about what actually counts as the reason. Such an argument, 15 Donald Davidson, 'Mental Events', in Essays on Actions and Events (Oxford: Clarendon Press, 1985), 207-25. 16 Far from treating the issue too lightly, this is a point central to many feminist analyses of the issue. The badness of rape is not located at the physiological level, which is one of the reasons consciousness raising may be required to discern instances of the category. My thanks to David DeGrazia for pressing this example, and to Alisa Carse for help in thinking through the issue of defining rape. 17 See Dancy's discussion of this point, Moral Reasons, ch. 4.
Moral Generalities Revisited
285
clearly, has tacitly assumed that moral considerations function atomistically. If, as is suggested here, they instead function holistically, then a set of considerations can function as a reason here—can truly function, and not simply be an incomplete rendition of a reason—and yet not count as such a reason in another context.
III
The suspicion persists for many that the difference between this picture and one that defends codification is simply one of degree and not kind. After all, once we agree that the moral supervenes on the nonmoral—once we agree that two objects cannot differ in moral properties without differing in nonmoral properties—then we have surely agreed that there are open sentences, of admittedly enormous complexity, which specify the moral as a function of the nonmoral. It is worth pausing to get clear about what is misguided in this response. Assume for purpose of discussion that admission of supervenience means that we can construct well-formed, if enormously complex, open sentences that specify the moral as a function of the nonmoral. Acknowledgment of such 'supervenience functions, as we might call them, does not dilute the particularist's claim, for such functions are not equivalent to the generalities she rejects. They are, to put it bluntly, the wrong type of generality, unable to serve the theoretic function that the notion of shapelessness concerns. As we have seen, the particularist argues that we cannot codify the contribution that nonmoral features make to the moral status of actions or situations. As I pointed out, we can put this claim in terms of what is often called the 'good-making' relation. We familiarly distinguish amongst the features of a situation or action those 'in virtue of which' it has the moral properties it does: that she made him cry, not that it occurred at 47.32 feet above sea level, was what made the action cruel. The particularist's claim is that the good-making relation cannot be cashed out in propositional form, for nonmoral features carry whatever import they have only holistically. Now, the concept of 'good-making' is not the most perspicuous concept (we'll be returning to it more than once); but whatever else we might say about it, it clearly involves a claim about explanation. To say of some nonmoral features that they and not others are morally salient—are those in virtue of which the situation or action has the moral status it has—is to make a judgement about what would be explanatory of that moral status.
286
Margaret Olivia Little
In the above example, pointing to the crying is explanatory of the cruelty— it illuminates its nature and helps those who are puzzled—in a way that looking up the topographical data does not. Those claiming that the moral is shapeless with respect to the nonmoral, then, are claiming that there are no codifiable explanatory generalities. But admission of supervenience functions does not obviate this claim; for the supervenience functions remaining in the face of holism will almost certainly not be explanatory. It's a familiar point in the philosophy of science that not all necessary generalizations are explanatory (or, as some would put it, law-like) generalities; and the supervenience functions relating the moral to the nonmoral, while necessary, are not the sort that would count as explanatory. It is not just that the functions would be very complicated, but that the complication would be of the sort—namely, radical over-specificity—that renders them incapable of serving the purpose at issue. To use Putnam's example, the geometric shape of objects supervenes on the physical positioning of their protons (an object can't change from square to circular without a rearrangement of its protons), but we will not find any explanation of why square pegs don't fit in round holes if we're looking at the level of proton positions without being able to help ourselves to any mathematical concepts: we will miss what unites the phenomena.18 Justo,weilnfdayhpboutwncisrelbyokgat the level of details about the physically sufficient conditions for cruelty. What explains—what would count as an illuminating answer to questions about why—it is cruel, will simply not be found at the level of detail the supervenience functions must encompass if they are to be accurate. Put bluntly, the notion of shapelessness at issue is not an extensional one. It is, instead, the notion that what explains a situation's moral status— what really does, and not just what we now realize were mistakenly taken to be explanations or good reasons for verdicts—cannot be codified. But this claim about shapelessness and explanation matters. For whatever else we might say about the sorts of generalization that have been advanced under the rubric of a moral principle, they were meant to be useful ones— generalizations capable of serving some epistemic-cum-theoretic function; and the supervenience functions remaining in the face of holism simply aren't the type that could serve such a purpose. 8 See Hilary Putnam, 'Philosophy and Our Mental Life', in Mind, Language, and Reality (Cambridge: CUP, 1975), 291-303; and Geoffrey Sayre-McCord's discussion of it in 'Moral Theory and Explanatory Impotence', in his Essays on Moral Realism (Ithaca: Cornell University Press, 1988), 256-81,276.
Moral Generalities Revisited
287
The point here tends to be missed because those impressed with the existence of supervenience functions tend tacitly to imagine a peculiar type of scenario, in which subjects are stipulated as already possessing the manual of supervenience functions and as having at their disposal full knowledge of the universe's nonmoral features. It is then pointed out that the manual would be epistemically useful because it would allow subjects to infer (indeed, to deduce!) the moral properties that various situations instantiate. But this is the completely wrong thought experiment for the present context. For one thing, when we return to any realistic scenario in which we know only a finite amount about the nonmoral features of the universe, possession of such a manual would immediately become useless. The supervenience functions contained in that manual, after all, tell us nothing about when non-deductive inferences are warranted: the contours of the former conditionals aren't remotely similar to the contours of rules or guidelines of warranted inference under uncertainty. More deeply, though, to imagine the manual readymade skirts the key question, which is: how one would rationally come to construct such a manual in the first place? Certainly, someone who could survey all possible situations and make note of all their nonmoral and moral properties—God perhaps—could derive the correct supervenience functions. But this assumes an ability to know the moral features already. Acceptance of the manual, that is, would be parasitic on already knowing the moral landscape, not a step in helping to discern it. But the whole point of searching for the generalizations we've advanced as moral principles has been to help us expand our moral knowledge given that we don't know each such instantiation individually. The fact is that, even if there were a fully accurate manual specifying the moral as a function of the nonmoral, we could have no rational means of constructing it, and no grounds for believing true any candidate we came across, given the radical complexity of supervenience functions remaining in the face of holism. 19 It is not just that the manual would be infinite in length. Given that the functions are not explanatory of any phenomena, the most we would have would be inductive evidence for bits and pieces of such functions and no evidence to support their claimed modality. 20 To 19
Compare Davidson, 'Mental Events', on the anomolism of the mental and the physi-
cal. 20 An exception to the generic limitation of inductive reasoning would be found if moral properties formed natural kinds: here, under currently prevalent theory, inductive a posteriori investigation can reveal necessary truths, a la water is H 20. But the idea of moral properties as natural kinds just is the possibility particularism denies.
288
Margaret Olivia Little
imagine that subjects already somehow possess the manual, then, gets things precisely backwards. Those who urge that the difference between particularism and codification is one of degree and not kind are looking at these doctrines at the wrong level. Considered extensionally, the difference in the claims they make is indeed one of (enormous) degree. That difference in degree, though, translates into a crucially important difference in kind when we consider the use to which their generalities can be put.
IV
Many who claim sympathy to particularist sentiments believe that those sentiments still leave room for some sort of codified law-like generalities linking moral and nonmoral properties. After all, it is thought, we can at least still say of all cases that ceteris paribus (roughly, 'all things equal'), or again pro tanto (roughly, 'in so far as this goes'), intentional falsehoods are breaches of fidelity and breaches of fidelity are wrong. Thus while it is true that arriving at moral answers requires judgement rather than mere application of algorithm, it is still judgement backed by an architecture of sorts— the generalities comprising that architecture are simply of a less ambitious sort than is sometimes supposed. Such thoughts are part of a broad sense, retained by many, that once we adjust the content of the generality, soften the quantifier by which it is bound, or narrow the scope within which it applies, we will recover context-independent, codifiable generalities. But this intuition is the lingering effect of the model the particularist wants to dislodge. Once we appreciate the argument behind particularism set out above, we will realize that its implications are more radical. To understand the real lesson of particularism is to understand that there is reason to doubt the existence of any codifiable generalities linking moral and nonmoral properties. Start with pro tanto claims: such claims assert what moral import a given feature would have when considered in abstraction from the import of other considerations that might be present. We are invited to imagine that there are no other moral variables, to assess the morality of the situation or action 'in so far as this feature goes'. If particularism is right, though, there is no reason to believe that any such moral claims, weak as they are, can be advanced with respect to nonmoral features such as 'causing pain'. The context-dependency of nonmoral features, we saw, extends to their very valence: whether a given such feature counts as any moral reason at
Moral Generalities Revisited
289
all—and if so, in which direction—is itself dependent on the background context. That is, not only can't one codify how the moral weight of a given feature stacks up against other moral considerations, it needn't have any moral weight to begin with, and certainly none of any given valence. The point about lies, for instance, is not just that they are sometimes permissible because breaches of fidelity can be justified by the presence of competing duties, but that not all lies count as such breaches in the first place. But this means that we can't issue any codified pro tanto claim about the moral import of a natural property. To return to the aesthetics analogy, we wouldn't think that the dab of red always constitutes a beauty-making feature wherever there are no other aesthetic properties: the aesthetic import—if any—that it carries is still dependent on the further particulars of the case. Just so, the fact that other features don't add up to moral claims needn't imply that the feature in question, such as lying, does. We are perhaps misled here by the fact that such pro tanto claims can indeed be advanced, even by the particularist, once we are firmly ensconced within the moral tier—once, that is, our claims concern the overall import of moral considerations, so identified (such as being cruel or being a duty). Thick moral features differ from nonmoral ones precisely because, so identified, they are guaranteed of carrying a given valence of moral significance (part of what it is to count as a moral feature, to earn the status as a moral feature so identified, is to count as a moral reason of a given direction). 21 Thus those who believe, say, that the category of duty exhausts the category of obligation can comfortably issue statements such as, 'In the absence of other duties, one has an obligation to be just' for the plain fact that an action is just means there is reason to do it. How justice weighs up with other duties is context-dependent, but that it is good-making is not. In contrast, the context-dependency of nonmoral features extends to their very valence. Even if we stipulate that other features do not themselves come together to ground duties or moral claims, then differences in those features can still change and influence the import that the targeted feature has.22 In the end, the most one can say is that, in contexts in which a
21 Here I part company with Jonathan Dancy, who urges that even thick moral concepts have variable moral valence. I think invariant valence here is an implication of the proper model of moral internalism. 22 To put it slightly differently: within the broadly defined 'context' in which other features ground no moral duties or claims, intentional falsehoods still form a morally varied lot.
290
Margaret Olivia Little
feature is good-making, it is good-making—not exactly late-breaking news.23 Classic renditions of ceteris paribus (roughly, 'all else equal') clauses are similarly denied us. As classically used, these are claims extending over open domains that concern the effects of isolated change or comparison. They indicate what would happen if some isolated feature or variable changed but everything else remained the same (we are invited to imagine the other variables or features, not absent, as with abstraction, but held constant). For instance, in describing Boyle's law, one states that 'all else equal, an increase in temperature leads to an increase in pressure' to control for changes in volume. The claim here is that whatever the values of the variables over which the ceteris paribus claim ranges, as long as they don't change one can assert with confidence the effect of another variable's change (whatever the volume might be, just so long as it now stays constant, we can judge the effect of increasing the temperature). But of course, this is just what the particularist denies about the moral import of natural variables. If we accept the doctrine of holism, the implication of changing one variable is influenced, in ways that defy codification, by differences in the other variables' values. We cannot isolate the implication of switching from truth-telling to lying by holding other variables constant, for it matters what the substantive content of those variables was in the first place. This point tends to missed because the kind of ceteris paribus claim just described tends to be conflated with another that particularists can happily make. Particularists can certainly talk about what moral status an individual situation or action would have if a given feature were changed in some specified way. They can issue judgements, that is, about what moral status a situation would have if it instantiated all and only the properties of the anchor case except that it had f instead of g—and they can use ceteris paribus as a way of referring to the 'all and only' that is held constant. The point, though, is that under holism such a comparison is not thought to 23 The temptation is strong here to think that we could revive pro tanto claims with respect to nonmoral features if only we made the abstraction sufficiently thoroughgoing: surely, if nothing else were at all morally relevant in a situation, there would be nothing to `keep' lying from being wrong. The problem is that, in that case, there would be nothing left to 'make' it wrong, either. That is, when we reject an atomistic model of how nonmoral features carry moral force, we agree that the reason-giving force of nonmoral features emerges as a function of the context in which it appears. To say that nothing else is at all morally relevant would be to say that nothing else is relevant even to determining the status of the feature in question. To abstract in this way, that is, would be to deny that anything functions as a context. But then, of course, we would be left with literally nothing to say about the moral force of the feature in question: under particularism, it has none on its own.
Moral Generalities Revisited
291
obtain in virtue of a recoverable codified law: there is precisely no cashing out in helpful propositional terms 'all and only the properties instantiated in the anchor case' (see again section 111). 24 This means that the ceteris paribus claims available to the particularist issue in enormously limited assertions: all they say is what would happen if this specific case were changed in some specified way, leaving it entirely open what that change would imply in another case. A bit of reflection shows how far distant such claims are from capturing the idea that lying, say, is somehow 'in its nature' wrong. For on this construction, ceteris paribus claims can be used truthfully to ascribe any moral import to any nonmoral feature. That is, far from being reserved to indicate the bad-making influence traditionally ascribed to lying or the goodmaking influence traditionally ascribed to alleviating pain, ceteris paribus claims as deployed under particularism broadcast the enormous range and diversity of moral significance. The claims, after all, are each anchored to a given case, in all its complexity and contingency. Depending on which case the comparison is made to, any feature may assume moral significance, from shoelace colour to the day of the week: after all, against a rich enough story, there are cases in which the change from Tuesday to Wednesday makes all the difference. Moreover, the valence of the contribution will vary claim to claim: against some given backgrounds, it is the change from truth- to falsehood-telling—not vice-versa—that will be good-making. 25 24 Indeed, invoking ceteris paribus clauses is useful for the particularist in such cases because it is a way of referring to context, where reference is needed precisely because the context cannot usefully be spelled out descriptively. 25 This demonstrates an important difference between moral particularists and scientific particularists—a point lost on many (e.g. Nancy Cartwright, How the Laws of Physics Lie (Oxford: OUP, 1983). Scientific particularists, of course, are also interested in the status of ceteris paribus clauses. In science, such clauses are often invoked to deal with the familiar epistemic distance between theory and its application; more specifically, they are used to modify conditionals to indicate that their content approximates conditionals holding in idealized models. For example, we might say `ceteris paribus, if an apple is dropped then it will fall' to indicate that we are discounting defeaters that would keep us from assimilating the situation to one covered in the pure model of physics. After all, the apple wouldn't fall when dropped if, say, someone reached out at that instant to grab it—a condition obviously not mentioned in the pure model itself, which talks about mass and energy, not apples and hungry agents. Scientific particularists are folks who argue (for reasons that would be familiar to us) that there is no cashing out the defeating conditions to which such clauses refer: while justification in science proceeds in virtue of approximating a pure model, deciding when the messy world of apples and agents sufficiently approximates that model takes judgement, not the application of deductive law. But the contextualism cited by moral particularists is more thoroughgoing than this sort of generic epistemic contextualism. The point of the argument about pro tanto claims was that there is no pure model of morality analogous to the models of physics or economics. We still have hold of
292
Margaret Olivia Little V
Moral particularism, then, is a more radical doctrine than many give it credit for. It is also, I now want to argue, less radical than many accuse it of being. Many worry that moral particularism, if true, would come at the cost of moral justification. I want to argue that this is based on a deep misunderstanding—a misunderstanding that, interestingly enough, many particularists themselves are guilty of. Let us turn, then, to what many regard as the $64,000 question: if we accept the argument presented above, what picture of moral justification follows? Particularists have famously emphasized the importance of attending to detail; more importantly, they have emphasized the possibility of coming to moral knowledge, not by invoking generalizations that allow us to infer moral conclusions from such details, but by seeing what moral properties such details together ground. 26 According to particularists, we can come to discern or interpret the moral nature of specific actions or individuals by exercising a sensitivity—a sensitivity that is perhaps analogous to a perceptual capacity, but is perhaps just a species of the more familiar 'faculty' we use to apprehend that something is a table, namely, the capacity or skill to apply concepts correctly. 27 On this view, to put it very schematically, we can apprehend that something falls under the classification cruel by attending to the complexities of the case, discerning what is salient, making appropriate discriminations, and employing a matured understanding of the concept. Such a method, clearly, does not proceed by subsumption under generality. Discerning that something is the case—whether it's that the action is cruel or that the ball is red—obviously involves subsuming the case under a concept, but it does not involve reaching a belief by invoking some generalization linking premises to conclusion. One of the central tenets of the particularist's moral epistemology, then, is that we can come to know the moral landscape by discernment. Now this tenet, of course, is the source of familiar controversy, and we can agree that the wrong picture of morality, that is—a picture that still imputes an intrinsic moral nature or moral valence to nonmoral properties—if we hold to an 'approximation model' of ceteris paribus clauses in morality. Whatever justification in morality ends up looking like, it will not proceed in virtue of approximating a pure model—for there is no such model in morality. 26 Iris Murdoch, Sovereignty of the Good, (Tel Aviv: Schocken, 1971); McDowell, NonCognitivism and Rule Following'; and Mark Platts, 'Moral Reality', in Geoffrey SayreMcCord (ed.), Essays on Moral Realism (Ithaca: Cornell University Press, 1988), 282-300. 27 See Warren Quinn, 'Truth and Explanation in Ethics', Ethics 96 (1986), 524-44.
Moral Generalities Revisited
293
one of the tasks particularists face is making good on defending its plausibility. But for many, the real stumbling block to accepting particularism lies not in suspicion of moral discernment, but in suspicion that moral discernment is the only thing particularism has up its epistemic sleeve. After all, its central claim is that there is no cashing out when inflicting pain, say, constitutes cruelty and when it constitutes kindness: how, then, short of seeing a case in all its richness, can we know what its moral status is? But we need to ask whether this suspicion is warranted. Does particularism mean that we cannot justifiably invoke moral generalities to expand our knowledge beyond the few cases that are available to us in all their richness? Certainly some particularists, in some moods, suggest that it does. Common particularist rhetoric can suggest that the only way to gain justified moral belief is to exercise discernment on a case at hand, and that moral generalities, though of pragmatic use, offer no epistemic warrant or justificatory weight. Thus in some passages of McNaughton's work, for instance, principles are abandoned and discernment is all we have: Moral particularism takes the view that moral principles are at best useless, and at worst a hindrance, in trying to find out which is the right action. What is required is the correct conception of the particular case in hand, with its unique set of properties. There is thus no substitute for a sensitive and detailed examination of each individual case. 28
And again, passages from both McNaughton and Dancy seem to close off the possibility of inductive moral knowledge: the presence of a nonmoral feature is thought to provide no warrant whatsoever for believing that some moral feature is present: The leading thought behind particularism is the thought that the behavior of a reason (or of a consideration that serves as a reason) in a new case cannot be predicted from its behavior elsewhere. 29 The particularist claims that we cannot know, in advance, what contribution any particular non-moral property will make to the moral nature of an action. We cannot know, in advance, whether it will be morally relevant at all and, if so, whether its presence will count for or against doing the action. 3°
Perhaps most telling, the repertoire of epistemic skills needed for moral wisdom seems comprised only of skills involved in reading and interpreting individual cases, not those skills involved in judging patterns, weighing the adequacy of inductive evidence, or assessing inferential moves: 28 3"
McNaughton, Moral Vision, 190. McNaughton, Moral Vision, 193.
29 Dancy, Moral Reasons, 60.
294
Margaret Olivia Little
Whatever account we give of the coherence of a moral outlook, our account of the person on whom we can rely to make sound moral judgments is not very long. Such a person is someone who gets it right case by case. To be so consistently successful, we need to have a broad range of sensitivities, so that no relevant feature escapes us, and we do not mistake its relevance either. But that is all there is to say on the matter. To have the relevant sensitivities just is to be able to get things right case by case. 31
Even past experience is said to be helpful, not because it gives one a sense of patterns that one might project forward, but because it offers one practice in—and hence hones the skill of—discerning the moral status of individual cases. 32 Such comments can all make it sound as though we must be agnostic about any moral situation not immediately before us—as though, until we are able to see or interpret a case for ourselves, there is nothing we can say. For on this picture, not only would there be no inferential moral knowledge, there would be no warranted moral presumptions to bring to the world. To say that stabbing is presumptive of cruelty, while obviously less ambitious than saying that stabbing provides grounds for concluding that cruelty has occurred, is still to agree that stabbing provides a kind of general epistemic warrant for cruelty (for instance, it shifts the burden of proof onto those who would deny cruelty in the face of stabbing); and it reflects a judgement of generality, for it reflects a judgement about the kind of context one is likely to encounter. The radical nature of such a stance is obvious. 33 We commonly regard as justified all sorts of moral inferences based on discrete and limited bits of information, such as inferring that a wrong was committed when we hear report of a robbery. And moral presumptions are an absolutely central part of our moral lives—not to mention moral aspirations. As Aristotle points out, the wise person inculcates dispositions of, say, not lying and not stabbing others, which is to say she carries the presumption that such actions will be wrong in the sorts of situations she's likely to meet. Morality in this respect, that is, is precisely unlike the aesthetic examples mentioned above: Dancy, Moral Reasons, 64 'Of course, a comparison with other cases may help us to decide how things are here, just as a long experience of car engines may help us to diagnose the fault this time. But this decision or diagnosis is still essentially particular. It would be surprising if a long experience in garages were no help to the mechanic; it would be surprising if a long and varied moral experience did not serve to sharpen one's sensitivity for the future. But in neither case is one's first question what one can say here that is consistent with what one has said elsewhere. The crucial question is how things are in the case before us.' Dancy, Moral Reasons, 63. 33 My thanks to Mark Lance for helpful and spirited discussions on this point. 31
32
Moral Generalities Revisited
295
I leave the house with no presumptions about whether the splashes of red I will see that day will be aesthetic- rather than unaesthetic-making, but I would think myself remiss to abandon presumptions about the moral relevance and valence of a range of nonmoral features. As it turns out, though, even the most radical particularist has other moods, for in fact virtually all particularists end up invoking moral presumptions if not moral inferences. For instance, particularists commonly advocate using moral principles as 'rules of thumb' (so long as we stand ready to abandon them in the face of specific information): but a rule of thumb about the moral status of stabbing just is a statement that stabbing offers some form of epistemic warrant, at least presumptive, for cruelty. To believe the rule's use justified—to regard some such rules as better than others—clearly involves projecting beyond any individual case to the kinds of contexts one thinks likely to be encountered. Or again, the idea that moral principles are useful pedagogic tools is tacitly premised on the idea that there are warranted presumptions. The judgement that a given principle such as lying is wrong' will help rather than mislead a moral novice reflects a judgement about the sorts of contexts she is likely to encounter, just as our agreement that it is better training to tell beginner drivers 'Never slam on the brakes' instead of 'Stomp on the gas whenever you see another car' reflects a judgement that the student will most likely be facing our world of crowded highways and not the post-apocalyptic world of Mad Max movies. There is an air of confusion, then, when particularist rhetoric cautions us not to travel beyond individual cases at the same time that it invokes as reasonable presumptions which, make no mistake about it, involve going beyond individual cases; or again, when the skills it regards as exhaustive of moral wisdom have nothing to do with judging patterns, even as it appeals to our commonsense that some rules of thumb are more useful than others; or again, when it casts suspicion on inferential moral knowledge at the same time it invokes presumptions, which of course differ from the former only in degree. Indeed, part of the confusion in the debate is the mixed message particularists send every time they emphasize the uniqueness of individual cases by giving examples of particular saliences 'such as "making her cry" . Implicit in not saying 'such as her shoelace being green '— which, after all, in the right context can be salient—is the idea of what tends to be salient—what, as Dancy puts it in another passage, is 'more commonly important'. 34 34
Dancy, Moral Reasons, 67-8.
296
Margaret Olivia Little
In fact, such rhetorical schizophrenia is unwarranted. The argument backing particularism is perfectly consistent with the idea that nonmoral features such as causing someone pain can be presumptive of moral properties such as being cruel, or, again, and more ambitiously, that evidence of the former can provide adequate grounds for inferring the presence of the latter. Indeed, one will be tempted to think it is not only if one is still wedded to atomism, here at the level of epistemic, rather than moral, reasons. Let me explain. Regarding stabbing as presumptive of cruelty, or again inferring the latter from the former, rests, of course, on judgements that one is justified in projecting forward previously experienced patterns of stabbing and cruelty. If we take seriously the particularist message, one may begin to wonder how such justification could ever be attained, however robust the previous patterns. After all, the particularist insists that there is no codifiable law-like pattern subsuming or connecting the individual instances of the N already experienced (in the way, say, that laws of chemistry bring together disparate instances of salt dissolving): how then do these individual instances add up to some aggregate phenomenon one could project forward? How are we to judge that the case around the corner isn't precisely the one in which the shoelace colour will matter? How could we be justified in thinking that the broad category of 'contexts in which lying is cruel', and not 'contexts in which lying is kind', is the one we are more likely to meet up with? Such a scenario will seem mysterious, though, only if we are still operating with a picture that believes such judgements must ultimately be backed by codifiable laws. On that traditional picture, we can project outward observed patterns only if the elements contributing to the pattern are deducible from or subsumable under some codifiable (perhaps statistical) law. Inferential knowledge would then be denied to the particularist, for how could we assess the likelihood of phenomena that show up as subjects in no law? But it is a fallacy to think that this is how justification proceeds— in ethics or anywhere else. Imagine that someone needs to get a message to my neighbour Joe, whom I've seen on my way home from the Metro virtually every workday evening for the past eight years, since I pass right by his front porch as he is ritually reading the paper. There have of course been a few exceptions— twice, for instance, Joe was mowing his back yard, since it was his first chance to mow in two drizzly weeks, and Joe's a bit compulsive when it comes to his lawn. None the less, we would take it as perfectly natural for me to offer to give him the message on the grounds that I know I'll see him
Moral Generalities Revisited
297
that evening. If asked, of course, I will readily admit that any number of circumstances—including Martians beaming him aboard a spaceship—could prevent him from being on his front porch this particular evening. My sense that I am none the less justified in thinking I'll meet up with Joe travels by way of familiar judgements—judgements that relevant patterns are sufficiently robust (his behaviour has amounted to a ritual), that the current context is not relevantly so dissimilar from the past as to interrupt projecting this pattern (we have not entered a period of nightly earthquakes, for instance), that certain defeaters (such as the Martian scenario) do not warrant investigation, and that relevant defeaters do not in fact defeat (Joe's grass was neatly cut when I passed by this morning). There is no cashing out the content of any of these judgements. There is no codifying the conditions that would affect whether Joe will be on the porch, and no notion that each potential influence, not to mention the interactions amongst them, is governed by some articulable law. It isn't as though I compute some probability function for Joe being on the porch by sifting through all the possible circumstances that could affect his presence there, weighting them by naturalistically specified probability functions, and toting up an end result. Instead, what it takes to make judgements of justification are, broadly put, epistemic skills. It is a skill to read the world—to know what, in the face of an infinite amount of change, would for a given purpose count as a sufficiently relevantly similar world, to know when a pattern is robust, to know how to navigate through patterns of competing influences, to determine which possibilities are epistemically relevant alternatives, to know when you know enough and when you don't, to know when you have entered a context in which previous experience no longer points the way. Such skills, like any others, are not exhausted by knowledge of some codifiable set of propositions (think of knowing how to drive): knowing how, as the saying goes, cannot be reduced to knowing that. 35 We make judgements here, then, as we do anywhere—by consulting our experience of patterns and our sense of current conditions, and then invoking our competency with the relevant epistemological concepts such as relevance, robustness, similarity. Only if we believe that epistemic reasons are supposed to function atomistically does this seem mysterious, for to say that the contours of the relevant considerations have no nonepistemic shape does not mean they have a shape over which we have no competency. 35 For a provocative description of skills in the moral life, see Dreyfus and Dreyfus, `What is Moral Maturity? Towards a Phenomenology of Ethical Expertise', in James Ogilvy (ed.), Revisioning Philosophy, (Albany: SUNY Press, 1992).
298
Margaret Olivia Little
Given a skills model of epistemic justification, then, there is ample room for the particularist to acknowledge that N can be presumptive of M, or indeed grounds for concluding M. Particularists who think their doctrine itself implies otherwise are missing the force of their own position on the holism of the evaluative. This is not to deny, of course, that a particularist might in the end become sceptical about moral judgements that are born of inference rather than discernment. Certainly, particularists as a class tend to be less impressed with the quality of our inductive evidence about the relations between N and M; and a truly radical particularist may well urge that the world is in fact too morally cacophonous for us to know—or perhaps even to presume—what's around the corner morally speaking. The point urged here is that such scepticism would reflect a substantive claim about the (dis)order of the world or the (in)adequacy of our evidence: it does not follow from the doctrine of particularism itself or the model that backs it. And whether such scepticism is warranted, let me point out, depends on the epistemic context or enclave inhabited. One person might live in a sufficiently predictable or simple microcosm of the world where stabbing and lying were virtually always wrong; while another might live in a microcosm so chaotic that the most familiar moral presumption becomes otiose. I can imagine a world in which one cannot presume not to stab the people one encounters (imagine again your favourite post-apocalyptic movie). But that doesn't mean it's our world. To say that the moral landscape cannot be codified is not to say that it is chaotic.
VI
I have argued that particularists can—and usually do—acknowledge a role for moral generalities of one stripe, namely, those involved in inductive generalizations. I now want to argue, perhaps more surprisingly, that there is room within particularism to advance explanatory or law-like generalizations between moral and nonmoral properties. To explain, let me return our focus to where we started, with the 'good-making' relation. It is a familiar aspect of our moral practice, I've said, that we distinguish amongst the features of a situation or action those 'in virtue of which' it has the moral properties it does. In doing so, we are saying that certain features have a special kind of salience: they are not just 'good-indicating' (providing helpful evidence of moral status), they are actually `good-making. Obviously, the notion of 'making' here is not the sort meant when we talk
Moral Generalities Revisited
299
of a carpenter making a chair: the claim is not that inflicting pain, say, causes cruelty. The idea, rather, put generically, is that the feature or features tagged as salient explain the action's moral status. On one model of how explanation proceeds, inherited from the Early Modern tradition, one phenomenon is explanatory of another when they are related under a codifiable law-like generalization (the amount of mass explains the energy discharged, e.g., given that the two forces are related algorithmically). Against this picture, it is puzzling how a particularist could preserve the good-making relation, puzzling how she could ever capture the idea that we regard some of an action's features as those that make it cruel or kind. But the particularist does not reject the good-making relation; she simply has a deeply different model of how the explanation that backs it can work. What model might that be? Some particularists, in their zeal to reject the traditional interpretation, have been led to claim that the alternative model must be one on which the good-making relationship has nothing to do with any sort of generality. 36 Therlationspcmyabutheinsc d:xplation here is 'stubbornly particular'.37 But this seems deeply wrong. Whatever else we want to say about claims of explanation, they are surely not reducible to claims about a particular case. 38 Explanation is a normgoverned activity of removing puzzles; it is thus an essentially epistemological activity. But, clearly, offering an explanation is not equivalent to giving just any sort of epistemic reason for a conclusion: it offers a particular kind of illumination; and it is nearly platitudinous that the kind of 36 Namely, Dancy (Moral Reasons, 104-6). According to Dancy, the relationship of good-making is a claim only about the instance at hand. More specifically, he argues that the good-making relation is the metaphysical relation of `resultance ' (see discussion pp. 73-4). He takes this relation at face value, but suggests that it might be spelled out as a token-token identity: when we say that the lying here 'makes' it cruel, we're saying that the lying instantiated in this particular instance constitutes or is identical to its cruelty. I won't be alone in finding this view an unhappy one. Token-token identities are certainly respectable when they are identities of objects or events (as in 'Patricia is identical to Helen's Auntie'), but this can't serve the purpose at hand. The good-making relation is precisely meant to discriminate amongst the properties of an object or event: it is meant to reach through to some but not others of an object's properties to designate them as salient. But, of course, it cannot be an identity of the moral and natural properties mentioned: that would be a type-type identity, and just the sort of modal generality particularists reject. The identity invoked here seems instead to be an identity between a very strange sort of relata, namely tokens of properties (as in Achilles' courage), otherwise—and infamously—known as tropes. 37 Dancy, Moral Reasons, 104. 38 Here I am much influenced—and much indebted—to many fruitful and enjoyable discussions with my colleague Mark Lance.
300
Margaret Olivia Little
illumination is one that involves situating the case in a broader pattern. As we saw when discussing the limitations of supervenience by alluding to Putnam's point about the square peg and round hole, explanation proceeds by showing what unites phenomena. To explain is to bring under some kind of general picture: if being N is here explanatory of being M, it is not just about this case. But it is a mistake to think that one can't accept this common sense view if one is committed to a holistic connection between the two realms. Indeed, the doctrine of holism demonstrates why the connection can still be general. In saying that explanations situate instances within broader patterns, we are saying, roughly speaking, that when we offer one phenomenon as explanatory of another we are asserting our confidence that the latter always follows from the first within a suitable constellation of possible worlds. The constellations invoked, of course, rarely extend to all possible worlds: we familiarly agree that a bridge's structure accounts for its strength even though it is logically possible to have such structure without such strength. The generic idea is that the conditional holds within a relevant or apt or informative constellation of worlds (where 'relevant, 'apt, and 'informative' are obviously the operative words). Thus when counterexamples are levied against a proffered explanation, they are sometimes meant to deny that the claimed conditional holds; but they are sometimes meant to deny that the constellation forming its scope is adequately apt or robust to ground the normative move we want to make on its behalf. It leaves out cases we precisely want to include, cases that form the backdrop to our thinking and questions about connections. Put abstractly, then, an explanation succeeds just in case a conditional tying the purported explicandum and explicans holds in a sufficiently interesting constellation of worlds—a constellation that is, crudely put, apt to the sense of the puzzle. We will communicate just in case those offering and those receiving the proffered explanation are tacitly loading in the same constellation. Now, according to the traditionalist picture of explanation, the contours of adequate constellations must be capable of being codified in practiceindependent ways, allowing us to recover laws joining the phenomena that are specifiable in context-independent ways. But once we appreciate the holism of reasons—here the holism of explanatory reasons, we can appreciate that there is no reason to think that success in explanation depends on such codification. While some explanations may proceed in this way, there is no reason to think they all must. For the crucial notions of 'apt' and 'informative' may be irreducibly practice-driven notions. There are explanatory
Moral Generalities Revisited
301
moves that require a background sense of what sorts of constellations are relevant and telling, what sorts of puzzles a given press alludes to. The contours of the constellation at issue need have no shape that is specifiable in nonmoral terms, no shape that is discernible to outsiders of our evaluative practices, in order to succeed as illuminating or informative. The fact that the scope of the generality can't be captured in finite terms doesn't mean we can't successfully invoke it, for we can share things, like understandings, concepts, and practices, that outstrip finite sets of propositions. That is, while the sense of 'relevance' and 'robustness' sometimes comes by piggybacking onto some codifiable generalization, there is no reason to think it always must. There are some explanatory practices that demand, not just hooking into a metaphysical relationship, but deployment of a skill—here the skill, not of judging particulars, but of judging what counts as 'apt' and 'informative'. There are some explanatory practices in which communication requires understanding a local sense of puzzle, a shared background sense of what sorts of constellations are available for tacit invocation, and an appreciation of which would count as illuminating in that given context of confusion. Indeed, explanation proceeds like this far more often than those suffering 'physics envy' acknowledge; it is the model, for instance, appropriate to the social sciences—something its practitioners would do better to realize, rather than losing the richness that comes with contorting their explanations to fit the narrow model of physics. Moral explanations will differ in some ways from the most general ones offered in the physical sciences. With the latter, there is just one constellation of worlds that can count as the appropriate backdrop to successful explanations, namely, those worlds in which the physical laws of the universe hold—thus the only sort of worlds, if our theories are right, that we're ever going to encounter. Under moral particularism, obviously, such broad universality does not hold. There will be many groupings of 'morally possible' worlds, if you like, in which intentional falsehoods are morally neutral or even kind—worlds, moreover, we have and will inhabit. With moral explanations, then, successful explanation is a more elastic notion than with, say, the explanations in the physical sciences. When we say of a given action that the infliction of pain made it cruel, we are not claiming that such infliction grounds cruelty in all physically possible worlds, or even all worlds we will experience. Rather, we are claiming that such infliction grounds cruelty in a particular group of cases we regard as telling or illuminating given the present sense of puzzle—a constellation whose capacity for illuminating is no less for being context-dependent. The explanation proceeds, that is, by tacitly subsuming the present case under a generality
302
Margaret Olivia Little
the contours of which relies on a shared sense of which constellations of worlds it would be telling, in the given context, to find a conditional. There is no one such constellation; rather, it depends on the context of confusion—on what is shared and what is puzzling. Once we are truly at ease with the idea of irreducible contextdependency, then, we can reintroduce into particularism a role for explanatory generalities beyond those invoked in pedagogy and heuristics. For while explanation has everything to do with generality, it need have nothing to do with codified generality. It is simply a false contrast to think that we must either talk about single cases or about codified generalities: the interesting, post-positivist terrain all lies in between. Particularists, then, can happily advance both inductive and explanatory moral generalizations. Lest this all start to sound too concessionary, let me emphasize that particularism born of holism is still a distinctive theory, its lessons hardly well worn. For the uncodifiable generalizations available under particularism are a far cry from the kind of generality that moral principles were traditionally thought to constitute. They permit no idea that natural properties, such as intentionally telling a falsehood, have some intrinsic or, if you like, metaphysically grounded moral nature. The point is worth underscoring. Sometimes, when those who acknowledge particularism's lesson retreat to the comfort of pointing out that 'usually' or 'for the most part ' lying is wrong-making, they speak as though this generalization still gets at the intrinsic moral nature of lying. The oddly persistent intuition seems to be that, while particularism reminds us that there are no exceptionless principles, we can still have faith that there is a codifiable pattern to the exceptions, that they are still driven by or get at the moral nature of lying or stabbing or writing cheques to charity. We see this, for instance, when people invoke the idea that 'in most contexts, it's cruel to lie' as some overarching comment about the nature of lying. But there is no one set of 'most contexts' or 'most worlds' or 'most actions'. There are any number of ways to divide up the terrain of possibilities into sets of unequal size, any number of differently composed majorities, and no reason to think 'if N then M' holds of each. And once we are called upon to specify the scope of such a claim, the issue of shapelessness rears its head once again. There will be constellations of worlds in which `for the most part' runs in just the opposite direction; and shapelessness claims that there is no designating in natural terms those constellations in which the proffered, more familiar direction holds. 39 39 One must pick and choose the worlds over which one's statistical claim ranges, that is, and the suspicion here is that one cannot specify that range in finite or meaningful terms
Moral Generalities Revisited
303
Once we accept particularism, then, statements about the 'usual' cruelty of lying, far from reflecting anything about some deep metaphysical nature of lying, simply reflect something about the epistemic context one contingently occupies. To put it another way, while acceptance of some moral presumption does mean one is committed to a certain generality—roughly, that within epistemically relevant alternatives, most N are M—this is not a statement about the architecture of the moral world; it is simply a reflection of N's local epistemic status, a way of restating that N is indeed here presumptive of M. And again, when we say that N is explanatory of M, it means the conditional 'if N then M' holds within sets of worlds that form locally helpful juxtapositions for explanation, given what we are likely to encounter and given the particular context of confusion. Generalities such as 'lying is usually wrong-making', in short, are not simply milder or less ambitious under particularism than with traditional moral principles, they are entirely different beasts. In sum, the model behind particularism yields a doctrine whose metaphysical implications, if you like, are far from modest. To accept the doctrine is to reject the idea that natural properties have any intrinsic moral natures, or even intrinsic moral valences. But while the metaphysical implications of particularism are radical, the epistemological implications can, depending on the circumstances, be harmless. To return to a famous clarification of Ross' view: ironically, particularism must eschew pro tanto for prima facie claims.
VII
Oliver Sacks tells a marvellous story about a man who suffered a curious sort of perceptual aphasia. 40 He lost the ability to recognize everyday objects like a glove, a hat, or people's faces. Devoid of the ability directly to perceive, phenomenologically to gestalt, the objects as such, he was reduced to trying to infer what they were from the discrete bits of information he had—premises about the crude shape he could see them to have, or inductive moves he could make by reflecting on where or with whom he without using evaluative language. The lesson about context-dependence is a lesson for deductive principles or law-like generalities that fall short of that level of guarantee. Just as the shape of the worlds in which universally, if N then M, has no nonmoral shape, the shape of worlds in which we can come to believe most Ns are Ms as a piece of theory has no nonevaluative shape. 4" Oliver Sacks, The Man Who Mistook His Wife for a Hat (New York: Perennial Library, 1987), 8-22.
304
Margaret Olivia Little
was. Since he was very smart, he arrived at the right answer surprisingly often; but clearly, his reliance on inference and presumption was a poor substitute for the ability to perceive for himself what the objects were. The particularist points out—rightly, in my view—that we tend to suffer a sort of moral aphasia. Lacking the training necessary for developing the skills needed to discern reliably and accurately the moral landscape, we end up relying on crude forms of inference. The reliance is so widespread, in fact, that theories start valorizing deployment of inference as the height of moral maturity rather than a crutch of the blind. This tradition, moreover, is dangerously self-reinforcing, for in overemphasizing the skills involved in subsuming under generality, skills of discernment and interpretation atrophy. In reality, with moral wisdom, as with any skill, it is a sign of maturity to be able to leave behind the guidebooks, cookbooks, and primers, and to exercise directly one's ability to judge. In issuing this crucial corrective, though, particularists must beware of committing the opposing sin. While the skills of discernment and interpretation are central to moral wisdom, they are not exhaustive of moral wisdom. Even the moral expert needs competencies of judging patterns and generalities, both because, in the face of limited information, we often need to rely on judgements about what tends to be salient, and, crucially, because judgements about what is good-making are judgements about explanation and hence generality. Certainly the importance of these skills is often overdrawn by the theory-loving philosopher. Just as a person may possess an unerring sense about cinematic quality without being an articulate movie critic, someone may be morally reliable and yet inarticulate about moral abstractions. But if the tradition valorizing inference can be said to mistakenly cast someone like Sacks's fellow as its moral epistemic exemplar, the moral exemplar of a thoroughly radical particularist is, in essence, a moral idiot savant—someone with an exquisite ability to see moral properties directly in the elements at hand, but at a loss when we ask him to make inferences or to explain why something is cruel rather than kind. The model backing particularism is insightful—indeed, of a piece with many strains of post-positivist philosophy elsewhere happily accepted; but its implications for the role of moral generality in epistemology has been misconstrued. Particularism is deserving of the name, for it insists on the importance of discernment, highlights our over-reliance on generalities, and views as folly any search for a moral architecture. While particularism dethrones moral generalizations, though, it hardly exiles them. We will think it does only when the lesson of holism is misidentified: the enemy its objections target is not generalization, but codification.
BIBLIOGRAPHY
ANAGNOSTOPOULOS, G., Aristotle on the Goals and Exactness of Ethics (Berkeley: University of California Press, 1994). ARISTOTLE, Nicomachean Ethics, 1.3 and v.10. AUDI, R., 'Ethical Reflectionism', The Monist, 76 (1993) 296-315. `Intuitionism, Pluralism, and the Foundations of Ethics', in W. SinnottArmstrong and M. Timmons (eds.), Moral Knowledge? (New York: Oxford University Press, 1996) 101-36. `Moderate Intuitionism and the Epistemology of Moral Judgement', Ethical Theory and Moral Practice, 1 (1998) 15-44, esp. 36-41. AUGUSTINE, `Love God and Do What You Will', Joann, vii, 8. BAKHURST, D., 'Ethical Particularism in Context', Ch. 7 in this volume. BALDWIN, T., 'Intuitionism and Common Sense', forthcoming in P. Stratton-Lake (ed.) Moral Intuitionism (Oxford: Clarendon Press, 2001). BARNES, J., Aristotle: Posterior Analytics (Oxford: Clarendon Press, 2nd edn., 1993), 192. BELZER, M., 'Reasoning with Defeasible Principles', Synthese, 66 (1986), 135-598. BENNETT, J. The Act Itself (Oxford: Clarendon Press, 1995), 80. BLACKBURN, S., `Rule-Following and Moral Realism', in S. Holtzman and C. Leich (eds.), Wittgenstein: To Follow a Rule (London: Routledge & Kegan Paul, 1981), 163-87. `Through Thick and Thin', Proceedings of The Aristotelian Society, Suppl. vol. 66 (1992), 285-99. `Securing the Nots', in W. Sinnott-Armstrong and M. Timmons (eds.), Moral Knowledge? (New York: Oxford University Press, 1996) 82-100, esp. 97-9. Ruling Passions (Oxford: Clarendon Press, 1998), 308-10. BLUNT, L., Moral Perception and Particularity (New York: Cambridge University Press, 1994). `Against Deriving Particularity', Ch. 9 in this volume. BOWDEN, P., 'Ethical Attention: Accumulating Understandings', European Journal of Philosophy, 6 (1998), 59-77. BROAD, C. D., Five Types of Ethical Theory (London: Routledge, 1930), esp. 282. BROADIE, S. W., Ethics with Aristotle (New York: Oxford University Press, 1993), esp. 18. BRODY, B., 'Intuitions and Objective Moral Knowledge', The Monist, 62 (1979), 446-56.
306
Bibliography
BROOME, J., Weighing Goods (Oxford: Blackwell, 1991), ch. 5. `Can a Humean be Moderate?' in R. G. Frey and C. Morris (eds.), Value, Welfare and Morality (Cambridge: Cambridge University Press, 1993), 51-73. BURTON, S. 'Particularism, Discretion, and the Rule of Law', Nomos, 36 (1994), 178-201. BUTLER, J., Fifteen Sermons Preached at the Rolls Chapel (1726). Repr. as Five Sermons (New York: Liberal Arts Press, 1949), esp. 45. CARRITT, E. F., A Theory of Morals (London: Oxford University Press, 1930), esp. 114. CARTWRIGHT, Nancy, How the Laws of Physics Lie (Oxford: Clarendon Press, 1983). CAVELL, S., The Claim of Reason (Oxford: Clarendon Press, 1979). COETZEE, P., 'Principles and Virtues—Or—Principles or Virtues', South African Journal of Philosophy, 4 (1985), 25-8. CRISP, R., 'Motivation, Universality, and the Good', Ratio, 6 (1993), 181-90. `The Dualism of Practical Reason', Proceedings of the Aristotelian Society, 96 (1996), 53-73. `Particularizing Particularism', Ch. 2 in this volume. DANDY, J. 'On Moral Properties', Mind, 90 (1981), 367-85. `Ethical Particularism and Morally Relevant Properties', Mind, 92 (1983) 530-47. `The Role of Imaginary Cases in Ethics', Pacific Philosophical Quarterly, 66 (1985), 141-53. `Caring about Justice', Philosophy, 67 (1992), 447-66. Moral Reasons (Oxford: Blackwell Publishers, 1993). `In Defense of Thick Concepts', in P. French (ed.) Midwest Studies in Philosophy, 20 (1995), Moral Concepts, 263-79. `Defending Particularism', Metaphilosophy, 30 (1999), 25-32. `Can the Particularist Learn the Difference between Right and Wrong?', in The Proceedings of the Twentieth World Congress of Philosophy, i: Ethics ed. K. Brinkmann (Bowling Green: Philosophy Documentation Center, 1999), 59-72. `The Particularist's Progress', Ch. 6 in this volume. Practical Reality (Oxford: Clarendon Press, 2000). DEVEREUX, D., 'Particular and Universal in Aristotle's Conception of Practical Knowledge', Review of Metaphysics, 29 (1986), 483-504. DREYFUS, HUBERT L., and DREYFUS, STUART, E., 'What is Moral Maturity? Towards a Phenomenology of Ethical Expertise', in James Ogilivy (ed.), Revisioning Philosophy (Albany: SUNY Press 1992). DWORKIN, G., 'Unprincipled Ethics', Midwest Studies in Philosophy, 20 (1995), 224-39. EWING, A. C., The Morality of Punishment (London: Kegan Paul, 1929), 166. FRANKENA, W., Ethics (Englewood Cliffs, NJ: Prentice Hall, 1963, 2nd edn. 1973), esp. 16-17,23-5.
Bibliography
307
R., 'Moral Relevance and Ceteris Paribus Principles', Ratio, 8 (1995), 113-27. FRIEDMAN, M., 'Care and Context in Moral Reasoning', in her What Are Friends For? (Ithaca: Cornell University Press, 1993). GARFIELD, J., 'Particularity and Principle: The Structure of Moral Knowledge', Ch. 8 in this volume. GAUT, B., 'Moral Pluralism', Philosophical Papers, 22 (1993), 17-40. GAY, R., 'Ethical Pluralism: A Reply to Dancy's "Ethical Particularism and Morally Relevant Properties"', Mind, 94 (1985), 250-62. GOLDSTEIN, I., 'Pleasure and Pain: Unconditional, Intrinsic Value', Philosophy and Phenomenological Research, 50 (1989), 255-76. HARE, R. M., Freedom and Reason (Oxford: Clarendon Press, 1963) esp. 18,38— 40. HOLTON, R., 'Principles and Particulars: Morals from Arithmetic', (forthcoming). HOOKER, B., 'Theories of Welfare, Theories of Good Reasons for Action, and Ontological Naturalism', Philosophical Papers, 20 (1991), 25-36. `Parfit's Arguments for the Present-aim Theory', Australasian Journal of Philosophy, 70 (1992), 61-7. `Ross-style Pluralism versus Rule-consequentialism', Mind, 105 (1996), 531-52. `Moral Particularism: Wrong and Bad', Ch. 1 in this volume. HURLEY, S., Natural Reasons (Oxford: Oxford University Press, 1989), 130-5, 235-7. IRWIN, T. H., 'Ethics as an Inexact Science: Aristotle's Ambitions for Moral Theory', Ch. 5 in this volume. JACKSON, F., Pettit, P., and Smith, M., 'Ethical Particularism and Patterns', Ch. 4 in this volume. JIWEI, C., 'The Logic of the Generality of Moral Principles', Philosophical Quarterly, 38 (1988), 305-14. JONSEN, A. R. and Toulmin, S., The Abuse of Casuistry: A History of Moral Reasoning (Berkeley: University of California Press, 1988). JuDsoN, L., 'Chance and "Always or for the Most Part" in Aristotle', in Aristotle's Physics, ed. Judson (Oxford: Clarendon Press, 1991), ch. 4. KAGAN, S., 'The Additive Fallacy', Ethics, 99 (1988), 5-31. The Limits of Morality (Oxford: Clarendon Press, 1989), esp. 17. KRIPKE, S., Wittgenstein on Rules and Private Language (Cambridge, Mass.: Harvard University Press, 1982). LEMOS, N., Intrinsic Value: Concept and Warrent (New York: Cambridge University Press, 1994). LITTLE, M., 'Moral Realism: Non-Naturalism', Philosophical Books, 35 (1994), 225-32. `Seeing and Caring: The Role of Affect in Feminist Moral Epistemology', Hypatia, 10 (1995), 117-37. FRAZIER,
Bibliography
308 LITTLE,
M., 'Virtue as Knowledge: Objections from the Philosophy of Mind',
Nous, 31 (1997), 59-79.
`Moral Generalities Revisited', Ch. 12 in this volume. `Wittgensteinian Lessons on Particularism', in Carl Elliot (ed.), Wittgensteinian Bioethics (Durham, NC: Duke University Press, forthcoming). LOUDEN, R. B., Morality and Moral Theory (New York: Oxford University Press, 1992), 88f, 103 f. LOUDEN, R. B., 'Aristotle's Practical Particularism', in Essays in Ancient Greek Philosophy, iv, J. P. Anton and A. Preus (eds.) (Albany: SUNY Press, 1991). McDowELL, J., 'Are Moral Requirements Hypothetical Imperatives?', Proceedings of the Aristotelian Society, Suppl. vol. 52 (1978), 13-29. `Virtue and Reason', The Monist, 62 (1979), 331-50. `The Role of Eudaimonia in Aristotle's Ethics', repr. in A. Rorty (ed.), Essays on Aristotle's Ethics (Berkeley: University of California Press, 1980), 359-76. Non-cognitivism and Rule-following', in Wittgenstein: To Follow a Rule, S. Holtzman and C. Leich (eds.) (London: Routledge & Kegan Paul, 1981), 141-62. Repr. in McDowell's Mind, Value, and Reality (Cambridge, Mass.: Harvard University Press, 1998), 198-218. `Aesthetic Value, Objectivity, and the Fabric of the World', in E. Shaper (ed.), Pleasure, Preference, and Value (Cambridge: Cambridge University Press, 1983), 1-16. `Values and Secondary Qualities', in T. Honderich (ed.), Morality and Objectivity: A Tribute to J. L. Mackie (London, Routledge & Kegan Paul), 110-29. `Comments on "Some Rational Aspects of Incontinence", by T. H. Irwin', Southern Journal of Philosophy, suppl. 27 (1988), 89-102. `Might There be External Reasons?', in J. Altham and R. Harrison (eds.), World, Mind, and Ethics: Essays on the Ethical Philosophy of Bernard Williams (Cambridge: Cambridge University Press, 1995), 68-85. `Two Sorts of Naturalism', in R. Hursthouse, G. Lawrence, and W. Quinn (eds.), Virtues and Reasons: Essays in Honour of Philippa Foot (Oxford: Clarendon Press, 1995). `Some Issues in Aristotle's Moral Psychology', in Cambridge Companion to Ancient Thought: Ethics, ed. S. Everson (Cambridge: Cambridge University Press, 1998). MACINTYRE, A., After Virtue: A Study in Moral Theory (London: Duckworth, 1981; 2nd edn. (revised), 1984). Whose Justice? Which Rationality? (London: Duckworth, 1988). Three Rival Versions of Moral Inquiry: Encyclopaedia, Genealogy, Tradition (London, Duckworth, 1990). MCKEEVER, S. and RIDGE, M., 'The Many Moral Particularisms', forthcoming. MCNAUGH TON, D., Moral Vision (Oxford: Blackwell, 1988).
Bibliography
309
`An Unconnected Heap of Duties?', Philosophical Quarterly, 46 (1996), 433-47. `Moral Intuitionism', in Hugh LaFollette (ed.), Blackwell Guide to Ethical Theory (Cambridge, Mass.: Blackwell Publishers, 2000) 268-87. MCNAUGHTON, D., and RAWLING, P., 'Unprincipled Ethics', Ch. 11 in this volume.
D., 'Are Moral Principles Really Necessary?', Australasian Journal of Philosophy, 41 (1963), 163-81. MOORE, G. E., Principia Ethica (Cambridge: Cambridge University Press, 1903) MITCHELL,
esp. 28-30. The Sovereignty of Good (Oxford: Blackwell, 1970), esp. 32-3,44. Metaphysics as a Guide to Morals (London: Chatto & Windus, 1992), esp.
MURDOCH, I.,
302,493. T., 'Fragmentation of Value', in his Mortal Questions (Cambridge: Cambridge University Press, 1979), 128-41. NUSSBAUM, M., The Fragility of Goodness (Cambridge: Cambridge University Press, 1986). Love's Knowledge (New York: Oxford University Press, 1990). `Why Practice Needs Ethical Theory: Particularism, Principle, and Bad Behaviour', Ch. 10 in this volume. O'NEILL, 0., Towards Justice and Virtue: A Reconstructive Account of Practical Reasoning (Cambridge: Cambridge University Press, 1996). PARFIT, D., Reasons and Persons (Oxford: Oxford University Press, 1984), 429. `Reasons and Values', Proceedings of the Aristotelian Society, Suppl. vol. 71 (1997), 99-130. PRICHARD, H. A., 'Does Moral Philosophy Rest Upon a Mistake?', Mind, 21 (1912), 21-37. Repr. in his Moral Obligation (Oxford: Clarendon Press, 1949). `Duty and Interest', repr. in Moral Obligation, 21. RACHELS, J., The Elements of Moral Philosophy (New York: Harper Collins, 1993), 114-16,120-38,180-93. RAWLS, J. A Theory of Justice (Cambridge, Mass.: Harvard University Press, 1971), 34. RAZ, J., 'The Central Conflict: Morality and Self-interest', in R. Crisp and B. Hooker (eds.), Well-Being and Morality: Essays in Honour of James Griffin (Oxford: Clarendon Press, 2000), 209-38. `The Truth in Particularism', Ch. 3 in this volume. READER, S., 'Principle Ethics, Particularism, and Another Possibility', Philosophy, 72 (1997), 269-92. Ross, W. D., The Right and the Good (Oxford: Clarendon Press, 1930). Foundations of Ethics (Oxford: Clarendon Press, 1939). SAYRE-MCCORD, G., 'Organic Unities and Actual Duty in Moore and Ross', forthcoming in P. Stratton-Lake (ed.), Moral Intuitionism (Oxford: Clarendon Press, 2001). NAGEL,
310
Bibliography
T. M., What We Owe Each Other (Cambridge, Mass.: Harvard University Press, 1998) 199,225,246,299. SCHAUER, F., Playing by the Rules: A Philosophical Examination of Rule-Based Decision-Making in Law and in Life (Oxford: Clarendon Press, 1991), 77-8, 135-66. SCHEFFLER, S., Human Morality (New York: Oxford University Press, 1992), 43-51. SHAFER-LANDAU, R., 'Moral Rules', Ethics, 107 (1997), 584-611. SHERMAN, N., Making a Necessity of Virtue (Cambridge: Cambridge University Press, 1996), ch. 6, esp. 262-4,268-72. SHINER, R. A., `Aisthesis, Phronesis, and Nous', Philosophical Studies, 36 (1979), 377-87. `Ethical Perception', Apeiron, 13 (1979), 79-85. `Aristotle's Theory of Equity', in Justice, Law, and Method, ed. S. Panagiotou (Edmonton: Academic Printing, 1987), ch. 12. SIDGWICK, H., The Methods of Ethics, 7th edn. (London: Macmillan, 1907). SINNOTT-ARMSTRONG, W., `Some Varieties of Particularism', Metaphilosophy, 30 (1999), 1-12. STEWART, J. A., Notes on the Nicomachean Ethics (Oxford: Clarendon Press, 1892). STOCKER, M., Plural and Conflicting Values (Oxford: Clarendon Press, 1990). STRATTON-LAKE, P., 'Can Hooker's Rule-consequentialist Principle Justify Rossian Prima Facie Duties?', Mind, 106 (1997), 751-58. Kant, Duty, and Moral Worth (London: Routledge, 2000). TEMKIN, L., 'Rethinking the Good' in J. Dancy (ed.) Reading Parfit (Oxford: Blackwell, 1997), 290-345. TORBJORN, T., 'In Defence of Theory in Ethics', Canadian Journal of Philosophy, 25 (1995). WALKER, M. U., Moral Understandings (London: Routledge, 1997). WALLACE, J., Ethical Norms, Particular Cases (Ithaca: Cornell University Press, 1996). WIGGINS, D., 'Deliberation and Practical Reason', Proceedings of the Aristotelian Society, 76 (1976), 29-51. Repr. in Wiggins's Needs, Values, Truth (Oxford, Blackwell, 1987), 215-37. `Truth, Invention and the Meaning of Life', Proceedings of the British Academy, 62 (1976), 331-78. Repr. in Wiggins's Needs, Values, Truth, ibid., 87-137. `Incommensurability: Four Proposals', in R. Chang (ed.), Incommensurability, Incomparability, and Practical Reason (Cambridge, Mass.: Cambridge University Press, 1997), 52-66. WILLIAMS, B., 'Conflicts of Value', in Alan Ryan, ed., The Idea of Freedom: Essays in Honour of Isaiah Berlin (Oxford: Oxford University Press, 1979). Ethics and the Limits of Philosophy (Cambridge, Mass.: Harvard University Press, 1985), 93-119,185-7. SCANLON,
Bibliography
311
`Acts and Omissions, Doing and Not Doing', in R. Hursthouse, G. Lawrence, and W. Quinn (eds.), Virtues and Reasons: Essays in Honour of Philippa Foot (Oxford, Clarendon Press, 1995), 331-40. See esp. pp. 332-3. Repr. in Williams's Making Sense of Humanity (Cambridge: Cambridge University Press, 1995), 56-64, esp. 57. `What Does Intuitionism Imply?', in J. Dancy, J. Morayscik, and C. C. W. Taylor (eds.), Human Agency: Essays for J. 0. Urmson (Palo Alto: Stanford University Press, 1988). Repr. in Williams's Making Sense of Humanity (Cambridge: Cambridge University Press, 1995), 182-91. WINCH, P., 'The Universalizability of Moral Judgements', The Monist, 49 (1965), 196-214. WITTGENSTEIN, L., Philosophical Investigations (Oxford: Blackwell Publishers, 1953). WOODS, M. J., 'Intuition and Perception in Aristotle's Ethics', Oxford Studies in Ancient Philosophy, 4 (1986), 145-66.
INDEX
absolutism 20-1, 163 aesthetics 170-1, 289, 294-5 agent-neutral value 209, 214-16, 218-19, 222, 225-6 agent-relative value 138, 219-20 anthropocentricity of moral properties 161 apodeictic discourse and justification 201-4, 264, 267 approximation 291-2 n. 25 Aquinas 228, 233 Aristotelian moral theory 25, 32, 158-9, 176, 180, 192, 198, 210 n., 276 Aristotle 24 n. 6, 26-7, 31, 39, 100-29, 161, 189, 193, 195, 198, 203, 228, 232-4, 237-8, 242-4, 246, 253, 294 Australian rules football 187-8 Baier, A. 194, 230, 232, 243, 246 Baron, M. 195 n. 35 Bentham, J. 233, 242 bigotry 180 n. 5, 201 Blackburn, S. 279 Borgmann, A. 201 bravery 109-11, 114-16 Brecht, B. 252 bridge laws 278 Brutus 239 care ethics 205, 209 Carnap, R. 97 Cartwright, N. 291 n. 25 casuistry 120-1, 128-9, 199 n. 40 ceteris paribus claims 200, 277, 288, 290-1 Child, W. 15 Chomskyans 197 Cicero 234, 238-9, 244, 250, 253-4 codification 25-7, 31-2, 80, 161, 265, 276, 279, 283, 285, 288, 291, 296-9, 300, 304 coercion 9-10, 270 coherence of agents' lives 174-5 consequentialism 2-3, 205, 215, 217-22, 224-6 see also utilitarianism
context-dependence 157-8, 165, 174, 260, 267, 280, 288-9, 301-2, 303 n. 38 see also context-sensitivity; situationrelativity context-sensitivity 130-1, 259 see also context-dependence; situationrelativity contractualism 2-3, 215 counterfactuals, in moral analysis 45, 64-5, 187, 189 n. 23, 261, 274 Crisp, R. 138 cruelty 26, 38, 40, 173, 273, 278-86, 289, 292-6, 299, 301-2, 304 Dancy, J. 5, 8, 12-13, 24, 32, 34-40, 42-3, 46-7, 49, 57-70, 80, 96, 102, 157, 160-74, 176-7, 181, 206, 257-9, 261-3, 266-7, 272-5, 280, 288, 289 n. 21, 299 n. 35 Davidson, D. 274, 283 defeating conditions 14, 19, 59, 68, 96, 146, 198 n. 40, 262, 265, 271, 291 n. 25, 297 deliberation 45, 57, 61, 63-4, 66, 117-19, 123-7, 145, 246 deontology 138, 173, 179 Dewey, J. 229 Diamond, C. 230-2, 243, 246, 248 disjunctive commonalities 83, 92, 95, 190 n. 25 doctrine of the mean 101, 115-16, 118 Duhem, P. 281 emotions 39, 226, 237-8, 243 see also passions; sentiments the Enlightenment 159-60, 165, 167, 228, 230, 247, 255 ethical theory, attacks on 227, 230-1 marks of 230-6, 240 vs. rules 236-8 evaluative riders 268 exceptions 105, 109 n. 16, 111, 114, 120-1, 302
314
Index
explanation 34-5, 43-4, 55-6, 113, 151-4, 159-60, 200, 285-6, 298-304 familial devotion, moral status of 206-7, 210, 212-14, 218-20, 222-5, 235, 239, 242-3, 245 family resemblances 83, 87 feminism 247, 284 n. 16 Flanagan, 0. 194 forms of life 201, 203 frequencies vs. normative regularities 106-13 full ordering of values 141-2, 145, 149-50 fun 37-8 games 83, 123-4, 187, 190 n. 25, 193, 198, 204 generalities, as summaries 111, 116, 122-4 generosity 211-13, 221-3, 225 Gewirth, A. 208, 215 good-making 7, 49, 80, 140, 280, 285, 289-91, 298-9, 304 Goodman, N. 274 grammer, patterns in 91, 184 n. 13, 198 Grotius, H. 228, 250, 253-4 Habermas, J. 194 n. 30, 215 Hare, R. 160, 162, 164, 199 n. 40, 213, 218-20, 225, 257-66 Harman, G. 36 n. 42, 281 harming, moral status of 3-4, 9-13, 19, 36 n. 42 Hart, H. 88 n. 5 Hegel, G. 244 Heidegger, M. 201-3 heuristics 198, 278, 302 Hobbes, T. 233 holism 13-15, 36, 96-9, 130-9, 141, 151-5, 164-5, 172, 259, 267, 273, 280, 283, 286-7, 298, 300, 304 in choice 141, 143, 147-50 of epistemic reasons 14, 97, 281-2, 296-7 of moral reasons 96, 98, 135, 164, 168, 174, 263, 280, 284-5, 290, 302 of value 137-9, 141-2, 147, 281 honesty 28, 38, 40, 45, 159, 200 n. 42, 273 Hume, D. 198, 228, 232-3, 243 Hypatia 228 idiot savant 304 illegality, moral relevance of 39, 259
impartialist moral theory 205-6, 208, 210, 213-17, 222, 226 impartiality 207, 209-11, 217-18, 220, 222, 224, 243 incommensurability 30-1, 72-4, 77, 166 indifference of independent alternatives 142, 144 inexactness 100-1, 105-6, 110-13, 115-18, 129 intelligibility: of the evaluative domain 50, 53, 55 of reasons 66-7 of value 60 intuitionism 167, 256, 260-3 fat 262-3, 266, 268, 273-5 thick 261, 263, 266, 268, 270, 272-5 thin 262-3, 266, 268, 272-3, 275 invariance in morality 135-7, 261-2 Jackson, F. 181, 187 n., 189 n. 23, 190 n. 25, 193, 283 n. 14 Jonsen, A. 199 n. 40 justice 9, 37-40, 45, 109, 260, 266, 269, 273 Kant, I. 160, 189, 195 n. 35, 207, 209, 227, 229, 232-3, 235, 237-8, 242-4, 246-7, 249-54 Kantian moral theory 2-3, 162 n. 10, 172, 179-80, 194, 205, 225-6, 250 Kripke, S. 191 law 16, 121, 164, 241, 247, 249-50, 253-4 law-like generalities 277-8, 286, 288, 296-9, 303 n. 38 literature 241, 245 n. 13, 246, 248 Little, M. 80, 97 love 207, 214, 217-23, 235-6, 243 lying, moral status of 37-8, 135, 267-8, 283, 289-91, 295, 302 McDowell, J. 24-7, 29, 52, 80, 93 n. 10, 102-3, 122, 160-2, 165-6, 170, 179-81, 185-9, 191-4, 196, 198, 203 n. 49, 262-6, 274 MacIntyre, A. 157-8, 160, 165-9, 173-4, 176-7, 230, 246 MacKinnon, C. 254 McNaughton, D. 80, 293 Marcus Aurelius 228 mathematics 26, 195, 264, 283 Melville, H. 40, 70
Index metaethics 79, 168-9 Mill, J. 233, 236, 243, 249, 251-2, 254 monism of value 31, 167-8 Moore, G. 35 n. 38, 88, 90, 139-41, 167 Moorean moral theory 93, 99 moral aphasia 304 moral character 73-8, 116, 176, 196, 237, 253 moral commitment 172-3, 175 moral dilemmas 244-5 moral discernment 29, 66, 80-1, 241, 292-4, 298, 304 see also moral perception moral dispositions 16-18, 172-3, 175, 182, 246 moral education 15, 24, 26, 28, 58, 160 n. 8, 195, 199-200, 203 n. 49, 234, 241, 243, 247 see also moral pedagogy moral inference 294-6, 298, 304 moral judgement, indispensability of 5, 29, 31-2, 47, 159, 253, 256, 273 moral justification 262, 264, 292 see also moral knowledge moral knowledge 181, 193-4, 202, 204 see also moral justification moral motivation 33, 42-3, 194-6, 211-13, 221 moral pedagogy 58, 195, 199-200, 278 n. 2, 295, 302 see also moral education moral perception 27 n. 14, 48, 63, 101-4, 115, 123, 161, 165, 179, 183, 197-8, 232, 249-50 see also moral discernment moral presumptions 294-6, 298 moral principles 2-4, 102 n. 5, 191, 201, 275, 278 n. 2 strong 264 weak 268, 271, 275 moral prodigies 195 moral properties, as sui generis 88, 90-1, 93, 99 moral psychology 74, 168-9, 171-2, 175, 179, 194, 203 n. 49, 238, 241, 243, 246 moral realism 165, 176, 276 moral significance, vs. moral reasons 172-3, 175 narrative 159, 166-7, 174, 241 naturalism 11-12, 40 n. 51, 141 Newtonian mechanics 35
315
Nietzsche, F. 159 non-cognitivism 161-2 non-linear dynamic cognitive theory 189 n. 23 non-naturalism 11, 52, 90, 161, 279-80 normative priority of principles vs. particulars 102-4, 112-13, 119, 122-5 nous 101, 124 Nussbaum, M. 27, 31,102 objectivity 282 O'Neill, 0. 179-201 organic wholes 35 n. 38, 47 n. 73, 139-40 pain, moral status of 37-8, 67-8, 131, 134, 136, 139, 147-8, 173, 280, 288, 291, 293, 296, 299, 301 Parfit, D. 145 n. 15 partialist moral theory 205, 208 partiality 208-9, 214-15, 218, 220, 226, 241 particularism, as ontological vs. epistemological thesis 179, 181-2, 277, 303 passions 194-6, 238, 243-4, 246, 250-4 see also emotions; sentiments pattern 79-93, 95-6, 99, 161, 190 n. 25, 193 n., 278, 281-3, 296-7, 300, 302 perception 118-29 personal relationships, moral status of 208, 210, 219-20, 235, 242 Pettit, P. 181, 187 n., 189 n. 23, 190 n. 25, 193 n., 283, n. 14 phronesis 101, 160 n. 8, 161, 166 see also prudence physics envy 301 pleasure, moral status of 6-8, 11, 13, 38, 57-60, 67-8, 97, 163-4 pluralism 2-3, 31, 45-7, 82, 95, 164, 177, 216, 220, 244-5 polarity 8, 12, 18, 131-2, 135, 139, 155 see also valence post-modernism 167 post-positivism 281, 302, 304 preference rankings 146-7 Prichard, H. 167 prima facie duties 4, 32, 46-7, 80, 131, 163-4, 171, 266, 271, 274, 303 Pritchard, M. 195 private language argument 191-2 pro tanto moral claims 4, 9-11, 18-19,135, 257, 277, 288-9, 290 n. 23, 291 n. 25, 303
316
Index
projectibility 83, 89, 95, 296 promise-keeping, moral status of 9-10, 17-21, 47, 68-9, 79, 125, 269-72 prudence 38, 101, 123-4, 129 see also phronesis Putnam, H. 94, 286, 300 Quine, W. 281 racism 94, 194 Railton, P. 219-25 rape 284, 249, 252 Rawls, J. 233-4, 249, 256, 259, 261 reasons 12, 33, 44, 59, 61-5, 67, 69, 74-5, 134, 138-9, 152-3, 155 primary vs. secondary 266-7, 271 ultimate vs. non-ultimate 37-40, 47 reflective equilibrium 167, 169, 234 relativism 41, 166, 245 response-dependence 92-4, 99, 193 n. 29 resultance 140, 299 n. 35 Ross, W. 46-7, 131, 160, 163-4, 167, 171, 198 n., 257, 264, 266-71, 274-5, 303 Rossian generalism 3-6, 7, 17, 19, 21, 32, 44, 47, 267 Rousseau, J. 233 rule-fetishism 169 rule-following 25-6, 32, 160 n. 8, 178, 186, 191 n. 27, 192, 263-4, 267, 283 rule-nihilism 181, 189 rules 24-5, 27, 29, 31, 170, 179, 186-8, 189 n. 23, 191 n. 26, 192, 196-200, 238-41, 269 rules of thumb 56, 295 Russell, B. 229 Sacks, O. 303-4 sadism 7, 8 salience 28-30, 122, 125-6, 165-6, 173, 241, 285, 295, 298-9, 304 Sartre, J. 42 Scanlon, T. 138, 215 science, generalities and explanation in 34 n. 36, 35, 108, 111-13, 125, 129, 152, 197, 286, 291 n. 25, 301 Sellars, W. 178, 200 semantic competency 81, 86, 93, 96, 99, 190-1, 282-3 Seneca 228, 231-2, 236, 241, 250, 254 sentiments 194-6, 203 n. 49, 214, 218-19, 221, 246 see also emotions; passions
sexism 94, 247 sex relations 248-51, 254 sexual harassment 252, 254 shapelessness 40 n. 51, 83, 93, 96, 99, 162, 266-7, 269, 274, 279, 285-6, 302 Sidgwick, H. 233-4, 236, 242, 244 silencing 46 n. 68 situational appreciation 122-9, 176 situation-relativity 1, 27, 29 see also context-dependence; contextsensitivity skills 26, 120, 126, 158, 169-70, 283, 292-8, 301, 304 Slote, M. 138 Smith, A. 233, 243, 253 Smith, M. 181, 187 n., 190 n. 25, 193 n., 283 n. 14 Socrates 37, 228, 234, 240 Spinoza, B. 233 statistical generalities 107, 109, 296, 302 n. stealing, moral status of 3-5, 10, 19, 200 Stoic moral theory 228, 239-40 Stoics 231, 233, 238, 242-3, 251 supervenience 6, 50-5, 8 91, 140-1, 187 n. , 257-8, 263, 265-6, 280-1, 285-7, 300 switching arguments 6, 145 systematicity 232-3, 250, 252, 254, 278 Tannsjo, T. 165-70, 174 Tännsjo, Taylor, C. 160 n. 8 teleological regularities 106-9, 113 Thesiger, W. 144 `thick' moral terms 82, 165, 172, 180-2, 260-1, 266, 269, 272-4, 289 `thin' moral terms 81-2, 181, 258, 274 Toulmin, S. 199 n. 40 tropes 299 n. 35 two-level view of morality 213-16 umpires 187-8, 191, 198 Unger, P. 194 universalisability 34 n. 37, 40-2, 47, 70-1, 73, 77-8, 162-3, 179, 180 n. 5, 217, 235, 258 Urmson, J. 259, 264, 266, 271 `usual' generalisations 100, 105-7, 109-15, 117, 121, 302-3 utilitarianism 28-30, 39, 79, 84, 97-8, 218, 220, 235, 237, 244, 246 see also consequentialism
Index valence 97-8, 257-68, 272-4, 280, 288-9, 290-1, 295, 303 see also polarity value, connection to reasons 138-9 Vere, Captain 40-2, 70-6 vertigo 264 virtue and happiness 109-11, 116 virtue theory 2-3, 28-9, 39, 45, 159, 176, 179, 182, 205 n., 209, 220 war 250-1
317
`whirl of organism' 25-6, 283 Wiggins, D. 30-1, 40, 122-3, 126-7, 176 Williams, B. 180, 210 n., 220, 229 n. 9, 230, 233-4, 242-3, 246, 248, 254, 260 Winch, P. 40-1, 49, 70-7 Wittgenstein, L. 24-5, 27, 32, 83, 87, 103, 186, 189-93, 198, 200, 203, 263-4, 282-3 Wittgensteinian epistemology 180, 185, 194, 198
11 11 1111123883 6 9 780198
