Knowing Better: Virtue, Deliberation, and Normative Ethics [Hardcover ed.] 0199570418, 9780199570416

Citation preview

Knowing Better

OXFORD PHILOSOPHICAL MONOGRAPHS Editorial Committee William Child, R. S. Crisp, A. W. Moore, Stephen Mulhall, Christopher G. Timpson OTHER TITLES IN THIS SERIES INCLUDE

Potentiality and Possibility: A Dispositional Account of Metaphysical Modality Barbara Vetter Moral Reason Julia Markovits Category Mistakes Ofra Magidor The Critical Imagination James Grant From Morality to Metaphysics: The Theistic Implications of our Ethical Commitments Angus Ritchie Aquinas on Friendship Daniel Schwartz The Brute Within: Appetitive Desire in Plato and Aristotle Hendrik Lorenz Plato and Aristotle in Agreement? Platonists on Aristotle from Antiochus to Porphyry George E. Karamanolis Of Liberty and Necessity: The Free Will Debate in Eighteenth-Century British Philosophy James A. Harris

Knowing Better Virtue, Deliberation, and Normative Ethics

Daniel Star

1

3

Great Clarendon Street, Oxford, OX2 6DP, United Kingdom 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. Oxford is a registered trade mark of Oxford University Press in the UK and in certain other countries # Daniel Star 2015 The moral rights of the author have been asserted First Edition published in 2015 Impression: 1 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, by licence 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 work in any other form and you must impose this same condition on any acquirer Published in the United States of America by Oxford University Press 198 Madison Avenue, New York, NY 10016, United States of America British Library Cataloguing in Publication Data Data available Library of Congress Control Number: 2014959431 ISBN 978–0–19–957041–6 Printed and bound by CPI Group (UK) Ltd, Croydon, CR0 4YY Links to third party websites are provided by Oxford in good faith and for information only. Oxford disclaims any responsibility for the materials contained in any third party website referenced in this work.

To my mother, Jan, and to the memory of my father, Kol

Preface This book grew out of thinking about two strongly held convictions, which many of us share, that appear to conflict with each other. On the one hand, sophisticated philosophical reflection is unnecessary for ordinary virtue. Whether we are considering the kindest and most courageous of our friends and loved ones, or more distant ethically admirable historical figures, it seems clear that, although far from perfect, such people are able to be good, and reliably so, in a way that does not depend on them being guided by ethical theories or principles of a particularly abstract kind (indeed, we might often think they are virtuous or ethically good people despite their commitment to flawed principles, at least with respect to some of the principles they seem committed to). On the other hand, the reflective interrogation of arguments for and against particular first-order ethical theories (and their different parts) seems to be a highly fruitful enterprise—or, at least, to a person who immerses him- or herself in this philosophical enterprise in a sincere fashion, it will not seem that it is about questions that have either easy answers or no answers at all. To reach a point of knowing such questions have no satisfactory answers would itself require an extremely significant philosophical achievement. Every attempt to provide principled answers to ethical questions runs up against obstacles, yet one’s conscience directs one to think that, at least in many cases, there must be such answers. Some philosophers seem willing to claim—even outside of rhetorical spaces that may pragmatically require us to assert our theories as if we know them to be true—that their particular consequentialist or Kantian principles of right action are simply correct. Such claims seem terribly premature at this time, and appear to demonstrate a lack of epistemic humility in relation to an extremely important topic. These two convictions seem to be in tension because it is difficult to see how the process of critical, philosophical reflection on very general, fundamental ethical principles could end in us endorsing principles that happen to be the same as those that good people who are not philosophers are already following. Yet it is just as difficult to see how such people could really count as good people, and reliably so, if they are not

viii

PREFACE

being guided by or responding to such principles. One might try to shrug off the appearance of conflict by claiming that such people unconsciously or tacitly follow the principles we could all eventually come to accept intellectually, but this seems like a truly incredible possibility when one recognizes that such principles build on an extremely diverse and complex array of surprising arguments and counterexamples (many of which, in all likelihood, are still unanticipated). Ethics does not merely describe our prior commitments, but instead forces us to consider taking on new commitments. So we have a problem. To bring out the sense in which there is a prima facie conflict here it can help to have a particular ethical theory in mind. Hedonistic utilitarianism—merely to use a fairly simple example for the sake of illustration—might be construed as a theory that says the normative reasons that are able to justify the actions of a good person must all have to do with the bringing about of states of affairs that contain pleasure or pain (on this picture, these are the only practical reasons). Yet the normative reasons that good people might typically cite as guiding their actions seem to cover a multitude of other considerations. One might think that utilitarianism, considered alone, has the resources to adequately respond to this concern. Utilitarianism might direct people to falsely believe that, or act as if, all sorts of facts are reasons for action; doing so might actually make people more likely to act rightly than they would be if they were to try to focus in their deliberations only on matters concerning pleasure and pain. Or it might be said that utilitarianism, properly understood, can allow that many facts other than facts directly having to do with pleasure and pain are reasons, albeit derivative reasons, in virtue of standing in appropriate causal or constitutive relations to the fundamental reasons that are facts about pleasures and pains. Despite the fact that I am not a utilitarian, I have some sympathy with the content of these responses (after all, all plausible ethical theories need to allow room for there to be derivative reasons, as well as room for the good to sometimes come apart from the true). However, these responses do not really solve our puzzle. Neither response gives us any reason to think that the virtuous person is ever more than haphazardly and luckily getting things right, whereas the first of our convictions appeals to reliability, and might even be said to involve the idea that the virtuous are knowingly getting things right. Good, rational people are guided in

PREFACE

ix

their actions by knowledge; at least, they do not just stumble occasionally (or even stumble over and over again) on the right answers. One might deny that there is any problem here at all by claiming that there are no ethical truths of the kind assumed by either one or both of the convictions described above. Most generally, one might hold to an ethical nihilism, or error theory, that would deny that there are any ethical truths, or a subjectivism or relativism that would make whatever ethical truths there are vary in a way that does not allow for a high degree of independence between the attitudes of particular agents, or their cultural milieus, and the ethical truths that apply to them. I cannot here argue against such views, but will assume, for the sake of argument, that they are false. Nor will I be concerned with comparing the robust realist and quasi-realist accounts of ethical truths that are the natural alternatives to the antirealist options just mentioned. It suffices for my project that I be permitted to assume there are ethical truths that are objective, and can be known. In the next chapter, I will try to motivate the thought that many of these truths take the form of objective principles yet unknown, and that we should not be satisfied with another view that attempts to avoid the problem I have described (by effectively denying the first of our convictions), but at too high a cost, viz ethical particularism. From this brief description of the problem I grapple with here—the problem of squaring two common, significant convictions—it may already be apparent that a solution to it will be concerned with: (1) the general relation of ethical theory to practice; (2) the nature of normative reasons (the considerations that guide and justify people’s actions); (3) the nature of virtue (the dispositions required for acting well); and (4) the knowledge that is required for acting well. Each chapter of this book corresponds to one of these topics. Although the chapters have been crafted to be read in order, each could, in fact, be read in isolation from the rest without this causing any particular difficulties; there is some repetition across chapters (although not a lot), and this will benefit readers who wish to pursue the option of reading a chapter out of order. I have no doubt already raised many questions in the reader’s mind. Chapter 1 is dedicated to describing the basic problem in more detail, in a way that hopefully addresses some of these questions, and to sketching my solution to the problem. Part of what is required for that solution to

x

PREFACE

work is that we construe ethical thinking as having a two-level structure—one for ethical practice, and another for philosophical ethical theory. I acknowledge that the idea that there are two levels of ethical thinking is not, in itself, a new idea. It has a particular history in the consequentialist tradition, and I will briefly explore a recent episode in that history. I contend this tradition has not provided us with a satisfactory account of the two levels of ethical thinking, because it has not paid sufficient attention to the need to credit non-philosophical agents with the knowledge of the virtuous. For all that, consequentialism, construed as a theory of right action, might be correct. I aim, for the most part, to be neutral on the question of which of several general first-order ethical theories is correct, although, in Chapter 3, I argue against a claim that some take to be fundamental to virtue ethics, in order to clear the way for my own account of virtue. It will also become clear that since I am inclined to think we possess much quite ordinary ethical knowledge, I am also inclined to be suspicious of theories (such as hedonistic utilitarianism) that appear to conflict too extensively with widely shared moral convictions. In Chapter 2, I provide a new argument for the account of normative reasons that I favor, reasons as evidence. A number of arguments have previously been provided by myself and a coauthor, Stephen Kearns, for this account. A key reason for focusing on the new argument here is that discussion of it, and the considerations concerning the nature of normative authority that lead to the articulation of one of its main premises, helps to bring out what I take to be a key attraction of this account of reasons. It is, I believe, an account of reasons that best reconciles the two roles that the concept of a reason is meant to play. Reasons justify action, by explaining why certain acts are right, and they also guide our deliberations and action. Reasons as evidence best explains how there can be two levels of reasons—one whose primary (but not sole) role is to justify, and one whose primary (but not sole) role is to guide us, and best explains how these two levels are related. I draw freely here on Joseph Raz’s well-known service account of authority, without attempting to be loyal to it, and argue for a more general, minimalist account of authority, which applies as much to reasons for belief as it applies to reasons for action. Chapter 3 provides a new theory of virtue that utilizes this account of reasons and is neutral when it comes to disagreements between

PREFACE

xi

consequentialist and deontological ethical theories. This theory of virtue accepts the traditional thought that the virtuous always act on good reasons, yet departs from tradition, insofar as it makes virtue less ethically demanding (hence the ideal of being virtuous more realistic), opens up space for fundamental ethical principles regarding right action that are independent of the virtues, and, most controversially, has it that the practical virtues are essentially much more epistemic in nature than traditional accounts of virtue would allow. I criticize accounts of virtue that tie the right or the good too closely to virtue; I do not deny that the concepts of virtue, right, and good are related, but I argue that virtue is more directly tied to reasons, and connections to the right and the good are indirect. I also argue that, absent some very special conditions (which I make explicit), to act virtuously is to act blamelessly. Due to the fact that virtue does not require access to fundamental ethical principles and reasons, moral ignorance can sometimes exculpate (this is much more contentious than the widely accepted view that non-moral ignorance can exculpate). Simply put, a virtuous person is a person who does her best to respond to reasons, and in so doing, does her best to act rightly. It is not the case that her actions are always right, nor that the relevant character states are best understood as being virtues because of good outcomes that follow from possessing them. My main claim about the essence of being virtuous may seem like a reasonably uncontroversial alternative to the claims of the other views I consider, but combining it with reasons as evidence leads to a strong conclusion that I call the primacy of the epistemic (practical reason is largely epistemic in nature), in order to contrast it with the broadly Kantian idea of the primacy of the practical. I end this chapter by considering an important objection to the proposed account of virtue, which is that it fails to adequately fit with many of our moral intuitions, insofar as it allows that an epistemically conscientious agent might be classified as virtuous even though she might have radically misleading evidence concerning what she ought to do (the fact that her evidence is misleading might well lead her to do many bad acts, out of moral, and not just non-moral ignorance). I discuss two possible responses to this objection, the more attractive of which requires us to add an additional feature to the account of virtue, a feature I will provide additional reasons for adopting in Chapter 4. This addition to the thought that the virtuous do their best to respond well to

xii

PREFACE

reasons is that the virtuous draw on a substantial body of ordinary ethical knowledge in their deliberations. This is not an ad hoc addition, since it simply makes explicit a widely shared, pre-theoretical assumption about the virtuous. Throughout much of the book I rely on intuitive judgments that certain facts count as evidence that we ought to do certain acts, and that they do so in weak or strong ways. The fact that a person is in considerable pain in front of me is, intuitively, strong evidence that I ought to help him or her, while the fact that I have promised to return a book to a particular person seems to provide weaker evidence that I ought to return the book to them (especially in a case where these reasons conflict because one can only do one of these two things). In Chapter 4, I try to say more about what exactly might make a fact evidence that one ought, and what might explain why it is evidence of a certain strength. I suggest that a variation on Timothy Williamson’s popular account of evidence is particularly well-placed to provide a general explanation of evidence-that-one-ought that fits together well with reasons as evidence. I do not offer new arguments in favor of this knowledge as evidence account of evidence; rather I suggest that the closely related idea (defended by Williamson, as well as others) that knowledge is the norm of practical reasoning and action lacks, by itself, the resources to explain why it is appropriate to take some facts to be reasons, even though it is not appropriate to act on them (when they are outweighed). Reasons as evidence provides the missing ingredient for this account of the normative connection between knowledge and action. And I contend that when we attempt to elucidate what is involved in having a reason, knowledge is an excellent candidate for a necessary condition (knowledge of the fact that is the reason) for possessing a reason. It is then natural to suppose that a fact can only be a reason for an agent if it is possible for the agent to know that fact. A problem can be posed for this way of providing an explanation of evidence-that-one-ought. One might worry that Hume’s famous claim that it is impossible to validly infer an ought from an is will generalize to evidential relations. I suspect that if I were to stop with the explanation of evidence-that-one-ought at the point where I simply say it is knowledge that explains such evidence, this would seem to many to be unsatisfactory for reasons related to Hume’s stricture. It is not obvious that this is a genuine problem, but supposing that it is, I think the natural thing to do

PREFACE

xiii

at this point is to focus on background ethical knowledge. The solution to this problem, assuming one is needed, is fortunately the same solution needed for the separate problem encountered at the end of Chapter 3. In both cases, one can appeal to basic ethical knowledge of a sort that it is highly plausible that non-philosophical virtuous agents already possess. It is necessary that this knowledge be of a non-inferential variety (otherwise it does not solve the problem involving Hume’s stricture). Fortunately, there are good candidates for non-philosophical non-inferential ethical knowledge, and I discuss the general forms such knowledge might take. I admit that I am not in a position to demonstrate that we have such knowledge (other than by saying things like “of course I know that lying is normally wrong”). However, even if it is the case that no such knowledge is ultimately available, we might still be in a position where it is rational, in a minimal sense that goes with being epistemically blameless, to take ourselves to have it, and act on the reasons that this apparent knowledge (in partnership with more mundane empirical knowledge) appears to generate. Finally, I address a concern that reasons as evidence and alternative accounts of reasons that focus on the rightmaking role of reasons might, in the end, just be talking about very different things.

Acknowledgments I am extremely grateful to all of the philosophers I was fortunate to receive advice from during the gestation of ideas that led to the writing of this book, as well as after various versions of various parts of it were written. It is quite likely that I forget to mention below some people who deserve to be thanked for asking me excellent questions, making useful suggestions, or simply providing me with helpful support along the way—I apologize in advance if you are one of these people. I am, in the first instance, particularly indebted to Robert M. Adams, Krister Bykvist, Roger Crisp, Brad Hooker, and Ralph Wedgwood, all of whom generously provided me with extensive feedback on my doctoral dissertation (a distant ancestor of the present work; there is no overlapping text). I feel especially lucky to have had so many helpful and interesting conversations with Roger and Ralph over the years, and they have both been superb teachers and mentors. In addition, John Broome, David Charles, Derek Parfit, Helen Steward, Joseph Raz, and Timothy Williamson all taught me a great deal at Oxford. With respect to the title of this book, thanks must go to Jessica Wolfendale for suggesting it during a conversation with her about my project. I am also grateful to her for many other conversations that came with the friendship we developed after I returned to Australia from Oxford to take up a research fellowship at the Australian National University. For this fellowship, I’d like to thank the Center for Applied Philosophy and Public Ethics. Christian Barry, David Chalmers, Steve Clarke, Daniel Cohen, Jeanette Kennett, Holly Lawford-Smith, Declan Smithies, and Nicholas Southwood all helped make Canberra a very welcoming place, conducive to doing good philosophical work. Conversations about my project with Nic Southwood were especially helpful, which will come as no surprise to philosophers who have been fortunate to have had the opportunity to discuss their work with him. I am also grateful to everyone who has provided me with advice or assistance in the philosophy department at Boston University. I received useful feedback on part of Chapter 3 from Paul Katsafanas and David Liebesman. Daniel Dahlstrom, Aaron Garrett, Charles Griswold,

xvi

ACKNOWLEDGMENTS

Manfred Kuehn, and David Lyons have all been helpful senior mentors. Closer in career stage, Walter Hopp and Susanne Sreedhar have proven themselves to be two of the most supportive colleagues anyone could possibly hope for. This is also an appropriate place to thank Candice Delmas for commenting on the material that I wrote for a coauthored paper during the period she was a graduate student in the department, and that I have drawn on here (details are provided below). I would like to thank John Gardner, University College, Oxford, and the Oxford Centre for Ethics and Philosophy of Law, for providing me with a H.L.A. Hart Fellowship, which enabled me to stay and work in Oxford during the summer of 2012. I also appreciate the semester of teaching relief provided to me in 2011 as part of a research fellowship by the Boston University Center for the Humanities. I have given talks related to material that appears in this book in a number of venues. I appreciate the feedback I received from audiences at Australasian Association of Philosophy Conference, La Trobe University; Australian National University; Boğaziçi University; First Annual Normative Ethics Workshop, University of Arizona; Florida State University; Fordham University; Harvard University’s Safra Center for Ethics; New York University; Princeton University; Queen’s University, Ontario; Rocky Mountain Ethics Conference, University of Colorado; Stockholm University; Syracuse University; University of Kent; University of Maryland; University of Massachusetts, Boston; University of Montreal; University of Oxford; University of Texas, San Antonio; University of Vermont; and Washington University, St. Louis. I have benefitted a great deal from conversations about my work with a number of philosophers. In particular, I would like to take this opportunity to thank Brian Ball, John Brunero, Garrett Cullity, Terence Cuneo, Tyler Doggett, Julia Driver, Geoffrey Ferrari, Simon Keller, Errol Lord, Julia Markovits, Hille Paakkunainen, Michael Smith, and Kurt Sylvan. Ofra Magidor has been a particularly supportive friend since the days when we were both philosophy graduate students at Oxford, and Andrew Reisner has been a particularly supportive friend in North America. At a critical juncture, Kate Manne met up with me in Boston to discuss drafts of a couple of chapters. I am also grateful to Clayton Littlejohn for providing helpful comments on a close to final draft of Chapter 4, and Daniel Elstein for providing helpful comments on part of a close to final draft of Chapter 3.

ACKNOWLEDGMENTS

xvii

I owe a special debt of gratitude to my good friend Stephen Kearns, who continually inspired and challenged me throughout graduate school, and has continued to inspire and challenge me since then, although we naturally see less of each other these days. The reasons as evidence thesis I put to work here is one that he and I hatched together not long before I finished my graduate studies in Oxford (it did not feature in my dissertation). In our coauthored papers to date, we have defended that thesis in a fairly narrow fashion, whereas here it is presented as a central component of a larger account of things. That being said, this book owes a great deal to conversations that I have had with him over the years. I count myself extremely fortunate to be married to the art historian extraordinaire Emine Fetvacı, who provided me with much crucial support and assistance while I was working on this book. The fact that she managed to do this while she was pregnant and going up for tenure (successfully), as well as during the period in which we began raising a child together, has left me without words to describe how grateful I am to her for her quite extraordinary kindness and patience. Our daughter Dilara has provided her own form of support by being such a happy and affectionate child (long may her happiness continue!). I am dedicating this book to my mother, Jan, and the memory of my father, Kol. My father died while I was still a teenager, well before I even considered the possibility of becoming a professor. He was himself at one time an academic (he received a PhD from the University of London), and it would have been wonderful to have been able to talk to him about so many matters, as one adult to another, but sadly that was not to be. Fortunately, my wonderful, extremely capable mother (Member of the Order of Australia, no less, for her own accomplishments) has always been there for me, as have my siblings, Alan and Jenni. I am glad to have this opportunity to thank the anonymous referees for Oxford University Press, for their useful comments, as well as Peter Momtchiloff, for being a patient and helpful editor. Thanks must also go to two journals that are allowing me to reproduce material from papers published in their pages: Chapter 1 draws on, but advances considerably beyond, an Oxford Studies in Normative Ethics paper, “Two Levels of Moral Thinking” (Star 2011), and Sections 3 and 4 of Chapter 2 draw on sections of a coauthored paper—sections that I authored alone— originally published in Jurisprudence (Star and Delmas 2011).

Contents 1. Two Levels of Ethical Thinking 1. 2. 3. 4. 5. 6.

Introduction Hare and Williams on Two Levels of Moral Thinking Two Levels of Normative Ethical Reasons The Virtuous Williams’s Concern Ethical Particularism

2. The Authority of Reasons 1. 2. 3. 4. 5. 6. 7.

Introduction Normative Reasons From the Service Conception to the Minimalist Conception A Challenge The Authority of Reasons, as Evidence The Razian Insight and a New Argument for Reasons as Evidence Derivative, Preemptive, and Instrumental Reasons

3. Virtue 1. 2. 3. 4. 5. 6. 7. 8.

Introduction Virtue-responsive Evil Demons Reasons, the Virtuous, and Modesty A Reduction Plan The Reasons of Virtue and the Primacy of the Normative Reasons as Evidence, the Virtuous, and the Primacy of the Epistemic Comparing Analyses of Particular Virtues A Problem and a Solution

4. Knowing Better 1. Introduction 2. The Norm of Practical Reasoning 3. The Norm of Practical Reasoning, Reasons as Evidence, and Knowledge as Evidence 4. Ethical Knowledge 5. The Reasons that There Are 6. Conclusion

References Index

1 1 8 12 21 25 31 36 36 38 40 46 51 55 64 68 68 71 80 85 91 97 100 102 106 106 110 115 120 125 135 139 145

1 Two Levels of Ethical Thinking 1 Introduction The purpose of this chapter is to introduce and begin defending a twolevel account of ethical thinking that does justice to both of two very plausible claims. These are claims which pull our thoughts in different directions, and may appear to be inconsistent: (1) People can be ethically virtuous without the aid of philosophy, where we understand people to be ethically virtuous only if they non-accidentally (or knowingly) act for genuine normative reasons, and typically work out what it is that they ought to do on the basis of considering such reasons. (2) Philosophers engaged in the project of articulating and defending normative ethical theories are not wasting their time when they search after very general ethical principles which could not be discovered or, at least, justifiably accepted in a direct fashion through non-philosophical thinking, and which (amongst other things) specify the reasons that justify the acts that virtuous people do (when they are acting rightly), as well as provide criteria for determining what it is that people ought to do. I hope that the first part of the first claim—that people can be virtuous without philosophy—will strike the reader as clearly true, at least on the assumption (which I allow myself ) that it ever makes sense to talk of people as being virtuous at all.1 This claim is partly motivated by the idea that it is 1 Despite much recent skepticism regarding the virtues, which has its origins especially in certain findings in social psychology, the ideal of virtue remains attractive. Arguably, this skepticism undermines certain traditional accounts of the virtues, rather than



TWO LEVELS OF ETHICAL THINKING

important to avoid an odious intellectual elitism which would have it that only philosophers can hope to be good people. The clarification of what it is to be ethically virtuous provided in the rest of the first claim simply gives expression to a highly attractive idea that can be traced back to Aristotle: virtuous people are not mentally disconnected from the ethical reasons for which they act, and are not only luckily getting things right; rather, a virtuous person does the right thing for the right reason.2 We might well wish to express this thought by saying that a virtuous person knowingly does the right thing. Although this is a stronger claim, it is appealing because it captures a common thought that the virtuous are knowledgeable, and an appeal to knowledge may provide the basis for a good explanation of the reliability of the judgments of the virtuous. The other claim, (2), seems highly attractive, at least to many philosophers, because normative ethics appears to be a valuable research program where we are making genuine, if slow, progress in articulating and defending fundamental ethical principles. This research program suggests that, in all likelihood, the correct ethical principles are not only such that we could not have begun by thinking of them as obviously true, but are also such that they clearly reflect and demand a degree of intellectual sophistication and complexity that is only possible to achieve and place trust in after extended philosophical reflection.3

contemporary theories that do not set the bar for ethical virtue too high, and that appreciate that the virtues are commonly fragmented, fragile, and lacking in the type of unity that the traditional doctrine of the unity of the virtues supposes them to possess (see Adams 2006, for an excellent example of such a contemporary account, and for references to some of the relevant literature). 2 I say the virtuous are not mentally disconnected from the reasons for which they act, when it might be simpler to say they are not unaware of the reasons for which they act, because the first is a weaker claim. The crucial thing is that virtuous people act on genuine normative reasons. I will say more in Chapter 4 about what might be involved in acting on a reason. 3 I do not mean to deny that it is, in some broad sense, possible to stumble across sophisticated ethical principles and understand them. However, what we look for in normative ethics is not mere acquaintance with some such principles, but good reasons to believe them. And I am assuming throughout that it is not necessary for us to be concerned with cases of people coming to know fundamental ethical principles through testimony, since such cases, assuming they are possible (I do think they are possible, although some philosophers have argued that it is a special feature of moral knowledge that they are not possible), would all require that some people know the fundamental ethical principles in a way that is not based on testimony. This is why I include the qualification “in a direct fashion” in (2).

TWO LEVELS OF ETHICAL THINKING



Considered together, our two highly plausible claims appear to give rise to an inconsistency, because it seems that either the virtuous cannot non-accidentally act on genuine ethical reasons—because we will only know these reasons when we do philosophy and discover the nonobvious principles that tell us what our genuine ethical reasons for action are—or we do not need philosophy to discover the correct ethical principles, for the virtuous can already know them.4 In order to reconcile and make good on both of our two major claims, I will suggest that we need to adopt a particular way of thinking about virtue, as well as a particular two-level account of normative reasons. Here and in later chapters, I will try to show that each of these is independently attractive, quite apart from the role they can play together in avoiding the problem that I just described. Although this chapter is intended to be schematic, it should at least become clear here how adopting these accounts of virtue and reasons might enable us to avoid the apparent inconsistency in the above pair of claims, and also put us in a good position to identify which types of ethical statements are good candidates for knowledge prior to philosophical reflection, and which are good candidates for knowledge only after philosophical reflection. I aim to show that there is actually no inconsistency or tension involved in accepting this pair of prima facie very attractive 4 Another option is to deny that there are any true and non-obvious ethical principles waiting to be discovered, which is to deny (2) (see, for example, Dancy 2004). I will focus on particularism in the last section of this chapter. Not everyone will think that the apparent inconsistency of the two central claims constitutes a problem, because not everyone will find both claims attractive, but I think a great many people would do so, after due reflection. It is a very interesting question how W.D. Ross’s pluralist deontology (as outlined in Ross 2002, 16–64), which purposely stays very close to commonsense morality, should be interpreted in relation to (2). One might well take Ross to be denying (2), insomuch as it can seem like he is not offering up new (justification for) ethical principles at all. It may appear that Ross is best interpreted this way when one focuses on: (i) his claim that it is impossible to provide principles that specify what we ought to do when prima facie duties conflict and (ii) his descriptions of particular prima facie duties, which are such that they correspond very closely to ethical duties or reasons recognized by ordinary, non-philosophical ethical thinking. On such a reading, there is not much left for us to do in normative ethics, in relation to questions about the fundamental reasons and principles of ethics, at least, and (1) seems straightforwardly true. This account of things is quite unsatisfactory, for reasons that should become apparent in the following discussion. However, there is no reason to suppose that more sophisticated pluralist deontological theories that take their inspiration from Ross need accept (i) and (ii), so they need not reject (2). A pluralist deontologist need not accept that no principles can be provided to resolve conflicts, nor need she accept that all fundamental reasons are to be located in commonsense morality.



TWO LEVELS OF ETHICAL THINKING

claims. There is no need to repudiate or weaken either claim. I will end this chapter by saying why I think that the account of reasons I favor is well placed to capture the grain of truth in particularism. This is important because particularism challenges the second main claim. A separate challenge to the first claim—that virtue does not require its possessor to respond well to genuine reasons—is, in effect, dealt with in Chapter 3, where I argue for an account of virtue that puts reasons responsiveness at center stage. Perhaps you are not sure what to make of the claim that normative ethics is a valuable research program, or find it implausible that philosophy might make it appropriate to articulate and defend sophisticated ethical principles that go beyond pre-philosophical ethical commitments. If so, I recommend carefully considering some of the particular ethical problems, and productive discussions of them, that arise in the normative ethics literature. I take normative ethics seriously, not because I am convinced that it is a part of philosophy that has, to date, been particularly successful at finding convincing solutions to moral problems (it is certainly consistent with (2) above, and perhaps appropriate, to think that normative ethics is still in its infancy),5 much less because I desire to boost the credentials of philosophers who work in this area (as someone might if they had the aim of establishing a class of professional moral experts). Rather, I am impressed by the serious and often surprising nature of the philosophical problems that arise in this field. Let me here mention just three examples that we find in late-twentieth and early-twenty-first century normative ethics. It is not essential that every well-informed reader think that all three examples are, in fact, good examples of genuine philosophical problems that have consequences for how we think about the fundamental reasons of ethics. Perhaps some readers will think they know of better examples, in which case they should feel free to focus their mind on such examples, and instead ask, about such examples, whether they think it is realistic to suppose

5

As Derek Parfit is keen to emphasize in the concluding sentences of Reasons and Persons (1987, 454): “Disbelief in God, openly admitted by a majority, is a recent event, not yet completed. Because this event is so recent, Non-Religious Ethics is at a very early stage. We cannot yet predict whether, as in Mathematics, we will all reach agreement. Since we cannot know how Ethics will develop, it is not irrational to have high hopes.”

TWO LEVELS OF ETHICAL THINKING



virtuous people must know what they ought to do, or even what the fundamental reasons in play are, in the relevant circumstances.6 First, consider the literature, replete with a multitude of carefully chosen thought experiments, that has grown up around a perceived need to develop principles to replace the traditional doctrines of double effect, and doing and allowing. I am thinking, in particular, of the sophisticated deontology of the form that gets its initial inspiration from Philippa Foot (2002, but originally 1967), and is pursued in earnest by Judith Jarvis Thomson (1976, 1985), Warren Quinn (1993), and Frances Kamm (2007), amongst others. The reader might think it tempting to scorn the fascination with so-called “trolley cases” that one finds in this literature. However, the common use of such cases there should not be dismissed as reflecting some bizarre taste for strange, entertaining scenarios that none of us are ever likely to encounter (although these philosophers often do demonstrate a good sense of humor). Such a reaction would be superficial. Rather, we are talking about very careful, sophisticated attempts to understand distinctions and associated claims and principles that arguably capture some part of the basic structure of moral reality. It is true that philosophers who reject this type of deontology or nonconsequentialism are likely to find many of the judgments about cases that are appealed to in this literature somewhat suspicious. Nonetheless, I think one can (and should) charitably appreciate what is being attempted here, and when one does so, it is clear that the apparent philosophical problems in this area, and the principles developed in order to deal with them, are subtle and sophisticated (this is nowhere more true than with respect to Kamm 2007). Second, consider the growing literature on the nature of well-being, and the several distinct theories of well-being that we find there (each of which comes with its own philosophical arguments and problems). Following Parfit (1987, 493–502), but simplifying a little, one might think we have arrived at the present situation. Hedonism gives way, for many contemporary philosophers, under considerations concerning the undesirability of a fake life lived within an experience machine. The alternative preference satisfaction theory flounders when we consider 6 Here, as elsewhere in this book, the question is not one concerning whether the virtuous might sometimes lack knowledge of non-ethical facts, but whether they might lack knowledge of some fundamental ethical facts or principles.



TWO LEVELS OF ETHICAL THINKING

people who want nothing more than to count blades of grass (and are fortunate to get to count them). Objective list theories face the concern that the goods that might be placed on any objective list, apart from pleasure and preference-satisfaction, seem to be goods that some possible agents might not want at all, and it is hard to see how such agents would have their lives improved by receiving such goods. Finally then, it is very tempting to look towards hybrid theories that claim that, in order for goods to benefit us, such goods need to feature on an objective list, but they also need to be wanted or enjoyed if they are to count as making a person’s life better. This is a very recent and now popular way of setting up the options in this area. The first three theories have a long history, and I do not mean to suggest that it is clear that the useful story of progression through theories just told objectively corresponds to a progression towards the truth. All of these theories still have their defenders, and this general research program continues to flourish (see, for instance, Heathwood 2006 and Bradley 2009). Insofar as ethics is concerned with questions to do with how our acts may affect the welfare of ourselves and other people, the project of locating fundamental normative reasons must hang, at least in part, on determining which account of well-being is correct. Since it is far from obvious which is correct (or whether some further account is instead correct), it is far from obvious what many of our fundamental ethical reasons actually are. Yet, presumably we would not expect an ordinary virtuous person to possess the correct theory of well-being. Third, it is, in this context, very much worth mentioning the nonidentity problem. As Derek Parfit (1987, 351–79) famously demonstrated, we do not harm or benefit future people when our actions are such that they are responsible for bringing into existence individuals who would not have existed if it were not for these actions, and yet, intuitively, we may still act wrongly in many such cases. The philosophical problem, as Parfit himself states it, is to identify what the “moral reason” is that explains why it is wrong to act in ways that will cause some future individuals to have significantly worse lives than some different individuals might have had if we had acted differently (after all, it is reasonable to suppose that so long as their lives are still clearly worth living, such individuals cannot reasonably regret being brought into existence).7 Interestingly, the term “moral reason” is one that Parfit himself uses repeatedly in his discussion of the non-identity problem (e.g. 1987, 363). It is grist for my mill that he 7

TWO LEVELS OF ETHICAL THINKING



The last two of these three examples are often associated with consequentialism, but clearly nonconsequentialists should also be interested in working out what the correct theories here are. And some consequentialist theories (especially rule-consequentialist theories) might leave adequate room for recognizing the distinctions between different types of reasons provided in the literature of the first example. Not only do examples such as these indicate that knowledge of our fundamental ethical reasons is a hostage to successful moral philosophy, but they may also have serious repercussions for answering the question of how we ought to act in a wide variety of practical circumstances. The study of possible justifications for entering into wars, or for engaging in particular types of conduct when at war, has clearly been influenced by the literature of our first example. Since the question of how it is best to live our lives is of obvious practical significance, I need say nothing more about the practical consequences of the second example. With respect to the third example, the climate change crisis presents us with a highly complex, large scale moral problem, a feature of which is that our choice of policies for responding to this problem will affect the identity of future individuals. This means that the non-identity problem is potentially a philosophical problem with immediate, wide scale practical ramifications.8 Throughout this book, I intend to remain fairly neutral with respect to debates between the main theoretical options in normative ethics. The guiding thought here is simply that, given the state of contemporary ethics, we have good reason to think that whether the correct ethical theory is consequentialist or not (deontological or not, etc.), it will be a explicitly views his aim as being one of locating the highly non-obvious fundamental moral reason (or reasons) at play in such cases, and that he takes this to be a seriously difficult problem to solve. 8 I say “potentially” here, because Parfit contends that we should accept a “No Difference View,” according to which we should view cases where we harm existing individuals as morally on a par with cases where we affect the identity of future individuals in ways that brings about equally bad outcomes from an impersonal perspective (1987, 368–9). However, John Broome (2012, 61–4) has argued that, when we look at the climate change crisis, there is a serious moral difference between our obligations to people recently born and future generations of people, with respect to the part of morality concerned with justice (which is not to be equated with the whole of morality): partly in virtue of the non-identity problem, our generation does not owe compensation to people not yet living as a matter of justice. Since, on this view, the fundamental reasons in play will not be as weighty as on the No Difference View, there will be cases where this view will deliver a different verdict concerning what we ought to do.



TWO LEVELS OF ETHICAL THINKING

theory that is intellectually sophisticated and non-obvious, from a nonphilosophical perspective, in many of its fundamental details.9 It is also worth bearing in mind that I ultimately wish to focus on ethical thinking, and ethical reasons, broadly construed, and not just moral thinking and moral reasons. Ethical thinking might properly take into account both moral considerations and considerations of prudence. I have more to say about this in Section 3, below.

2 Hare and Williams on Two Levels of Moral Thinking A good place to begin thinking about how there might be two different levels of ethical thinking is with an interesting exchange between R.M. Hare and Bernard Williams concerning the account of two levels of moral thinking outlined in Hare’s Moral Thinking (1981). The book begins by distinguishing between an “intuitive” level and a “critical” level of moral thinking. On the first level, we are all said to follow various simple prima facie rules, confining our reasoning to how these best fit the everyday situations that we encounter. On the second level, it is said to be appropriate to take an impartial utilitarian perspective and assess the prima facie rules themselves, and revise them when necessary, as well as attempt to work out what to do when they conflict or otherwise provide inadequate guidance. In order to make good decisions about which rules to endorse on the critical level (rules that may sometimes be very limited in scope), Hare claims it is necessary to survey the preferences of every person that will be affected by particular actions, placing oneself in the position of each of these people in turn. Hare recognizes that this is an ideal that we can only ever hope to imperfectly approach, rather than completely realize. He distinguishes between the ideal types of “proles” and “archangels”; the former make moral decisions only on an intuitive level—if they make good decisions in a reliable fashion, this is only because they have, either directly or 9 At the same time, there seems to be a growing consensus in normative ethics that the correct ethical theory will not be too distant from commonsense morality in its prescriptions. I suspect that hedonistic act-utilitarianism is too distant from commonsense morality to be a theory that any of us could end up believing with sufficient epistemic justification— there may be no good epistemic route from here to there—however, my arguments here are not aimed at utilitarianism.

TWO LEVELS OF ETHICAL THINKING



indirectly, received a good education from archangels—while the latter only ever make decisions that are based on an accurate view of all of the consequences of all of the actions that are available to them. Actual human beings generally fall somewhere in between these two extremes (Hare 1981, 45). Williams expressed a number of concerns about this account of moral thinking in “The Structure of Hare’s Theory” (1988). Noting that Hare’s account of the two levels of moral thinking is somewhat like Sidgwick’s, Williams contends that it improves upon Sidgwick’s account by avoiding the arguably odious “Government House utilitarianism” that would have it that there are two classes of people—an elite class who apply the correct ethical theory to particular practical issues, and a class of commoners who are best kept ignorant of the correct theory (Williams 1988, 188). Nonetheless, Hare is said to inherit from Sidgwick a fatal tension in his account of the two levels of moral thinking: the intuitive level is meant to correspond to ordinary non-philosophical moral judgments; however, “the more the theory represents the intuitive reactions as merely superficial, provisional, and instrumental, the fewer appearances it saves: it does not explain what people do feel and think, but suggests something else in the same area that they might usefully feel and think” (1988, 190, emphasis added). Intuitively, there does seem to be something right about this verdict, but it is not at all obvious what the best approach to saving the appearances is, once we give up on Hare’s approach, and if we take the philosophical project of articulating and defending ethical principles seriously. In any case, Williams argues that ordinary agents are not generally able to adopt the position of an archangel, even to the limited extent Hare contends that they are, and that ordinary agents are not generally inclined to accept the sometimes radical revisions to prima facie rules which utilitarians would have them undergo. It seems to me that Hare’s failure to give an account of the intuitive level of moral thinking that actually lines up with and appropriately explains our ordinary moral judgments is connected to two other closely related aspects of Hare’s project: his explicit rejection of any reliance on particular substantive moral judgments in his basic methodology, in favor of instead appealing only to formal conditions said to be required for the proper use of moral language; and his failure to provide any space within his project for legitimate claims about moral knowledge, despite



TWO LEVELS OF ETHICAL THINKING

he himself suggesting his project has knowledge as its aim. First, with respect to his methodology, Hare (1981, 12) writes: The appeal to moral intuitions will never do as a basis for a moral system. It is certainly possible, as some thinkers . . . have done, to collect all the moral opinions of which they and their contemporaries feel most sure, find some relatively simple method or apparatus which can be represented . . . as generating all these opinions; and then pronounce that that is the moral system we must acknowledge to be the correct one. But they have absolutely no authority for this claim beyond the original convictions, for which no ground or argument was given. The “equilibrium” they have reached is one which might have been generated by prejudice, and no amount of reflection can make that a solid basis for morality.

For Hare, utilitarianism itself is to be justified not through reflection on our moral intuitions, but through observing, and thinking about what follows from features that are necessary for the proper use of moral language (universalizability and prescriptivity). It seems to most of us now that such formal linguistic conditions cannot play the substantive normative role that Hare wanted them to play (I take it that this is the general consensus). Hare’s attempt to ground the fundamental principle of normative ethics on prescriptivity and universalizability alone, while philosophically heroic, can be challenged both with respect to the relevant claims about what the necessary features of moral language are, and, more crucially than this, with respect to the steps that lead from these linguistic claims to the adoption of the utilitarian principle, in particular. In any case, Hare’s rejection of any reliance on the substantive moral judgments that make up our “original convictions” takes for granted that these judgments are likely to have been generated by prejudice (to be fair, he says “might have been generated by prejudice,” but the tone of the above quotation is one of damning skepticism). This is already to cast first-level moral thinking in a very weak role indeed, and the argument against (reflective) “equilibrium” (which clearly has Rawls in mind) ignores the possibility that the process that leads to such equilibrium might be better thought of as one that builds on incomplete knowledge, rather than on mere opinions. It is telling that the epigraph to Hare’s book is a quotation from Plato (Meno 98b; Hare 1981, 1): “And truly I too speak as one who knows not—only guesses. But that there is a difference in kind between right opinion and knowledge, this, it seems to me, I do not guess; but of the few things, if any, that I would claim to know, this is one.” This is an

TWO LEVELS OF ETHICAL THINKING



intriguing choice of epigraph because Hare, in fact, has very little to say about knowledge in his book. In particular, there does not seem to be any room in the theory outlined therein for moral knowledge. When he does speak of knowledge he is concerned with knowledge of people’s preferences (and he then speaks of “ . . . the sense of ‘know’ that moral thinking demands,” 1981, 96), not with knowledge of what people ought to do. We are expected to leave behind our ordinary moral opinions and come to see that prescriptivism is correct, and that the most sensible prescription (one that fully appreciates the universalizability feature of moral language) is ultimately going to be a purely utilitarian prescription, which is not truth-apt.10 That’s all utilitarianism can be. Hare’s rejection of descriptivism leaves us no room to claim that we could ever come to know any moral claims, let alone know that we ought to (adopt prima facie rules that) impartially maximize utility. Hare’s Platonic rejection of any reliance on ordinary, substantive moral judgments (mere “opinion”) goes hand in hand with his adoption of a normative ethical theory that conflicts with such judgments to too great an extent. I would like to suggest at this point that it is only by construing a class of first-level judgments as examples of basic, 10

Michael Ridge has suggested to me that, in saying this, I might be accused of misstating Hare’s view, since in “Some Confusions about Subjectivity,” originally published in 1976, Hare writes: “We can use the words ‘right’ and ‘wrong’ when speaking of moral statements that people have made. Some opponents of non-descriptivism have made a great deal of play with this fact, as they have with the similar fact that we also use the words ‘true’ and ‘false,’ in certain contexts, of moral statements. But there is no reason why a nondescriptivist should not readily admit that these words are used in speaking of moral statements, provided at any rate that he is prepared, as most non-descriptivists have been since Stevenson, to allow that moral statements do have, as one element in their meaning, what is usually called ‘descriptive meaning’ . . . For it may be this element to which we are adverting when we call a moral statement true or false” (Hare 1989: 26). Ridge thinks that Hare and Stevenson here anticipated certain elements of quasi-realism. That may be so, and quasi-realists may certainly have sophisticated things to say about moral knowledge (I see no obvious reason to suppose my own account of ethical thinking is inconsistent with sophisticated quasi-realism). However, to say, as Hare does, that just part of a moral statement is truth-apt is a long way from saying that the whole of a moral statement is truth-apt, and knowing the descriptive meaning component of a moral statement (on the assumption that one is attempting to follow Hare’s outmoded way of distinguishing between descriptive and prescriptive meanings) is a long way from knowing the moral statement (or better, the proposition expressed) as a whole. The last may be impossible if part of the statement is essentially prescriptive and a different part of the statement is essentially descriptive. In any case, it is not at all clear what the descriptive meaning component of the fundamental utilitarian principle actually is, or whether that principle would even have very much descriptive content, assuming prescriptivism were correct.



TWO LEVELS OF ETHICAL THINKING

unreflective knowledge that we can both make sense of the appearances, so far as ordinary agents are concerned (Star 2008), as well as make sense of the project of normative ethics, since this project can be most plausibly construed as one that is guided by an ideal of knowledge of fundamental ethical principles that builds on pre-philosophical knowledge (if we assume that a purely coherentist account of epistemic justification is not an option, which I think nearly all of us do these days, then it is important that we start with some moral knowledge). Hare’s particular failure should not lead us to conclude that normative ethics—construed as a philosophical enterprise that aims to discover highly general and non-obvious explanatory ethical principles—is a hopeless project (contra Williams). Various ethical theories, whether consequentialist or deontological, may be compatible with those of our first-level ethical judgments that are good candidates for knowledge, insofar as they are compatible with certain core judgments regarding right and wrong action (that it is always wrong to torture a person merely for fun, for example), and certain core judgments regarding normative reasons (for example, that there is always some reason to be kind to people wherever possible, and that there is normally some reason not to lie). It may be that utilitarianism does not fall into this category—that is, it may be the case that Williams is right that utilitarianism is too revisionary, and explanatorily too distant from many of our ordinary moral judgments—but this should not deter us from developing consequentialist or deontological theories that might better fit together with our most fundamental firstlevel ethical judgments (nor from aiming, ultimately, to come to know that one particular theory is correct).

3 Two Levels of Normative Ethical Reasons In order to successfully develop a more satisfactory two-level account of ethical thinking, it will be necessary to: (1) provide an account of ethical reasons that divides reasons into two types: those that non-philosophers can non-accidentally follow, and those that philosophers can discover (and, ideally, come to know); (2) provide an adequate explanation of how these two types of reasons are related; and (3) provide a plausible account of virtue, according to which being minimally virtuous will depend only on first-level thinking—that is, thinking about and responding to first-level

TWO LEVELS OF ETHICAL THINKING



reasons. Achieving the first task will provide a solid basis for avoiding an odious elitism, and for establishing a more attractive division of labor between non-philosophers and philosophers, and achieving the third task will place virtue squarely with non-philosophers. The second task is also crucial, because it is one thing to posit two categories of reasons, and quite another to successfully avoid being in a position where one might justly be accused of trying to have one’s cake and eat it too. In this section, I will begin to undertake the first two tasks, and in the next section I will begin to undertake the third task. The new account of reasons that will enable us to achieve these tasks has at its heart a unified and informative analysis of normative reasons that has been argued for by myself and Stephen Kearns elsewhere (see, especially, Kearns and Star 2008, 2009). We call this analysis reasons as evidence. It can be stated most succinctly as follows. R: Necessarily, a fact F is a reason for an agent A to ϕ if and only if F is evidence that A ought to ϕ.

This analysis of reasons explains what it is for a fact (or true proposition) to be a reason in terms of that fact being evidence for the truth of an ought proposition. In the case of reasons for action, “ϕ” is to be replaced by a particular type of act (e.g. “keep a promise”). In the case of reasons for belief, “ϕ” is to be replaced by a cognitive event such as “believe p” (or “suspend belief in proposition p”), for some p. With respect to normal, purely non-pragmatic reasons for belief, one might also say that any fact that is evidence that I ought to believe some proposition p is evidence that I ought to believe p because it is evidence that p (if there are no pragmatic reasons for belief, this will be true more generally). A simple example will help in understanding R, as it applies to reasons for action. It will be useful to return to this example later. Simple Example (SE): You are hurrying to meet a friend, Jed, who you promised to meet in a few minutes for coffee, and you come across a very sick stranger who needs your assistance to get to a hospital. It is apparent to you that providing such assistance will take quite a bit of time, so it is not possible to both help the stranger and keep your promise to Jed. On the basis of considering the reasons that apply to you, you decide to help the stranger.11

11

This, it should be noted is very similar to an example provided by Ross in his classic discussion of prima facie duties (2002, 18).



TWO LEVELS OF ETHICAL THINKING

A natural way of describing your decision to help the stranger in SE is to say that you recognize that there is a reason to meet your friend, as well as a reason to help the stranger, and you judge that the reason to help the stranger is a stronger or weightier reason than the reason to meet your friend. Another way of describing your situation is to say that you recognize that there is evidence that you ought to meet your friend and evidence that you ought to help the stranger, and you judge that the evidence that you ought to help the stranger is stronger or weightier than the evidence that you ought to meet your friend. Reasons as evidence has it that the relevant claims about evidence and reasons are extensionally equivalent. We can speak, as I just did, of it being the case that there “is” a reason, just as we can speak of it being the case that there “is” evidence. Similarly, we can speak of it being the case that a person “has” a reason, just as we can speak of it being the case that a person “has” evidence. If a stranger outside my front door urgently needs my help, but I am inside and unaware of that fact, there is still a reason for me to help the stranger (there is evidence that I ought to help the stranger), but this is not a reason (evidence) that I possess. There may well be general constraints on which facts can count as unpossessed evidence, and such constraints as there are will also be constraints on unpossessed reasons. Plausibly, a fact F can only be evidence for me that p if an idealized counterpart of myself would have this evidence (that is, would stand in appropriate relations, cognitively speaking, to F and p). Instead of R, one might instead choose to initially focus on the following claim: A has a reason to ϕ and that reason is F iff A has evidence that he ought to ϕ and that evidence is F. One could then define what it is to be evidence that one ought to ϕ as follows: F is evidence that A ought to ϕ iff an appropriately idealized counterpart of A would have evidence that he ought to ϕ, and that evidence would include F. Finally, one could specify what it is for there to be a reason (that is, derive R from the preceding claims): F is a reason for A to ϕ iff F is evidence that A ought to ϕ. Although for most purposes, it is sufficient to focus on R as a claim about a necessarily true biconditional, I think that what, in fact, explains the truth of R is not some further truth; rather, I take it that reasons as evidence is ultimately best understood as an account of the real definition

TWO LEVELS OF ETHICAL THINKING



of a normative reason.12 According to reasons as evidence, thus understood, all reasons for action are evidence that I ought to do something in particular.13 Going beyond this core analysis, we might additionally suppose that some of the facts that are reasons also play a fundamental right-making role (or wrong-making role); they make it the case that particular acts are right (or wrong). These fundamental reasons are the reasons that we aim to specify in normative ethical theories, along with universal principles concerning what we ought to do. Suppose, just for the sake of simplicity, that the correct ethical theory is hedonistic utilitarianism. This theory would have it that the only fundamental reasons are facts about pain and pleasure. If an act would increase pleasure in the world, then the fact that this act would increase pleasure is a fundamental reason to do it, while if an act would increase pain in the world then the fact that this act would increase pain is a fundamental reason not to do it. Furthermore, one ought to do those acts that increase the balance of pleasure over pain (according to this theory). Now consider the fact that one has made a promise in SE. At the level of basic theory, hedonistic utilitarianism says nothing at all about promises. Yet, very plausibly, given contingent facts about the world, the fact that one has promised to do something is (quite strong) evidence, in a situation where one is provided with an opportunity to fulfill a promise, that one ought to fulfill the 12 As we said in Kearns and Star (2009, 219), “we also believe that the best explanation of the truth of all these principles is that the property of being a reason and the property of being evidence of an ought are identical.” 13 Gert (forthcoming) suggests an addendum to reason as evidence. He thinks that it is important to distinguish between justifying reasons and requiring reasons, since, on some occasions, reasons may justify my doing any of a range of permissible acts, none of which I am required to do. He suggests that the defender of reasons as evidence can very naturally capture this distinction by claiming that facts that are evidence that it is not the case that one ought not ç are also reasons, of the justifying type. I think it might well be fine for defenders of reasons as evidence to accept this friendly suggestion. However, I am not sure that they need to. So far as supererogatory acts are concerned, it is possible to distinguish between what is morally required and what one ought to do, and specify that an act is supererogatory if the total evidence that it is morally required is weaker than the total evidence (with respect to morally relevant considerations alone) that one ought to do it (all evidence that an act is morally required is also evidence that one ought to do it, but the reverse does not hold). And, so far as cases where it might seem like it is not true to say that one ought to do any particular act are concerned, we might hope to capture the relevant intuitions by focusing on rationality (or reason), rather than reasons; often one does not have sufficient evidence that one ought to do anything, but it is still rationally appropriate to do the act that the strongest evidence that one ought favors (if time constraints permit, this act may be an act of gathering more evidence).



TWO LEVELS OF ETHICAL THINKING

promise. What ultimately makes the fact that one has promised evidence that one ought to fulfill one’s promise is that it is generally true that fulfilling a promise leads to an increase in pleasure or a decrease in pain. To use a different example, consider any case where one is thinking of telling a lie. The normative theory we are here assuming to be true, simply for the sake of illustration, says nothing about facts that are lies. However, very plausibly, the fact that an act is a lie is (quite strong) evidence that one ought not do it, and is thus also a derivative reason not to do it. What ultimately makes it evidence that one ought not lie, assuming it is, is that it is generally true that lying leads to an increase in pain or a decrease in pleasure. Of course, discovering fundamental reasons will be no simple task, but we are all familiar with derivative reasons, as we are all familiar with much mundane evidence concerning what we ought to do (and ought not do). It is partly through thinking carefully and systematically about such mundane evidence (often construed as “intuitions”) that philosophers are able to develop good ethical theories that provide us with deep explanations of what is right and wrong. The direction of explanation is the reverse of the direction of discovery—fundamental reasons explain derivative reasons, but it is only through first encountering derivative reasons that we are able to discover fundamental reasons. Furthermore, many non-philosophical agents may be in a position to respond directly to good (reliable) evidence concerning what they ought to do. This, I would suggest, is the position that the virtuous typically find themselves in. To illustrate using the considerations just discussed: on the assumption that utilitarianism is true (contrary to what I actually think), a virtuous agent will justifiably recognize that there are reasons to keep promises and not to lie, even if they have never heard of, let alone thought carefully about, utilitarianism; they will be able to do this because they are able to recognize ordinary evidence that they ought to keep promises, and ordinary evidence that they ought not lie (e.g. the fact that it upsets people when they discover that they have been told a lie, or the fact that they themselves feel bad when they have been lied to, for instance).14 Such evidence might itself be an input into subsequent

14

One might also consider the possibility that the correct ethical theory specifies quite a long list of fundamental reasons, then consider an agent who reliably acts well in virtue of

TWO LEVELS OF ETHICAL THINKING



attempts to discover, articulate, or defend the correct normative ethical theory, even though it would not ultimately play a role in the content of that theory.15 I believe that only reasons as evidence can adequately explain how it is that the direction of explanation can be the reverse of the direction of discovery in the way I have just described. At least, this seems to be the only account of reasons that might hope to do so in a way that is true to the appearances (that is, to the ethical reasons we encounter in our ordinary lives), at the same time as providing room for there to be the fundamental reasons that sophisticated first-order ethical theories would posit. In particular, it can hope to do this in a way that is promising when we think about ethical epistemology: a crucial challenge for alternative accounts of reasons is to explain how it is that we can know, or be justified in following, derivate reasons, without possessing any knowledge of fundamental reasons. Reasons as evidence does not hide the key to solving this puzzle: we can start off being cognizant of reasons, in virtue of the fact that we start off with evidence concerning what it is that we ought to do, and these reasons, as evidence, come ready to also lead us to deeper knowledge through reflection. It would be too much to expect me to demonstrate here that it is impossible for any account of reasons to do as well as reasons as evidence on this score, but I can provide a telling criticism of a central class of alternative accounts. A popular idea is to think of a normative reason as being at least part of a certain type of explanation: it might, for instance, be part of an explanation of why one ought to do a particular act, or part of an explanation of why an act is good or desirable.16 The the fact that they follow fewer types of derivative reasons than there are types of fundamental reasons. 15 Even Kantian theories, which take promise-keeping and honesty very seriously, do not typically take principles concerning lying or promise-keeping to be fundamental. One might think lying always demonstrates a failure to respect the rational autonomy (or dignity) of other people, but even then it is standard on such theories to understand the failure to respect rational autonomy as explanatorily (more) basic. 16 John Broome defends a sophisticated version of the first view (for more extensive commentary on Broome, see Kearns and Star 2008, 2015), and Joseph Raz and others have either flirted with or adopted the second view (Brunero 2013 and Way 2013 provide helpful critical discussions). It is also quite common to think of reasons as right-makers (or wrongmakers). “Right-maker” is a term of art: it might either be used to refer to one side of some kind of explanatory relation, in which case the criticism above applies (unless one also accepts reasons as evidence), or be used to refer to one side of some kind of objective



TWO LEVELS OF ETHICAL THINKING

problem with accounts of this form is that parts of explanations are not well suited to play the role in practical reasoning that normative reasons play. I take it that two truisms about normative reasons are: (1) that they are typically weighed against each other; and (2) that they are typically inputs, rather than outputs, in pieces of practical reasoning. In relation to both (1) and (2), reasons contrast with oughts—both of these claims are not true of oughts, since judgments about what one ought to do typically come as conclusions to pieces of practical reasoning. Reasons are typically upstream of oughts. Explanans, however, are typically downstream of explananda; one starts with something to be explained and then one looks for ways in which it might be explained. In practical reasoning, one cannot start with explanations of something one ought to do, because by paying attention to reasons one is trying to work out what one ought to do! On the other hand, reasons as evidence fits together very well with (1) and (2). One might be tempted to respond to this criticism of explanationbased accounts of normative reasons by saying the following: the facts that are reasons in virtue of playing an explanatory role can play the right kind of role in practical reasoning without being appreciated or understood as (parts of ) explanations by agents engaged in practical reasoning.17 However, I see little reason to accept this response as satisfactory, since the only way this alternative picture could be correct is if nonphilosophers were able to generally and reliably identify the right facts as reasons without thinking of them as playing an explanatory role, where the right facts happen to be those that would also be independently identified as explanations by philosophers, and we have yet to be told by what means non-philosophers are able to do this (until we are provided with a plausible story as to how this could occur, any strong correlation here should strike us as miraculous). metaphysical relation, such as a grounding or constitution relation. I cannot do justice to the second option here, but my concern with it would be partly metaphysical and partly epistemological: how are we to understand the metaphysical relation in question (especially given that right-makers can clash with other right-makers, hence they do not always make right), and how do right-makers, on this understanding, admit of both derivative and fundamental kinds (keeping fixed the desideratum that right-makers of the derivative kind come first in the order of discovery, and second in the order of explanation)? 17 Michael Ridge responded to the criticism in this way when I raised it in response to the presentation at the Seventh Annual Metaethics Workshop that was a precursor to McKeever and Ridge (2012).

TWO LEVELS OF ETHICAL THINKING



Suppose that I am right that reasons as evidence is the best account of reasons to place at the heart of a two-level approach to understanding reasons. It might be objected that this is neither here nor there, because all this talk of there being two levels of reasons—one for the ordinary virtuous person and one for the philosopher—already hopelessly clashes with the non-philosophical commitments of ordinary ethical agents; ordinary agents will wish to deny that there is a deeper explanation of our ethical reasons waiting to be provided. Contrary to what some have claimed,18 I do not believe commonsense morality contains any such commitment to there being no deeper philosophical explanations of ethical truths. I agree that commonsense morality would have us save many of the appearances, and thus might well be thought to conflict with some ethical theories (act-utilitarianism, for example). However, this claim should not be confused with the claim that commonsense morality rules out the possibility that deeper explanations of ethical truths might be provided. On the contrary, non-philosophically minded people who are virtuous, or in the process of becoming virtuous, might plausibly be thought to be committed to thinking of the ordinary, virtuous person as being epistemically modest about many ethical truths (including any possible fundamental ethical principle); in any case, immodest commitments are not worth saving. An analogy may be helpful here: a general ontological theory in metaphysics that would have it that all that exists are quantum particles or forces would clash with commonsense ontological commitments and knowledge claims (e.g. that one is sitting on a chair-sized object). However, there is no reason to think that such commitments would conflict with a general ontological theory that would have it that the most fundamental constituents of reality are quantum particles or forces, and that would also contend that it is a useful and important project to attempt to establish how it is that the objects we are ordinarily more aware of (chairs, people, etc.) manage to exist in ways that are grounded in such fundamental facts. Likewise, a general ethical theory that would have it that all normative reasons are of the fundamental kind cited in the theory would clash with many commonsense ethical commitments and associated knowledge claims.

18

See Zangwill (2011).



TWO LEVELS OF ETHICAL THINKING

It should be said that, for the sake of simplicity, I have so far been focusing, and will mostly continue to focus on the reasons of morality, rather than the reasons of ethics more broadly construed, but the contrast between the perspectives of ordinary deliberation and ethical theory is ultimately not to be viewed as so restricted. Following Williams (1993, Ch. 1), I construe “morality” to be narrower in scope than “ethics” (at least conceptually). Most crucially, ethical principles will include principles of prudence or self-interest.19 SE involves a clash of other-regarding moral reasons, but R is clearly very general in scope—the “ought” in R is intended to be the most general practical ought when considering reasons for action, and not the “ought” of morality in particular. Much actual practical deliberation focuses primarily or wholly on the consideration of prudential reasons, and this remains true even when we restrict the class of agents we are considering to the virtuous (at least within a wide range of contexts). In saying this, I am most definitely not saying—in fact, I would deny— that reasons generally appear to us in practical deliberation with the labels “prudential” and “moral” attached to them (or with any markers that would make inferences as to how the reasons are to be thus categorized quite straightforward). It may be that the correct ethical theory will clearly separate prudential and moral reasons of a fundamental kind, but it is a virtue of reasons as evidence that it harmonizes well with the observation that one can ordinarily weigh reasons that have a moral grounding against reasons that have a prudential grounding, since both will simply be evidence that one ought to act one way or another; one can weigh such reasons without being able to categorize them. When we restrict ourselves to the consideration of prudential reasons, it is important to distinguish between fundamental reasons, which prudent agents might well be ignorant of, and derivative reasons, just as it was important to distinguish between these two types of reasons when considering moral reasons. Perhaps the correct theory of well-being is a 19 And it is ultimately one of the key tasks of ethical theory to provide an account of how the reasons of prudence and the reasons of morality are related: whether morality and prudence are ultimately the same thing, or at least always coincide (a position we now rightly think has little going for it, I believe), morality always dominates prudence, prudence always dominates morality, each sometimes dominates the other (the most interesting of all options to pursue, I would say), or neither is capable of dominating the other (the last position is a way of understanding Sidgwick’s doctrine of the “dualism of practical reason”).

TWO LEVELS OF ETHICAL THINKING



hybrid theory that combines an objective list of goods with a hedonistic requirement, such that an item of the kind that is on the list (knowledge, friendship, achievement, etc.) only counts as good for a person if the person takes pleasure in receiving the item (Parfit 1987, 501–2). Or perhaps it is a simple hedonistic theory, a preference-satisfaction theory, or yet some other theory. As I suggested above, we are here talking about a difficult, vibrant, interesting, and ongoing philosophical research program. Ordinary agents cannot be expected to know the correct theory of prudence, yet this need not prevent them from being able to act prudentially by responding to evidence concerning what they (prudentially) ought to do.

4 The Virtuous A satisfactory account of virtue needs to meet the following three desiderata (amongst other things): (1) the acts that the virtuous typically perform are, at least very often, ones that the correct ethical theory would say that they ought to perform; (2) it is admitted that the virtuous can be, and often are, ignorant of the ultimate explanations provided by the correct ethical theory, as well as ignorant of the facts such explanations pick out (and they can be ignorant that these facts are reasons); and (3) the virtuous always act for genuine normative reasons (i.e. they are responding to normative reasons when they act). The second and third desiderata correspond to the first of the main claims provided right at the beginning of this chapter. The first desideratum asks us to think of the virtuous as responding, in some relevant sense, to directions grounded in the fundamental right-makers (and wrong-makers), and to not allow too much room for the virtuous to be responding to derivative reasons in a way that would lead them to act contrary to how they objectively ought to act. It might seem that it is difficult to develop an account of virtue that meets all three desiderata, just as it might have initially seemed difficult to avoid inconsistency without denying either of the two highly plausible claims with which we began. Here is a summary of the account of virtue that I think best meets these desiderata, and is also attractive in its own right. Rather than follow neoAristotelian virtue ethicists like Rosalind Hursthouse (1999), who construe virtuous agents as ethically omniscient (at least whenever the relevant non-ethical knowledge is in place, and one is in a situation where



TWO LEVELS OF ETHICAL THINKING

there is a determinate answer to the question “what ought to be done?”), we should think of them as agents who track and respond reasonably well to pro tanto derivative reasons, guided by knowledge of what is normally right and wrong. They must also possess knowledge of what they always ought (not) to do, all-things-considered, in a large range of paradigmatic cases (for example, one always ought to help people in need when this imposes no significant cost on anyone, and it is always wrong to torture people merely for fun). We should think of virtuous agents as being very skilled at weighing (derivative) reasons, but also as often lacking knowledge of true normative principles that might allow them to work out what they ought to do in very difficult cases, where the reasons they encounter are of a similar weight. This is to say that good agents possess the virtue of normative epistemic humility. They do not usually take themselves to have all the answers to ethical questions. Anyone who takes him- or herself to always have such answers typically thereby demonstrates a vice. In difficult cases, the virtuous might benefit from the provision of an ethical theory that specifies universal, highly general principles that tell them what they ought and ought not do, but, very intuitively, their inability to bring to mind or justifiably endorse such a theory does not detract from their virtue. After all, to be virtuous one need not be maximally virtuous. This account, while admittedly sketchy at this stage, appears to meet the specified desiderata. Notice, in particular, how desideratum (1) is met since, on this account of virtue, the virtuous respond to derivative reasons, and the analysis of reasons that was provided above connects both derivative and fundamental reasons directly to ought facts. One can respond to the directive of fundamental reasons (as it were) indirectly, via derivative reasons/evidence, while acting in the same way that one would act if responding to fundamental reasons directly.20 I can respond to the 20 When Mackie (1977) presents his famous argument from relativity (now often called “the argument from disagreement”), he also presents a “well-known counter to this argument” (37). His argument depends on first recognizing that there are many moral disagreements between people in different societies, and then accepting that the best explanation of this disagreement is that differences in moral codes are caused by differences in ways of life, and not the other way around. The counter to the argument that he considers is that there may be some principles which are recognized in all societies, and that many of the variations we see in moral codes are due to differing applications of these principles to differing circumstances. His reply to this objection is that “people judge that some things

TWO LEVELS OF ETHICAL THINKING



fact that an act would cause someone pain indirectly by responding directly to the fact that the act would be a lie, without for a moment taking into account the fact that it would cause pain.21 My response—in this case, choosing not to lie—would be the same regardless of whether I was well-placed to focus on facts about lying, or facts about pain, as reasons. To use a more positive example: I might take my reason to cheer you up when you are worried to be the fact that you are worried (this is what I think to myself ), but it could be that by directly responding to the fact that you are worried I am indirectly, and appropriately, responding to the fact that you are in pain (or responding to the fact that your rational autonomy is being undermined, say). It needs to be admitted at this point that, on the account of reasons introduced above, derivative reasons can sometimes mislead us, in a way that direct access to fundamental reasons, were it always possible, would not. Some lies may not cause pain or undermine rational autonomy (or align with any other fundamental reasons), and sometimes agents may be led to do the wrong thing by responding to reasons that are evidence concerning what they ought to do. I actually think this is the right thing to say, even though some might take this to be a problematic feature of

are good or right, and others are bad or wrong, not because—or at any rate not only because—they exemplify some general principle for which widespread implicit acceptance could be claimed, but because something about those things arouses certain responses immediately in them, though they would arouse radically and irresolvably different responses in others. . . . the argument from relativity [thus] remains in full force” (37–8). The present approach to normative reasons and virtue provides a way of responding to this reply: ordinary virtuous people may not be directly aware of the deep principles that specify the features that always make certain acts right and certain acts wrong, but they may, in a loose sense, be responding to these features anyway, via responding to evidence concerning what they ought to do. 21

Remember that this is meant to be a simple example. If the example seems like one where it would be easy for a virtuous person to actually directly attend to the fundamental reason in play, bear in mind that there are good reasons to think that the correct normative ethical theory will actually posit quite sophisticated facts when it comes to at least some fundamental reasons. Perhaps the fact that an act would respect rational autonomy would be a good candidate for a fundamental reason that would better make the point here (since it is plausible that ordinary virtuous agents do not think thoughts about rational autonomy, but rather focus on considerations to do with honesty, kindness, courage, etc.). I should add that it is not necessary for me to deny that some of the fundamental reasons that ethical theories fix on are also reasons that ordinary virtuous agents may sometimes directly respond to, and pain is a particularly good candidate for such a reason (yet it is still unusual for ordinary agents to take it that other reasons, such as those provided by considerations concerning lying, are themselves reasons grounded in facts about pain).



TWO LEVELS OF ETHICAL THINKING

my view. It is the right thing to say because it is implausible to think either that ordinary virtuous agents always have direct access to fundamental reasons (given that they do not, they will sometimes get things wrong), or that ordinary virtuous agents generally do not respond to genuine reasons when they act. At a number of places in this book, I will provide additional considerations that should seriously mitigate concerns about the fact that derivative reasons can mislead on my approach.22 Here is an objection to the preceding account of virtue: suppose Joe happens to believe that facts about the relative positions of stars in the Milky Way are evidence concerning what he ought to do, not just in relation to particular decisions someone might need to make if they were a scientist in an observatory (for there are non-bizarre ways that the positions of stars could count as reasons for action), but, quite bizarrely, in relation to all of his ethical decisions. Furthermore, just suppose that, even more bizarrely, through some huge cosmic coincidence, these facts really are reliable evidence concerning what Joe ought to do, whenever he makes any kind of practical decision whatsoever. Now, notice how it seems quite incredible to think that Joe could count as an ethically virtuous agent because he reliably tracks and responds to such facts when making practical decisions.23 Although this may initially seem like a good objection, it falls apart as soon as one clearly distinguishes between two versions of the imagined scenario: Joe may have formed his commitments concerning what is and is not evidence concerning what he ought to do in either an epistemically irresponsible or epistemically responsible fashion. If he formed his beliefs about the relevant evidence in an epistemically irresponsible fashion (which seems much easier to imagine than the alternative), then we shouldn’t think he is virtuous at all. To be sure, I didn’t previously mention this as a condition for possessing ethical virtue, but it now seems we need to add this condition to our account of

22 In particular: in Section 6 below, I discuss the possibility of sometimes coming to discover that a feature that is usually a derivative reason is not a reason at all in some contexts, even whilst one remains ignorant of the fundamental ethical reasons in play; in Chapter 2, I present an argument for R that springs from some thoughts about the authority of derivative reasons; and, in Chapter 3, I discuss an original thought experiment that demonstrates that both subjective and objective rightness come apart from virtuous acts. 23 I am grateful to Howard Nye for raising this objection.

TWO LEVELS OF ETHICAL THINKING



virtue, and I do not see any reasons why we should not.24 On the other hand, if Joe somehow managed to form his beliefs about the relevant evidence in an epistemically responsible fashion, then I think we would properly view him as ethically virtuous. At this point, I am in a position to provide a summary statement of the new two-level account of ethical thinking. On a first level, ethical thinking involves the collection and weighing up of basic forms of evidence concerning what one ought to do—an appreciation of and an ability to weigh derivative reasons. It seems that most of us already possess such evidence, or are close to possessing it, and may also possess certain virtues by responding well to certain types of evidence (e.g. the evidence that we classify under “kindness”). On a second level, ethical thinking involves essentially philosophical thinking that moves from paying attention to such evidence in a way that guides our actions to thinking about it systematically in order to postulate and, ideally, come to know ultimate explanations for the rightness or wrongness of particular acts, on this basis. Ideally, such thinking will vindicate some of our evidence concerning what we ought to do, as well as enable us to recognize that some of it is misleading. And, ideally, it will also take us beyond minimal virtue, in enabling us to determine what we ought to do in many difficult cases, where we previously only saw (derivative) reasons of roughly equal weight.

5 Williams’s Concern Let me now return to Williams’s comment that Hare’s two-level approach to moral thinking is problematic because “the more the theory represents the intuitive reactions as merely superficial, provisional, and instrumental, the fewer appearances it saves: it does not explain what people do feel and think, but suggests something else in the same area that they might usefully feel and think.” One might feel some sympathy with Williams here, yet be unsure about what precisely the problem is. The problem is not meant to be just that utilitarianism may direct us to

This fits with the fact that I am happily heading in the direction of viewing most of the practical virtues as reducible, in large measure, to epistemic materials, a thought that will be made more precise in Chapter 3. If enkrasia is itself a virtue (it is not at all obvious that it should be thought of as such), then it is an example of a virtue that cannot be understood in such a reductive fashion. 24



TWO LEVELS OF ETHICAL THINKING

act differently than ordinary moral thinking directs us to act—of course, we might worry that this will happen (too often), but it is not the concern that Williams is trying to put his finger on. Suppose the correct ethical theory gives the tick to enough of our ordinary judgments about what we ought to do, and to enough of our ordinary judgments about which factors we ought to pay attention to when we are trying to work out what we ought to do, even though those factors are not the fundamental reasons that the theory picks out. Williams is suggesting it may do this in a manner which conflicts with the way ordinary ethical judgments appear to us when we reflect on them. How so? Think about the way a principle such as “lying is wrong” might direct one’s ordinary judgments in cases where one might be tempted to lie.25 One might think that the principle will be justifiably accepted. If one is committed to the principle in an ethically and epistemically conscientious manner, it won’t seem just luckily and provisionally correct. Yet if act-utilitarianism is correct, and we fix on agents who have never encountered utilitarianism, the fact that an ordinary “prole” commitment to a rule happens to get the tick from the theory is simply a matter of contingent luck. Williams does not mention knowledge or epistemic warrant, but I think that one reason we might find ourselves assenting to the claim in the quotation provided above is that we take ourselves to know many of our ordinary ethical judgments (or the simple principles behind them). This would explain very well why we are uncomfortable with the thought that the judgments we are considering are merely luckily right, for luck undermines knowledge: if our core ethical beliefs are only luckily true then they are not knowledge, or warranted. I will return to the idea that knowledge guides the virtuous agent’s actions in the last two chapters of this book. I make these claims about knowledge here, in relation to the somewhat opaque criticism from Williams, because I think they provide one plausible interpretation of the worry that Williams is trying to put his finger on. But now let me mention another interpretation which more closely fits the theme of the present chapter. Instead of focusing merely on the epistemic credentials 25 The example is meant to be a principle that has a generic moral claim as its content. When I go on to say it will strike one as possessing a degree of necessity, I do not mean that it will strike one that, necessarily, lying is always wrong.

TWO LEVELS OF ETHICAL THINKING



of ordinary moral judgments, we might focus instead on the epistemic credentials that an ethical theory might possibly possess to those who defend it, and how those credentials are meant to be related to the credentials of our ordinary judgments. We might wonder what authority theoretical ethical principles should have for us, such that it might be correct to give up ordinary moral judgments in the light of such principles. The point here is not to block the idea that we should give up some of our ordinary moral judgments in the light of better principles—that may well be true—but to ask what it would be like to reside in a satisfactory epistemic position to generally rationally take ethical theoretical principles to be more authoritative than the contents of one’s ordinary, pre-theoretical moral judgments. Here are two options: first, one might think that ordinary moral judgments have no particular epistemic credentials and that one is able to come to accept the favored theoretical principles by a more reliable method. One version of this idea would be Hare’s linguistic prescriptivism, but there is also the intuitionist alternative here: one might somehow simply come to see that a theoretical ethical principle is selfevidently correct, perhaps as a result of first establishing that commonsense morality is incoherent (this is roughly what Sidgwick attempts to do in the Methods of Ethics). The second option, which is the one I favor, is this: we might think there must be an epistemic route from here to there. Let me explain. Even prior to discussing the contentious topic of ethical knowledge, one might view the ordinary agent as being in a position where they are confronted with much evidence concerning ethical truths,26 and such evidence will need to be considered alongside any theoretical principles on offer—the reason we might find the act-utilitarian principle unattractive, for instance, might be that it strongly conflicts with this body of evidence (for it would require us to act in a different way than we would ordinarily judge, with good justification, that we should act, in too wide a range of possible cases). It does this, even if contingent facts happen to provide for the lucky coincidence of utilitarianism giving the tick to many of our ordinary judgments about the world as it now happens to be, since this evidence also affects our judgments about many counterfactuals, and it is 26

In Chapter 4, I will look favorably on the idea that evidence depends on knowledge, but I am not relying on that view here.



TWO LEVELS OF ETHICAL THINKING

when we consider these counterfactuals that we are struck by ways in which the seeming necessity of ordinary moral commitments clashes with the verdicts on many acts that utilitarianism gives us (this is Williams’s point about the judgments not striking us as provisional). Admittedly, this fact of conflict may not be ultimately decisive. I cannot here demonstrate that there is not a rational epistemic route by which all of the evidence that speaks against act-utilitarianism can come to rightly be seen as misleading. All I mean to emphasize here is that ordinary agents possess a body of evidence that strongly suggests that there is no such route.27 I think the second option should appear more attractive to us than the first. I am not claiming that it is not possible to have a priori intuitions about very general, fundamental ethical principles that might rationally overturn ordinary moral convictions. But I am certain I ought to be ethically and epistemically very careful about accepting that anyone has ever had one of these intuitions in relation to fundamental ethical principles. So much of importance hangs on this, practically speaking, and such epistemic claims are controversial. After all, I am straightforwardly confronted by so many relatively mundane items of evidence concerning what I ought to do (that are normative reasons, on the theory defended herein) which do not appear to depend on contentious a priori intuitions about very general, fundamental principles. This second option, in combination with reasons as evidence, provides another dimension to our solution of the puzzle of this chapter. Ordinary judgments about ethical reasons are not just superficial and instrumental, insofar as they are judgments about genuine normative reasons that virtuous people can act on in ignorance of fundamental ethical theory. And virtuous people are not just luckily getting at the ethical truth, because such reasons are good evidence (on the assumption that such reasons are reliably indicating truths about what they ought to do, and are not generally misleading).28 27 This is a more robust observation if one grants, as I think one ultimately should, that we possess moral knowledge that, as a body, conflicts with act-utilitarianism, and that it wouldn’t be rational to destroy such knowledge on the way to coming to believe an actutilitarian principle. 28 We can distinguish between two levels of luck here. If the ordinary evidence is not generally misleading, this is enough to establish that the virtuous person is not just luckily getting things right. But we might think it is just lucky that the ordinary evidence is not

TWO LEVELS OF ETHICAL THINKING



Let me emphasize that the crucial issue here is not whether actutilitarianism is correct (if I am wrong about the lack of an epistemic route in relation to this theory, that would not undermine my main point). It is rather that Hare’s two-level model does not fit well with the way we conceive of virtuous people, insofar as we think of such people as reliably responding to genuine normative reasons (at least once we reject his idea that everyone already has an aspect of themselves that is a utilitarian angel; this idea does not seem true). To bring out the main concern, consider another theory: rule-consequentialism, in the particular way it is defended by Brad Hooker (2000). One of Hooker’s clear motivations for defending rule-consequentialism over act-consequentialism is that the first fits much better with many of our ordinary moral judgments. Hooker argues that it is possible to avoid the old criticism that ruleconsequentialism must collapse into act-consequentialism under the pressure of a general consequentialist intuition that one ought to always bring about the best state of affairs. The way to avoid this problem, he claims, is to not commit oneself at all to the general consequentialist intuition (this should be easy if that commitment is not an ordinary, pretheoretical judgment), but to rather begin by accepting the commitments of commonsense morality that are best summed up by a Rossian account of pro tanto duties, and only then ask for a principle that might explain why all of the pro tanto duties are, in fact, pro tanto duties. The motivation that might lead to acceptance of the rule-consequentialist principle might thus merely be a motivation to find a deeper explanatory principle. I am sympathetic to Hooker’s approach. I don’t think it runs afoul of the provisional contingency worries that seemed to be behind William’s comment on Hare (or, at least, it stands a good chance of avoiding them, given the vagueness of the notion that ordinary agents are committed to some degree of necessity or non-contingency in their commitments). However, I think Hooker’s way of defending rule-consequentialism is still lacking a crucial element. We are still left wondering what the

generally misleading (assuming it is not). I think that this higher-level luck does not undermine knowledge (contra Setiya’s principle K [2012, 96]), and should not worry us here. That being said, the possibility that the ordinary evidence is generally misleading is one that it is appropriate to take seriously when considering moral skepticism; as I indicated in the Preface, I am putting concerns about moral skepticism to one side in this book.



TWO LEVELS OF ETHICAL THINKING

epistemic status of the convictions we begin with is—perhaps they are the mere prejudices Hare refers to, in which case focusing on them in order to locate a deeper explanatory principle might just lead to increasing the amount of rubbish we believe. Hooker’s aim is not to give an account of the exact epistemic status of the ordinary “convictions” we begin with (he too eschews talk of “intuitions” [2000, 15], but he does indicate that he takes the convictions behind our commitments to a list of pro tanto duties to be “independently credible”—a term he thinks of as a good replacement for “self-evident” [2000, 13]). I take myself to be going further here. I have suggested that there are two levels of normative reasons, and that ethical theory is in the business of telling us what our fundamental normative reasons are. One objection to this view is that we might instead think of ethical theory as providing substantive explanations of why the reasons ordinary agents encounter are reasons, and as not pointing to any reasons itself.29 This is one way we might interpret Hooker’s rule-consequentialism, for instance. Under this interpretation, the fact that a rule that directs us to keep our promises, say, is one of the rules that belong to a set that together better satisfy the rule-consequentialist conditions itself provides no reason at all to act in conformity with the rule, whereas the fact that an act would be the keeping of a promise does provide us with a reason. This strikes me as quite wrong, even on the assumption that the motive for accepting the rule-consequentialist principle is purely explanatory (it tells us why we have certain pro tanto duties). To bring out why this is wrong, suppose I am talking to a reliable guardian angel, who tells me that rule-consequentialism is correct, and that an act I am considering doing is required by one of the rules that meet the rule-consequentialist conditions (roughly, one of the rules which is part of the set of rules that will, if most people follow them, make things go best), but does not tell me what the rule is. Wouldn’t we be inclined to accept that the fact that the act is required by one such rule is a reason? To bring out why the alternative first mentioned seems wrong, just imagine responding to the guardian angel’s suggestion that this fact might be a reason to act by saying “I understand that this fact explains why this rule is one that it is right to follow, but even though

29

I am grateful to Kate Manne for raising this objection.

TWO LEVELS OF ETHICAL THINKING



I assume rule-consequentialism is correct, this fact, that it is a rule in the set that will produce the best consequences is no reason at all to perform the act.” So far as I can see, this point generalizes to other ethical theories, and the relation they would stand in to our reasons, if correct.

6 Ethical Particularism Now that I have provided a way of resolving the apparent inconsistency with which I began, while not giving up on either of the two main claims provided at the beginning of this chapter, I will finish by suggesting that the proposed two-level framework may have an additional attractive feature. The account of reasons that I introduced above is particularly well-placed to explain many of the key judgments that drive ethical particularists, and the (limited) truth of the holism of reasons thesis that results, in large part, from reflection on such judgments.30 The explanation of many of the judgments that motivate ethical particularists is one that will, in fact, lead us away from particularism. I am thinking, in particular, of the judgments that Jonathan Dancy (2004) relies on when he claims that the holism of reasons thesis makes it is impossible to articulate interesting, true moral principles. If we separate reasons out into our two classes of derivative and fundamental reasons, we will be in a good position to deny that there is holism on the level of fundamental right-makers and wrong-makers. Derivative reasons (as evidence) have an holistic structure—and to the extent that this is so, the particularist is on to something important when it comes to describing the deliberations of the virtuous, who may be ignorant of fundamental reasons—but now we can see that the particularist may be confusing intuitions that involve derivative reasons with the intuitions that we can come to have about fundamental reasons. Fundamental reasons have an atomistic structure on the

30 McKeever and Ridge (2013), who do not themselves accept reasons as evidence, agree with me that reasons as evidence is in a particularly good position to provide an explanation of the holism of reasons thesis, although their suggestion for how it might do so is different than the one I wish to provide. They recommend that defenders of reasons as evidence exploit a distinction between background information that is not itself evidence, but may partly explain the strength of evidence, and evidence itself. I do not rely on this distinction below.



TWO LEVELS OF ETHICAL THINKING

account I sketched above, since they always count in favor, or always count against, particular acts.31 Once again, it might help if we use an example. Very plausibly, there is normally a reason not to lie, and on many individual occasions one may correctly judge that there is a reason not to lie. Particularists like to point out that it is also very plausible to suppose that one can correctly judge that there is no reason not to lie when one is in a situation where one is playing a game with friends that everyone knows requires lying in order to win (to use one of Dancy’s examples). Examples such as this are taken to be evidence in favor of the holism of reasons thesis, since they seem to suggest that reasons are not generally invariant—that they can change their “valency” from situation to situation in a way that always depends on other features of particular situations. Yet, on the present two-level account of reasons, the fundamental wrong-making feature in cases where it is wrong to lie (such as disrespect of another’s rational faculties, or pain, or whatever features would be cited in the correct ethical theory) can instead be thought of as being part of an atomistically structured class of fundamental reasons. There is no reason not to lie when playing a game, precisely because no fundamental wrong-making feature is present in such a case.32 This last point seems perfectly generalizable, at least in principle.33 However, one might worry that since I claim that the virtuous do not need to be aware of fundamental reasons (and that they often are not), an adequate explanation of why an ordinary virtuous agent might 31 I am ignoring a complication here. McKeever and Ridge (2006) have argued that it is possible to reject particularism about ethical principles whilst still maintaining that the structure of reasons is holistic through and through. While this might be true, merely as a point about the logical relations between the bare particularist position and the bare holist position, I believe holism is, in fact, explanatorily redundant in ethics once one rejects particularism about ethical principles. I discuss some reasons for thinking this is so that do not depend on reasons as evidence in Star (2007). 32 I follow Crisp (2007) in thinking that, whatever we say of derivative reasons, fundamental reasons (he calls them “ultimate”) are atomistic. Nothing that I have said guarantees that this is so, but I think that once we see that the examples particularists favor can be described as really being about derivative reasons, we are left with little reason to think fundamental ethical reasons need be holistic. 33 When I say it is perfectly generalizable, I have, in the first instance, reasons for action in mind, rather than reasons for belief. The reasons we have for being attracted to atomism in ethics may have no parallel in general epistemology. This being said, I provide some reasons for thinking that there may also be some atomistic reasons for belief (outside of ethics) in Section 5 of Chapter 4.

TWO LEVELS OF ETHICAL THINKING



appropriately judge that there is no reason at all not to lie in a situation that involves a game that requires lying to win should not appeal to the absence of a fundamental reason. Since they may be ignorant of what the fundamental reasons are, virtuous agents may not be in a position to conclude that a derivative reason has been silenced in some context by way of drawing an inference from a judgment that a fundamental reason is absent. I think this worry can be adequately responded to as follows. What ultimately explains why there is no reason not to lie in the game case is the absence of a fundamental reason not to lie, but what explains how the agent can judge that he has no reason not to lie may be something quite different. Assuming reasons are evidence (in the exact manner specified above), we can explain “silencing” on the level of derivative reasons by utilizing the notion of an epistemic defeater. The ordinary virtuous agent ordinarily takes the fact that an act will be a lie to be evidence that he ought not do it, but in certain cases this evidence can be defeated by other evidence that he possesses, such as strong evidence (or knowledge) that everybody has agreed to have fun by lying within the boundaries of a game. This role for epistemic defeaters when it comes to practical reasons is simply an illustration of a general consequence of reasons as evidence. This is an approach to analyzing reasons that brings with it the resources for explaining the holism of reasons (derivate reasons, that is) in terms of the holism of evidence. It is also, fortunately, an approach that provides an explanation of why the strength of a derivative reason and the strength of an underlying fundamental reason do not need to be added together in those cases where an agent is aware of both reasons. One might worry that any highly promiscuous account of reasons for action, such as reasons as evidence, will lead to a counting problem, and perhaps then to an unattractive solution to this problem that supposes that some reasons have no weight at all (see McKeever and Ridge 2006, 132–3; and Star 2007). Here is the explanation of why, on reasons as evidence, one cannot always aggregate the strengths of reasons: some items of evidence are independent, while some items of evidence are not independent. It is not rational to simply add the strength of two dependent reasons together (assuming one has evidence that they are not independent). For example, if I look out the window and notice for the first time that there appears to be a green car parked outside, I have



TWO LEVELS OF ETHICAL THINKING

intuitively encountered a strong reason to believe that there is a green car parked outside; if, a moment later, a reliable friend looks out the window and has a similar experience to mine, he might report to me that there is a green car outside; absent the fact that I had already looked out the window, her testimony might provide me with a strong reason to believe that there is a green car parked outside; however, I am not now entitled to simply add the strength of the reason her testimony would have provided prior to my own observation to the strength of the reason that observation actually already provides to believe there is a green car parked outside, as if I now have twice as much reason to believe this. And if I tell a second friend that there is a green car outside, and I know that he does not check this for himself, but instead simply repeats back to me twenty seconds later that there is a green car outside, I cannot thereby rationally increase my confidence in the truth of this proposition, at least not by very much (the strength of the reason that his testimony would provide in isolation from my prior experiences is now almost completely silenced). In becoming aware of fundamental reasons (in general, if not always, through philosophical reflection), the relevant sophisticated agents will become aware that certain derivative reasons are not independent of certain fundamental reasons. Although it is no easy task for epistemology to explain how we are able to make good judgments concerning the degree to which some items of evidence are dependent on some other items of evidence, it is clear that we all do this very often (however imperfectly) when it comes to ordinary reasons for belief. Whatever explains how we are able to do this in mundane cases involving reasons for belief will also explain how we are able to do this when it comes to reasons for action, assuming, as we are, that the latter are also evidence. As I see it, the type of ethical particularism that would deny that there are highly general, foundational ethical principles concerning our fundamental reasons for action and what we ought to do is, in fact, a position of last resort. Normative ethics may still be in its infancy, yet we already have many candidate principles that are worth taking seriously, even if only to work at improving on them. Given that an appeal to the holism of reasons does not suffice to show that there are no true, highly general ethical principles, it is difficult to see how anything other than pessimism concerning the project of normative ethics can be the basis for the view that we will never make any progress on this front.

TWO LEVELS OF ETHICAL THINKING



Similarly, a Rossian pluralist deontological theory that would have it that all we are able to do when reflecting on the ethical domain is construct a list of pro tanto duties that are very close to commonsense commitments, and which would merely have us rely on a capacity to form “considered opinions” when it comes to judgments about what we ought to do when they conflict (Ross 2002, 19), offers a great deal less than we might reasonably hope for. Having said this, it should be noted that particularists have done much to show that the structure of ethical reality is complex (see, especially, Dancy 2004), and some of the distinctions that they have drawn may be distinctions that generalists in normative ethics need to pay attention to when developing their theories. Regarding deontological pluralism, contemporary or future pluralists may be able to say much more than Ross did about the nature of fundamental pro tanto duties, and related fundamental goods, and may be able to provide principles concerning what we ought to do when they conflict (inspired, perhaps, by much of the rich reflection on substantive ethical matters that one finds in Ross’s work). Insofar as we should all be concerned to aim for the ideal of knowledge of fundamental ethical principles, we should all be concerned to promote intellectual debate between defenders of different ethical theories; the continuing development of sophisticated forms of deontology is to be welcomed, as is the continuing development of sophisticated forms of consequentialism. This concludes the overall sketch of a solution to the problem we began with. There is clearly much more to be said. In the remainder of this book, the more to be said will partly involve demonstrating that the solution sketched above is satisfactory, and will partly involve providing arguments for relevant but more refined claims about the nature of reasons (Chapter 2), the nature of virtue (Chapter 3), and the relation between knowledge and reasons (Chapter 4).

2 The Authority of Reasons 1 Introduction In the previous chapter, I claimed that it is crucial that those of us of a broadly ethical realist persuasion (including quasi-realists) provide an adequate account of the general relation that the fundamental rightmakers and wrong-makers posited by first-order normative ethical theories stand in to the generally more mundane considerations that ordinary non-philosophical agents take to be crucial in their deliberations. I argued that for such an account to be adequate, it is essential that it make clear how—and not merely assert that—ordinary ethical thinking is in a position to be reliable in relation to fundamental ethical facts (when ordinary agents are ignorant of such facts, or would at least be unjustified in accepting them, since they require sophisticated philosophical reflection and arguments to uncover and defend). The ethical realist must explain why many of the mundane considerations crucial to ordinary ethical thinking fully deserve to be acknowledged as genuine normative reasons—for those who are ignorant of the relevant fundamental normative reasons—while, at the same time, leaving room for philosophically sophisticated and highly general ethical principles to still be true. Here I outline and defend an account of normative reasons that is especially well-suited to take on this explanatory burden.1 1 It would be nice to be able to show conclusively that this account of reasons is the only account that might adequately take on this explanatory burden, but arguing for that stronger conclusion would require very extensive analysis of all of the alternatives, which is beyond the aims of this book. In any case, it is in virtue of explicitly essential properties of the favored account of reasons that it plays the explanatory role I have in mind; it will at least be more difficult for defenders of other accounts to establish that their favored accounts can play this role, since, so far as I can see, any attempt to do so will require them to develop arguments that go beyond essential features of their favored accounts. This is certainly straightforwardly true of Tim Scanlon’s and Derek Parfit’s accounts of

THE AUTHORITY OF REASONS



The account of normative reasons that I favor was first developed and defended against various criticisms in papers that I coauthored with Stephen Kearns, where we dubbed it reasons as evidence (Kearns and Star 2008, 2009, 2011, 2013a, 2013b, 2015).2 In the papers we wrote together, Kearns and I mainly aimed to defend a fairly narrow thesis, i.e. the specific biconditional analysis of reasons R, introduced in Chapter 1 (in simple terms: reasons are facts that are evidence with respect to what one ought to do, or ought to believe, etc.). Throughout this book I put that thesis to work in my own, more expansive way. The present chapter provides a new argument for the most contentious, right-to-left reading of R—this argument takes off from a starting point unexplored in the papers I wrote with Kearns, and that is Joseph Raz’s pioneering work on practical authority. I argue that Raz’s service account of authority can and should be tweaked in a very natural way to produce a minimalist account of authority. Doing this enables us to appreciate that there is a unity to normativity across the practical and epistemic domains. It also clears the ground for the utilization in the new argument for the right-to-left reading of R of a Razian insight concerning the function of reasons in relation to reason (the general capacity to respond to reasons, where this may involve, amongst other things, rationally appreciating, weighing, reasoning about, and acting on reasons). In effect, this new argument provides an explanation of why facts that are evidence concerning what one ought to do are also reasons for action. In short, the answer is that such facts are reasons for action because they are authoritative in relation to reason, and they are authoritative in relation to reason because of the service they fulfill in being an essential guide to right action; a guide that is as reliable as it is possible to find without offending against reason.

normative reasons (since they explicitly reject the possibility of providing any analysis of reasons), and Kearns and Star (2008, 2015) provide grounds for thinking it is also true of John Broome’s account of reasons. 2

See also Star (2010, 2011) and, for an independently developed, similar account of reasons that is defended using quite different arguments, see Judith Jarvis Thomson (2008). Critics of Kearns and Star include Broome (2008), Brunero (2009), McNaughton and Rawling (2011), McKeever and Ridge (2012), Fletcher (2013), McBride (2013), and Setiya (2014).



THE AUTHORITY OF REASONS

2 Normative Reasons I take it that a basic category of normative facts are facts about what one ought to do, ought to believe, ought to care about, etc. (and ought not do, ought not believe, ought not care about, etc.) In normal circumstances, one ought to help people in need, and ought not believe that the world is flat. Within the category of ought-facts, some such facts are more fundamental than others: if hedonistic act-utilitarianism were true, for instance, then the most general practical normative fact would be that one always ought to act so as to maximize the total amount of hedonic value in the world;3 the fact that one ought to now help a particular person in need would be a derivative normative fact, whose truth would be explained by contingent, wholly non-normative, facts (to do with pleasures, pains, and the causes thereof), in conjunction with the utilitarian principle that states the fundamental practical normative fact. I take it that it is one of the central tasks of normative ethical theory to articulate and defend principles that state the most basic or fundamental normative facts.4 There are also facts about what normative reasons there are to do one thing or another. The concept of a reason stands to the concept of ought in the following two ways: (1) judgments about reasons are typically inputs to practical reasoning, whereas judgments about oughts are typically outputs; and (2) reasons are typically weighed against each other, whereas oughts are (at least typically) verdictive, or all-thingsconsidered. I take these to be two basic truisms about the relation of normative reasons to oughts that are independent of any particular account of reasons. In SE (introduced in Chapter 1), you might weigh the reason you have to meet Jed against the reason you have to stop and 3 To keep the example simple, I am assuming also that moral requirements always trump prudential requirements. As I indicated in Chapter 1, I am not, in fact, committed to this assumption. 4 It may be that basic oughts can be further explained (e.g. in terms of the most general facts about what is good or bad, or in terms of the notion of a moral requirement, or in terms of both moral and prudential requirements, etc.). I do not mean to rule out such a possibility, but I do mean to contend that ought facts are metaphysically more fundamental than reason facts. If basic ought-facts come apart from a class of fundamental reasons that we might term fundamental moral right-makers, as they will on some accounts of the relation of morality to prudence, then it is also one of the task of normative ethical theory to provide us with principles that link moral requirements and the requirements of prudence to fundamental ought-facts.

THE AUTHORITY OF REASONS



help a person in need. You might rationally judge that one of these reasons is stronger than the other, then conclude that you ought to help the person in dire need (rather than meet Jed). In attending to normative reasons, you are attending to essential guides for determining—as best as you can, anyway—what, more fundamentally, you ought to do. I take it to be an elementary fact about the phenomenology of deliberation in relation to such cases that one does not typically view the force of these reasons as residing inside of oneself (whether in one’s will or desires), but rather in external facts having to do with other people being in need, promises being made to others, etc.5 Crucially, normative reasons, like oughts, have a quality of external authoritativeness about them.6 I think that this is where one should start when thinking about authority, i.e. with the external authoritativeness of normative facts. The fact that a person in front of me is in dire straits is a reason to help her whose normative authority rests, at least in part, in this very external fact. If I asked “but why help her?” you would be right to say, “can’t you see she needs your help!” or “can’t you see, she is in pain!”7 The fact of her pain is (or gives rise to) a directive that I must, if I am reasonable, appropriately respond to.8 On the account of reasons that I favor, the practical authority that reasons strike rational people as possessing is 5 Even the Humean theory of practical reasons, which claims that all normative reasons are to be explained, at least in part, by the conative psychology of the agent for whom they are reasons need not deny this. For a recent defense of the Humean theory that attempts to do justice to the external authoritativeness of (many) reasons, see Schroeder (2007a). 6 Defenders of a Humean theory of practical reasons, at least assuming it is defended in a form that possesses the kind of sophistication that is required to make it plausible, need not deny this, nor need they view this quality as completely superficial, for they can agree with non-Humeans that the authority of reasons lies outside of one’s conscious self, since, according to them, this authority will (in many cases, at least) be seen to reside in external facts, even though it is explained by conative mental states (states that may, in any case, be distant, opaque, and inflexible, compared to one’s present, occurrent desires or will). 7 It may be tempting to view only the second example as citing a genuine normative reason, but, as indicated in Chapter 1, I take it to be an outstanding virtue of reasons as evidence that it explains why both claims can be citing reasons and why we need not worry that this will lead to a counting problem where we might think we are entitled to add the weight of non-independent reasons together. 8 Since reasons are (at least typically) pro tanto, they can be outweighed, so it is possible that the appropriate response to a reason will not be to act as the reason directs me to act, but that will be because of countervailing considerations, and it would generally be appropriate for me to recognize, either in my deliberations or after the fact, that an overridden directive of which I am aware does still have some claim on me (assuming i were to consider the reason, which is itself something I might not have good reason to do).



THE AUTHORITY OF REASONS

inherited from the practical authority of a large class of ought-facts (the true ought propositions regarding actions).9 But what is practical authority?

3 From the Service Conception to the Minimalist Conception According to Joseph Raz’s service conception of practical authority, practical authority must be distinguished from mere de facto authority (which has to do with the presence of coercive powers), and is normally justified if: (1) the subject is likely to better conform with reasons which already, independently, apply to her if she intends to accept the authority’s directives and treat them as valid, rather than not do so; and (2) the relevant situation is one where it is better that the agent conforms with the reasons that externally apply to her, rather than decide for herself what to do without the aid of the authority.10 Raz dubs the first of these conditions the normal justification thesis (NJT), and the second the independence condition. I will focus on the NJT here. On this picture, agents that are subject to an authority that satisfies these conditions will find themselves with new, derivative reasons for action. These “preemptive” reasons, as Raz calls them, are reasons that follow directly from the authority’s directives and are to be considered without considering independent reasons that might conflict with them: a reason of this type “is not to be added to all other relevant reasons when assessing what to do, but should exclude and take the place of some of them.”11 For example, a person driving down a highway might, very 9 Broome (2013) also takes ought-facts to be more fundamental than reasons. On his account of reasons, as parts of explanations of ought-facts, it might be best to say that reasons are not genuinely normative—or, at least, that the concept “reason” is not a normative concept. Explanations are not themselves normative. I instead think we should say that reasons are genuinely normative, but that their normativity (the normativity of evidence that one ought) is inherited from the normativity of ought-facts. 10 Raz (2006, 1014). What Raz would take to be an exact statement of the independence condition is left a little unclear. 11 Raz (1986, 46). The relation between preemptive reasons and “exclusionary” reasons is complicated. Suffice to say that Raz thinks that it is usually fine to focus on the above type of characterization of preemptive reasons alone when considering the nature of authority, but he also thinks that he can tell a deeper story about preemptive reasons where we understand them to be “protected” reasons, where a protected reason is a fact that both constitutes a first-order reason to do a particular act and an exclusionary reason “not to fail to [do this

THE AUTHORITY OF REASONS



plausibly, be thought to have a preemptive reason to drive no faster than a sign-posted speed limit, where this reason also speaks against her attempting to independently determine the fastest speed at which it might be safe for her to drive, i.e. from considering non-authoritybased reasons with respect to the issue of determining the fastest speed at which it might be safe for her to drive. Very plausibly, such a person better conforms with independent reasons that already apply to her in the long term by treating the speed limit as authoritative than she would if she considered non-authority-based reasons, with respect to the issue of determining the fastest speed at which she might drive. And on Raz’s account of authority this means not merely that she should treat the speed limit as if it provides a genuine reason for action (although she should do that), but also that the speed limit provides a genuine reason for action—a reason that can be classified as preemptive, since it is paired with an exclusionary reason that rules out the consideration of certain facts as reasons—in virtue of the fact that the person who follows it better conforms with independent reasons in the long term.12 So much for the basic idea at the heart of Raz’s account of authority. We might next ask: What is the fundamental role of the concept of authority meant to be?13 Raz tells us that the problem that he is focusing particular act] for a certain range of excluded reasons” (Raz 2010, 298). For a definition of “exclusionary reason,” as a second-order reason to refrain from acting on a first-order reason, see Raz (1990, 39). 12

This example (a common one in the literature) is not meant to illustrate the only way in which the directives of political authorities can provide preemptive reasons. Political authorities provide preemptive reasons that enable subjects to better conform with prior reasons in a number of ways, especially: (1) when partly arbitrary decisions are required (e.g. deciding which side of the road everyone should drive on); (2) when establishing or maintaining conventions or laws that are needed as solutions to coordination problems (e.g. prisoner’s dilemmas); (3) when they are in a better epistemic position to see how some prior ends might be best pursued; and (4) when their powers put them in a better position to achieves some prior ends (Raz 1986, 48–51, 75). 13 I am using “concept,” “conception,” and “role of the concept” as follows: first, I follow Rawls and others in using “concept” to refer to any particularly central idea that all genuine participants in a philosophical dispute agree is the idea that they are providing alternative accounts of, and in using “conception” to refer to any one of a number of differing accounts of the same idea (so Rawls takes it that his opponents will share the concept of justice, but will disagree with him as to whether justice as fairness is an accurate conception of justice; see Rawls 1999, 5); second, to give an account of the role of a concept is not, in itself, to provide the correct conception of that concept, but to endeavor to make sure that disputants have all fixed on one particular concept, by providing some account of the role that the concept plays in our thinking—if the reference of a concept is not first fixed in this way,



THE AUTHORITY OF REASONS

on when writing about authority is simply “the problem of the possible justification of subjecting one’s will to that of another, and of the normative standing of demands to do so” (Raz 2006, 1003; emphasis added). This statement of the basic issue would appear to be a very good place to start.14 Raz thinks that to claim that someone or something is an authority in relation to oneself is, first and foremost, to claim that they stand in a certain normative or justificatory relation to one’s will (or merely purport to, in the case of merely de facto authority). Raz claims that this feature of practical authority sets it apart from epistemic authority. Whereas the will plays an essential role in relation to action, it plays no such role in relation to belief, for we do not ordinarily choose what to believe (Raz 2006, 1034). I do not deny that there is a difference here, but I am suspicious of the idea that the right place to begin analyzing the concept of authority is with a statement of the problem of authority that entails, from the get-go, that there is a radical break between practical and epistemic authority. I would like to suggest instead that although the above way of pinning down what exactly the problem of authority is meant to be initially seems promising, the problem might instead be better stated in a more general way that involves simplifying Raz’s statement as follows: the problem of authority is the problem of the possible justification of one being subject to directives originating outside of oneself. In the case of forming beliefs based on testimony, for example, the problem takes the particular form of determining when it is rational to believe the assertions of another, and the will of the other is relevant insofar as it might be directed by a desire to straightforwardly inform one of some truth, or, alternatively, by a desire to mislead (much more could be said here, of course). I contend that whether or not one’s own will plays a role in one being subject to the will of another is a secondary matter, so far as authority is concerned. I will call the conception of authority that takes the problem of authority to

then, for all we know, we might be talking past each other altogether, i.e. engaging in a merely verbal dispute (since it might then be the case that we are each using different concepts, rather than comparing different conceptions of a single concept). 14 Darwall (2009, 2011) also thinks this is the right place to start, although he further contends that practical authority can only be justified from the second-person standpoint, which we take up whenever we address or acknowledge a moral claim or demand. I defend Raz against Darwall in the next section of this chapter.

THE AUTHORITY OF REASONS



have this broader scope (and is otherwise similar in essence to the service conception) the minimalist conception of authority. Once one carefully distinguishes between being legitimately subject to directives that originate outside of oneself, and being legitimately subject to the will of another in a way that involves one’s own will, it becomes very attractive to think of the type of authority that involves a relation between the wills of individuals as being simply a particular species of normative authority in general. It seems sensible to view practical authority as normative authority with respect to action in particular, since “in a way that involves one’s own will” appears to be a mere qualification of a more general, attractive statement of the province of normative authority (a qualification that captures the distinctive role the will plays in relation to action, rather than belief formation). What about the move from “subjecting . . . to that of another[’s will]” to “being subject to directives originating outside of oneself ”? Again, I take this to be a move that simply hinges on noticing that there is a more general problem in this territory than one that centers on the will of another.15 Raz would not accept this move of focusing on normative authority in general in order to understand practical authority, in part because he thinks one type of normative authority, epistemic authority (or “theoretical authority,” to use his term), is very different from practical authority, in the following way: unlike practical authority, epistemic authority goes merely with being authoritative on some subject matter, rather than with having authority over another.16 If he were right about this, he 15 Once one makes this move, I believe it becomes unproblematic to confer authority (with the power to provide preemptive reasons) even on alarm clocks, when they are suitably prepared by agents who rely on them to ensure that they get out of bed on time. Darwall (2009) contends that Raz cannot adequately explain why alarm clocks do not have practical authority, and are not thus a source of preemptive reasons; Raz (2010) argues in his reply that he has no problem ruling out alarm clocks; I would instead happily bite the bullet (in relation to some alarm clocks, owned by people particularly prone to irrationality in the early morning). It is worth noting that Raz’s theory of practical authority was initially designed with legal and political institutions in mind—one way in which institutions differ from people is that they do not, strictly speaking, have wills. Plausibly, this fact may provide grounds for thinking Raz’s account of authority ultimately collapses into the minimalist account (this is my bone of contention, rather than Darwall’s—I think Raz does a good job of responding to Darwall’s reasons for supposing alarm clocks pose a problem for his account). 16 Raz (2006, 1034). Stephen Darwall (2006, 12) likewise contends that epistemic authority is fundamentally different than practical authority, but he locates the difference in the first being fundamentally third-personal, unlike the second, which, is essentially second-personal (on his view).



THE AUTHORITY OF REASONS

would also be right to resist my move of viewing practical authority as simply a species of normative authority in general; indeed one might then suspect that statements concerning “epistemic authority” and statements concerning “practical authority” involve the deployment of different concepts of authority altogether (this would be surprising, since it would suggest that the word “authority” is ambiguous, when it does not appear to be). To make his case that epistemic authority does not involve having authority over another, he uses an example, but what he says about this example strikes me as misleading at best. Raz (2006, 1034) writes, “For example, some people are authorities on eighteenth-century farming methods, but they do not have authority over anyone. I know nothing about eighteenth-century farming methods and should take what they say as authoritative, but they do not have authority over me.” I think it is true that, on the assumption that one knows nothing about eighteenth-century farming methods and one has no reason (or insufficient reason) to seek to form any beliefs about eighteenth-century farming methods (apart from superficial or indirectly relevant beliefs, such as that one is not interested in eighteenth-century farming methods), what an expert on such matters has to say about them will have no bearing, and plausibly should not have any bearing, on any of the beliefs one will form. However, suppose one suddenly becomes interested in eighteenth-century farming methods; assuming that one is then in the business of forming beliefs about eighteenth-century farming methods (the reader can fill in the story to the requisite level of seriousness), the experts on the topic surely do indeed now have authority over oneself, with respect to the beliefs one forms about eighteenth-century farming methods. Context matters when considering who has authority over oneself. Practical authority is no different from epistemic or theoretical authority in this respect. No military sergeant presently has authority over your actions (let us assume); but, presumably, there is a possible context in which a military sergeant would have such authority over you. Since practical authority is no different than epistemic authority in this respect, I believe Raz fails to establish that only practical authority is authority over someone. In the same passage, Raz also claims that: (1) the distinction between merely de facto authority and legitimate authority does not exist in the case of theoretical authorities; and (2) only in the case of practical

THE AUTHORITY OF REASONS



authority do we say that someone “has authority” (rather than just “is an authority”). I would suggest that (2) is an interesting, but ultimately superficial fact about our language (admittedly, it does sound a little odd to say “that teacher has the authority to make little Jack believe that 1 + 3 = 4”). So far as (1) is concerned, I am a little surprised to find Raz making what seems, on the face of it, to be a straightforwardly false claim. Some Creationists are de facto epistemic authorities on the topic of fossils, for they strongly influence the beliefs of large numbers of followers who take them to be authoritative on the topic of fossils, but they nonetheless fail to be epistemic authorities de jure. Raz might wish to claim that the correct thing to say about this example is that Creationists are not epistemic authorities of any kind, not even de facto; however, this also seems false—if you were writing a book on Creationist teachers you might quite naturally write, “their statements regarding fossils were authoritative with respect to the beliefs of their followers”; and you would be expressing a different kind of thought if you wrote, “their statements regarding fossils were wrongly taken to be authoritative by their followers.” If the reader still suspects there is a crucial difference here, then perhaps that is because he or she is misled by the fact that, in many contexts, epistemic authorities may disagree on some topic, or the fact that, in many contexts (such as in university classrooms), it is important for adults to think and reason about even the most seemingly settled of questions. That these facts can only mislead at this point should be obvious if one emphasizes a point that Raz himself makes much of, viz that beliefs are not formed in a direct, voluntary matter. It may be that the will can play an important role in placing oneself inside or outside of a context where a particular expert has epistemic authority over one’s beliefs (this may sometimes be rational to do, and sometimes irrational), but the will cannot (rationally) play a direct role in resisting epistemic authority, once one is in the relevant type of context. It is true that a good theory of authority will need to adequately respect the autonomy of individual agents, but this is equally true in both the practical and epistemic domains: when it comes to the preemptive reasons that play a central role in Raz’s service conception of authority, for instance, Raz does not want us to think that such reasons provide us with absolute duties to defer to others, or could ever be so strong or all pervasive that they would undermine our rational capacities (since the



THE AUTHORITY OF REASONS

whole point of such reasons, on his account, is to allow us to better conform with reason in the long term). On the contrary, individuals will need to be able to rely on their own capacity for rational judgment to ensure that their suspension of any attempt to respond to certain types of reasons directly (rather than through the derivative reasons that authorities provide) is not inappropriate. Similarly, the ability to respond to epistemic authorities appropriately requires both a capacity to trust others—to take them at their word—as well as a capacity to be able to judge when it is appropriate to suspend trust and call evidence into question.

4 A Challenge Stephen Darwall (2009, 2011) agrees with Raz that practical authority is radically different than epistemic authority, but he rejects the service conception of authority. As part of a defense of an alternative, secondpersonal account of practical authority, Darwall provides a series of apparent counterexamples to the claim that the NJT lays down the conditions for determining when a putative authority’s directives actually provide people who are subject to that authority with derivative, preemptive reasons. These putative counterexamples, if they succeed at all, are as much counterexamples to the minimalist conception of authority as they are to the NJT.17 To take one such example: even though it may be desirable for me to treat an alarm clock’s authoritative voice recording (“You must get out of bed now!”) as a binding order, this does not mean that its directive actually preempts the reasons I have to stay in bed. I do not, Darwall claims, now possess a preemptive reason to get out of bed that would exclude consideration of reasons to fall back to sleep. Nor does it follow that the alarm clock has acquired authority over me or that I can be blamed for my failure to comply with its directive. What the alarm clocks lacks, according to Darwall, is the second-personal standing to issue directives to people, as well as the capacity to hold them accountable, so it simply cannot have the capacity to create preemptive reasons. Furthermore, he claims that it will make no difference to

17 Actually, this last claim is not quite true. I need not be especially committed to the idea that authorities create preemptive reasons (although I happen to think they sometimes do); but I am committed to the idea that they provide agents with derivative reasons in a way that is even more general than the NJT would allow. This will be made clearer below.

THE AUTHORITY OF REASONS



our judgment regarding the lack of such a standing, and the practical authority that would go with this standing if it were there, if we instead imagine that I have hired a person to come by my bed and provide me with an “authority experience.”18 We might here simply deny the verisimilitude of Darwall’s intuitions, and hold that the alarm clock, or at least a person doing the job of an alarm clock, could have practical authority over a person. Suppose Fred were to hire a personal life coach to help him improve his work ethic and discipline. The coach’s orders to get out of bed at 7 am might then satisfy the NJT and be genuinely binding on Fred. Darwall would reply that it would be very odd to think that we can hold Fred morally accountable to his coach, and that the coach thus lacks the standing to create preemptive reasons.19 I believe the intuitions Darwall relies on here do not withstand critical scrutiny, unless (like Darwall) one has already accepted that interpersonal moral authority is the one true kind of practical authority. In his discussions of Raz’s conception of authority, Darwall denies that talk of authority and accountability, rights and duties, is appropriate in cases involving prudence, but does not provide sufficient, non-questionbegging reasons to cordon off practical authority to the exclusive sphere of interpersonal morality. One could, on the contrary, quite plausibly criticize Fred for failure to comply with his coach’s orders, from the point of view of prudence (rather than morality), and construe the coach as possessing the capacity to create preemptive reasons to the extent that they are in Fred’s own interest, despite the fact that we might not go so far as to blame Fred for any failure to follow his coach’s directives. Fred might thereby be taken to have sufficient reason to accept his coach’s directives as legitimate, and the coach in turn would have practical authority over Fred. In defense of the minimalist conception of authority, I would further add that since such a coach would have practical authority over Fred, I don’t see any reason to suppose an alarm clock could not have practical authority (in the most general sense) over Fred.

18

The examples, as well as the quoted phrase, are from Darwall (2011, 113–16). One might also be concerned that the fact that the life coach must have been willingly employed to give directives, and did not just appear out of nowhere to give such directives, is playing a role in relation to our intuitions. I address this concern below. 19



THE AUTHORITY OF REASONS

As Darwall is aware, Raz considered in earlier work how best to respond to a variation on the criticism that Darwall provides.20 Raz considers a case of a man who goes to a cooking class and decides not to follow his teacher’s instructions at some point during the class. To the criticism that it does not look like the teacher has the authority to demand that the man in the cooking class follow his instructions—even when his directives are the kind that should provide preemptive reasons according to the NJT—Raz replies that we can explain away this intuition. What is needed, he thinks, is for us to bear in mind that being able to make autonomous decisions is plausibly an important aspect of well-being (one might add that spontaneity and creativity further constitute important goods that would be ignored if the teacher were to be dictatorial in his response). However, Darwall (2009, 151) thinks that even if we alter the case to make it one where all the man who attends the class wants to do is to cook as well as possible, we still will not come to accept that the teacher possesses the authority to demand compliance with his directives. I would like to provide a different response that complements and strengthens, rather than contradicts, the response I just took from Raz: I think it is important to bear in mind that what might be fueling counterauthority intuitions in such cases (that is, cases that involve prudential, rather than moral, reasons), to the extent that people might be inclined to share Darwall’s intuitions, may be general ignorance about what is genuinely in the best interests of other people. Plausibly, we are rarely, if ever, in a particularly good epistemic position vis-à-vis the truth about what is best for other adults’ lives, especially when we limit our attention to cases where what is at issue are various ways of improving another individual’s life.21 This is, in part, because we are very unsure 20

See Raz 1986, 64–5. What is at issue in the cases we are focusing on is not an obvious threat to another person’s life or well-being—since, in such cases, people might, very plausibly, be thought to possess knowledge about what is in another person’s self-interest—but, instead, ways to improve a healthy person’s well-being. In cases where it is really clear that some act will radically undermine a person’s well-being, I do not share Darwall’s intuition that a directive cannot be authoritative. For example, when I yell at you to look out for the ice giving way on the river under your feet, my demand that you do so is, I think, quite authoritative (assuming we are not talking about a case of rational suicide), and this demand can be authoritative even if you are a complete stranger to me. Darwall might well reply that you are under no obligation to me to look out for the ice giving way; even so, you ought to pay attention to what I am saying, and you ought to look out for the ice giving way, and that, I would contend, is enough to make my demand practically authoritative. 21

THE AUTHORITY OF REASONS



about what is the correct, principled way to weigh the factors that contribute to well-being, and even whether well-being depends on preference-satisfaction, or the balancing of goods on an objective list, or factors that might be specified by some other general theory of wellbeing. We also very often have at least some degree of ignorance about all the prudentially relevant non-evaluative facts—for example, the relevant preferences of the person we are considering.22 However, if we simply imagine a teacher being in a state of complete knowledge about what is actually best for her adult student, and the adult student being epistemically justified in accepting that the teacher is knowledgeable in this way, and the value of making autonomous choices (or some relevantly similar value) not being something that is at issue, then I submit that it would not be counterintuitive to think of her as being a genuine practical authority in relation to the student (the type of authority the student really ought to follow, regardless of whether moral blame is an appropriate attitude in a case of noncompliance). Of course, given Raz’s extremely attractive contention that part of what makes a life go well is the exercise of autonomy,23 the (ideal) teacher who is in a position to actually know what is best for an adult student in some particular context would very often thus know that the student should make his own mistakes or creative decisions—this knowledge could itself be the basis of a legitimate, authoritative directive to be self-guided (“Be yourself!”).24 And if the teacher knows that what is best for her student is that he make his own decisions, she would also rightly 22 There are multiple roles preferences might play in this context. They are obviously relevant to a preference-satisfaction account of well-being (or to a hybrid account of wellbeing that combines a preference-satisfaction criterion with an objective list criterion, in order to distinguish between good, bad, and neutral preferences). A hedonist who rejects this account is still likely to think that having one’s preferences satisfied (at least when one knows they have been satisfied) will strongly positively correlate with one experiencing pleasure. And an objective list account of well-being might include autonomy as a value, in which case realizing one’s specific preferences (or some subset of one’s preferences, such as those that one has control over) will be thought to be generally valuable because doing so enables one to exercise one’s autonomy. 23 See Raz 1986, 390–5 and Raz 2006, 1015–16. 24 It is not clear what sense Darwall can make of such a directive, and how it appears authoritative to us: the teacher is clearly not undermining or interfering with the student’s autonomy in this case. To flesh out the case a little, we might add that following the teacher’s directive—that the student be self-guided—will lead the student away from a tendency to make bad faith decisions, hence make him better conform with reason in the long term.



THE AUTHORITY OF REASONS

conclude that it would be a mistake for the student to fail to conform to the directive to be self-guided. The student would be demonstrating bad faith were he to pretend that the teacher was actually prompting him to choose X over Y, rather than to make a decision for himself with respect to X and Y. I am suggesting here that one’s intuitions are likely to shift if one attends to the case just described, and if one thinks about it in relation to practical authority in general. Focusing on a teacher who possesses this kind of knowledge should lead anyone who begins by sharing the intuitions that help Darwall build his case against Raz away from those intuitions. Once such a reader’s intuitions have shifted, I would also suggest that he or she consider the possibility that not only agents in the right kind of epistemic position can possess practical authority in cases involving the prudential reasons of other agents, but that even alarm clocks (as in the earlier example that Darwall discusses) might possess such authority, in virtue of them reliably issuing directives that make agents better conform with reasons that already apply to them. If I am right about this, even the directives of alarm clocks can constitute preemptive reasons. There is a further dialectical move open to a defender of Darwall at this point. My example of the lifestyle coach, or, indeed, Raz’s example of the cooking teacher, might be thought to be problematic examples to use when arguing against Darwall, precisely because they are cases where a contractual relationship is entered into. It could be conceded by someone sympathetic to Darwall’s approach that such coaches or teachers do possess practical authority, but that they only do so in virtue of the contractual relationship or promise entered into at the beginning of the relationship. And contracts or promises are clearly phenomena that are well-suited to admit of a second-personal analysis, if anything is. Suppose I distinguish between two cases: (1) unbeknownst to you, a stranger has somehow been collecting highly accurate information about you, and he now comes to you to provide you with a potentially lifechanging directive; and (2) you have voluntarily employed a lifestyle coach to collect highly accurate information about you, and having collected this information, he now comes to you to provide you with a potentially life-changing directive. It is tempting to suppose that even if Darwall were to concede that the putative authority in (2) is a genuine authority, while denying that the putative authority in (1) is a

THE AUTHORITY OF REASONS



genuine authority, he would still be in a dialectically stronger position than Raz, because he could explain the difference between (1) and (2) in second-personal terms (there is a contract or promise at play only in (2)), and because many people might intuitively judge that the stranger in (1) has no practical authority over you, no matter how accurate and detailed his information is about you. However, such a conclusion would actually be much too quick, because Raz has available to him a perfectly good alternative explanation as to why the person in (1) would not count as a practical authority. For Raz, it is crucial that certain epistemic conditions be met by any person who is being directed to follow a putative authority. What is lacking in (1), and is present in (2), is epistemically well-grounded trust in the putative authority.25 I take it that expert lifestyle coaches and cooking teachers that we might actually employ will indeed be capable of being excellent epistemic authorities in the relevant domains, otherwise we would not (or should not, anyway) employ them. In any case, Raz could claim that their directives are only authoritative if we meet appropriate epistemic conditions, and these conditions will normally not be met in cases like (1); hence, I have an alternative explanation to rely on for the judgment that the stranger in (1) is not a practical authority. Given my earlier suggestions concerning how the intuitions that favor Darwall might be undermined, I think the dialectic now favors Raz, and the minimalist account inspired by Raz’s account of authority, rather than Darwall.

5 The Authority of Reasons, as Evidence I have proposed that we should accept a minimalist conception of practical authority, according to which the problem of authority is the

25 Raz thinks that there are epistemic conditions that must be met by any people that might be thought to be subject to an authority, in order for the putative authority to count as legitimate with respect to those people (see, for example, Raz 2006, 1025–6). Spelling out these epistemic conditions in a precise way is a very tricky business for the Razian: if he makes them too weak, it will seem implausible that there can be any practical authority in play (I fear that the mere condition of knowability, provided in Raz 2006, has this weakness), but if he makes them too strong, then it will turn out that what is really at play is mere epistemic authority, rather than practical authority (see the very interesting conclusion to Raz 2010, 300–1). I do not believe that the version of the minimalist conception of authority I defend suffers from this problem, since it is not merely epistemic authority that is epistemic on this conception— practical authority is epistemic too.



THE AUTHORITY OF REASONS

problem of the possible justification of one being subject to directives originating outside of oneself. Now it can be seen that this is simply the problem of how it is that some facts provide agents with normative reasons, where the focus is on agents themselves and their rational acceptance of certain facts as reasons, where attending to reasons can lead agents to make correct judgments about what they ought to do, and where these reasons do not (in any straightforward or obvious way, at least) rest on properties merely internal to the agent. We are especially interested in how it is that many facts that are not plausibly construed as normatively basic (that is, not basic ought facts or fundamental rightmakers) provide agents with normative reasons for action, but we might also say that all normative reasons are facts that constitute pro tanto directives for rational agents. Some such facts will be facts about commands issued by entities (e.g. people or institutions), and in such cases verbal directives may provide normative reasons of a derivative kind, but we should not begin by restricting the notion of a directive to that of a verbal directive. And we might go a long way to answering the general problem of normative authority if we can explain, in general, why certain facts, and not others, count as reasons. This leads one to ask: which facts are normative reasons, exactly? According to some philosophers, such as Tim Scanlon and Derek Parfit, the concept of a normative reason is basic and unanalyzable, so we can never hope to provide a general informative explanation of which facts count as reasons.26 According to reasons as evidence, a fact is a reason to do a particular act just in case this fact is evidence that one ought to do the particular act, and a fact is a reason to believe a particular proposition just in case this fact is evidence that one ought to believe the particular proposition. Recall from Chapter 1, that R is the precise claim that is at the heart of this account. R: Necessarily, a fact F is a reason for an agent A to ϕ if and only if F is evidence that A ought to ϕ.

26 Parfit (2011, 31) writes, “Facts give us reasons . . . when they count in favour of . . . But ‘counts in favour of ’ means roughly ‘gives a reason for.’ Like some other fundamental concepts . . . the concept of a reason is indefinable . . . it cannot be helpfully explained merely by using words.” See also Scanlon (1998).

THE AUTHORITY OF REASONS



R provides the basis for an informative and unified analysis of reasons for action and reasons for belief (and this is one of its selling points):27 in both cases, one is referring to evidence concerning a (putative) oughtfact. This is one way in which this thesis fits together very nicely with the minimalist conception of practical authority, which views practical authority and epistemic authority as simply two species of normative authority. The second way in which the reasons as evidence thesis fits together with the minimalist conception is that it helps provide an informative answer to the problem of the possible justification of one being subject to directives originating outside of oneself. The justification is said to rest in the fact that one is dealing with evidence, and working out what one ought to do is all about responding appropriately to the evidence concerning what one ought to do. At this point, it might be objected that this analysis of reasons is not at all well-positioned to be part of an answer to the problem of authority, on the grounds that evidence, even evidence concerning what one ought to do, is not something external to oneself. It is certainly true that philosophers have tended to concentrate on a notion of evidence that ties or relativizes evidence to particular subjects, even though they may admit, when careful, that there is another way the word “evidence” is commonly used in ordinary language, such as when we say that the evidence that Smith is the murderer (the bloodied knife, say) has not yet been discovered by anyone.28

27 Some other accounts of reasons are unified, and some are informative—it is a particular strength of the present account that it is both of these. An example of an informative analysis of reasons is the Humean theory of practical reasons, which explains reasons in terms of desires. It is not so tempting to analyze reasons for beliefs in terms of desires, but it is only by doing so that the Humean might also hope for a unified analysis of reasons. 28 Williamson (2000, 195) also considers a bloodied knife, focusing on the way in which referring to an object (the knife) as evidence might be thought to cause problems for his view that evidence consists of true propositions. I agree with his reasons for thinking that the bloodied knife does not in fact cause any problems for his view (R also has it that facts, or true propositions, are evidence)—when we refer to the bloody knife as itself evidence, this is shorthand for a large number of propositions concerning the knife, that Jones’s fingerprints are on it, etc. (particularly those propositions that might fit with the competing hypotheses of the prosecution and the defense in a court of law, as Williamson says). I am instead interested here in the fact that we seem to be able to properly describe an undiscovered bloody knife (or an undiscovered fossil, never seen by a human being) as evidence.



THE AUTHORITY OF REASONS

Four things can be said in response here. First, even when we restrict our attention to evidence that is possessed by particular people, there is an important, non-controversial sense in which such possessed evidence is independent of the will, insofar as it concerns propositions that it is rational for one to proportion some positive degree of credence to, not simply any propositions at all that one might happen to wish to accept as evidence. This may be enough for my immediate purposes, since my main claim here is that the experience of being confronted with evidence typically has a property of authoritativeness.29 Second, Timothy Williamson (2000) has argued for a very attractive account of evidence according to which it is not possible to think of possessed evidence as purely subjective; in particular, he contends that a person’s evidence for a proposition is simply the content of their knowledge. Since knowledge is factive, possessed evidence turns out to be objective in a very robust sense—it consists of objective truths. Of course, Williamson’s account of evidence is contentious. For my purposes here (as distinct from my purposes in Chapter 4), his account of (possessed) evidence is worth mentioning simply because it shows that it is not incoherent to view evidence as something that is quite objective, in one very important sense, even using the notion of evidence that always connects or relativizes evidence to particular agents. Third, I believe, in any case, that it is important that we start to pay more attention to the unpossessed notion of evidence (such as becomes apparent when we talk about as yet undiscovered evidence); it is presently out of favor with philosophers, but its prevalence in ordinary thought and language suggests that there is much room for further philosophical work here. Finally, I have so far been speaking of possessed and unpossessed notions of evidence (and one can also speak of parallel possessed and unpossessed notions of a reason), as a way of leaving it up in the air as to whether we are really talking about different concepts, or merely different conceptions of evidence here. In fact, I think that there is only one concept generally employed here, and it is a concept that can be deployed to capture both the possessed and unpossessed usages of the word “evidence.” Support for this contention comes from the observation that we ordinarily talk both about having evidence and about there being evidence, 29

This echoes the point I made above about how even a Humean theory of practical reasons needs to leave room for the property of authoritativeness.

THE AUTHORITY OF REASONS



just as we talk of having reasons and there being reasons. For example: the fact that there is a bomb in a building is a reason to leave it immediately/ evidence that one ought to leave it immediately, even if one has no reason to leave it immediately/evidence that one ought to leave it immediately. Kearns and I have argued that this mapping of being/having across reasons and evidence is a consideration that counts in favor of adopting the reasons as evidence account of reasons.30 I hope to have now done enough to demonstrate that reasons as evidence can fit together harmoniously with the minimalist conception of authority. Evidence concerning what one ought to do or believe seems prima facie to exhibit the property of authoritativeness required for the minimalist conception to work. Still, I need to provide more in the way of an argument for reasons as evidence, and that is what I’ll do next. I’ll argue that evidence concerning ought is particularly well-placed to play the authoritativeness role, and that it is in virtue of this fact that the particularly contentious, right-to-left reading of R is true.

6 The Razian Insight and a New Argument for Reasons as Evidence To a skeptical reader, the minimalist conception of authority might be thought to provide little more than a slightly novel way of asserting that some facts have the status of being reasons, and of underlining that, both as reasons for belief and reasons for action, such facts guide and justify our actions in a normative fashion. I don’t think such skepticism is warranted, at least in part because the minimalist conception, as I construe it, holds to a powerful element of Raz’s service conception; an element which, when reconfigured, we might term The Razian Insight (for want of a better name). To see how this is so, we can compare a canonical statement of the Normal Justification Thesis with a simple variation on it (the underlined words are the only words that vary).

30

Kearns and Star (2008, 38; and 2009, 234–5). Feldman (2004) thinks of evidence that one possesses as available evidence—he considers several versions of this view, arguing in the end that available evidence is evidence that one is thinking of. On this view, which I must reject, unpossessed evidence will still be evidence for particular (non-idealized) agents.



THE AUTHORITY OF REASONS

NJT: “ . . . [the] way to establish that a person has authority over another person involves showing that the alleged subject is likely better to comply with reasons which apply to him [independently] if he accepts the directives of the alleged authority as . . . binding and tries to follow them, rather than by trying to follow the reasons which apply to him directly.” (Raz 1986, 53) NJT*: the way to establish that a fact has authority over a person involves showing that the alleged subject is likely better to comply with reasons which apply to him [independently] if he accepts the directives of the alleged authority as . . . binding and tries to follow them, rather than by trying to follow the reasons which apply to him directly.

NJT* is simply NJT modified to echo the transition from the service conception to the minimalist conception in Section 3. I take it to be an essential element of that conception. In combination with Raz’s thought that the directives of authorities (people, institutions, or, on the minimalist conception, facts) can be the source of certain derivative reasons, NJT* preserves his deep insight that many reasons are only reasons in virtue of the fact that they empower individuals to better conform with non-derivative reasons. On the minimalist conception of authority, as on the service conception, authority is not a mere synonym of “normativity” (to respond to the skeptical worry just expressed), but a property that certain entities have in virtue of a triadic relation they stand in, to the rational capacities of particular agents that are concerned with responding to reasons (that is, reason), on the one side, and to more fundamental normative reasons, on the other side.31 When we focus on the authority of reasons as directives, it is their guidance function that we especially pay attention to. Authoritative directives ask us to pay attention to their content (and, in the case of the exclusionary reasons that go together with preemptive reasons, to not pay attention to certain other reasons). If we now focus on the question of how derivative reasons are generated according to the minimalist account (rather than the service account), we arrive at the following idea. The Razian Insight: If non-ideal agents would, in the context of deliberation, best conform with reason by responding to facts of a certain general kind as reasons (rather than by attempting to respond directly to fundamental reasons), then facts of that general kind are (derivative) reasons. 31 Of course, the fundamental reasons also have the property of being authoritative, but, given the precisification of authoritativeness just provided, they have this property in a somewhat more trivial fashion (in virtue of a relation they stand in to themselves, as well as the rational capacities of agents).

THE AUTHORITY OF REASONS



I speak here of conforming with reason, and in doing so, I purposely echo Raz.32 “Reason” is here meant to be a name for that part of our rational nature that is concerned with fallibly responding to reasons (including both reasons to believe and reasons to act), especially, however indirectly, fundamental reasons.33 When we think about what reason permits and requires, we are thinking about what rationality permits and requires in relation to what agents take to be their normative reasons. On the extremely attractive assumption that rationality as a whole is not conflicted (i.e. that there cannot be rational requirements that can conflict, such that rationality would require A and not B, yet also require B and not A at the same time), our judgments about what rationality as a whole might say is impermissible may also be relevant to judgments about whether something conforms with reason—if either reason (i.e. that part of rationality concerned with reasons) says X is impermissible or rationality as a whole says X is impermissible, then X does not conform with reason.34 If this seems confusing, just bear in mind that instead of using the phrase “better conform with reason” above I could have used the phrase “better conform with fundamental reasons in a way that is not irrational” (and made corresponding changes in the description of each step of the argument that I present below). The ruling out of irrational responses does much to help guarantee the truth of the Razian Insight: if responding to some general class of facts as reasons could only help me conform with reasons that already apply to me—so that I would be doing a better job by the light of reason itself, in a way that is not irrational—why wouldn’t such facts themselves get to count as genuine normative reasons? Reasons speak in favor of various acts or beliefs and, in so doing, provide essential guides to right action or belief (to doing what we ought to do, or believing what we ought to believe). If some general class of facts can deliver on this score, without our responding to such facts as reasons leading us to commit any rational

32

See, especially, the essays in Raz (1999). On a strong version of this view, our rational capacities simply have the function of responding appropriately (if often indirectly) to fundamental reasons (in which case “reason,” in the sense I intend when speaking of a capacity, will be equivalent to “rational capacities”), but I do not need to commit myself to this strong view. 34 Also: if rationality as a whole says X is required, then reason can not say X is not required; and if reason says X is rationally required, rationality as a whole can not say X is not required. 33



THE AUTHORITY OF REASONS

offence, it seems such facts have done enough to earn the status of being reasons. To the extent that the reader found him- or herself attracted to assent to the minimalist conception of authority in the above discussion of that conception—perhaps because they initially found Raz’s service account attractive, and were then convinced by the considerations I provided for broadening that account (which included reflection on some examples)— the reader may be willing to allow that the Razian Insight is a foundational insight. In any case, I must confess that at this point I do not know what more I can say to defend it. Philosophy being the business that it is, some will no doubt take the idea in question to be neither foundational, nor an insight. Still, it appears to me that I have hit bedrock at this point. Before I provide the argument that makes use of the Razian Insight as a premise, let me briefly say something more about arguments for R. As I have noted, a number of arguments have been presented in previous papers. In “Reasons as Evidence” (2009), in particular, Kearns and I explicitly presented six arguments for R. As soon as we started presenting these arguments, we found it much easier to obtain agreement from people that the left-to-right reading of R is true—or, at least, relatively unobjectionable—than obtain agreement concerning the right-to-left reading of R.35 It seems relatively straightforward to think that all facts that are normative reasons to X are also evidence that one ought to X, but much more striking to think that all facts that are evidence that one ought to X are also normative reasons to X. For this reason, I am focusing on the second of these claims (R being the conjunction of both of them). In doing so, I don’t mean to say that the other claim can’t be challenged, much less that it can’t be defended. Here, for instance, is one argument that I still stand by for the claim I will be passing over which comes from the earlier coauthored paper: (1) reasons play a central role in reliable practical reasoning that issues in conclusions concerning what one ought to do; (2) if reasons are not evidence that one ought, then they can’t play this role; (3) therefore, reasons are evidence that one ought (2009, 224–5).

35 John Broome once told us, some time ago, that he had no objections to the left-toright reading (of course, by reporting this, I do not mean to attempt to hold him to an informal comment).

THE AUTHORITY OF REASONS



Now I am in a position to introduce the new argument for the rightto-left reading of R (i.e. for the claim that all facts that are evidence concerning what one ought to do are reasons). Here it is: 1. If non-ideal agents would, in the context of deliberation, best conform with reason by responding to facts of a certain general kind as reasons (rather than by attempting to respond directly to fundamental reasons), then facts of that general kind are (derivative) reasons (The Razian Insight). 2. Non-ideal agents would, in the context of deliberation, best conform with reason by responding to a certain kind of fact as reasons, namely facts that are evidence concerning what they ought to do. 3. Conclusion: Facts that are evidence concerning what agents ought to do are reasons. I have already discussed the first premise of this argument. I am now left with a need to defend the second premise. I think I can do this by asking: what might the alternatives be to responding to evidence concerning what one ought to do as providing reasons, in a situation where one is not able to respond directly to fundamental reasons (due to ignorance of them—this is what makes agents non-ideal)? Here are some alternatives: (1) an agent could respond to her desires; (2) an agent could respond to her judgments concerning what is good or best; (3) an agent could do nothing; (4) an agent could somehow respond to the fundamental reasons (even though ignorant of them); and (5) an agent could form, and act on, whatever beliefs about reasons will effectively lead her to respond to the fundamental reasons (independently of whether such beliefs are true or well-justified).36 No doubt this list is not exhaustive, but these seem to be the clearest alternatives. I will now argue that none of them are good alternatives. Doing so will increase the attractiveness of the second premise of the main argument, as I will be continually comparing the option highlighted in that premise with the 36 Another alternative would be to think that the agent should only use evidence that she presently possesses as a guide. This is an unsatisfactory alternative in light of the fact that there are cases where an agent is culpable for not possessing evidence (e.g. I might not possess evidence that ingesting a certain medicine will make me unfit to drive, since I fail to read a very clear message to that effect written on the medicine jar; intuitively, the unpossessed evidence that is provided by the message on the label provides a consideration which, if I were to attend to it, would ensure that I better conform with reason).



THE AUTHORITY OF REASONS

alternatives. Of course, this strategy leaves me vulnerable to critics who might argue that some further alternative to that premise is preferable to it, but that does not mean the strategy is a bad one to employ here. Imagine an ethically conscientious agent attempting to determine what to do, on some particular occasion. I suggest we do this not because we are trying to provide an account of reasons for conscientious agents in particular, but just because if (1) to (5) are not good alternatives for conscientious agents to the option provided in the second premise (evidence that one ought), then that will be enough to rule them out as good alternatives in general. We should imagine that this agent is also either very uncertain that she has cognitive access to any relevant fundamental reasons (during the window of time that is open for deliberation), or even justifiably believes that she does not have such access. This is necessary to avoid the agent confusing derivative and fundamental reasons (of course, I don’t think that ordinary agents usually classify reasons into these categories, but that doesn’t matter for our purposes). And we should imagine that it is in fact true that this agent doesn’t have any straightforward cognitive access to any relevant fundamental reasons. Let me further specify that she is aware that one option on the table is for her to respond to evidence concerning what she ought to do— she is to compare the alternatives to this option. Now this agent considers option (1), i.e. the possibility of responding to her desires in a way where she can draw the inference that she ought to F (or, at least, that by F-ing she will be getting as close as she can to acting as she ought) on the basis of her desires to achieve the end that F will lead to her achieving (which, we may suppose, are stronger, as a set, than her desires for any other end). The problem is that, as a conscientious agent, she will be aware that desires are often irrational or biased states, and certainly not always as they ought to be—it would be one thing if she were to respond to desires that she ought to have, but she is aware that, since she doesn’t have access to any fundamental reasons, she doesn’t have access to any fundamental reasons to desire. Two possible thoughts might next occur to this agent: (i) she might think she is a virtuous or fully rational person, so her desires are reliable (having been shaped to avoid bad biases, etc.); or (ii) she might consider what she would desire if she were virtuous or fully rational. The first of these options is certainly not going to provide a general alternative to evidence that one ought, since it will only apply to the virtuous or fully

THE AUTHORITY OF REASONS



rational agent (assuming that those who know they are virtuous are allowed to infer that they ought to F on the basis of desiring to F). In any case, it collapses into the view that she should respond to evidence concerning what she ought to do. If she asks herself why she should follow her desires, her thought that she should do so because she is virtuous or fully rational can be understood as resting on the thought that facts about the desires of the virtuous or the fully rational are good reliable indicators (i.e. evidence) concerning ought-facts—if they were not reliable in this way, why ought they be followed? Alternatively, if the agent we are imagining is not virtuous, but is considering trying to aim for what she would desire if she were virtuous, or fully rational, she will need to examine evidence for thinking that her idealized self would desire the end that requires her to F. Now she might think of such evidence in either of two ways: (a) she might think her idealized self would be in a position to do as she ought by responding appropriately to fundamental reasons, perhaps partly in virtue of having consistent and coherently unified desires (Smith 1994), or (b) she might think her idealized self would not be in such a position, but she would at least be fully rational (her non-ethical beliefs would be impeccable and her desires would be consistent, coherently unified, etc.). If she accepts option (a), it seems that, in her present position of ignorance of the fundamental reasons, she should think that evidence concerning what her idealized self would want her now to do just is evidence concerning what she now ought to do. We have a ready explanation available here: the reason why one might think one should defer to one’s ideal self is most plausibly thought of as being that one thinks one’s ideal self is in a better position to recognize what it is important to do; that is, that she is better placed to respond well to the relevant evidence. If, on the other hand, she accepts option (b), she is now left comparing evidence concerning what she ought to do (the option that, for the sake of comparison, we put on the table at the beginning) and evidence concerning what she would desire if virtuous or fully rational, where this, ex hypothesi, can come apart from being in a position to recognize fundamental reasons. In this case, the option of acting on merely more consistent and coherent desires (or more virtuously shaped desires) will not seem like a good alternative to acting on evidence concerning what she ought to do, since such desires (or virtuous dispositions) could not be more reliable (in relation to the relevant



THE AUTHORITY OF REASONS

ought-facts) than all her relevant evidence concerning what she ought to do.37 The next option is to suppose that the conscientious agent who is ignorant of fundamental reasons might aim to act on (2) judgments about what is good or best. The impetus for saying that agents are generally ignorant of fundamental reasons came from the idea that ordinary agents are ignorant of normative ethical theory. This means we should not take them to be committed to a view that all reasons are evaluative at base. They will, if they are appropriately epistemically cautious, be open minded about issues that divide consequentialists and deontologists. Hence, if they have a choice between acting on evidence concerning what they ought to do and evidence concerning what is good (or best) in some context, if this evidence points in different directions, it seems appropriate for them to act on the evidence concerning what they ought to do. Having said this, I do not need to deny that evidence that an option is good (or will realize some good) is also evidence that one ought to act on this option.38 My point is that in the cases where these classes of evidence come apart it seems appropriate to act on evidence that one ought. Alternative (3) is to do nothing. It seems clear that, in general, doing something that is based on evidence concerning what one ought to do is rationally preferable to doing nothing. It must be preferable because it is not as though doing nothing is itself something that the evidence concerning ought needs to rule out—such evidence may, in fact, direct one to do nothing (or, more precisely, to wait until one has gathered more evidence, or until a certain event occurs, or to simply sit tight in order to play it safe). However, if a rational assessment of such evidence does not direct one to do nothing, surely it is preferable to do as the evidence suggests one ought. 37 Things would be more complicated were this agent in a situation where she had to compare relying on weak evidence that she ought to act in some way with relying on strong evidence that her internally rational (yet still possibly unreliable, in relation to what she ought to do) idealized self would want her to act in some other way. This complication simply introduces a special case, so it won’t matter in general. 38 I also do not need to deny that there are more general senses of “good,” “best,” and “better than” that we can use when talking about deontic matters, e.g. when we say that an act that conforms with duty is better than an act that does not conform with duty. Using “best” in a more general sense, I am arguing by elimination that it is best to act on evidence that one ought.

THE AUTHORITY OF REASONS



What are we to think of alternative (4)? It might seem very odd to suggest that when, due to ignorance, one is not able to respond to fundamental reasons, then one will still best conform with reason by responding to fundamental reasons anyway, but I suppose one might be misled into thinking that this is true if one were to only pay attention to reasons for action, and ignore reasons for belief. It might sometimes very luckily be true that it is possible for agents to act in conformity with fundamental reasons for action in a condition of ignorance (and without taking up the alternative of responding to evidence concerning what they ought to do) if they were to constantly act in ways that, by their own lights, they don’t have good reason to. However, this is clearly to act in ways that are irrational. Generally speaking, intentionally acting in ways that one does not take oneself to have reason to act will violate requirements of rationality with respect to the beliefs and intentions that precede one’s acts—roughly speaking, our intentional acts are explained by our intentions, and to the extent that intentions are formed rationally, they are formed by considering what one ought to do; deliberately forming intentions that are contrary to what one believes one ought to do will thus violate a rational requirement. This is enough to take this alternative off the table. There is now just alternative (5) to consider. I believe it suffers from the same fault as alternative (4), but I thought it worth including here because it might appear to be an attractive alternative to some consequentialists, in particular (the reader will recall that the present project aims to be neutral when it comes to first-order ethical theory). Some consequentialists might think that agents should form, and act on, whatever beliefs about reasons will effectively lead them to respond to fundamental reasons, independently of whether such beliefs are true or well-justified.39 For our purposes here, we need not deny that this is sometimes true (it is not difficult to imagine cases of great moral import where some degree of rational belief formation seems worth sacrificing). It is enough to point out that conforming with reason on the whole is not

39

Plausibly, most of the time consequentialism would direct people to form beliefs in a standard rational fashion, since it is extremely plausible that our rational belief forming capacities produce more good in the long term than attempts to circumvent or undermine them would (especially since too much circumvention will lead to long-term undermining of our capacities).



THE AUTHORITY OF REASONS

the same thing as doing what we ought to do—conforming with reason is fundamentally about us doing our rational best to follow the reasons, both epistemic and practical, that apply to us, in order to get as close as we can to always doing and believing what we ought to do and believe, but there is no guarantee that such guides to right action (despite being the only reliable guides that we have) will always lead us to the promised land.40

7 Derivative, Preemptive, and Instrumental Reasons Let me conclude with some clarificatory remarks regarding the issue of how a defender of the minimalist conception of authority and reasons as evidence might best construe the relation between derivative reasons in general and the subclass of such reasons that are preemptive reasons, and the relation between derivative reasons in general and instrumental reasons. I should emphasize that while Raz thinks reasons that depend on the directives of authorities are ipso facto preemptive reasons, I do not agree with him about this. Some derivative reasons arising from the directives of external authorities are preemptive, but not all are. In particular, I see little reason to think a reason need have the status of being preemptive, rather than merely derivative, when an agent is ignorant of the relevant more fundamental reasons and not even inclined to attempt to follow them. This is as much a criticism of Raz, as a clarification of my own view—since the whole point of preemptive reasons is to preempt consideration of various reasons, such reasons seem unnecessary in cases where an agent is not going to be led astray by attempting to consider more fundamental reasons. I can think of four general factors that might make a derivative reason preemptive: (1) someone might suffer from a relevant form of irrationality in a particular domain (e.g. the directives of alarm clocks can be thought to provide preemptive reasons for people who are irrationally tempted to stay in bed in the morning); (2) someone might have false beliefs about fundamental reasons, and would better conform with 40 And if there are situations where we can and should decide to be irrational for a good end, the leap into irrationality is a leap where we leave these guides (i.e. reasons) behind, at least temporarily.

THE AUTHORITY OF REASONS



reason by responding to derivative reasons (e.g. someone wrongly convinced a particular ethical theory is correct would be better off responding to more mundane evidence concerning what they ought to do than attempting to respond to the considerations their false theory says they should respond to); (3) someone might have true beliefs about the correct ethical theory, but the theory might require a degree of intelligence beyond his or her ken to be directly put into practice in a legitimate fashion (for instance, a consequentialist theory of right action may be correct, and, as is well known, attempts to put a consequentialist principle directly into practice—treating it as if it were a decision procedure— are, in many contexts, likely to make agents act contrary to the principle itself); and (4) in precisely the range of circumstances Raz originally had in mind with his more limited account of authority (listed in footnote 12 in Section 3 above), the directives of a political body might provide preemptive reasons—in line with the theory of reasons that has been defended here, this will be because such directives provide evidence concerning what one ought to do and there is an exclusionary reason present (i.e. evidence that one ought not consider certain other reasons/ pieces of evidence concerning what one ought to do).41 Sometimes none of these factors will be present, and then a reason will be merely derivative, rather than preemptive. I have been focusing on reasons for action, but it is an interesting question to what extent there are preemptive reasons for belief. Consider condition (1) from the previous paragraph. Roy Sorensen (ms.) asks us to consider a case where you receive a packet concerning which you justifiably believe (a) that it contains misleading evidence on some topic of interest; and (b) that, if you open the packet and read its contents, you will be swayed by the misleading evidence contained therein to rationally change your mind on the topic. Plausibly (but not uncontroversially), this is a case where you have a preemptive epistemic reason not to consider (or, more to the point, form beliefs directly based on) the 41 The fourth basis for preemptive reasons suggests that some derivative reasons are more fundamental than other derivative reasons. Returning to the speed limit example, recall that a driver is not meant to consider the state of the road in order to attempt to determine how fast it is safe to travel above the sign-posted speed limit. Facts about the state of the road are themselves derivative reasons (for they surely won’t be picked out as fundamental reasons by a correct ethical theory), but one must not consider them when considering driving above the speed limit, given the relevant preemptive reasons in play.



THE AUTHORITY OF REASONS

evidence contained inside the package (hence also a practical reason not to open the package)—you certainly have evidence that you ought not update your beliefs using the evidence inside the package, and this evidence seems to exclude consideration of the evidence inside the package. There may also be other types of preemptive reasons for belief (or for not believing).42 What about instrumental reasons for action? Some like to use the term “instrumental reason” in a very expansive way to include all derivative reasons for action, in which case I might appear to have defended a particular account of instrumental reasons (see, for instance, Kolodny, forthcoming). I don’t think this is a very helpful way of using this term, but nothing very substantial hangs on this. The more traditional notion of an instrumental reason is that of a reason that exists in virtue of some act being a means to an end that corresponds to another reason (e.g. if I have a reason to be in New York by the afternoon, then I have a reason to take a train to New York in the morning, or if I have a reason to do something creative, and writing now would be something creative, then I have a reason to do some writing now). Although instrumental reasons, according to this standard notion, will also provide evidence concerning what one ought to do (they are no exception to R), they seem dependent on either a causal or constitutive connection that is lacking with many derivative reasons. Consider a case of misleading evidence concerning a putative oughtfact, and suppose that fundamental practical reasons all have to do with protecting autonomy (for the sake of the example). Suppose I am warranted in thinking that whenever Jack grimaces he is in pain, hence that there is a reason to help him (that he is grimacing), but that, as it happens, Jack is cleverly deceiving me on this occasion and is not at all in pain. The fact that Jack is grimacing is thus a misleading reason to help him, i.e. it is misleading evidence that I ought to help him. This misleading reason cannot be a means to protecting Jack’s autonomy, since the act I do to attempt to help him does not either causally or

42

Late in the writing process, I discovered Linda Zagzebski’s (2012) very interesting book on epistemic authority (written around the same time as Star and Delmas 2011, as it happens). Zagzebski utilizes Raz’s theory of authority to provide room for preemptive reasons for belief (see, especially, 2012, 113–17). Unfortunately, I do not have space here to consider specific ways in which our views converge and diverge.

THE AUTHORITY OF REASONS



constitutively result in Jack’s autonomy being protected (even though we might suppose it would if he were genuinely in pain); or, if it does freakishly have such a result, this will involve too accidental a connection to make it legitimate to think of the fact that Jack grimaced as having really grounded an instrumental reason to help him. I think the correct lesson to take from this example is not that even misleading derivative reasons are instrumental reasons, nor that only misleading reasons are non-instrumental derivative reasons, but instead that we need some further condition to be met for a derivative reason (misleading or not) to count as an instrumental reason—roughly, that the agent is in a position to successfully use the relevant considerations in a piece of instrumental reasoning, given an end he is in a position to recognize. Instrumental reasoning is reasoning that proceeds on the basis of agents taking themselves to be authorities on which ends are to be pursued (authorities for the moment, that is), and instrumental reasons are best understood as reasons to take appropriate causally or constitutively connected means to such ends; reasons that exist with a particular force—when they do—in virtue of the fact that such an acceptance of oneself as an authority in relation to the ends in question happens, on such occasions, to be warranted.43

I have not said anything at all above about Raz’s “Facilitative Principle” (2005), which he takes to explain why instrumental reasons are reasons—roughly, because they facilitate the performance of an act that there is reason to perform. I took it to be more helpful for my purposes to start with his account of authority, and the Razian Insight my discussion of it located. Despite the fact that I have found inspiration in some of his work, I labor under no illusion that he would himself accept the way I have used his ideas here. 43

3 Virtue 1 Introduction The main claims of this chapter are as follows. First, it is not the case that right acts always accord with virtue (as virtue ethicists generally claim they do), and it is also not the case that good effects in the actual world are what ensure that particular character states count as virtues (as consequentialists tend to think). Second, rather than search for a tight connection of dependency between virtue, on the one hand, and the right or the good, on the other hand, we should instead simply characterize the virtuous as people who respond well to normative reasons in their actions and feelings, on the basis of virtues, and characterize particular virtues as involving stable dispositions to respond to reasons of particular types, where the types can be specified using non-ethical language. Third, given that we can understand normative reasons as evidence that we ought to act in one way or another (a claim I argued for in the preceding chapter), it follows from the previous claim that virtue can be characterized as being about responding to evidence concerning truths about what one ought to do, and not necessarily or essentially about responding to the deepest explanatory facts, or “right-makers”, concerning what one ought to do (so people may count as virtuous when only responding to derivative normative reasons). This means that we can understand the practical virtues as being largely grounded in the epistemic domain. Two general issues in ethics are particularly relevant here. The first is the issue of to what extent we need to reconcile normative ethical theories, especially (but not only) consequentialist theories, with ordinary pre-theoretical moral judgments and commitments. The second concerns how we might best understand the relation between virtue, on the one hand, and the good and the right, on the other. It is interesting to note that, in relation to both issues, many philosophers working in

VIRTUE



normative ethics have been moving in new directions. With respect to the first issue, one can observe a trend towards articulating and defending theories that non-superficially preserve more of our pre-theoretical moral judgments and commitments than traditional utilitarian or Kantian theories do.1 With respect to the second issue, there has been a movement away from ignoring virtue (a feature of much twentieth century moral philosophy) towards recognizing the importance of virtue, and of attempts to understand it, not merely amongst virtue ethicists, but amongst philosophers who do not accept theories that attempt to make virtue foundational to ethics.2 How are these two trends related? I will consider one important way in which the ideal of moving normative ethical theory in the direction of squaring with more of our moral intuitions (or knowledge) and the ideal of providing a positive, but non-foundational account of the virtues might seem to conflict. I will focus below on a now fairly standard consequentialist account of virtue delineated and defended by Julia Driver in her Uneasy Virtue (2001). Driver certainly takes virtue seriously, but I believe her theory of virtue conflicts too acutely with certain of our commonsense commitments concerning virtue, for it implies that, in certain circumstances, greed, apathy, cowardice, etc., can be virtues, and honesty, kindness, courage, etc., vices. The consequentialist who adopts this way of understanding virtue is moving in an opposite direction from the trend of reconciling consequentialism and our 1 A number of consequentialists now contend that there are agent-relative goods, and a number now support rule-consequentialism, instead of act-consequentialism, due, at least in part, to the fact that they take it that such forms of consequentialism (or teleology) are closer to commonsense morality than traditional utilitarianism is, in certain important respects. For a useful, though skeptical discussion of the first sub-trend see Schroeder (2007b). With respect to the second sub-trend, important examples include Hooker (2000) and Parfit (2011). 2 Examples of recent texts on virtue written by philosophers who are not virtue ethicists include Hurka (2001), Driver (2001), and Adams (2006). Of course, there are other trends than the ones I am focusing on. Virtue ethics, as an alternative account of right action, has suffered a large number of criticisms, but is still being defended by some, e.g. Swanton (2005) and Kawall (2009). Of course, some people who reject virtue ethics as an account of right action still call themselves virtue ethicists; like Anscombe (1958), they are often motivated by skepticism about the viability of the project of providing general principles of right action (Annas 2011 is a good example, although the source of her skepticism is not the same as Anscombe’s). Another important trend, which I do not have space to discuss, involves skepticism about the very existence of virtues, and possible responses to such skepticism (see Harman 1999, Doris 2002, and Part III of Adams 2006).



VIRTUE

commonsense moral judgments. I will argue for an alternative approach to understanding virtue that fits better with our pre-theoretical commitments, whilst still leaving room for consequentialism, as a theory of right action, to be correct. I will begin by discussing two evil demon cases involving virtue that are interesting to reflect on because doing so leads to the conclusion that we are right to avoid connecting acting virtuously with the right or the good too closely, and also because they bring out what is overly counterintuitive about the standard consequentialist way of thinking about virtue. The alternative conception of the virtuous that I will then sketch is one that purposely loosens the connection between acting virtuously, on the one hand, and right action or the bringing about of good outcomes, on the other hand. If, as I think, it is a mistake to tie virtue too closely to the right, it is certainly an easy mistake to make, since, when we reflect on our pretheoretical commitments concerning the virtuous, it does appear to us that they are generally attempting to act rightly. This commitment is better explained by the view that what is essential to being virtuous is that virtuous people do their best to respond appropriately to normative reasons. This is a view that leaves room for us to think that sometimes acting virtuously will involve acting wrongly (non-culpably), even when the relevant agents possess all relevant non-ethical information. I will begin by arguing for the perhaps surprising view that acting virtuously can come apart from both subjective and objective rightness. I will then make it clear why I think the standard consequentialist account of virtue is unsatisfactory,3 before moving on to develop my own account of virtue, partly out of materials provided by Daniel Elstein and Tom Hurka. I end the chapter by considering an objection to my positive account of virtue which focuses on its most radical aspect—that the practical virtues are best construed as being epistemic in nature.

3 Although I focus on Julia Driver, I also have so-called “global consequentialism” in mind. Of course, there are alternative ways of understanding virtue on a consequentialist theory that are not my target. In particular, Hurka (2001) provides an interesting account of virtue that construes them as having intrinsic value, rather than merely instrumental value. Although our accounts differ as to how best to analyze virtue, his account of value may be compatible with my account of virtue, as long as one does not follow him in calling the attitudes he takes to be intrinsically valuable “virtues.”

VIRTUE



2 Virtue-responsive Evil Demons Suppose you are an ordinary virtuous agent. Up until now, you haven’t lived the kind of life that involves making any huge life or death decisions, but you are kind to your friends and acquaintances, charitable to distant strangers in need, honest to everyone except those you think you can’t trust, etc. Now imagine that an evil demon comes to you and tells you that, for the rest of your life, he will kill five strangers every time you act in accordance with virtue, but won’t kill any people any time you act contrary to virtue.4 The demon makes it clear that he is not telling you that he will kill people when you act rightly (he expresses no opinion on the relation of his threat to such a possible threat), but rather only that he will kill them when you act as a virtuous person might in your circumstances (if he or she were acting in character).5 He makes it impossible for you to commit suicide—if this were an option, you might think you should immediately take it (although doing so might itself lead to the death of five people, the killing might stop there). On the assumption that the evil demon will keep his word, what are you to do?6 Given that I have specified that you start off virtuous, what might we predict you, as a virtuous person, will do in this situation? This will, in

4 An interesting variation on this thought experiment would be one where the demon threatens to harm your friends and loved ones whenever you act in accordance with virtue. I think the main conclusions I draw will also hold for this variation on the thought experiment; that is to say, not much hangs on whether the relevant outcomes are agentrelatively bad or merely agent-neutrally bad. 5 To act in accordance with virtue is to act in one of the ways a virtuous person might act if they were acting in character in your circumstances, and to act contrary to virtue is to act in one of the ways no virtuous person would act if they were acting in character in your circumstances. I have specified that you are virtuous, so why don’t I just specify that the demon simply focuses on the option of you acting virtuously? I wish to block off the option of you purposely acting on a bad motive (i.e. not acting virtuously, in a key sense), but still acting in accordance with virtue (still doing something the virtuous might do). 6 One answer I won’t be considering below is the following. Some deontologists might be tempted to say: it is up to the demon whether he kills people; by trying to demand that I act differently in response to his threat he is on a par with terrorists, and the correct morality or conception of virtue will tell me I need not listen to his demand at all. I believe this is simply not a credible moral view. It should not be confused with the weaker deontological view that I may properly refuse to follow an order to torture one innocent by an evil agent who, it is clear, will otherwise torture two innocents (“not through me” I might respond to such a threat, as Bernard Williams famously noted). For all I will say, this more moderate view might be correct.



VIRTUE

part, depend on considering what a virtuous person might reasonably predict a demon to do in response to various possible acts. Here is one thing we should not do when considering your options. We should not begin by assuming that acting in accordance with virtue is extensionally equivalent to acting rightly.7 I’m not saying we should simply assume that this is false; rather, we should begin by being open minded about this issue, and consider the demon’s threat from an initially neutral perspective. I will next demonstrate that if we start by being open minded about this question, we will end up concluding that there is a coherent strategy that the virtuous person can pursue in the imagined scenario, and it is a strategy that has the upshot of indicating that acting in accordance with virtue is not extensionally equivalent to acting rightly. This approach need not beg any questions against the virtue ethicist who maintains that acting in accordance with virtue is extensionally equivalent to acting rightly, since we are beginning from a neutral position. A refusal to allow us to begin in this way would beg the question, at least absent the consideration of persuasive arguments for thinking that the extensional equivalence claim holds. Since I think this claim is generally held to simply as an assumption, rather than as the result of any arguments, the present strategy seems an attractive one to pursue. In the scenario described above, I would contend that a virtuous person is likely to reason roughly as follows: “Anything I do will lead to the killing of at least five people. However, suppose I work to destroy my virtues by developing vices, and act on these vices as much as I can; then I will certainly minimize killings! Let me go to the television and start the process of developing apathy and sloth.” In the situation we are imagining, it is very attractive to think that a virtuous person would and should develop vices, and act on them. In particular, they should plan not to be virtuous in the future. Unless we accept a strong unity of the virtues 7 For this reason, I will not complicate matters by considering what it might be correct to say about our thought experiment if the extensional equivalence claim were correct. Three possible options that one might consider in that case are: (i) the demon’s threat makes no difference at all to what it is rational to choose to do, because the demon’s threat is simply incoherent; (ii) the demon’s threat makes no difference at all to what it is rational to choose to do, because it cannot rationally alter a ranking of available options (if every possible option is taken to include killings then this will effectively make no difference to one’s reasoning); and (iii) the thought experiment provides us with an example of a genuine paradox.

VIRTUE



thesis (I don’t believe we should), this won’t mean that they need destroy all of their virtues, but they will need to replace enough of them with vices to enable them to act viciously in a consistent fashion in the future, in the world as they might reasonably expect it to be. They will do their best to ensure that their future acts not conform with virtue. The acts that they carry out on the way to losing their virtues and developing vices will, unfortunately, themselves count as acts that are in conformity with virtue (given the peculiar context), hence will lead to the demon killing people, but at least they will be able to significantly minimize deaths, in the long term, by developing vices.8 If the virtuous person were to take the alternative view, that they should protect their own virtue even at the expense of many people being killed by another, they would, I think, strike us as being extremely self-indulgent, in a morally blameworthy fashion. So it seems that an at one time virtuous person could significantly minimize very bad outcomes by simply developing sloth, greed, etc., and then acting on the vices (watching television most of the time, failing to aid people, etc.). Perhaps sloth will not be enough, because it will be important to be acting in a way that a virtuous person would never act, and perhaps the presence of the vice of sloth is consistent with being a generally virtuous person; if so, then it will be necessary to develop further vices, up until the point that the previously virtuous person clearly isn’t virtuous any more. Still, very plausibly, the necessary degree of viciousness required will be nowhere near the degree of viciousness that would lead to acts that are on a par with the continual acts of killing that the demon will otherwise engage in.9 8 I am assuming throughout that the vices we are considering are not evil enough to lead the vicious agent to herself engage in killings (hence my examples of apathy and sloth). A variation on our thought experiment would be one where the demon simply offers you a pill which, if you ingest it, will instantly replace your virtues with non-extreme vices. 9 In what follows I will speak of the necessary strategy as being one of becoming vicious, as an alternative to being virtuous. By “vicious” I mean the opposite of virtuous (this follows a tradition in ethics, rather than our ordinary usage of the word). One might worry that I am ignoring an important middle ground here, since one might hope to fail to count as virtuous whilst still not counting (even weakly) as vicious. I grant that there is such a middle ground when we think in general terms about the virtuous and the vicious; however, in the present context, that middle ground threatens to shrink to a point where it becomes vanishingly small. The problem is that the agent who is ruining his or her virtue in order to prevent the demon from killing more people in the future needs to ensure that she will not, in the future, do any acts that a virtuous person might do in his or her



VIRTUE

We might have a different worry about the proposed strategy: perhaps sloth would be a virtue in a situation where one only develops sloth, and acts on it, in order to prevent killings. However, I don’t think this thought is right. If it were right, our thought experiment would have serious consequences for the way we understand virtue. It would have what we might call a strongly consequentialist consequence. Moderate consequentialists about virtue, like Julia Driver, think that a character state can be a virtue in one social world and a vice in another, since (on Driver’s view) a virtue is simply a character state that generally, systematically brings about good consequences in one’s environment in the actual world, while a vice is a character state that generally brings about bad consequences in the actual world.10 On this view, we are meant to consider the long term and widespread effects of particular character states, across individuals, and not just more limited effects. By contrast, we might say that a strong consequentialist about virtue would contend that a virtue can quickly become a vice for an individual (rather than across society in general) if it can be expected to generally bring about bad consequences from now on in the actual environment of that individual, and a vice might quickly become a virtue for an individual if it can be expected to generally bring about good consequences from now on in that individual’s environment. This view seems quite implausible. Assuming, as I think we should, that it is incorrect, no problem of this kind is generated in the case where one develops non-extreme vices in order to prevent the demon killing people. I believe that if we fix our minds on the person who becomes vicious, through cultivating laziness and apathy, etc., in order to significantly minimize the number of killings the demon will commit, we will judge that his subsequent actions (when he is lounging on his couch watching television all day, etc.) are both subjectively and objectively right,

circumstances. However, it is hard to imagine any acts in this future context that are such that no virtuous person would do them in order to prevent a large number of killings, even though they are non-vicious acts. Surely the virtuous person would be willing to commit pretty much any generally non-vicious acts in this special context. She might even be prepared to commit some vicious ones, which is why the imagined agent may need to be quite vicious indeed, in order to avoid conforming with virtue. 10

Driver (2001, xvii). Driver is influenced by Hume’s account of virtue, although she also distances herself from that account somewhat.

VIRTUE



although they do not conform with virtue.11 He does not act on any virtues, but he has not forgotten the demon’s threat, so he rationally judges that his vicious actions are right. And a knowledgeable impartial observer would judge that they are right, from an objective perspective, whilst also agreeing that they are vicious acts. This is an important upshot of our consideration of this first evil demon case, not least because it demonstrates the falsity of the type of virtue ethics that identifies right action with action that conforms with virtue.12 There is one small complication that I have been ignoring, but it won’t mean that we need to alter the above conclusion. I began this section by indicating that an act conforms with virtue when it is either virtuous or such that a virtuous person would do it if they were acting in character.13 Perhaps a virtuous person confronted by our demon might simply decide to stop acting in character? It is a little unclear what is involved in not acting in character, but let’s suppose that the virtuous person has the option of simply taking heavy doses of barbiturates, or other drugs that will significantly slow her down—we might think that this will ensure that she won’t act in character, and that, therefore, the demon will not kill people (apart from the people he kills when she first starts taking the drugs, since the first act of taking the drugs will conform with virtue). The use of drugs, or some very similar means, seems necessary here, because it doesn’t seem like a virtuous person could otherwise form very effective intentions to act out of character, in a way that wouldn’t 11 I am using the terms “subjectively right” and “objectively right” to refer to what might reasonably be considered to be right given the subject’s epistemic limitations, on the one hand, and what might reasonably be considered to be right, putting those limitations to one side, on the other hand (so the expected-value consequentialist focuses on one of these in giving an account of rightness, and the actual-value consequentialist focuses on the other). Subjective rightness does not simply correspond to whatever a person happens to think is right. I remain neutral here on whether there are actually two senses or concepts of “right,” or merely one sense or concept that corresponds to just one of “subjectively right” and “objectively right.” 12 That is, the type of virtue ethics defended in Rosalind Hursthouse (1999). Robert Johnson (2003) also uses thought experiments to attempt to demonstrate the falsity of this type of virtue ethics, but his differ from mine in focusing on people who are not virtuous from the get go, but are trying to become virtuous, whereas I am focusing on someone who begins by being virtuous, rightly becomes vicious, and rightly has no intention of becoming virtuous again (so long as the demon and his threat remain in the picture). 13 This ensures that my target is Hursthouse’s claim (which I am arguing is false) that “An action is right iff it is what a virtuous person would characteristically (i.e. acting in character) do in the circumstances” (1999, 28).



VIRTUE

require constant boosting by acts that would themselves count as conforming with virtue (if she had to remind herself to act out of character several times a day then people would be killed several times a day, for instance). It’s actually fine for my purposes if this alternative drug-taking solution would work, because here we are also imagining an agent acting in ways that are subjectively and objectively right (when the drugs ensure she spends all day in bed, etc.), but that don’t conform with virtue. However, this solution doesn’t actually seem very satisfactory. Either the drugs will run out at some point in the short term, in which case this solution looks quite a bit worse than the solution that involved cultivating vices (since more people will be killed when the virtuous agent is acting in character again), or we will want to say that the agent’s dispositions are changing over time in such a way that she is losing her virtues anyway, as a result of taking drugs for a long time, in which case this solution just becomes a variant on the solution that involved cultivating vices. In any case, worries about the “acting in character” qualification do not appear to prevent us from reaching the conclusion that rightness comes apart from acting in conformity with virtue. The virtue ethicist might be tempted to resist the idea that the person who has replaced their virtues with vices, in the extreme scenario we have been considering, will count as vicious. He might be tempted to say that this is an extraordinary case where one can count as virtuous when one is lacking in virtues and not lacking in vices. The virtue ethicist I am considering would then also say that replacing one’s virtues with vices will, in fact, provide no way to avoid the demon’s threat, for the demon would continue to kill people as this virtuous person without virtues acts in a way that would normally be considered vicious (for his or her acts would then count as acts that conform with virtue), but the virtue ethicist might be quite happy to bite this bullet. He might also then claim that there would be no need, after all, for a virtuous person confronted by our demon to start to destroy her virtues, since doing so wouldn’t do any good (since she’d still end up counting as virtuous, but without virtues, and her acts would therefore conform with virtue), hence that a decision to keep acting virtuously in the scenario under consideration wouldn’t be self-indulgent after all. And the virtue ethicist might then conclude that I have not managed to provide an example of a situation where acting in accordance with virtue comes apart from rightness.

VIRTUE



There is just one problem: this imagined response is actually completely implausible in its content, for it attempts to sever the connection between a person being virtuous at some time (not merely acting in conformity with virtue, note) and a person possessing virtues at that time. It seems, quite clearly, to be a necessary condition for being virtuous at some time that one possess virtues at that time. I very much doubt that the virtue ethicist would, on reflection, really wish to give up this condition; in fact, virtue ethicists typically want to make claims that are stronger than this one by saying something substantial about which virtues it is essential to possess in order to count as virtuous, and by claiming it is in virtue of possessing virtues (and certainly not in virtue of once upon a time having possessed virtues) that a person counts as virtuous. And the virtue ethicist is surely right to want to say such things. What might we say here about the thinner concept of the morally worthy act (a concept that is of particular interest to Kantians, but might also reasonably be thought to be of interest to the rest of us)? Very plausibly (and one need not be a Kantian to say this), not every morally worthy act is a virtuous act. A stingy man might perform a genuinely charitable act, and in so doing act out of character. On a common reading of the Groundwork, Kant would be likely to view such an act as particularly morally worthy, even though he also ultimately thinks that developing virtues will more reliably lead one to act in a morally worthy fashion (as he makes clear in The Metaphysics of Morals). If we focus on where we have ended up in the above discussion of our thought experiment, with our once virtuous, now purposely vicious agent (and with the evil demon still hanging around), we find that we now have a very interesting choice of two things to say about morally worthy acts. First, although it might be clear that an act can be morally worthy without being virtuous (i.e. without springing from firm, virtuous dispositions), it is not at all clear that an act can be morally worthy without conforming with virtue (i.e. without being an act that a virtuous person might perform in the same circumstances). Suppose all morally worthy acts conform with virtue; in that case, our thought experiment interestingly reveals that morally worthy acts sometimes must come apart from subjectively right actions, since the person who has become vicious in the scenario under consideration would seem to still be able to recognize he is doing the subjectively right thing when he is acting in a fashion that is



VIRTUE

not morally worthy. What’s more, he will recognize that it would be subjectively wrong of him to act in a morally worthy fashion. Of course, we might have already thought that morally worthy acts and subjectively right acts come apart to some extent, since there seem to be ordinary circumstances in which people are able to recognize they are acting rightly at the same time as acting on a bad motive, but the conclusion here is stronger: there may be situations where it is impossible to act in a morally worthy fashion and a subjectively right fashion at the same time. Suppose, on the other hand, that morally worthy acts can come apart from acts that conform with virtue. I won’t try to decide here whether this or the prior view is correct, although this second, alternative view appears unmotivated. In any case, if this second view is correct, something just as interesting as if it is incorrect will follow. It will follow that the vicious agent in the scenario we are imagining can very easily act in a morally worthy fashion by simply following his greedy, lazy motives, etc., just as he is now inclined to do.14 In fact, if this view is correct, the only way for him to stop acting in a morally worthy way, once he has become vicious, would be for him to start taking the very difficult path back to being virtuous—however, this would be incredibly evil in the situation we are imagining (it would be to purposely aim, in a way that would involve a lot of effort, to get back to a situation where the demon will resume killing many people). We would hope that nobody who started off virtuous would have overshot the mark to this extent, by making themselves as evil as this (since their only motive for becoming vicious to some extent was to avoid the demon’s killing)!15 In any case, the basic

14 Of course, those with more sympathy for Aristotle than Kant, when it comes to the question of what constitutes a good will, will already have thought this is a problem with Kant’s view of morally worthy acts: it might be true that a miser who overcomes his greed to be charitable is acting in a more morally worthy way than a person who is born naturally sympathetic and gives away money or other resources due to a natural propensity to do so, but it might also be true that a man who was once a miser but has worked very hard to become virtuous and now gives effortlessly performs charitable acts that are even more morally worthy than the charitable acts of the miser who gives with difficulty (I take it that this last thought is more Aristotelian than Kantian). 15 In any case, this evil would soon undermine itself, for as soon as the evil intention started disappearing—and this would happen as the agent started to approach a state of being virtuous again—the desire to become virtuous would disappear. We can imagine the evil agent might be able to avoid this problem if the demon now offered a pill that would instantly make this agent virtuous again. A further minor point: one might think that any acts the agent takes in order to become virtuous again will themselves conform with virtue,

VIRTUE



lesson to take away from this (if this second view is correct) is the strange, certainly unKantian, conclusion that it is possible to be in a situation where it is extremely easy to reliably act in a morally worthy fashion and would be extremely difficult to avoid reliably acting in a morally worthy fashion. Let me now briefly mention a different evil demon case, reflection on which speaks against even the moderate consequentialist account of virtue (since it does this, it follows that it also speaks against the strong consequentialist account of virtue). Imagine that, unbeknownst to any of us, a demon simply decides to kill five people every time anyone acts from a character state that we would normally think of as a virtue—he is not going to allow us to discover that he is doing this, let alone announce his intention to us.16 It might also be specified that this demon has already been doing this for a long time. There is no risk of this case appearing paradoxical or influencing any ordinary agent’s reasoning, since everyone is ignorant of the demon’s intentions. On a variant of the case, one might add that the demon produces very good outcomes whenever one acts from a character state that we would normally think of as a vice (but, again, we are never aware of this fact). Suppose, following the arrival of this demon, you act kindly or honestly, and that you do an act that would, outside of the imagined scenario, be considered to flow, quite uncontroversially, from the possession of a virtue. This act will lead to the demon killing people. On even the moderate consequentialist account of virtue (as well as the strong consequentialist account), this act will now demonstrate vices, given that kind and honest acts flowing from stable character states will more generally also lead to the demon killing people. On that account, this second demon has, in fact, made it impossible for you to act virtuously, and has transformed all of your virtues into vices. And on the variant of the so the demon will start killing people again straight away, even before the agent becomes virtuous again (so much the better, the evil agent might think). In fact, it appears that only some of these actions will conform with virtue, since there are many acts that non-virtuous agents might perform on the way to becoming virtuous that virtuous agents would never need to do, as Johnson (2003) points out. 16

John Skorupski (2004, 14) discusses a thought experiment that is very similar to this one, and has his demon produce the bad effects on another planet that we don’t know anything about, and Driver discusses a similar but less extreme example (2001, 79–81). Skorupski reaches a similar conclusion to mine. Driver is willing to admit that cases like this one will lead us to judge that her theory is counterintuitive in certain important respects.



VIRTUE

case where the demon also responds perversely to apparent vices by doing good, it has become impossible for you to knowingly act viciously, as all your apparent vices are really virtues. This is, to say the least, highly counterintuitive.

3 Reasons, the Virtuous, and Modesty These virtue-responsive evil demon cases have a number of important features. First, we have a firm judgment in the first case described that the virtuous person is morally required to give up her virtues for vices, given that this will minimize killings—to refuse to do so would be extremely self-indulgent and morally objectionable. Second, we determined that the best strategy to pursue in that case is one that demonstrates that acting in accord with virtue and acting rightly (whether subjectively or objectively speaking) are not extensionally equivalent. Third, the second evil demon case demonstrates that even the moderate consequentialist account of virtue has extremely counterintuitive implications. On Driver’s account of the virtues, honesty would not be a virtue at all in a very bad world where one would need to be dishonest most of the time to do good. Such worlds seem fairly easily conceivable. If we take a truly virtuous agent from a good society, and place him in a very bad one (or imagine his intrinsic duplicate in a bad society in another possible world), I contend, contra Driver, that he does not thereby immediately lose his virtues. This seems true for at least two reasons. First, the reader will recall from the discussion of Williams on Hare in Section 5 of Chapter 1 that many of our ordinary ethical commitments seem to be commitments to non-contingent truths of a limited kind; one might naturally take it that such truths include that if a person does an act that springs from kindness, her act is good, at least in one crucial respect (for being an act of kindness), and so on, for each of the character states we ordinarily take to be virtues. (The general claim here about the pro tanto goodness of acts that flow from virtues also happens to follow from the type of analysis of individual virtues that I provide below.) Second, our ordinary conception of virtue is tightly connected to judgments about praiseworthiness, and non-culpability. Very plausibly, if one is a virtuous agent (so one is capable of engaging in practical reasoning), and acting on one’s virtues has generally bad consequences,

VIRTUE



then this will not be through any fault of one’s own.17 Consider again the two evil demon cases discussed above. In one case, one knows about the demon and his dastardly intentions, and, in the other case, one is ignorant of the fact that such a demon is hovering nearby. Notice now that the very same conclusion regarding the relevant agent’s blamelessness seems to follow, regardless of whether or not the agent knows about the evil demon, although what the agent would choose to do will vary. If the agent is ignorant of the evil demon’s responses to his own “virtuous” actions (i.e. actions that spring from dispositions that would normally be considered virtuous) then he clearly should not be held accountable for the terrible behavior of the demon. On the other hand, if the agent knows about the evil demon and the way in which he will respond to the agent’s own behavior, then it would seem that the one thing the virtuous agent would not do is continue acting as before, i.e. on the basis of virtuous dispositions, as long as ceasing to so act is a possibility for the agent.18 She will replace her virtues with vices, and, so long as she does this, she is blameless. Driver might respond that all of these considerations rely too much on simple appeals to our intuitions, but I suspect most of us will find it very hard to shake the relevant judgments. Consider the stark contrast between the following two ways of describing what happens when a virtuous person encounters a virtue-responsive evil demon in the second of the two ways discussed above (i.e. without knowing it): (1) he suddenly becomes vicious and begins doing wrong through no fault of his own; and (2) he suddenly begins doing wrong through no fault of his own, while remaining virtuous. The second is surely a much more plausible description of what happens, and this means we should also reject the moderate consequentialist account of virtue.

17 It might be objected that the first evil demon case is a case where, if a virtuous person were to remain virtuous (rather than immediately start down the path of corrupting her own virtue), then she would be blameworthy. To this one can respond that a virtuous person would not protect her own virtue at the expense of people’s lives, so while this counterfactual might be true, it will remain the case that it will never be her own fault that bad consequences will result from her acting on her virtues (assuming she acts in character). 18 If, contrary to what I have suggested, it is not possible for the virtuous agent to stop acting on virtuous dispositions in full knowledge of the evil demon’s intentions then it still looks like she should not be blamed for her behavior, because she did her best to become the best kind of moral agent one could hope to be, and objectively achieved this end, yet, because of very bad fortune, thereby ended up in a position that made it impossible to do anything but behave in a way that brings about terrible outcomes.



VIRTUE

On the view I wish to defend, it is true to say that if a virtuous person acts in character then he or she cannot be blamed for any objective wrongness that attaches to his or her acts.19 It is possible that a (minimally) virtuous agent will sometimes act wrongly, and not merely due to ignorance of relevant non-ethical facts, and this is one way in which virtue and right action come apart. Imagine a case of a virtuous person thinking long and hard about what to do in a difficult moral situation, being somewhat ignorant of (some of) the relevant fundamental ethical reasons that feature in fundamental ethical principles, weighing derivative reasons, that, so far as she can tell, are of roughly equal weight, then acting on a reason that, it turns out, was actually outweighed by a stronger consideration. Practical deliberation that is guided by virtue seems, by its nature, to take it for granted that there may be an act that, more than any other act, one really ought to do in any situation. When deciding to act in a particular way in a situation where it is difficult to see what it is right to do, we expect the ordinary virtuous agent to himself grant that he may have got things wrong. We might name the virtue that is at play here “epistemic humility about the ethical.” This seems to be a particularly important type of epistemic humility. When agents are in such circumstances, a correct ethical theory might be very helpful, but, as I argued in Chapter 1, I think we should say that such conscientious agents may be virtuous even though they do not have access to such a theory. Still, it might be claimed that the use of the above thought experiments is not sufficient to undermine the moderate consequentialist account of virtue. After all, they are somewhat outlandish, and I haven’t marshaled an independent set of arguments against the thesis, but simply appealed to some very common judgments. I believe that such a response would fail to appreciate the relevant dialectical position. As I read her, Julia Driver provides us with two relevant things: (1) a particular account of 19 In other words, so long as one acts virtuously, one acts blamelessly (of course, it does not follow that this is the only way of acting blamelessly). Actually, earlier reflections on the first evil demon case suggest that even this one-way entailment claim is not quite right. The person who is lounging around, having cultivated sloth, in order to minimize the number of people killed by the evil demon, is acting rightly, blamelessly, and non-virtuously; suppose he were to now purposely choose to work at becoming virtuous again—we would then judge him to subsequently be acting in a virtuous but blameworthy fashion. In light of this case, the more precise claim I wish to defend is that so long as one acts virtuously then one acts blamelessly, unless one becomes virtuous by purposely failing to do what one correctly judges to be right.

VIRTUE



what an ethical virtue is; and (2) an argument for this account that concerns so-called “virtues of ignorance.” With respect to (1) alone, she is not at a dialectical advantage, if it is possible to provide a plausible alternative account of virtue that doesn’t conflict so sharply without pretheoretical intuitions (or even harmoniously conforms with them) and that is compatible with a consequentialist account of right action. I will begin the task of doing this in the next section of the chapter. With respect to (2), Driver is only at a dialectical advantage if she is right that there are virtues of ignorance, and that they speak in favor of a moderate consequentialist account of virtue. I will now argue that she is wrong about the main virtue that she focuses on when attempting to show that there are such virtues. If the advantages of (1) and (2) drop away, I think Driver is in a weak position, since we have seen her theory has counterintuitive consequences, so we will be left with little motivation to accept her theory of virtue, given that the alternative I sketch below is compatible with a consequentialist theory of right action. Other consequentialists, especially so called “global consequentialists,” might appeal to the theoretical virtues of simplicity and elegance that come with being able to apply consequentialism across the board not just to actions, but also character states (as well as anything else that seems even prima facie capable of being assessed in terms of consequences, e.g. epistemic norms). However, appeals to simplicity and elegance are probably only ever really persuasive when competing theories are roughly equally good in other respects. In any case, such an appeal could do little in itself to counter the commonsense judgments I have appealed to here. Driver bases her argument for a moderate consequentialist account of virtue on a claim that certain character traits constitute “virtues of ignorance” that essentially involve epistemic defects (and a claim that only a consequentialist account of virtue can explain why traits that involve epistemic defects can be virtues; I’ll just take this claim for granted). She focuses her analysis on the virtue of modesty. According to her, “for a person to be modest, she must be ignorant with regard to her self-worth. She must think herself less deserving, or less worthy than she actually is. . . . it would [thus] seem that this virtue rests upon an epistemic defect” (2001, 19). I think Driver is here overlooking the importance of the fact that not all examples of ignorance involve an epistemic defect. My ignorance of key features of fifteenth century British political history indicates no epistemic defect, unless I am supposed to be an



VIRTUE

expert on fifteenth century British political history, or have been ignoring other strong reasons to learn about this subject. Even when the subject is myself, ignorance of various facts about myself does not necessarily indicate any epistemic defect, and this can be true, I would suggest, even when that ignorance is willfully maintained (at least in certain respects). I would suggest that a better analysis of modesty would take modesty to involve a refusal to ascertain or pay attention to one’s own apparently good qualities. A modest person does not collect evidence concerning her good qualities, but rather believes that such evidencecollecting would be wrong (ceteris paribus). Suppose the modest person stumbles across evidence that she is excellent at some activity—e.g. she is shown a newspaper that she generally takes to be reliable, bearing a large headline announcing that she is the best pianist in the world. Does this now make it impossible for her to be modest? Perhaps not, as long as she is quick to change the subject with anyone who brings it up. Driver considers whether a person aware of her “self-worth” who avoids discussing it can be considered modest, and says “No. Modesty is something that is internal; it is basically an attitude of ignorance that one has to oneself ” (2001, 19). There does seem to be something right about the thought that modesty is internal (at least in the sense that one can be modest even if no external actions would reveal this fact), but I don’t think it follows from this that a modest person must be ignorant of her good qualities, since one internal activity she might engage in is a refusal to think about the fact that she has these qualities (if she has inadvertently discovered them); another is a refusal to infer from the possession of certain good qualities that she is better than other people in general. A reason to think this last thought is correct is that we do, in practice, often wish to classify as immodest those people who infer from the fact that they are better than others at one thing, that they are better in general.20 Driver thinks that alternative accounts of modesty, like the one I am proposing, must classify as modest certain acts that we instead take, in practice, to indicate “false modesty.” I admit that knowing about one’s

20 After writing this response to Driver on modesty, I became aware of Bommarito (2013). Bommarito provides a much more detailed account of modesty as a “virtue of attention” (his term), where virtues of attention are virtues specifically concerned with directing or controlling our powers of attention.

VIRTUE



good properties (even if one has attempted to avoid discovering them) makes it harder to act modestly, but I don’t think it makes it impossible to act modestly (in relation to those very properties). We would call a person who lies about being the best pianist (“no, really I’m not . . . ”) falsely modest, perhaps because a vice independent of immodesty is involved (e.g. dishonesty), but what about a person who cleverly changes the subject, rather than answer a question on the topic (this is also what someone who doesn’t know they are best but also doesn’t want to investigate the question would do)? It is worth emphasizing that Driver claims that it is a necessary condition of modesty that one be ignorant of one’s own self-worth (2001, 19). I think it is more accurate to say that immodesty typically involves a focus on less general qualities of oneself, and, to the extent that it does involve global judgments of worth, immodesty also involves comparative judgments about one being a better person than other people (when people make such judgments they are usually immodest in virtue of the fact that it doesn’t follow from being better in one respect, e.g. being a better pianist, that one is a better person; or, in some cases at least, simply in virtue of the fact that they lack epistemic humility).

4 A Reduction Plan Is it possible to provide informative, non-circular analyses of individual practical virtues in terms of something prima facie distinct from virtue? As we have seen, it has been claimed that virtues are character states that systematically bring about the realization of good outcomes in the world. And it has also been claimed that they are character states that, in various ways, promote the flourishing of the individuals that possess them (Hursthouse 1999, for instance, defends such a view). The aim here is to defend a new, quite different kind of reductive analysis, one that takes off from Daniel Elstein and Thomas Hurka’s attempt to analyze individual virtues in terms of a non-ethical element and a “thin” ethical element, viz goodness. Partly through thinking about the perspective of the virtuous, and partly through the examination of individual virtues, I will argue that although Elstein and Hurka are on the right track, a better recipe for analyzing virtue might be provided by focusing primarily on normative reasons, instead of on goodness. This is an option that they appear amenable to.



VIRTUE

In their excellent paper, “From Thick to Thin: Two Reduction Plans,” Elstein and Hurka (2009) attempt to provide disentangling, reductive analyses of thick virtue concepts in terms of separable non-ethical elements and thin ethical elements.21 For example, they suggest that courage might best be analyzed in the following way: “ ‘Act x is courageous’ can be analyzed as something like ‘x is good, and x involves an agent’s accepting harm for himself for the sake of goods greater than the evil of that harm, where this property makes any act that has it good,’ and where . . . the second ‘good’ is an embedded evaluation” (527).22 They contend that, following this model, it may be possible to analyze all (thick) virtue concepts in terms of a core descriptive (that is, non-ethical) component, and a thin evaluative component. Elstein and Hurka work up to the complex form of analysis that they favor for understanding (most) practical virtues by starting with two less complex forms of analysis that seem to work for some thick concepts. The first form of analysis, which I’ll call “Type I,” they describe as a “descriptively determinate two-part form” (518). An analysis of a concept that takes this form will be committed to viewing the extension of the relevant thick concept as wholly fixed by certain descriptive properties that are provided upfront, and the evaluation that is added as idle in determining the extension of the concept. To use their example, the pejorative term “Kraut” may be thought to have an identical extension to “German,” with the only difference between the two terms being that a further negative evaluation is part of the meaning of the first term, but not the second. Type I x falls under the extension of concept T iff “x has descriptive properties A, B, and C (for specific A, B, and C), and is good/bad/right/wrong for doing so” (518, emphasis added) Elstein and Hurka do not think this form of analysis can work for virtue concepts or terms. About this they agree with the main targets of their 21 Elstein and Hurka move fairly freely between talk of virtues, virtue terms and virtue concepts. I will not attempt to regiment things too much on this score, but I take it that ultimately we are most interested in what virtues are, rather than semantics or conceptual analysis (at least that is what I am interested in here). 22 All page numbers mentioned without a date of publication should be read as references to Elstein and Hurka (2009).

VIRTUE



paper, proponents of the so called “disentanglement argument.” The disentanglement argument presupposes that the reductivist will wish to claim that descriptive properties alone determine the extension of virtue concepts or terms. This argument has as one of its premises that claim that if this were so, then “one could know which actions the term would be applied to . . . not merely without oneself sharing the community’s admiration . . . but without even embarking on an attempt to make sense of their admiration” (McDowell [quoted in Elstein and Hurka 2009, 518–19]). The other main premise of the argument is that this appears to be impossible when we consider thick virtue concepts. Elstein and Hurka contend that there are two plausible forms of reductive analysis that the disentanglement argument is unable to get any purchase on. The first of these, which I will call “Type II,” might be described as a descriptively only partly determinate two-part form of analysis. Type II x falls under the extension of concept T iff “x is good, and there are properties X, Y, and Z (not specified) of general [descriptive] type A (specified), such that x has X, Y, and Z, and X, Y, and Z make anything that has them good” (521, emphasis added) Elstein and Hurka explain how this type of analysis works by providing a plausible analysis of distributive justice that takes this form, and contrasting it with the way in which distributive justice might implausibly be analyzed using the Type I form. According to an analysis of Type II, “x is distributively just” is (roughly) equivalent in meaning to “x is good, and there are properties X, Y, and Z (not specified) that distributions have as distributions, or in virtue of their distributive shape, such that x has X, Y, and Z, and X, Y, and Z make any distribution that has them good” (522). This analysis leaves open exactly what type of distributive pattern would count as just, providing room for both a desert theorist and an egalitarian (say) to share the same concept of justice but disagree on its extension (although they would both agree that entities that have no distributive profile fall outside the extension of “just”). Any putative analysis of distributive justice of Type I, on the other hand, would have the unfortunate consequence of making it impossible for the desert theorist and egalitarian to share the same concept of justice



VIRTUE

(521)—it would be as if one theorist about “Kraut” (the term as we know it) were to claim it refers to people of one nation, while another were to claim it refers to people of some other nation; no more than one of these two could count as a competent user of the term.23 Importantly, it is clear that proponents of the Type II analysis of distributive justice provided can agree with proponents of the disentanglement argument, such as McDowell, that “to know the intended extension of ‘distributively just,’ as used by some person or community, it is not enough to know the descriptive part of that term’s meaning; we must also know what evaluations they use it to make, that is, which properties they take to make distributions good” (522). Finally, we come to the Type III form of analysis, which might be described as a descriptively only partly determinate three-part form. Elstein and Hurka do not actually provide a schema for this form of analysis, so I will attempt to provide one myself that is based on their discussion. They offer two main examples. I have already provided their analysis of courage, and will summarize it again below. With respect to integrity, an act x is said to be an act of integrity just in case “ ‘x is good, and x involves an agent’s sticking to a significantly good goal despite distractions and temptations, where this property makes any act that has

23 One might suspect that this is too quick. When describing the way thick concepts work on a Type II analysis, Elstein and Hurka recognize that they need to rely on a distinction between the intended extension of a concept and the actual extension of a concept (522, especially fn. 12). A proponent of a Type I analysis is also free to use this distinction. Suppose we were provided with a Type I analysis of the concept of justice. Would it really be true that if this were a correct analysis of the concept of justice, the egalitarian and the desert-theorist could not possibly share this concept? Perhaps it would be sufficient to maintain that the intended extensions of the concept come apart, even though the actual extension does not. Of course, at least one of these theorists would be going more seriously wrong in their own understanding of the concept in question than if a Type II analysis were correct, but it is a live, interesting question in the philosophy of concepts whether it is possible to count as possessing a concept even when one goes very seriously wrong in reflecting on it, or, more generally, in judgments that utilize it (Williamson 2007 argues that a correct account of concepts must allow for the possibility of such errors). Here we may be misled by the focus on “Kraut” as an example of Type I, as this term may correspond to an unusually transparent concept. All this being said, the Type I form of analysis simply fails to reflect the thought behind the Type II form of analysis that it is part of our understanding of certain concepts that we cannot settle disputes using them by conceptual analysis, but must rather ask which features of the world we ought to value (or recognize as reasons), to paraphrase personal correspondence from Daniel Elstein.

VIRTUE



it good,’ and where the second ‘good’ indicates an embedded evaluation” (526). Here is what I take to be the general form of these examples: Type III (acts) Act x falls under the extension of concept T iff x is good, and x is the act of an agent with intentional property P of type A, where any competent use of T will involve an attribution of a property of type A to an agent, where an attribution of a property of type A consists of a descriptive component D (specified) and a thin evaluation regarding particular goods or evils (not specified), and where the possession of P by an agent makes any act that directly results from it good.24 Here is how this formula applies in the case of courage. Consider courageous acts. As Elstein and Hurka understand any such act, the act is good, and it “involves an agent’s accepting harm for himself for the sake of goods greater than the evil of that harm” (527). I take it that this means that the agent who is acting has a general property of accepting harm for himself in the way they describe. Any competent ascription of this property is partly specified in advance by describing the type of property we are concerned with (what we can know in advance, prior to any consideration of what the goods and evils in question are), and partly not specified, with respect to the content of the embedded evaluation involved in such ascriptions (of particular goods and evils). Notice that in both the analysis of integrity and the analysis of courage, the phrasing in question is “involves an agent’s . . . ,” followed by a very general description of what the agent is doing that is bound up with their intentional attitudes (“sticking to . . . despite . . . ” and “accepting harm . . . for the sake of . . . ”). Hence a variable for an intentional property in my formulation of Type III. Finally, the formula 24 It might appear highly objectionable to mention judgments involving the use of a concept on the right-hand side of a biconditional that is supposedly specifying the extension of this very concept. In fact, this is not objectionable, since it is straightforward to restate the claims here without any apparent circularity, as follows: (a) concept T is such that any competent use of T will involve an attribution of a property of type A to an agent, where an attribution of a property of type A consists of a descriptive component D (specified) and a thin evaluation regarding particular goods or evils (not specified); and (b) act x falls under the extension of concept T iff x is good, and x is the act of an agent with intentional property P of type A, where the possession of P by an agent makes any act that directly results from it good (mutatis mutandis for all the similar forms of analysis that follow).



VIRTUE

tells us that it is the possession of this property that makes acts that flow from it good. The reader might think this formulation of the Type III schema is overly complex. Why is it necessary to provide a place for competent attributions? It is necessary to refer to such attributions, in addition to a property attributed, in order to locate the embedded evaluation in the right place. It would be a mistake to instead claim that the property P involves the relevant evaluation. It is a very interesting question whether agents who act virtuously must themselves view their options “under the guise of the good”; whether acting courageously, say, involves viewing the harms in question as evils and the goods in question as goods. But this is not what is at issue here. What is it issue here, according to Elstein and Hurka, is that in order to be able to directly ascribe courageousness to any act (whether one’s own, or someone else’s) it is necessary to judge that the particular goal being aimed at is valuable enough to justify the particular harm being accepted. As they highlight, there can be much disagreement about such matters; we might all agree that someone who persists in maintaining a trivial hobby despite facing political dangers does not thereby demonstrate integrity, but we might find ourselves in disagreement if we wonder whether a religious leader who holds firm to his church’s ban on the ordination of women thereby demonstrates integrity (526). Unlike with thick concepts that are thought to be of Type I or Type II, the competent, direct use of concepts of Type III necessarily involves embedded evaluations. Crucially, these evaluations are thin rather than thick. Elstein and Hurka make it clear that they think that, when it comes to analyzing virtues, this is really where the action is: “ . . . this three-part analysis applies very widely, since many and even all virtues and vices involve a relation to some independently given moral consideration . . . ” (527, emphasis added). Despite the fact that Elstein and Hurka here take themselves to be analyzing virtues (as they indicate in the sentence just quoted), it is an interesting feature of their paper that they actually focus on providing analyses of virtuous actions (or patterns of the distribution of goods in the case of justice). This seems a little odd, but does not constitute a substantial problem for the approach they end up favoring, since it is fairly straightforward to see how one might derive a corresponding claim about the virtues of agents.

VIRTUE



Type III (agents) Agent s falls under the extension of concept T iff s possesses a stable disposition to do acts that are good and result from her possession of intentional property P of type A, where any competent use of T will involve an attribution of a property of type A to an agent, where an attribution of a property of type A consists of a descriptive component D (specified) and a thin evaluation regarding particular goods or evils (not specified), and where the possession of P by an agent makes any act that directly results from it good. It is worth noting that these formulations of Type III (acts) and Type III (agents) are neutral in relation to the interesting question of the relative priority of occurrent states and stable dispositions when it comes to understanding virtue (this is not mentioned in Elstein and Hurka’s paper). Hurka (2006) points out that ordinary moral thinking allows that an act may genuinely count as generous, given the right motive, without the agent in question possessing a corresponding stable disposition (e.g. one might act kindly without being a kind person). This will not be possible on theories of virtue that take stable dispositions to be prior to occurrent states; such theories hold that it is a requirement on virtuous actions that they must directly result from the possession of stable dispositions. A proponent of this kind of theory is free to specify that the descriptive component D in the above formulae will build in mention of the requisite background dispositions. Any act x that demonstrates integrity might then be considered to be an act such that x is good, and x involves an agent’s exercising a stable disposition to stick to a significantly good goal despite distractions and temptations, where this property makes any act that has it good, and where the second “good” indicates an embedded evaluation.

5 The Reasons of Virtue and the Primacy of the Normative An alternative type of reductive analysis to the one that Elstein and Hurka focus on providing, a type of analysis that I favor, starts with the following thought: perhaps individual virtue concepts are best analyzed by reference to normative reasons of various types, where



VIRTUE

the type is partly provided by the non-normative component of the relevant thick concept. In fact, although Elstein and Hurka focus on attempting to explain virtues evaluatively, they indicate that they are aware that an alternative, perhaps viable, view would be to focus on rightness (I take it they mean pro tanto rightness), and that there is an historical precedent for doing so in W.D. Ross’s account of virtue. They also sometimes mention reasons themselves in passing when describing particular virtues.25 Their primary goal in their paper is not, in fact, to establish the truth of the evaluative type of analyses that they provide of virtues, but rather to “to illustrate the general resources the reductive view [in general] has” (531), with respect to countering the claim made by Bernard Williams and John McDowell that thick virtue concepts cannot be analyzed by reference to thin ethical concepts and descriptive claims without falling afoul of the disentanglement argument. Elstein and Hurka acknowledge that the disentanglement argument works when faced with analyses of virtues that have the Type I form, but they think friends of the argument have overlooked the fact that there are ways of analyzing virtues, such as those they propose, that do not fall afoul of that argument, as they do not attempt to detach ethical judgments from the factors that determine the extension of virtue terms. So, their aim is not so much to establish that the example analyses they provide are correct, but rather to demonstrate that they are highly plausible. I, on the other hand, am not so interested in responding to the disentanglement argument, but am instead interested in trying to ascertain whether an analysis roughly of the kind they are proposing might actually be correct.

25 Ross’s account is complex because (it is said that) he recognizes that the desire to do what is right is one of the forms that virtue can take, and that desires to bring into being things that are good and to produce pleasure in others are also forms that virtue can take (530). Elstein and Hurka also recognize the option of focusing on right and wrong when discussing a reductive analysis of “lewd” owed to Allan Gibbard (525), and they themselves analyze the vice of selfishness by reference to wrongness (522–3). In addition, they also initially rely on reference to reasons when analyzing courage. In fact, in the discussion of courage they appear to carefully hedge their claims: “In many though not all cases these reasons concern the goods that can be achieved by risking the harm. . . . So for these cases ‘act x is courageous’ can be analyzed as something like ‘x is good, and . . . ’ ” (527, emphasis added). I am ignoring such hedging, since I am focusing on comparing particular concrete analyses.

VIRTUE



I will soon provide some reasons to think that it is better to analyze the virtues in terms of normative reasons, than in terms of goods and evils, but first: how precisely should one analyze virtue concepts if one does wish to shift to a focus on normative reasons? I will here focus on the Type III form of analysis, and not discuss Type II.26 Type IIIR is Type III modified in just the way we need. I provide both a formulation for acts, and a formulation for agents. Type IIIR (acts) Act x falls under the extension of concept T iff x is good, and x is the act of an agent with intentional property P of type A, where any competent use of T will involve an attribution of a property of type A to an agent, where an attribution of a property of type A consists of a descriptive component D (specified) and a judgment regarding particular normative reasons (not specified), and where the possession of P by an agent makes any act that directly results from it good. Type IIIR (agents) Agent s falls under the extension of concept T iff s possesses a stable disposition to do acts that are good and result from her possession of intentional property P of type A, where any competent use of T will involve an attribution of a property of type A to an agent, where an attribution of a property of type A consists of a descriptive component D (specified) and a judgment regarding particular normative reasons (not specified), and where the possession of P by an agent makes any act that directly results from it good.

26 Since Elstein and Hurka think the Type II analysis might have something going for it in relation to certain virtues, such as distributive justice, let me suggest the following alternative form of analysis. As applied to distributive justice, it says that, whenever one is able to realize a distribution of goods that has the right distributive profile, the fact that this act will realize such a distribution is a (strong) normative reason to do so.

Type IIR x falls under the extension of concept T iff there are properties X, Y, and Z (not specified) of general [descriptive] type A (specified), such that x has X, Y, and Z, and, on any occasion when it is possible to act so as to realize X, Y, and Z together this fact (that one’s act will realize X, Y, and Z together) always provides a (strong) normative reason to do so.



VIRTUE

Here is an attractive picture of the way in which virtuous agents deliberate which I believe should lead us away from focusing on goods and evils to focusing on reasons when attempting to understand virtue. Virtuous agents find themselves faced with various reasons for action, and, when considering different options, they weigh reasons against each other. In saying that they weigh reasons for various options, I do not mean to say virtuous agents always need to engage in some process that takes an extended period of time; often they may simply respond on the basis of an immediate judgment concerning the reasons that apply to them. When considering more difficult cases, such weighing will be more thoughtful and take more time. On the basis of considering reasons for action, virtuous agents reach judgments concerning what they ought to do on specific occasions, and their actions are in line with these judgments (perhaps they sometimes act without forming a judgment as to what they ought to do, but one might take it that they are then still normally in a position to form such a judgment, even if only retrospectively). Why think that the virtuous focus primarily on reasons in their deliberations, rather than on goods and evils? First, there will be many facts that the virtuous will wish to count as reasons even though these facts appear to have no direct connection to the good. For example, a virtuous person will take the fact that they promised to do something as a reason to do it (within certain limits, since the immorality of the content of a putative promise might defeat its reason-giving force), independent of any good that might follow from keeping the promise. Even if fulfilling a promise could always be expected to produce some good, or even if it is the case that the institution of promise keeping is itself best justified in a consequentialist fashion, the strength of a reason to fulfill a promise will generally not straightforwardly correspond to the degree of goodness that lies in fulfilling it.27 I return to this general point about the limits of focusing on goodness when considering particular virtues below. Second, an account of virtue that focuses on goods and evils is less ecumenical in

27 Daniel Elstein has reminded me that it is possible for a consequentialist to claim that promise keeping acts are non-instrumentally good. I do not actually think that this is the best option for the consequentialist, given the availability of very plausible instrumental justifications in this area, but my main concern here is with the way things appear to the ordinary virtuous agent, and it does not seem to me correct to say that non-philosophical virtuous agents are, in general, dedicated to viewing promise keeping in evaluative terms.

VIRTUE



relation to normative ethical theories than one that focuses on reasons; the first fits more comfortably with consequentialism than deontology, whereas the second is neutral. Given that we may be more sure of many of our claims about (derivative) reasons than we are about fundamental ethical principles, it is a good thing that the preferred account is neutral in this way. Third, even if we restrict ourselves to cases where one might be more tempted to think a virtuous agent could act well just by focusing directly on goods and evils, we need to bear in mind that one of the key ways (perhaps the key way) we distinguish between evaluative and deontic matters is by recognizing that it is only the second that involves a constraint concerning what we are able to do. If a stray asteroid is about to destroy Boston, it might be a very good thing indeed if I stopped it from doing so; however, given my complete lack of superpowers, there is no reason for me to do so.28 Even if all reasons coincide with actual goods and evils (contrary to what I believe), it is the facts in question qua reasons that virtuous agents consider most directly in their deliberations. If the above picture of the way agents deliberate is correct, then we have located an important reason for preferring the Type IIIR form of analysis to the form of analysis proposed by Elstein and Hurka (I will provide some further reasons when discussing particular virtues below). A Type III approach attributes to agents a consideration of goods and evils, while a Type IIIR approach attributes to agents a consideration of normative reasons. Now, as I earlier indicated, Type III formulations do not specify that agents are viewing their options under the guise of the good (even though they are considering things that are goods and evils), and similarly Type IIIR formulations do not specify that

28 Actually, if reasons as evidence is correct, “reason implies can” is a more complicated principle than this suggests, although “ought implies can” need be no more complicated than on other theories. If I know that I cannot ϕ, then I have no reason to ϕ, as evidence that I cannot ϕ will undermine any evidence that I ought to ϕ (assuming ought implies can is a background principle). If I don’t know that I cannot ϕ, but have some evidence to that effect, then the strength of the reason to ϕ (evidence that I ought to ϕ) will weaken in proportion to the strength of the evidence that I cannot ϕ. I could say much more about complexities in this area (what I just said already goes beyond what Kearns and I said about this issue in Kearns and Star 2009, 235–6; I now think it is more important to defend a version of the “reason implies can” principle than I did at the time), but the crucial point here is that there can still be close connections between abilities and reasons on the reasons as evidence account.



VIRTUE

agents are viewing their options under the guise of reasons (even though they are considering facts that are reasons). This means that it does not strictly follow from the truth of the above reasons-focused description of the way the virtuous deliberate that Type IIIR formulations are more accurate than Type III formulations.29 Nonetheless, I do think the truth of that description does makes it much more likely that Type IIIR formulations will be more accurate than Type III formulations. In any case, it is best to build in the general claim that virtuous deliberation involves responding to reasons by further modifying our favored forms of analysis as follows. Type IIIR* (acts) Act x falls under the extension of concept T iff x is good, and x is the act of an agent with intentional property P of type A, where any competent use of T will involve an attribution of a property of type A to an agent, where an attribution of a property of type A consists of a descriptive component D (specified) and a judgment regarding particular normative reasons that the agent is understood to be responding to (not specified), and where the possession of P by an agent makes any act that directly results from it good. Type IIIR* (agents) Agent s falls under the extension of concept T iff s possesses a stable disposition to do acts that are good and result from her possession of intentional property P of type A, where any competent use of T will involve an attribution of a property of type A to an agent, where an attribution of a property of type A consists of a descriptive component D (specified) and a judgment regarding particular normative reasons that the agent is understood to be responding to (not specified), and where the possession of P by an agent makes any act that directly results from it good.

29 One might have a view that ties together reasons and goods, so that wherever there is a reason there is a good (or an evil). This could be a buck-passing account of goodness, or a goodness-first account of reasons. I think both of these are problematic (see Way 2013 for a relevant discussion of both). In any case, reasons as evidence is inconsistent with the first, and probably also inconsistent with the second, at least if an account of the second type extends beyond fundamental reasons, or focuses on actual value rather than expected value.

VIRTUE



6 Reasons as Evidence, the Virtuous, and the Primacy of the Epistemic In Chapter 2, I defended reasons as evidence, which provides the following analysis of reasons. R: Necessarily, a fact F is a reason for an agent A to ϕ if and only if F is evidence that A ought to ϕ.

As I have said, the reasons as evidence account of reasons provides a unified and informative analysis of reasons for action and reasons for belief: in both cases, one is referring to evidence concerning a putative truth about what one ought to do or believe. This account has come under attack from a number of critics, and, with Stephen Kearns, I have already published responses to some of these critics (see Kearns and Star 2011, 2013a, 2013b). The thesis has been standing up well to criticism. According to R, a fact is a normative reason for action just in case it is evidence that I ought to do a particular act. We can now consider what happens to the claims about the deliberations of the virtuous provided above when we bring this analysis of reasons into the picture.30 The result is a more informative account of what is involved in being virtuous: the basis of being virtuous consists in sensitively attending to evidence concerning what one ought and ought not do, and acting in line with the conclusions one arrives at concerning what one ought to do on the basis of considering such evidence.31 Appropriately responding to such evidence will sometimes involve spontaneous, seemingly non-intellectual reactions, and sometimes, in more difficult cases, involve the active intellectual weighing of evidence. An analogy with ordinary perception suggests itself here: normally, we spontaneously form beliefs about our physical environment on the basis of evidence received through perception, but sometimes we need to stop and think about what we should believe, given such evidence. Note that this is intended to be an account

30 I have focused mostly on action throughout, but we might think it is crucial to some virtues (e.g. kindness) that certain emotions (e.g. of sympathy) are a necessary part of their manifestation. The reasons as evidence account of reasons can very naturally be extended to include reasons/evidence that one ought to feel certain emotions. 31 I mean for “concerning” to be broad enough to include evidence that the agent ought to ϕ, evidence that it is not the case that she ought to ϕ, evidence that she ought not ϕ, and evidence that it is not the case that she ought not ϕ.



VIRTUE

of what it takes to be virtuous, and not an account of what it takes to be maximally virtuous. Virtue clearly admits of degrees, and it is likely to be true that all maximally virtuous agents need to have access to the correct fundamental ethical principles, even though it is a crucially important aspect of the proffered account of what it takes to be virtuous that virtuous agents in general need no such access.32 How might we go about understanding particular virtues, now that reasons as evidence and the evidence-centered conception of the virtuous are on the table? I am now in a position to propose that particular virtues are stable dispositions to respond well to particular classes of pieces of evidence concerning what one ought to do.33 In line with this contention, we might modify Type IIIR* to refer to evidence concerning what we ought to do. This results in the following two formulae. Type IIIE (acts) Act x falls under the extension of concept T iff x is good, and x is the act of an agent with intentional property P of type A, where any competent use of T will involve an attribution of a property of type A to an agent, where an attribution of a property of type A consists of a descriptive component D (specified) and a judgment regarding specific evidence concerning what the agent ought to do that the agent is understood to be responding to (not specified), and where the possession of P by any agent makes any act that directly results from it good. Type IIIE (agents) Agent s falls under the extension of concept T iff s possesses a stable disposition to do acts that are good and result from her possession of intentional property P of type A, where any competent use of T will involve an attribution of a property of type A to an agent, where an attribution of a property of type A will consist of a descriptive component D (specified) and a judgment regarding specific evidence concerning what the agent ought to do 32 There is no reason to think that the account of the virtuous just sketched is incompatible with consequentialism (as a theory of right action). The consequentialist need not lose anything by giving up on the idea that virtue itself needs to be analyzed in a consequentialist fashion. By giving up on that ambition she can hope to find more common ground with her opponents. 33 And vices are stable dispositions that lead one to fail to appropriately attend or respond to evidence concerning what one ought to do.

VIRTUE



that the agent is understood to be responding to (not specified), and where the possession of P by an agent makes any act that directly results from it good. We have arrived at a view that takes the practical virtues to be essentially epistemic in nature. One important qualification seems necessary at this point, and that is that a strong will is required if a person is to count as a virtuous agent. Although epistemic akrasia is a genuine phenomenon, it is a very different phenomenon than practical akrasia. It is all too easy not to do what one judges one ought to do, whereas it is generally difficult not to believe what one judges one ought to believe (think of the knowledge we are constantly gathering through ordinary perception). Apart from the need for a strong or resolute will, we might say that, on the view we have arrived at, the practical virtues share the aim of the intellectual virtues (which we have not attempted to analyze); that is, they share the aim of having one stand in an appropriate relation to truth. Plausibly, the ideal relation to stand in to truth is one of knowledge, and following evidence is the ideal when knowledge is absent. The practical virtues aim to have one stand in an appropriate relation to ethical truths, and we might sometimes be able to further distinguish between moral virtues that aim to have one stand in the appropriate relation to moral truths, and prudential virtues that aim to have one stand in an appropriate relation to prudential truths. In hereby affirming a thesis we might call the primacy of the epistemic, so called to suggest a contrast with a common idea that goes by the name of “the primacy of the practical” (or “the primacy of pure practical reason” in Kant’s Critique of Practical Reason), I take it that virtuous agents are aiming to act rightly (with kindness, courage, modesty, honest, etc.) in a way that essentially involves appropriately hitting on ethical truths, as well as doing the acts and feeling the emotions that, insofar as we possess reason (the capacity defined in Chapter 2), concrete facts demand of us.34

34 It is important that we not hypostatize or fetishize the truth (as some might worry we would if we were to talk of pursuing the True). The truth that one ought to help a nearby person because she is suffering (in a case where there are no outweighing reasons, to keep things simple), is not something practically separable from his or her actual suffering. Even though (as we will see in Chapter 4), I think fundamental ethical principles of right (and wrong) action are best understood as more than just statements that universally quantify over instances of right (or wrong) acts, I also hold to the widely shared view that they are necessary



VIRTUE

7 Comparing Analyses of Particular Virtues We arrived at the present way of analyzing virtues by first focusing on the idea that what is essential to virtue is that the virtuous respond to reasons well, then combining that ordinary thought with reasons as evidence. The current proposal for analyzing individual virtues follows on naturally from what came before, and so might be supported by appeal to the preceding arguments. Still, we can also ask: how does the present approach compare to Hurka’s approach of focusing on the good when analyzing particular virtues? Which approach seems to provide the best form of analysis? I won’t be able to establish for certain that my favored approach is clearly superior for understanding particular virtues, but I will now provide a few reasons to think it is preferable. First, consider Elstein and Hurka’s analysis of courage: “x is an act of courage” is equivalent to “x is good, and x involves an agent’s accepting harm or the risk of harm for himself for the sake of goods greater than the evil of that harm, where this property makes any act that has it good” (527). Consider two soldiers fighting in a war, on opposite sides. Suppose they are very roughly similar individuals, and are such that, without knowing much about the war in which they are fighting, we would be inclined to judge them both courageous. Now suppose that one of them is fighting on the unjust side and the victory he is aiming for and reasonably takes to be good would not, in fact, be good. It seems wrong to claim that this soldier is not, after all, courageous. If we focus instead on the soldier’s reasons, and understand evidence concerning what one ought to do as constituting reasons, then we can make sense of the thought that he is indeed courageous (assuming he is not forming his beliefs in an epistemically irresponsible fashion). Elstein and Hurka recognize that there is an “intensional” alternative to the “causal” view of virtue that they adopt, an option previously adopted by Ross and Rashdall (530). They claim that our everyday understanding of virtue does not “mandate either one of these views in preference to the other” (530), but here I beg to differ. The discussion of the radical consequences of Driver’s view of virtue when considering the second of the two evil demon scenarios described above was meant, in

truths, and it is this second feature that ensures that there is no odious way in which one might pull apart contingent oughts from the contingent facts that they supervene on.

VIRTUE



part, to bring out how it is part of our everyday understanding of virtue that what matters for virtue is what might be rationally expected to result from our actions (given our evidence), rather than what actually results from them.35 Second, consider Elstein and Hurka’s analysis of integrity: “x is an act of integrity” is equivalent to “x is good, and x involves an agent’s sticking to a significantly good goal despite distractions and temptations, where this property makes any act that has it good” (526). I think this is a somewhat odd way of understanding integrity. It is more common to instead reach for talk of principles when attempting to understand integrity. An act shows integrity if it involves sticking to a principle one accepts in the face of distractions and temptations, one might say. This contrasts with the focus on good goals in the above analysis. How might we bring reasons into the picture here? In this case, it is attractive to alter the analysis to say that what is crucial to an act of integrity is that it involves sticking to a commitment to an ethical rule or principle in line with evidence that the agent is responding to that he or she ought to do so.36 Third, consider Elstein and Hurka’s analysis of distributive justice: “x is distributively just” is (roughly) equivalent to “x is good, and there are properties X, Y, and Z (not specified) that distributions have as distributions, or in virtue of their distributive shape, such that x has X, Y, and Z, and X, Y, and Z make any distribution that has them good” (522). Here I think it is worth pointing out that the view that what is crucial to understanding distributive justice has to do with working out which distributions are good (and which bad) is actually a somewhat controversial view.37 Deontologists, in particular, may wish to resist thinking of 35 Of course, a variation on the Elstein and Hurka analysis of courage might focus on expected value, instead of actual value. I would then point to a second virtue of my account of courage, which is that it can more easily handle cases where agents demonstrate courage at least partly by following deontic principles that they are deeply committed to, rather than by focusing on goods and evils. 36 It is not the fact that a principle is one that I accept that would make an act of sticking to it an act of integrity. Rather it is that I reasonably take the principle to be correct or highly reliable, hence stick by it. Of course, the principle one holds fast to might theoretically itself take a consequentialist form, but I think it is more likely that the principles the virtuous in general are committed to are deontic principles that take a ceteris paribus form. 37 Here is an example of something that matters here. Consider a distribution of goods across a society that is an egalitarian distribution, but is also such that it would make everyone very badly off. The fact that the distribution is egalitarian might well be thought to provide a (fundamental) reason to favor bringing it about, but then it is compelling to think



VIRTUE

distributive justice in these terms—for a just distribution might well be thought to concern goods without itself being a good—however, they are likely to be more than happy to view the fact that a distribution has a certain shape as a reason to bring it about. A focus on reasons, rather than goods, when analyzing distributive justice, promises to be more general, and more ecumenical. The consequentialist can still insist that certain distributions qua distributions are good (as this may do important theoretical work), without requiring the virtue of justice to be such that the just person need view just distributions as good. But it seems close to unintelligible to suppose that the just could view certain distributions as just without taking it that they have a reason to bring about such distributions when they are able to.

8 A Problem and a Solution It is a general fact about evidence that it can be misleading, without any blame for this fact attaching to the agent who possesses the evidence in question. I might, for instance, be blameless for having misleading evidence that it is going to rain tomorrow. I might know that a generally reliable website recently reported that it is going to rain tomorrow, and, unfortunately, have no idea that the team of people in charge of updating the website are suffering from food poisoning, have not been doing their job properly for a few days, and that it isn’t going to rain tomorrow. The following concern thus naturally arises in relation to my account of virtue, since it is ultimately centered on evidence concerning what one ought to do: what are we to say about a person who is epistemically conscientious when it comes to gathering evidence concerning what they ought to do, across all relevant domains, yet, due to factors beyond their control, ends up with misleading evidence concerning what they ought to do? We might imagine that this evidence is so misleading that the person we are considering ends up intentionally acting in all sorts of objectively very bad ways. We should not restrict ourselves to cases where merely

that there is a much stronger reason to favor not bringing it about (it would make everyone very badly off). Way (2013) seems right to say that such a distribution, even if just, is not good to favor, because it is not good in any respect (either non-instrumentally, or in terms of what might be brought about by favoring it).

VIRTUE



non-ethical evidence is misleading, since the problem here needs to be distinguished from the demand that expected-value consequentialists face of needing to admit that when people are doing what they ought to do, they will often do things that are very bad due to faultless ignorance of relevant non-ethical facts (e.g. a patient has a yet undiscovered disease, with the consequence that a penicillin injection will kill him). This pill may not be too bitter to swallow, but the problem I’m concerned to address here is more worrying, because we are imagining that the reasons/evidence-that-oneought in question may be ethically misleading, due to no epistemic fault on the part of the agent we are considering. Perhaps this person intentionally, and not just accidentally, kills many innocent people, at least partly on the basis of misleading evidence, responsibly acquired, that favors a false moral view. If we are analyzing the virtues using the Type IIIR*/IIIE forms of analysis we will need to claim that many acts that fits this description are good acts, but this may seem quite strange.38 And it may likewise appear quite odd to claim that a person could generally do acts of this kind (by responding to evidence that has been collected well, yet is radically misleading), and still get to count as an ethically virtuous person. Two mutually exclusive responses are possible here. One is to bite the bullet and do one’s best to make it seem plausible that such a person is, despite some of our judgments to the contrary, an ethically virtuous person. The second, which I suspect will appear more attractive to most readers, is to supplement the account of virtue so far discussed in a way that is both natural (rather than ad hoc) and will prevent agents who end up intentionally doing many bad things due to the presence of misleading evidence from counting as virtuous.39

38 One option is to sever the link between virtuous action and good action, by dropping those parts of Type IIIR* and IIIE that specify that the actions in question are good. Pekka Väyrynen (2013, 40–3) contends that this aspect of the Elstein and Hurka approach to analyzing thick concepts is problematic, while the element of embedded evaluation is not problematic. His arguments for severing the tie between global evaluations of goodness and thick concepts (which go beyond the pages referred to here) are complex, and there is no space to discuss them here. In any case, we can move from the question about what it takes for a virtue attribution to be correct to the question of what it takes for someone to count as virtuous, and I think many people would find it difficult to accept the thought that an agent who systematically acts badly (objectively speaking) might still count as virtuous. 39 I do not mean that the agents imagined intentionally end up doing bad things under the description “bad things” (they may not realize what they are doing is bad), but that they



VIRTUE

First, it is not as obvious as it might appear at first blush that the person who is epistemically conscientious and strong-willed, who has done a wholly rational job of gathering, processing, and acting on evidence concerning what they ought to do, has acted wrongly when they end up doing bad things. There is a temptation, at least, not to morally blame them for the bad acts they carry out, assuming we really hold in place the idea that they are epistemically blameless for the position that they are in. And this gets some purchase when we think about moral luck. If we also imagine someone in epistemically much less demanding circumstances who isn’t anywhere near as epistemically conscientious, but ends up coming to the right, epistemically warranted moral conclusions, we are inclined to think that if such a person gets to count as virtuous, surely our unlucky but much more epistemically conscientious agent also gets to count as virtuous. Although I think this idea has some things going for it— for instance, it may provide a way to resolve the moral luck problem—and although I think that virtuous people sometimes act wrongly through no fault of their own due to moral ignorance (in very difficult cases where failure to do the right thing is a result of not knowing the principles of the correct ethical theory),40 I hesitate to cut the connection between virtue and the good (as loose as I have suggested it is) in the direction this line of thinking about very bad cases of misleading evidence would take us.41 In any case, there is an alternative. We might ask of the agent we began by imagining: what is it he lacks that we think virtuous agents must possess? I take it that a very plausible answer to this question is ethical knowledge. Virtuous agents must know that pain is bad, that killing is (normally) wrong, that lying is (normally) wrong, that helping people in need is right, etc. We need not imagine that the virtuous have very complex, theoretical knowledge about more fundamental ethical principles. intentionally do things that are in fact bad, such as intentionally killing innocent people (and not in order to save however many lives one might think would justify these killings). 40 The general view that moral ignorance, like non-moral ignorance, can exculpate is defended by Zimmerman (1997) and Rosen (2003). Harman (2011) and Weatherson (2013) argue (wrongly, I think) that moral ignorance should be treated differently than non-moral ignorance in this respect. 41 This is one reason I am reluctant to endorse Michael Zimmerman’s (2008) prospective view, according to which what one ought to do depends on the evidence one possesses, where this is not limited to evidence regarding non-ethical matters, but includes evidence regarding ethical matters. It should become clearer in the next chapter why it is that I do not think that what one ought to do supervenes on the evidence that one possesses.

VIRTUE



We need only assume that the virtuous are influenced and guided in responding to evidence by some fairly simple, basic ethical knowledge. This does not seem like an ad hoc addition to the account of virtue so far sketched, since I think this is something many of us would already be inclined to think about the virtuous. Furthermore, I think it may be the case that ethical knowledge is intimately connected to the presence and strength of the reasons, or evidence concerning what one ought to do, that the virtuous encounter. Of course, this might not seem to be the case on a purely subjective Bayesian account of evidence (in relation to evidence that one ought in general), but we may actually have good reasons to reject such an account of evidence, in favor of one that is centered on knowledge (Williamson 2000). If pre-theoretically accessible ethical knowledge is intimately connected to the presence and strength of the reasons/evidence-that-one-ought that the virtuous encounter, as I will suggest it is in the next chapter, the Type IIIR* and Type IIIE formulae defended above may provide a suitable recipe for analyzing the practical virtues in general.

4 Knowing Better 1 Introduction Reasons are typically prior to oughts in at least one way: they are normally upstream of oughts so far as practical deliberation is concerned. When one is not sure what one ought to do (which is true much of the time), one weighs reasons, and, on that basis, one comes to ascertain, or at least get as close as one can to determining, what it is one ought to do.1 An appreciation of this truism might lead a person to think that reasons must be metaphysically prior to oughts, but it does not follow from reasons being prior to oughts so far as reasoning is concerned that they are metaphysically more fundamental than oughts. Reasons as evidence is an account of reasons that is particularly well placed to explain how we might sometimes come to know what we ought to do on the basis of reasoning with, or weighing, reasons, and this is despite the fact that it is an account of reasons according to which they are not metaphysically basic. On this account of reasons, as we have seen, a fact (or true proposition) is a reason just in case, and in virtue of the fact that it is evidence that one ought to do a particular act (Kearns and Star 2008, 2009).2 It is fairly uncontroversial that, at least with respect to much human knowledge, we come to know true propositions by considering evidence

1 Kearns and I (2008, 2015) argue that this truism about reasons is a serious problem for any account of reasons, such as John Broome’s, that claim that reasons are (parts of ) explanations. 2 In our coauthored papers, Kearns and I only argued for the truth of the biconditional connecting reasons, on the one side, and evidence and ought, on the other, and not for the stronger identity claim I am endorsing here, although we did suggest that one might argue from the biconditional to the stronger claim using inference to the best explanation.

KNOWING BETTER



that makes those propositions more or less likely to be true.3 In conjunction with the claim that reasons are evidence that one ought to act in certain ways, this common, independent observation about much of our knowledge—that it is arrived at through the consideration of evidence— entails that there is a simple to mention (if far from simple to further elucidate) route to gaining particular bits of knowledge about what one ought to do: one can come to know what one ought to do on particular occasions by considering (often, by weighing) reasons for action. Of course, it is not always possible to come to know what one ought to do by considering practical reasons, but it very plausibly sometimes is. This is important, because here we have one good reason for thinking that it is no mystery that we are sometimes able to know what we ought to do while being completely ignorant of any theory that would provide fundamental explanations or principles concerning what we ought to do—if there are correct ethical principles waiting to be discovered, one might have thought one can only come to know what one ought to do by first knowing these very general principles. Reasons as evidence is in an extremely good position to explain both (a) how it is that there can be fundamental ethical principles that specify what our fundamental normative reasons for action are; and (b) how it is possible to act virtuously by responding directly to (derivative) reasons for action, and generally thereby be able to come to know what it is one ought to do, all the while being completely ignorant of the fundamental ethical principles that we are only able to discover by engaging in philosophical normative ethics.4 A critic might at this point accept that, in principle, the favored picture of reasons uncovers a route to coming to know what one ought to do

3

Although what it is for a proposition to be more or less likely to be true is far from uncontroversial. I take it that we are talking about epistemic probability here. In limiting my claim to much human knowledge, I am trying to make this a claim that evidentialists and non-evidentialists will all accept. 4 I began to explore this view in Star 2011. How do I square the view that reasons are not metaphysically basic, but are explained by evidence and ought with the view that normative ethical theory is in the business of specifying our fundamental reasons for action? I sketched an answer to this question in Chapter 1 that hinged on the claim that fundamental reasons are facts that both explain what one ought to do and also evidence concerning what one ought to do, but it might appear to be a problem for my account that this leaves open the possibility that fundamental explanations are just a separate class of metaphysically basic reasons, since it might be thought that this is what other philosophers have meant by “reason.” Perhaps we are talking past each other? I address this concern in Section 5 below.



KNOWING BETTER

when ignorant of ethical theory, but might insist that this route will still only seem like a plausible one, so far as actual (realistic) practical deliberation is concerned, if it is possible to provide a plausible story about how it is that the ordinary practical deliberator ever really has the evidence needed for knowledge acquisition concerning what he or she ought to do. Of course, a normative error theorist might get off the boat straight away just because they think there are no true ought propositions waiting to be known, but this is not a reaction I wish to address here.5 Rather, I have in mind the critic who does not find the idea of the weighing of evidence leading to knowledge problematic in general, but, rather, finds talk of weighing evidence that one ought to ç rather mysterious. Two questions naturally arise at this point: What makes a fact evidence that one ought to ϕ? And what determines the strength of the evidence that one ought to ϕ (which, on the favored account of reasons, is the strength of a reason to ϕ)? These are two different questions,6 but I will propose an answer to the first question that also happens to answer the second. It utilizes a well known, but far from uncontroversial account of evidence, and that account of evidence is not one that I have the space to separately defend. The point is to provide an attractive way of answering these questions that takes them seriously. The account of evidence I have in mind is Timothy Williamson’s: one’s evidence consists in known true propositions. Or, more precisely, on a slight variant of Williamson’s thesis that I favor: evidence consists in propositions that agents are in a position to know, while the evidence that an agent possesses consists of known propositions. This knowledge as evidence account of evidence is often paired with another attractive idea, and that is that knowledge is the norm of action: that a reasonable person aims to act on the basis of knowledge, and not on the basis of weaker mental states, such as belief. We will see that this

5

I provide a response to the error theorist in Star (2010), although my response there is not of the kind that attempts to demonstrate that error theory is incorrect. 6 As are the two parallel questions about reasons (the answers to which are equivalent to answers about evidence-that-one-ought so far as I am concerned, but not according to others): what makes a fact a reason? and what determines the strength of a reason? Mark Schroeder’s (2007a) attempt to provide a very different answer to the first of these questions about reasons than to the second is an example of how, on alternative approaches, they may be treated quite differently.

KNOWING BETTER



account of the relation between our epistemic states and action has been criticized for being underdeveloped and difficult to assess, since examples that have been used to either support or criticize it are not decisive. The solution I will provide for answering the two questions mentioned above also, importantly, provides a basis for shoring up and filling out the claim that knowledge is the norm of action. In what follows, I will be referring to certain core claims. Here they are: RAE: p is a reason for A to ç iff p is evidence that A ought to ϕ. (Also, p is a reason/evidence only if p.) RAE*: p is a reason that A has to ç iff p is evidence that A has that A ought to ϕ. RAEB*: p is a reason that A has to believe q iff p is evidence that A has that A ought to believe q. (This is RAE* applied to beliefs.) EOB*: p is evidence that A has that q only if p is evidence that A has that A ought to believe q.7

In previous chapters, I discussed and defended RAE and RAE*. The primary focus was the first of these claims, but the primary focus in the present chapter is the second of them. The previous formulations of these claims focused on facts (which, like many others, I take to be true propositions), whereas here it will simplify the presentation of the argument I am about to provide if the focus is simply on propositions.8 This will not make for a difference in substance, as long as one bears in mind that the propositions in question all need to be true. I should also say that I take all of the claims listed above to be necessary truths. RAEB* is simply RAE* as applied to beliefs. EOB* is a highly plausible claim that fits together nicely with RAEB*, but acceptance of it does not require a commitment to the reasons as evidence account of reasons. It says simply that if a person possesses a proposition as evidence for the truth of some other proposition then the first proposition is also evidence that they ought to believe that other proposition.9 If the evidence 7 EOB* is the partner of EOB: p is evidence that q only if p is evidence that one ought to believe q. 8 So RAE is equivalent to R (the formulation used in previous chapters), assuming that facts are true propositions and that RAE is to be restricted to cases where the propositions (that are reasons) are true. 9 Why is this not a biconditional claim? A biconditional claim is stronger than I will need in the argument I provide below, and it would rule out the possibility that there are pragmatic reasons for belief. A pragmatic reason for belief, on the current view, is a true proposition that is evidence that one ought to believe some other proposition, where this is so not (merely) in virtue of the fact that it is evidence for the truth of that other proposition. Sometimes



KNOWING BETTER

I possess speaks in favor of the proposition that it will rain today, then it clearly speaks in favor of the proposition that I ought to believe it will rain today. What will at least typically explain why some fact is evidence that I ought to believe a proposition will be that it is evidence for the truth of that proposition, and, in any case where I have evidence that a proposition is true, this will surely suffice for me to have evidence that I ought to believe the proposition—the relevant fact will thus play two roles. An attractive answer to both the question of what makes a fact evidence that one ought to ϕ and the question of what determines the strength of the evidence that one ought to ç (/reason to ç) is that it is the content of knowledge that one is in a position to possess that does both these things. I intend to argue that this is the case, but I will first argue for the less contentious claim that if you have a reason then you know the fact that is that reason. And after I have defended my answers to these two questions, I will explore the possibility that it might be necessary for someone who defends the favored answers to these questions to accept a more controversial claim, viz that the possession of reasons always partly depends on background ethical knowledge.

2 The Norm of Practical Reasoning Let us start by considering the following claim, which is sometimes called “the epistemic norm of practical reasoning,” but which I think of as a more general norm that tells us when it is reasonable, or (in other words) rationally appropriate to act on the basis of a proposition.10 pragmatic reasons for belief are contrasted with evidence in a way that suggests they are wholly non-evidential. I take this to be a mistake. If I have some evidence that I would be much happier if I were to believe in God, then it seems plausible (although by no means obvious) that I thus have some evidence that I ought to believe in God, although the evidence in question is not evidence that God exists. Of course, one might think there are no pragmatic reasons for belief (as opposed to reasons to bring it about that one believes certain propositions), and nothing here hangs on any claim to the effect that such reasons do in fact exist. Why not just “appropriate,” or just “rational”? In a general sense of “appropriate,” it might be appropriate to reason irrationally (it seems justifiable to do so when this will save many lives, for instance). And in some sense of “rational” it might be rational to act on a proposition that is not known (even if it is never justifiable to do so, it will often be excusable). Regarding this second claim, I have been influenced by Clayton Littlejohn (2012, esp. 118–20). I take the property of being rationally appropriate to be similar (or, perhaps, identical) to the normative property of fittingness. It is tempting to here use the term “reasonable,” instead of “rationally appropriate,” although this usage of “reasonable” 10

KNOWING BETTER



KA: It is rationally appropriate for A to treat p as a reason for action just in case A knows p.

This can be broken down into two claims: KA1: If it is rationally appropriate for A to treat p as a reason to ç then A knows p. KA2: If A knows p then it is rationally appropriate for A to treat p as a reason to ϕ.

I take treating p as a practical reason to involve responding to p by acting on p, or using p as a premise in practical reasoning (the latter may consist in nothing more than p being weighed alongside or against other reasons).11 A close variant of KA has been defended by Timothy Williamson, with the formulation “One knows q iff q is an appropriate premise for one’s practical reasoning,” (2005, 231), and a slightly more distant variant has been defended by John Hawthorne and Jason Stanley (2008, 578). As these authors point out, KA (or a very similar claim) seems to capture many ordinary judgments about when it is appropriate for a person to use a proposition when reasoning or acting. To provide a simple example: if a senior colleague at Boston University knows that I plan to attend a talk at Harvard that starts at 4 pm, and notices that I am still in my office at 3.30 pm, she might ask me why I am still there. I might try to defend myself by saying it only takes twenty-five minutes to get to Harvard, to which it seems appropriate for her to respond “you don’t know that.” Hawthorne and Stanley (2008, 572) suggest that the same kind of response is appropriate if one considers being offered a cent for a lottery ticket, when the lottery has not yet been held—one does not know that one will lose the lottery, hence it is not appropriate to sell one’s ticket for a cent. It is generally recognized that examples such these are does not straightforwardly follow the ordinary language usage of this term, nor align with the way the term is used elsewhere in philosophy, e.g. in political philosophy. 11 In the literature, the norm is often formulated with just the use of a premise in practical reasoning in mind, but sometimes reasoning would involve one thought too many—it is sometimes appropriate to act on reasons without reasoning. Also, suppose one knows p at time t1, when one is reasoning, but fails to know p at t2, after some practical reasoning that relies on p has occurred, but before an action based on p has occurred; if one has ceased to know p at t2, it would be in the spirit of KA to deny that it is appropriate to act on p at t2, even though it was appropriate to reason using p at t1 (I’m not certain that this is what should be said in this case, but my formulation of KA leaves this option open). My use of the phrase “using p as a premise in practical reasoning” simply follows the literature on the knowledge norm—it is important not to be misled into thinking that practical deliberation, which I have suggested often just involves weighing reasons in order to come to a conclusion about what one ought to do, must take a syllogistic form.



KNOWING BETTER

not decisive, since there are alternative explanations available, but they nonetheless illustrate the prima facie plausibility of KA. One reason one might doubt that knowledge is a necessary condition for a proposition to be relied on when it comes to practical reasoning and action is that it might seem that justified true belief is sufficient. Jessica Brown (2008, 1141) suggests that in a Gettier case that involves a person taking a train to meet someone for lunch, it does not seem inappropriate for her to (plan to) take the 12.20 pm express even though, unbeknownst to her, a hacker has replaced the internet timetable she consults with the timetable from last season (it just so happens the express train still leaves at 12.20 pm). Brown recognizes that there is room here for the defender of KA1 to claim that what should be said instead is that this person reasons and acts blamelessly, even though still inappropriately. I take it that this is the right thing to say here.12 Many pre-theoretical judgments seem to speak in favor of KA, even if not decisively so. That being said, it has been claimed by a number of authors that cases like the following (from Brown 2008, 1144–5) generate judgments that speak against knowledge as a sufficient condition for it being appropriate to rely on a proposition in reasoning and action: A student is spending the day shadowing a surgeon. In the morning he observes her in clinic examining patient A who has a diseased left kidney. The decision is taken to remove it that afternoon. Later, the student observes the surgeon in theatre where patient A is lying anesthetized on the operating table. The operation hasn’t started as the surgeon is consulting the patient’s notes. The student is puzzled and asks one of the nurses what’s going on: Student: I don’t understand. Why is she looking at the patient’s records? She was in clinic with the patient this morning. Doesn’t she even know which kidney it is? Nurse: Of course, she knows which kidney it is. But, imagine what it would be like if she removed the wrong kidney. She shouldn’t operate before checking the patient’s records.

Brown claims that the obvious appropriateness of the nurse’s last comment shows that the surgeon’s knowledge that it is the left kidney that is 12 Brown disagrees: “When I consider the question of whether it is appropriate for her to rely on the relevant proposition in practical reasoning, it seems that it is so, even recognizing the distinction between blamelessness and impropriety in acting. She justifiably and truly believes that there is an express train at 12.20 pm, and she has no reason to doubt this” (1142). I agree with Littlejohn (2013) and Williamson (forthcoming) that the view Brown is expressing here does not properly distinguish between justifications and excuses.

KNOWING BETTER



diseased is not enough to license the surgeon acting so as to remove the left kidney, and she comes to the same kind of conclusion about a number of cases that have the same general form. However, such cases are not actually counterexamples to KA, despite initial appearances. Jonathan Jenkins Ichikawa points out that “Brown’s argument is valid only on the assumption that that the disease is in the left kidney would be a sufficient reason for operating without checking the charts.” But why assume this? Why not suppose instead that a fact that needs to be treated as a reason in order to justify the surgeon in proceeding with the operation is “that the chart indicates that the disease is in the left kidney”? (both quotes from 2012, 51). And there are many other facts that might be relevant reasons here. For instance, I would think that strong evidence that one ought to check a patient’s chart before operating is likely to be provided by hospital regulations. That being said, the conclusion such examples do entitle us to reach is that KA alone provides no guidance when it comes to determining which known facts are reasons, or which of one’s reasons it is appropriate to act on. The principle provides an inadequate statement of the connection between knowledge and rationally justifiable action (even assuming that the spirit of the principle is correct, as I am doing here). This should not seem all that surprising if we pay attention to the fact that the right-hand side of KA does not make use of any ethically evaluative or normative concepts (this is a reason for thinking that Brown is on the wrong track when she suggests mental states with a better pedigree than knowledge might be what is needed):13 it would indeed be surprising if we could determine when action is appropriate through a principle that contains no reference to values, reasons or oughts.14 One might think that a fairly obvious reason why KA fails to provide us with adequate guidance as to when it is appropriate to treat a proposition as a reason in practical reasoning is that some of our 13 Of course, one might say that the concept of “knowledge” is itself a normative concept, but even if this is true, it is in itself a purely epistemically normative concept (of course, I think knowledge and evidence play a practically normative role, but it is necessary for playing this role that they be connected to practical oughts, as reasons as evidence claims that they are). 14 Hawthorne and Stanley implicitly acknowledge this when they defend a version of KA that builds in reference to preferability (2008, 578). I criticize their variation on KA below.



KNOWING BETTER

knowledge will, in many contexts, have no bearing on action whatsoever. My knowledge of the number of planets in the solar system, or of many of the capitals of many of the countries in the world is generally practically irrelevant, given that I am not a scientist or a designer of maps. Defenders of KA might well respond to this rather simplistic point by saying it rests on a rather uncharitable reading of “appropriate”; they might say that all that is meant is that known propositions are the right kinds of things (play the right kind of role) to feature in practical reasoning if and when they are practically relevant. This may be fair as an interpretation of the intent of the defenders of KA, but making this feature of KA explicit does leave one thinking it is a little less informative than one might have originally hoped it would be (this is not a criticism of KA1, but is a criticism of KA2). Would it not be a good thing if we could narrow down the conditions for appropriate action? I am about to suggest how we might narrow down KA in the right way, but before I do so, I wish to mention another problem with the principle. Suppose a satisfactory condition for determining when knowledge is practically relevant to decision-making has been provided. We should still be very concerned that modified versions of KA will not have the resources to distinguish between merely (outweighed) pro tanto reasons and pro tanto reasons that point in the same direction in relation to action as correct judgments about what ought to be done, all things considered, i.e. those reasons that it is actually appropriate to act on. Given that it clearly is not rational to act on reasons that it is rational to judge are outweighed, the principles by itself is inadequate for deciding which of one’s reasons to act on. Hawthorne and Stanley begin by defending KA1 (2008, 577), then defend a variation on KA that purports to provide necessary and sufficient conditions for when it is appropriate to treat a proposition as a reason for action: “Where one’s choice is p-dependent, it is appropriate to treat the proposition that p as a reason for acting iff you know that p” (578). They take a choice between options to be p-dependent just in case the most preferable option conditional on p is not the same as the most preferable option conditional on not-p (578). Given this, their principle may be a reasonable candidate for a principle that tells us when a reason may rationally be acted upon, but it is most certainly not a good candidate for a principle that tells us when a fact may be treated as a reason, since it

KNOWING BETTER



fails to pick out reasons that are outweighed.15 This principle might have the potential to explain the connection to reasoning and action for reasons that it is rational to act on, all things considered, yet it fails to explain why, when one is reasoning, one should treat reasons that are outweighed as reasons. So, on the one hand, we have the simpler KA, which cannot explain why it is not rationally appropriate to act on reasons that are outweighed, and, on the other hand, we have Hawthorne and Stanley’s variant, which cannot explain why it is rationally appropriate to deliberate by considering reasons that (one has yet to realize) are outweighed.

3 The Norm of Practical Reasoning, Reasons as Evidence, and Knowledge as Evidence I would suggest that RAE* can provide the missing ingredients. Here, my main claims are these. First, it is rationally appropriate to treat a fact merely as a reason (outweighed or not) just in case it is known and it is evidence that one possesses that one ought to do a particular act. Second, it is rationally appropriate to act on a reason to ç just in case (a) one knows one ought to ç, or (when one does not know one ought to ç) (b) the evidence that one possesses concerning what one ought to do makes it most probable that one ought to ç, relative to one’s other options, or no less probable that one ought to ç than do any other act.16 Cases of type (a) are those where the evidence that one possesses concerning what one ought to do will warrant drawing the conclusion that one ought to ϕ, while cases of type (b) will be those where one is not warranted in concluding that one (objectively) ought to ç, but where action is nonetheless required and one must thus do the best one can to follow the goal of acting as one ought to. This is a claim about when it is rationally appropriate to act, and not about when it is rationally appropriate to form a belief, since the option of suspending belief must be taken into

15

They appear to admit as much in response to a comment from Selim Berker (Hawthorne and Stanley 2008, 578, fn. 6). 16 One should bear in mind that often the practical reasons we possess at one time will most favor engaging in evidence-gathering activities, with the aim of (getting as close as one can to) determining what one ought to do at a future time.



KNOWING BETTER

account when considering the appropriateness, or otherwise, of belief formation, but it has no analogue in the case of action.17 Now, RAE* will only be able to play this role of filling out KA if it is always true that whenever one has a reason, one knows the fact that is the reason. Fortunately, RAE*, KA1 and one other extremely plausible claim together entail that all possessed reasons are known. This argument is provided below, from (1) to (5). The argument achieves two important aims: first, it delivers the very interesting result that all reasons to act that agents possess require knowledge (of the facts that are the reasons); second, it guarantees that RAE* is in a position to provide the much needed extra dimension of specification that has so far been lacking in discussions of KA. (1) It is rationally appropriate for A to treat p as a reason to ç only if A knows p (KA1). (2) p is a reason that A has to ç iff p is evidence that A has that A ought to ç (RAE*). (3) p is a reason that A has to ç only if it is rationally appropriate for A to treat p as a reason to ç (an extremely attractive claim). (4) p is a reason that A has to ç only if A knows p [from (1) and (3)]. (5) p is evidence that A has that A ought to ç only if A knows p [from (2) and (4)]. We have seen (1) and (2) already. (3) says that if one has a reason to do an act then it is rationally appropriate to treat that reason as a reason to do that act. This should be uncontroversial, at least on the assumption (which is not peculiar to RAE*) that one having a reason to ç is not equivalent to there simply being a reason for one to ç, for the first, but not the second, involves standing in a relation to the reason where one is able to use it in practical reasoning or attempt to directly act on it.18 17 One might literally do nothing when asleep or paralyzed, and one might ordinarily be said to do nothing, in a different sense, by purposefully staying still. Under neither construal does the option of doing nothing stand in the same relation to the option of behaving in some other manner as the option of suspending belief stands in to the option of believing a proposition. 18 Clayton Littlejohn has suggested to me that (3) might seem false when we consider cases where if a person attempted to act on a reason, they might do so rather badly. To use his example, an insensitive clod might have reason to comfort a grieving friend, but might have such a poor understanding of how to do so that, were she to comfort her friend, she would do more harm than good. Such cases are not, in fact, counterexamples to (3), since

KNOWING BETTER



Although the argument does not require (3) to be a biconditional in order to be valid, one might think that the biconditional variant of (3) actually simply provides a definition of what it is to have a reason (I will say more about this below). When we examine the concluding steps in this valid argument, we see that (4) provides a partial explanation of what it is to have a reason (it is, in part, to know the proposition that is the reason), and (5) provides a partial explanation of what it is to have evidence (it is, in part, at least, to know the proposition that is the evidence). All possessed reasons/evidence-that-one-ought are known. If one wishes to endorse a claim close in essence to KA, one should either accept RAE* or complement KA with an alternative to RAE*, which might then be critically compared with RAE*. On this test, it might be hoped that RAE* will perform particularly well. In any case, friends of KA have not proposed a satisfactory alternative to RAE* (that a satisfactory way of filling out KA is missing in the literature is the lesson of Ichikawa 2012). An argument for a (close variant of ) the more controversial half of Timothy Williamson’s (2000, Chapter 9) account of evidence, E=K, can also be provided by assuming that KA1 and RAE* generalize to hold for reasons for belief (permitting ç to be replaced by “believe q” in (1) to (5)), as well as reasons for action, and then adding one more premise:19 (6) p is evidence that A has that q only if p is evidence that A has that A ought to believe q (EOB*). (7) p is evidence that A has that q only if A knows p [from (5) and (6)] (call this EK*1).

I take it that to treat some proposition as a reason does not necessarily involve acting on it (or acting on some conclusion that follows from reasoning with it in isolation from other reasons); the clod can appropriately treat the fact that her friend is grieving as a reason to help her friend by while still that other reasons outweigh, or otherwise preclude her from acting on, this reason. 19 Kearns and Star (2008, 37) take a virtue of generalized RAE to be that it provides a unified account of reasons for action and belief, and Hawthorne and Stanley (2008, 577) make an exactly parallel claim about practical and theoretical rationality, with respect to generalized KA. There is, it must be admitted, a simpler argument for (7) which moves straight from the generalized version of (1) and uses just one other very plausible premise (viz that one’s evidence includes a proposition only if it is rationally appropriate to treat the proposition as a reason for belief [p is evidence that A has that q only if it is rationally appropriate for A to treat p as a reason to believe q]). Nonetheless, it remains interesting that one might alternatively get to (7) through generalized KA and RAE*. It is worth emphasizing that the argument up to (5) does not depend on the generalizing assumption.



KNOWING BETTER

One interesting additional point needs to be made about the arguments from (1)–(5), and (1)–(7): it is possible to weaken these arguments by replacing “knows” with “has a justified true belief ”—or, even weaker, “has a true belief ”—throughout. As one weakens KA1, it becomes harder to deny that it is true. Hence I think that, at the very least, it must be a consequence of RAE* that if a proposition is part of the evidence one possesses that one ought to do a certain act then one truly believes that proposition, and (on the generalizing assumption) that if one has some evidence, one truly believes the fact that is that evidence. Of course, I also want to say more than this, because I accept KA1 in the form originally stated, but I think it is good to bear in mind that even a weaker premise will have interesting consequences for the defender of reasons as evidence. Now let us consider the following more general claims. EK*: A has evidence that includes p iff A knows p (EK*2: A knows p only if A has evidence that includes p). EK: p is evidence for A iff A is in a position to know p.20

EK and EK* say that evidence can be identified with the content of the knowledge one is in a position to possess. Williamson (2000, Chapter 10) argues that the strength of an item of evidence is the degree to which it increases the probability that a proposition is true, given the content of one’s knowledge. If RAE and RAE* hold (and they, as we have seen, provide reason to accept EK and EK*), it is natural to adopt this account of the strength of evidence (that one possesses) as an account of the strength of the reasons (that one possesses).

20 “In a position to know p” might be taken to be equivalent to “p and p is propositionally warranted,” or might instead be taken to be equivalent to a more expansive condition—at the limit, p simply being knowable in principle; closer to home, p being knowable within time constraints dictated by the context of action. One of the more expansive conditions is required if RAE is to square with many ordinary ethical judgments. Neither EK or EK* are exactly Williamson’s E=K (A’s evidence includes p iff A knows p), as formulated, but this may only be a superficial difference, insofar as one might read Williamson as meaning by “evidence” what is here meant by “evidence that one has” (in which case, EK* and E=K are identical). So far as EK is concerned, Williamson might decline to accept that what is here thought of as unpossessed evidence is evidence (unless, perhaps, I were to adopt the restricted reading of “in a position to know”); still, he should at least allow that there is a commonplace way of speaking that would have it that there is such unpossessed evidence, as in, “The evidence we need to solve the crime is out there somewhere. If only we could find the bloody knife!”

KNOWING BETTER



Finally, let us return to (3) above. As I noted in passing above, it is very plausible to suppose that the biconditional variant of (3) tells us precisely what it is to have a reason. To have or possess a reason is for it to be rationally appropriate to treat a fact that is a reason as a reason, either by acting directly on that fact or by using it as a premise in practical reasoning (it may not be appropriate to do both).21 And given the conclusion I have come to about when it is rationally appropriate to treat a fact as a reason, it follows that having a reason involves knowing the fact that is the reason, and that fact being evidence for the truth of an oughtproposition. However, there may seem to be a counterexample to the claim that to have a reason is for it to be appropriate to treat it as a reason by reasoning with it, or acting on it. Suppose I know p, and p is evidence that I ought to ç, but suppose also that my knowledge that p cannot, at least in my present circumstances, be raised to the level of consciousness. p might thus be thought to be a reason that I possess, but not one that it is appropriate to treat as a premise in practical reasoning or act on. I think this fails as a counterexample. If there are cases when one is unable to either reason with or act on a known proposition, then, very intuitively, those are also cases where one does not possess the reason in question.22

21 I talk of practical reasoning here, but I do not think such reasoning must be contrasted with theoretical reasoning. On the view that reasons are evidence much, perhaps all, practical reasoning is theoretical reasoning (but not all theoretical reasoning is practical reasoning, since not all reasoning is oriented to working out what one ought to do). If reasoning from a belief about what one has most reason to do, or ought to do, to an intention counts as practical reasoning, or if sticking to one’s intention in action (i.e. not being akratic) counts as practical reasoning (I doubt it does), then there are parts of practical reasoning that are not theoretical. If actions themselves, and not just judgments concerning what one ought to do, are conclusions of practical reasoning, as some think, then the drawing of these conclusions would be one part of practical reasoning that is nontheoretical. 22 I am very open to the possibility that (as Stanley 2011 argues) know how involves knowledge of propositions that are often not available to reflective consciousness (which we might think is required for reasoning), and that it is appropriate to act on. I am less certain that there are subconscious processes that count as forms of practical reasoning. We have been assuming throughout that in order for it to be true that it is appropriate to treat p as a reason, it does not need to be true that it would be appropriate to reason with it and appropriate to act on it, but only that it needs to be true that it would be appropriate to reason with it or appropriate to act on it. Therefore, if acting on know how involves acting on propositional knowledge this is enough to show that it is sometimes appropriate to act on known propositions that have not risen (or perhaps cannot rise) to the level of reflective consciousness.



KNOWING BETTER

4 Ethical Knowledge Sometimes we seem to be able to come to know what we ought to do on the basis of considering our reasons, that is on the basis of considering evidence concerning what we ought to do. Our reasons for action are typically themselves facts like that she is in pain or that she asked me to help her. Reflecting on this fact, the question naturally arises: how could a body of knowledge whose content is wholly non-normative in character potentially deliver up evidence that could be the sole basis of knowledge that has normative content (knowledge one ought to ç)? We are now in the territory of Hume’s worries about the impossibility of deducing an “ought” from an “is.”23 Of course, we are not talking about deduction in the present context, and one might think that only deductive inferences that move from an “is” to an “ought” are illegitimate, or doomed to failure. It seems less problematic to think a non-ethical fact can be evidence for the truth of a normative proposition.24 Still, it does seem difficult to imagine coming to know a normative proposition on the basis of wholly non-normative evidence. Testimony might be thought to provide a route to knowledge that is an exception here (at least on some accounts of testimony): I may sometimes come to know I ought to do an act through the testimony of a reliable person, and the fact that a reliable person tells me I ought to do an act is not itself a normative fact. However, even if testimony is a source of knowledge of normative truths in some cases, it can hardly be the basis of knowledge concerning what we ought to do in general. I suggest that basic ethical knowledge must play a crucial background role when we weigh reasons with the aim of coming to know what we ought to do, if we take the Humean worry on board. Plausibly, the strength of the evidence that one has that one ought to ç is the epistemic probability that one ought to ç, conditional on the content of one’s ethical and non-ethical knowledge. This does not contradict anything that was said previously, but simply brings out that we should not 23 I do not think this is a problem for reasons for belief, because it seems highly plausible that the following bridge principle is necessarily true: if p is evidence that q then p is evidence that one ought to believe q. At least, I hope that the reader will agree that this is a clear candidate for a much more straightforward answer to the Humean worry than is available for reason for actions. 24 As I pointed out in Star 2010.

KNOWING BETTER



assume the knowledge whose content we are conditionalizing on is all non-ethical in nature. So far as unpossessed reasons are concerned, when explaining the strength of a reason (the evidence that one ought to ç), what will count are propositions, ethical and non-ethical, that one is in a position to know. One might take basic ethical knowledge to be a priori ethical knowledge. What is important for our purposes is that we take basic ethical knowledge to be non-inferential in nature. This does not commit us to thinking that such knowledge is something that is generally come by through reflection. In fact, there are good reasons to suppose that ordinary agents come by much of this knowledge pre-reflectively (see Star 2008). They need not engage in (much) reflection on the principles that we might take to be the content of such knowledge. Such principles will commonly reside in the background when it comes to practical deliberation, leaving the phenomenology of reasons to be such that reasons individually appear to be more or less weighty, and often appear to pull us in various directions. The reasons (evidence that they ought to act one way or another) that agents typically confront will not, from their perspective, be factored into ethical and non-ethical components. We can allow that agents do not generally focus their attention on principles when deliberating (this explains some of the appeal of ethical particularism), even if we take it that the ethical principles that agents are, in fact, committed to work in tandem with non-ethical knowledge to determine the strength of reasons that agents possess. When we consider the types of ethical principles that ordinary nonphilosophical deliberators might reasonably be thought to know, I think two classes of such principles stand out as the most promising. Consider these candidate forms for basic ethical knowledge for ordinary agents: (1) Conditionals, e.g. If an act is a lie then one ought not do it (or Lying is always wrong). (2) Biconditionals, e.g. An act ought not be done iff it is not welfare maximizing. (3) Defeasible generalizations, or generics: e.g. Lying is wrong (or Lying ought not be done). (4) “That’s It!” principles: e.g. Lying to a friend, and that’s it, is wrong (Holton 2002; McKeever and Ridge 2006, Chapter 6).



KNOWING BETTER

Principles of type (1) don’t seem like plausible candidates for ordinary moral knowledge at all, given that the most common examples of things ordinary agents might be tempted to suggest are always right or always wrong are not always right or always wrong. This is one thing that reflection on difficult ethical problems reveals. Lying is not always wrong (even a strong deontological ban on lying must give way when we consider the possibility of saving many lives by lying). Principles of type (2) seem even less plausible candidates for ordinary ethical knowledge—they are a type of principle we are interested in when doing ethical theory, but part of the point of the present project is that it is not at all reasonable to suppose people start off committed to principles like this (or, if some people do, they are unjustified in accepting such principles, since they did not come to accept them on the basis of extended philosophical reflection). Candidate principles of types (3) and (4) will hence be the best candidates for ordinary moral knowledge. It does seem very plausible to suppose that all of us know, for instance, that lying is (generally) wrong and that lying when doing so will harm someone—and that’s it!—is wrong.25 The defeasible generalizations that are found in generic moral claims seem to roll off the tongue easily, and this is evidence that we have identified an appropriate class of candidates. Here is a test: ask a nonphilosopher (i.e. someone who does not have theoretical commitments that might get in the way), “is lying wrong?” When she answers “yes” to this question, point out that it doesn’t seem wrong in a case where one can prevent someone being killed by lying, and she is likely to both admit this is true, and take you to have responded inappropriately to her original statement (“I wasn’t saying lying is always wrong,” she might say). Although generic propositions are good candidates for the basic moral knowledge we are looking for, there is a problem we must confront when it comes to using them to explain the strength of reasons. Consider the generic claim that cows have four legs. Given my commitment to this generic claim, what am I meant to think about the probability that the next cow I come across will have four legs? The general issue here is one of trying to determine how defeasible generalizations probabilify. I do not have any general solution to this problem, but I can suggest a model for The antecedent of “That’s it!” principles might be taken to end with “and there are no other ethically relevant facts.” 25

KNOWING BETTER



the ethical domain. Plausibly, defeasible generalizations might be ranked (this idea is from Horty 2012), and rankings influence probability distributions. We might well say, stealing is worse than lying, and lying is better than killing, so stealing is better than killing, where we understand these claims as generic. This doesn’t commit us to thinking all stealings are morally worse than all acts that are the telling of lies and that all acts that are the telling of lies are morally better than all killings. There is another way in which we might take the weight of reasons to be determined by principles of a certain kind. At one point in The Right and the Good, W.D. Ross claims that although we can know our general duties, we can never know what we ought to do when such duties conflict; about this we can merely have hope to have a “considered opinion” (2002, 19). Notice that, as a blanket statement, this seems quite wrong (I will use the language of reasons, rather than duties, in what follows, but the statement seems wrong either way). If I can know that when I have promised to meet a friend to see a movie then I have a (fairly weighty) reason to meet my friend, and I can know that when I come across a stranger in urgent need of help I have an (extremely weighty) reason to help them, then surely I can know that when I have both made such a promise, and come across such a stranger, and am aware that I can only help the stranger by breaking my promise, then I ought to help the stranger (thus, unfortunately, breaking my promise). Here is a highly specific principle that corresponds to the final judgment: “When one has promised to meet a friend and one comes across a stranger in urgent need of help, and one cannot do both things, and that’s it, then one ought to help the stranger.” The “That’s It!” here functions to indicate that one is assuming that there are no other ethically relevant considerations in play. As Holton (2002) emphasizes, such judgments leave it completely open what effect the addition of further reasons in other contexts might have on verdicts about what ought to be done (or, even, which facts count as reasons) in those other contexts.26 26 Holton does not himself talk of “That’s It!” principles, but focuses instead on the role of “That’s It!” judgments in practical reasoning. I take it to be a very straightforward move to instead see “That’s It!” clauses as themselves parts of principles of a particular kind. McKeever and Ridge contrast such principles with their “default principles,” e.g. “(P) For all actions (x) and all facts (F): If F is a fact to the effect that x would be pleasant and no other feature of the situation explains why F is not a reason to x, then F is a reason to x” (2006, 120). As this quotation makes clear, McKeever and Ridge focus on principles concerning reasons, whereas Holton is quite explicitly concerned with moral verdicts in general. Here I am interested in “That’s It!” principles concerning what one ought to do, rather than



KNOWING BETTER

One may reasonably suppose that ordinary agents possess a body of ethical knowledge of principles that is very specific in this way, and that such knowledge, residing in the background when deliberating, does much to determine the weight of reasons at the foreground of deliberation. Certain features are likely to appear across many such principles within a body of knowledge of this form, and there will be considerable variation in the extent to which such features will align with conclusions within this body of knowledge concerning what one ought to do; this variation may (when partnered with relevant non-ethical knowledge) explain the relevant probability judgments. For example, many relevant “That’s It!” principles might mention the option of telling a lie, and many relevant “That’s It!” principles might mention the option of breaking a promise; when faced with a novel situation in which an agent can only avoid telling a lie by breaking a promise (a promise not to reveal a certain fact, say), the weights of the relevant reasons may, in large part, be determined by how often all of the relevant “That’s It!” principles permit (or forbid) lying, and how often they permit (or forbid) the breaking of promises. An analogy with expertise at playing chess may help to make these claims seem plausible. According to Dreyfus and Dreyfus (1992), the expert chess player draws on a vast amount of experience of particular moves in particular games, rather than any very general rules, when spontaneously judging what moves to make in new games. Even players that fall just short of being complete experts will, on their account, draw on their vast store of experiences in a way that results in various options appearing to come with different weights: “The proficient chess player, who is classed a master, can discriminate a large repertoire of types of positions. Experiencing a situation as a field of conflicting forces and seeing almost immediately the sense of a position, he sets about calculating the move that best achieves his goal” (1992, 116).27

“That’s It!” principles concerning what one has reason to do. Note that we need not view the “That’s It!” part of a “That’s It!” principle as actually having an imperative form. This would add an unnecessary complication to my claim that it is attractive to suppose we have propositional knowledge of such principles. Rather, as I said above, I take “That’s It!” to here be shorthand for “and there are no other ethically relevant facts.” 27

I do not wish to endorse the larger Dreyfus and Dreyfus account of moral expertise. In particular, I think their attack on intellectualism is unsuccessful.

KNOWING BETTER



It would be nice to be able to provide more here in the way of a detailed model for how either a ranking of defeasible generalizations will, in practice, influence the weights of reasons (i.e. the strength of the evidence in particular contexts that one ought not lie, ought not kill, ought to assist one’s friend, etc.), or a more detailed story of how exactly “That’s it!” principles can, when many are known together, influence the weights of reasons, but these are extremely difficult tasks that are beyond the scope of the present project. My main aim here is simply to clarify why I take it to be quite plausible to suppose that ordinary agents have background ethical knowledge which plays a crucial role in determining the (strength of ) evidence that they possess that they ought to act in one way or another. I am far from certain that the Humean worry that motivated this suggestion in the first place is really justified, but if it is, then it is very tempting to suppose that the moral knowledge that I suggested in the previous chapter is necessary to think of as being an essential feature of virtue (knowledge that may also belong to many agents that are not yet virtuous) can also factor in to our assessments of the reasons that we possess.28

5 The Reasons that There Are I said at the beginning of this chapter that I take oughts to be more metaphysically fundamental than reasons. How do I square the view that reasons are not metaphysically basic, but are explained by reference to evidence and ought with the view that normative ethical theory is in the business of specifying our fundamental reasons for action? It might be thought to be a problem for the account of reasons I have been defending 28

Kieran Setiya (2012) does not think that ethical knowledge needs to be possessed in order for agents to possess evidence that an act is right or wrong (he focuses on “right,” rather than “ought”), so I think he would reject this Humean worry. However, he does think that a bridge principle between the ethical and the non-ethical must be knowable. If the Humean worry is not really justified then it is not actually necessary for present purposes for us to posit that ordinary agents possess a body of pre-theoretical background ethical knowledge which plays a crucial role in determining the presence and strength of the reasons for action that they possess (even though I still think it is attractive to suppose that we do possess such a body of ordinary ethical knowledge). This would mean that the ethical knowledge that, I suggested in Chapter 3, virtuous agents must possess might be knowledge that they arrive at merely through considering practical reasons (since reasons are evidence that one ought, they may come to have much knowledge about what they ought to do on the basis of such evidence), rather than prior to any consideration of practical reasons.



KNOWING BETTER

that fundamental reasons on this account belong to a distinct class of metaphysically basic reasons. This would make it difficult to address a concern I have heard raised on a number of occasions.29 It might appear to be the case that reasons as evidence and more traditional accounts of reasons (Broome’s or Scanlon’s, for instance) are not really alternative accounts of reasons, but simply have different subject matters. If I ultimately need to introduce a class of reasons that perform a very different function than reasons in general do on my account, and this is the same function that reasons play on other accounts, perhaps I was always talking about what it takes for something to be a reason* and not what it takes to be a reason. To this criticism, I actually have a number of things to say, but ultimately it may be the case that unless I can do more to show that fundamental reasons are metaphysically of the same kind as derivative reasons, the other things I can say will not be sufficient. Still, here is what I wish to say before I directly address the issue of the metaphysical distinctiveness of fundamental reasons: let’s begin by (1) noticing that there is significant overlap in which facts actually count as reasons, on both my account of reasons and my opponents’ (at least if my opponents’ accounts are not extremely counterintuitive); and (2) identifying truisms or platitudes accepted by pretty much everyone when it comes to thinking about normative reasons. I won’t spend much time on (1), as it seems clear that if my neighbor Skip needs help taking a ladder downstairs and I am ignorant of the correct ethical theory, I still have a reason to help him take the ladder downstairs (etc.), rather than it being the case that the only reasons that there really are come into view when we discover the correct ethical theory. I take it that both Scanlon and Broome, for instance, will wish to agree with me about this, and about many other ordinary reports as to the reasons that actually apply to people.30 Someone with an account of reasons like Broome’s could insist that only the most fundamental reasons are really reasons, since only fundamental reasons provide the genuine explanations for why we ought to act in various ways, but this would come at the cost of making most 29 David Chalmers put this worry to me when I first started comparing reasons as evidence with Broome’s (2013) account of reasons, as did Max Barkhausen more recently. 30 However, I don’t think the alternative accounts of reasons will be able to as straightforwardly explain how it is we can begin by knowing or reliably responding to derivative reasons before we come to know what the fundamental reasons are. Reasons as evidence is particularly well-suited for that job, I have argued.

KNOWING BETTER



ordinary reports about our reasons simply false. It seems much more plausible for the Broomean to start with fundamental reasons and then explain how derivative reasons also exist by using a transmission principle of some kind (or maybe simply by allowing that there are lighter explanations of reasons than the heavy duty ones that provide fundamental explanans).31 If many of the same things count as reasons on my account as on other accounts, this strongly suggests that we are not simply talking past each other. One may go further. What are the truisms or platitudes that we (participants in this conversation about the nature of reasons) all accept?32 Here are three candidate truisms: (i) reasons at least typically have weights and can be weighed against each other, whereas oughts at least typically cannot be weighed against each other; (ii) we reason with reasons, in order to work out what we ought to do; and (iii) reasons can make it the case that (or explain why) we ought to do certain things. If both I and my opponents accept these truisms, that’s also compelling evidence that we are not talking past each other; this will strongly speak in favor of the idea that we are using the very same concept of a reason when we disagree with each other about certain further claims. Our dispute is not merely terminological. That being said, Kearns and I (2008, 2015) have argued that Broome’s account of reasons, unlike reasons as evidence, does not fit together well with truism (ii).33 Admittedly, a critic of reasons as evidence might try the same general strategy to 31 The addition of a transmission principle to the Broomean account of reasons may come at a significant cost. If the best version of this theory says that there are some reasons that are parts of weighing explanations and some reasons that are not themselves explanations or parts of explanations, the overall theory will be less elegant and unified than one might have hoped it would be. To put the point another way, under this interpretation, the Broomean account of reasons is saying that fundamental reasons for action are of a different metaphysical kind than derivative reasons for action. 32 Kearns (forthcoming) considers the possibility that there is no one type of entity that plays all the main roles philosophers hope (normative) reasons can play, and he does this by carefully considering a list of possible roles. I am somewhat more optimistic, and take the three truisms I mention here to be the most crucial for ensuring we all have the same subject matter in mind. 33 A key claim here is that the truism that practical reasoning typically starts with reasons and ends with a conclusion about what one ought to do is, if one accepts Broome’s account, equivalent to the claim that practical reasoning starts with explanans (as reasons are parts of explanations of ought facts on Broome’s account) in ignorance of explandum, then moves to a conclusion as to the explandum—it seems that parts of explanations are the wrong kinds of things to be inputs to pieces of practical reasoning.



KNOWING BETTER

attack reasons as evidence, using (iii), instead of (ii). Such a critic may ask: can reasons as evidence provide a good account of how fundamental reasons explain or make it the case that one ought to act in one way or another without it ending up being true that such reasons are of a fundamentally different metaphysical kind than derivative reasons? In Chapter 1, I highlighted that it is an attractive feature of reasons as evidence that, whatever else we say about the facts that are fundamental reasons, we can say that, just like the facts that are derivative reasons, they are all evidence that one ought to act in some particular way. This may go a long way to assuaging the present concern. Perhaps it does not go far enough. In order to further address this concern, I will now consider the possibility that there may be basic, indefeasible (although not necessarily non-overrideable) pro tanto epistemic principles in the practical realm. It might be true that the fact that an act causes pain is always pro tanto evidence that one ought not do it, just as the fact that something appears to visual perception to be a certain way may always be pro tanto evidence that (one ought to believe) it is that way. On this approach, fundamental practical reasons are still facts that count as reasons in virtue of the fact that they are evidence concerning what one ought to do, and an explanation is provided as to how such reasons can be in the makingright/making-wrong business; an explanation that does not require us to think of them as essentially right-making/wrong-making (avoiding the concern that we have in mind a fundamentally different kind of entity). Let us explore the present proposal in a little more detail, by first considering reasons for belief. It is important to distinguish between generally defeasible reasons and always indefeasible reasons (in the end, given my more general theory, this will amount to the same distinction in reality as that between derivative and fundamental reasons).34 When we think about reasons for belief, it is very plausible to suppose that sometimes evidence may be disabled, silenced, undermined or (I will say) defeated, rather than overridden, while at other times evidence will be

34

I was tempted to describe this as a distinction between prima facie reasons and pro tanto reasons, but it is natural to think of a prima facie reason as something that merely appears to be a reason, but was never really a reason. The point here is that some reasons are defeasible (or, in other words, capable of being silenced), while others are not, and not that some facts that are taken to be reasons were never really reasons.

KNOWING BETTER



overridden but not defeated; and both of these are distinct from evidence no longer being present (due to merely external factors), or no longer being salient. Suppose when I first look at a newly encountered object, it appears to be a solid cube. I gently turn it around and notice that while the front part of the object is certainly very straight, the back is curvy. At this point, my reason to believe that I am looking at a cube is not merely overridden, it is completely defeated. It is not as if I might think to myself, well, I have some reason to believe that this is a cube, and some reason to believe that it is not, and the reason to believe that it is not is stronger than the reason to believe it is; on the contrary, I take it that I no longer have any evidence at all that it is a cube. I no longer believe it is a cube. Now I squeeze the object and it changes shape completely (perhaps it is made of moist clay). There is now also no longer any evidence to believe I am holding a cube, nor to believe that I am holding an object that has a side like a cube, but there is still evidence to believe that I was holding an object that had a side like a cube. In destroying the shape that the object originally had, I change the evidential situation in a way that is neither providing me with an example of a defeated reason, nor presenting me with an overriding reason (rather, I simply change the subject matter). An indefeasible pro tanto reason to believe is a reason to believe that can be overridden, and that might vanish as a result of a change in one’s situation, but is also such that while the subject matter remains the same, it will not cease to be a reason to believe. Very plausibly, visual perception in general provides us with an example here. I approach an instance of the Müller-Lyer illusion. It visually appears to me that two lines are of different lengths. My visual perception certainly provides me with a reason to believe that this is so. Now I measure the lines, and my measurement exercise reveals that the lines are of the same length. I stand back and look at the illusion. I contend that the natural thing to say in such a case is that my visual system continues to provide me with evidence that the lines are of a different length, but the reason to believe thusly provided to me is overridden, rather than silenced/disabled/defeated by the reason (or reasons) to believe that they are of the same length which I encountered when I measured the lines.35 And 35 This will be a case where one knows the lines are of the same length, and given the modified version of KA that I defended above, such knowledge might, of course, make it appropriate to act in ways that depend on the lines being of the same length. Jonathan



KNOWING BETTER

simply removing the illusion from my presence would be a different thing again (it is not to be confused with my reason for belief being defeated). Assuming my description of this case is correct, what would be the general epistemic principle that would explain why visual perception provides indefeasible but overridable reasons to believe? Very naturally the principle might be said to be: if visual perception presents an object as possessing a property P (during an interval of time), then one has an indefeasible pro tanto reason to believe that the object has property P (during the same interval of time).36 Or, in a formulation of our principle where we focus on evidence: if visual perception provides evidence that an object has property P (during an interval of time), then one has indefeasible pro tanto evidence to believe that the object has property P (during the same interval of time). What might explain why this principle is true? I’m not exactly sure, but one possibility is that the principle is true in virtue of the fact that visual perception is highly reliable (or highly reliable in normal worlds), with respect to being knowledge producing. Perhaps any source of evidence that is highly reliable in this way will ipso facto be a source of indefeasible pro tanto reasons. But perhaps this is not a satisfying enough explanation. It is not essential to my project to locate the right kind of explanation for a pro tanto principle concerning visual perception. I merely began here because such a principle seems plausible, and it seems reasonable to suppose that the principle need not be a brute truth, but might instead have the potential to be further explained. My goal here is simply to suggest that defenders of reasons as evidence can help themselves to the above distinctions when characterizing the difference between fundamental and derivative reasons for action. It isn’t

Dancy (2004, 74) has suggested that in a case where one has taken a drug that makes blue things appear red and red things appear blue, and one knows that one has taken this drug, the fact that an object appears blue is no reason at all to believe it is blue, but is merely a reason to believe that it is red. I think this is a mischaracterization of this case, although I agree with Dancy that there are cases where reasons for belief are defeated or silenced, rather than overridden. 36 And, if visual perception would present an object as having a property P (during a certain interval of time), for an appropriately idealized counterpart, then there is an indefeasible pro tanto reason to believe that the object has property P (during the same interval of time).

KNOWING BETTER



essential that the truth of the indefeasible pro tanto principles regarding practical reasons be explained in the same way (in terms of reliability, say). Let’s consider the options for two different types of ethical theory. The consequentialist might claim that the fact that an act can be reasonably expected to cause pain (or whatever other kind of bad outcome one might wish to use as an example) is an indefeasible pro tanto reason, as it is indefeasible evidence that one ought to act so as to remove or diminish the pain. What might naturally explain why such facts are always evidence that one ought to act so as to remove or diminish pain? The fact that such acts always make the world better in at least one respect. What might the deontologist instead provide as an explanation of the truth of favored principles specifying indefeasible pro tanto reasons? Deontologists can refer to duties. If acts of keeping promises, say, always align with a fundamental duty of fidelity (as Ross claimed), then this explains why the fact that an act would be a keeping of a promise is indefeasible pro tanto evidence that one ought to do it. The objective grounds of a pro tanto duty in any particular concrete case will be the knowable facts (knowable according to whichever conditions of idealization we ultimately decide are correct) that are evidence that one ought to act as the principle specifying the general duty suggests one ought to act. And the strength of the fundamental practical reasons in favor of an option, in any context, will be the strength of the evidence that an appropriately idealized agent would possess that they ought to act on that option. It might be thought that the idea that there are epistemic principles (capturing claims about invariant pro tanto evidence) concerning action, in particular, is rather mysterious. However, assuming that there is a correct set of fundamental ethical principles, and assuming reasons as evidence, it is not difficult to explain why there would be such true epistemic principles concerning invariant pro tanto evidence that one ought to act in particular ways. Surprisingly, this may be even less mysterious in one way than in the case of pro tanto epistemic principles regarding what one ought to believe. When one considers the reliability of perception, one would need to fill out much more of a story to be confident that what explains the relevant pro tanto principle (assuming it true) is that there is a pro tanto duty to believe as one perceives, or, alternatively, that it is always good (in at least one respect) to believe as one perceives. It’s not clear that these are decent enough explanations.



KNOWING BETTER

And merely reaching for reliability in knowledge formation as an alternative explanation, as I did above, might not seem satisfactory, either. However, when one considers reasons for action, it is very plausible to suppose that what explains any fundamental evidential relation is either something about the good, or a pro tanto duty (or some mixture thereof ). And here is why. Suppose the hedonistic act-utilitarian principle is necessarily true. Then the following conditional is necessarily true: if an act would increase pleasure in some respect, then that fact about the act is one that any direct application of the fundamental ethical principle (that one ought to maximize happiness) must factor in. What does it mean for an application of the theory to factor this in? It means that any decent attempt to apply the theory to a particular situation would need to take notice of the relevant fact (that the act in question would increase pleasure in some respect, or, on a deontological theory, be the meeting of a fundamental pro tanto duty). And to say that any decent attempt to apply the theory to the particular situation (by an idealized counterpart, it is natural to imagine) would need to take note of the relevant fact just is to say that the relevant fact is evidence that one ought to do the relevant act—the pleasure increasing or duty conforming act. When we ask after what evidence there is concerning practical oughts, we are asking about idealized counterparts who possess all the fundamental non-inferential ethical knowledge, as well as all the relevant nonethical knowledge, and we are asking what evidence such counterparts would possess, concerning what we ought to do. Such idealized counterparts know the fundamental ethical principles that would lead to certain facts counting as evidence one ought.37 Recall that the point of this discussion has been to attempt to show that, on reasons as evidence, derivative and fundamental reasons can be of the same fundamental metaphysical kind. I took it that it may not be enough just to say that the fundamental reasons are the particular facts 37 I recognize that ultimately more will need to be done to spell out the exact idealization conditions that are appropriate for understanding unpossessed evidence that agents ought to act in a particular ways (i.e. unpossessed normative reasons). In particular, I have not addressed the extent to which the limitations of actual agents and time constraints might both limit the extent to which fundamental ethical principles and relevant non-ethical facts are construed as knowable. On a common picture, the fundamental ethical principles are all knowable, in the relevant sense, as they are all knowable a priori for rational agents, while time constraints, etc. may limit which non-ethical facts get to count as knowable. I am by no means certain that this picture is correct, but I defer discussion of it to another occasion.

KNOWING BETTER



that explain why an act is right, and that they always also get to count as evidence that one ought to do the act in question—this claim is, I believe, true (as long as one does not understand the explanations in question to be complete explanations), but it may appear to leave open the possibility that the philosopher who views reasons as evidence and the philosopher who views reasons as right-makers are simply talking past each other. To seal off this possibility, I have argued that the defender of reasons as evidence may be in a good position to demonstrate that evidence can also be in the explanatory business, by putting pro tanto principles concerning evidence at center stage (e.g. the fact that an act will cause pain always provides evidence that one ought not do it). One more commitment needs to be made explicit, for without it, I fear this strategy will not work: we must construe the central elements of normative ethical theory— fundamental ethical principles concerning what we ought to do, and (possibly) fundamental pro tanto principles of duty or (possibly) fundamental pro tanto principles of goodness or badness—as substantially true principles, and not merely as universal generalizations that are made true by their instances. To see why I say this, consider hedonistic act-utilitarianism once more. This theory seems to consist (at a minimum) in principles concerning what is fundamentally good and bad (pleasure and pain), and a principle concerning what ought to be done (agents always ought to act so as to maximize the balance of pleasure over pain). There are two ways one might understand this ethical theory. One might take it that all that these principles do is describe or summarize a regularity in the ethical landscape: it so happens that every instance of pain is something bad. Alternatively, one might take these principles to be doing genuine explanatory work. A complete explanation of why hitting Johnny would be wrong would include not merely the fact that doing so would cause pain on this occasion, but also the principles just mentioned (assuming, as we are here just for the sake of providing a simple example to discuss, that hedonistic act-utilitarianism is true). It is only if this second interpretation is correct that fundamental reasons (e.g. the fact that hitting Johnny would cause pain) will not be metaphysically more basic than oughts, which, as I indicated at the beginning of this chapter, is something required by the version of reasons as evidence that is more than just a biconditional claim. The second interpretation is also required in order to avoid the following unattractive circularity: if items of evidence that



KNOWING BETTER

one ought are themselves the basis of complete explanations of what one ought to do, then evidence that one ought metaphysically grounds that which it is evidence of. Fortunately, the second story about what normative ethical principles do is more attractive than the first. If all putative fundamental ethical principles did was describe the ethical landscape (or, more precisely, merely universally quantify over ethical properties and the not merely ethical properties they supervene on), then I think they, and the ethical theories they constitute, would seem much less important to us than they do. Why would anyone resist the idea that fundamental ethical principles are in the substantive explanation business? One source of resistance might be a moral fetishism concern (inspired by Smith 1994). Expressing this concern, someone might say that surely a good person need not be motivated by fundamental moral principles (following a de dicto desire to act as they ought), but should rather be motivated by contingent reasons that (fully) explain (de re) why they ought to act in one way or the other. In the present context, this concern actually has no good target. This concern might well work to undermine any view of moral motivation that has it that the good person is motivated by the things that fully explain why they ought to act as they ought to act. However, it should have been clear right from the beginning of this book that I am keen to reject such a view of moral motivation. In particular, in Chapter 3, I argued that the virtuous person needs to be good at following and appropriately responding to derivative reasons (non-fundamental evidence concerning what they ought to do), and will typically be ignorant of fundamental ethical facts—such derivative reasons are a long way from being fundamental, substantive explanations. Finally, as I said at the end of Chapter 1, particularism of the kind that would deny that there are any correct fundamental ethical principles should, I think, be understood to be a position of last resort: we should accept it only if normative ethics ends up failing as a research program centered on articulating and defending very general ethical principles. If, however, such particularism were to turn out to be correct, the reasons as evidence account might be thought to have an even easier time of things, because now it will not be essential to capture the domain of talk about fundamental right-makers and wrong-makers. There will simply be a wide range of explanations (some still more shallow than others) of

KNOWING BETTER



why certain acts ought or ought not be done, and the various explanans will each get to count as evidence that one ought, in various contexts (for if E explains why one ought to ç, then E is evidence that one ought to ç).

6 Conclusion I have throughout been assuming a broad ethical realism, and, in the present chapter, that we do have a reasonable amount of ethical knowledge (although we don’t know any principles of the form that first order ethical theory aims to uncover and defend). These assumptions will seem particularly contentious to moral skeptics. It may seem as though the place we have ended up at in the present chapter is one that makes the position defended a hostage to fortune. If we only have practical normative reasons when we have ethical knowledge, and it is only rationally appropriate to act on knowledge, then, if, contrary to what we have been supposing, the moral error theorist turns out to be right, we have been acting inappropriately whenever we have acted on the basis of propositions that we took to constitute ethical knowledge.38 However, I earlier distinguished “rationally appropriate” (or “reasonable”) from a narrower notion of “rational,” and it is important to emphasize that, even if moral realism (or a quasirealism that aspires to give us all we want in terms of objectivity and knowledge) is false, we may still have been acting rationally, in this narrower sense, all along.39 In any case, I have reached a point where I am prepared to conclude that an ethical realist should accept that ordinary agents are in a position to possess fairly simple background ethical knowledge of a kind that plays a crucial role in explaining why facts count as reasons for them (and that they do possess some such knowledge, which plays a role in

38 Having said this, I actually find myself fairly certain that I know that very many of the particular things that human beings have done to each other over the millennia are wrong, and if I do know that such acts are wrong (and similarly know, on the other hand, that some of the acts of people who acted well in risky circumstances where others were doing evil acts, e.g. those who helped Jewish people escape from Nazi Germany, were right), then I am warranted in rejecting error theory on this basis, am I not? 39 Even if one were not happy to describe the ongoing reliance on untrue moral beliefs that I have in mind “rational,” I think one would at least wish to view it as epistemically excusable (here I have Williamson, forthcoming in mind).



KNOWING BETTER

explaining why they possess certain reasons).40 I have not explained how exactly it is possible for there to be such knowledge, nor provided a story concerning the mechanisms by which it arises. It might thus be felt that I have left the most serious philosophical problems in this territory unaddressed. However, I take myself to have achieved some significant philosophical goals despite not addressing this issue.41 If it were not for the fact that it would be extremely pretentious, I might have subtitled this book a “a prolegomena to any future moral epistemology.” Still, a critic who is focusing on my claims concerning ordinary pretheoretical ethical knowledge, and who is intent on pressing a concern that I have achieved too little in this book, might say the following:42 the point of this book was supposed to be that there is a puzzle as to how ordinary virtuous agents might reliably respond to genuine normative reasons while being ignorant of fundamental ethical principles (which are nonetheless discoverable via philosophical thinking), but if you are going to end up needing agents to actually possess basic ethical knowledge (of what they normally ought to do, etc.), why is your detailed solution preferable to a simpler solution that begins by positing such knowledge? This alternative approach might contend that the way that agents get to be virtuous while being ignorant of fundamental principles is simply by possessing and being guided by the kind of relatively unsophisticated ethical knowledge I end up positing anyway. This criticism overlooks the fact that a focus simply on claims about non-fundamental ethical knowledge—as plausible as I do think it is to suppose we have such knowledge—would not tell us very much at all about how exactly the multitude of particular facts that virtuous agents take to provide reasons for action get to count as genuine ethical reasons that they are entitled to respond to when ignorant of fundamental reasons for action. We have seen that reasons as evidence (which I presented a new argument for in Chapter 2) provides us with an explanation of the authority and good standing of the derivative reasons we encounter prior to

40 An unexplored alternative would be to adopt an objective Bayesian framework (as Williamson 2000 does, in a fashion), where this is understood as extending to unpossessed evidence that one ought, and deny that knowledge of ethical principles has any particular role to play with respect to objective evidential relations concerning evidence that one ought. 41 I would also add that those of us with externalist leanings in epistemology are entitled to view this as less of a pressing problem than internalists are (Star 2008). 42 Kris McDaniel put a version of this criticism to me.

KNOWING BETTER



discovering any of the fundamental reasons much sort after in normative ethics. Furthermore, it turned out to be necessary for properly addressing the problem I started with to develop an account of virtue that takes more seriously the thought that the virtuous may be ignorant of fundamental reasons than other accounts of virtue do, yet does not give up on the desideratum that the virtuous act reliably on the basis of genuine normative reasons. This account turned out to be both commonsensical under one description (roughly, the essence of virtue is being responsive to reasons), yet surprising in its upshot that the practical virtues rest, to a very large extent, in our epistemic capacities. Finally, in the present chapter I argued that reasons as evidence provides a much needed, highly plausible way of filling out the attractive idea that knowledge is the norm of action. Merely asserting that virtuous people might be able to act on incomplete ethical knowledge would offer much less than this package of ideas and arguments.43 All of the tasks of this book were undertaken in a concerted effort to both secure our judgment that virtuous people who are not philosophers typically know better than us philosophers what reasons for action it is appropriate to pay attention to in practice and to undergird our growing confidence in the philosophers’ project of articulating, defending, and, ultimately, securing a perspective from which we might knowingly survey the most general and fundamental ethical truths.

43 A further way to see that the criticism being responded to here misses its target is to pay attention to the fact that the two main reasons for positing ordinary ethical knowledge in Chapters 3 and 4 were, in the sections where I first raised them, not taken to be clearly decisive. A number of key aspects of the present project could, in fact, have been defended without claims about ethical knowledge; still, I hope the reader found the project as a whole more interesting for not having been completed with fewer materials.

References Adams, R. (2006). A Theory of Virtue. Oxford: Oxford University Press. Annas, J. (2011). Intelligent Virtue. Oxford: Oxford University Press. Anscombe, G.E.M. (1958). Modern Moral Philosophy. Philosophy 33(124): 1–19. Bommarito, N. (2013). Modesty as a Virtue of Attention. Philosophical Review 122: 93–117. Bradley, B. (2009). Wellbeing and Death. Oxford: Oxford University Press. Broome, J. (2008). Reply to Southwood, Kearns and Star, and Cullity. Ethics 119(1): 96–108. Broome, J. (2012). Climate Matters: Ethics in a Warming World. New York: W.W. Norton & Company. Broome, J. (2013). Rationality through Reasoning. Malden, MA: Wiley-Blackwell. Brown, J. (2008). Knowledge and Practical Reason. Philosophy Compass 3(6): 1135–1152. Brunero, J. (2009). Reasons and Evidence One Ought. Ethics 119(3): 538–545. Brunero, J. (2013). Reasons as Explanations. Philosophical Studies 165(3): 805–824. Crisp, R. (2007). Ethics Without Reasons? Journal of Moral Philosophy 4(1): 40–49. Dancy, J. (2004). Ethics Without Principles. Oxford: Oxford University Press. Darwall, S. (2006). The Second-Person Standpoint: Morality, Respect, and Accountability. Cambridge, MA: Harvard University Press. Darwall, S. (2009). Authority and Second-Personal Reasons for Acting. In: D. Sobel and S. Wall (eds), Reasons for Action, pp. 134–154. Cambridge: Cambridge University Press. Darwall, S. (2011). Authority, Accountability, and Preemption. Jurisprudence 2(1): 103–119. Doris, J. (2002). Lack of Character: Personality and Moral Behavior. Cambridge: Cambridge University Press. Dreyfus, H.L. and Dreyfus, S.E. (1992). What is Moral Maturity? Towards a Phenomenology of Ethical Expertise. In: J. Ogilvy (ed.), Revisioning Philosophy, pp. 111–132. Albany: State University of New York. Driver, J. (2001). Uneasy Virtue. Cambridge: Cambridge University Press. Elstein, D.Y. and Hurka, T. (2009). From Thick to Thin: Two Moral Reduction Plans. Canadian Journal of Philosophy 39(4): 515–536.



REFERENCES

Feldman, R. (2004). Having Evidence. In: E. Conee and R. Feldman (eds), Evidentialism, pp. 219–241. Oxford: Oxford University Press. Fletcher, G. (2013). A Millian Objection to Reasons as Evidence. Utilitas 25(3): 417–420. Foot, P. (2002). The Problem of Abortion and the Doctrine of Double Effect. In: Virtues and Vices and Other Essays in Moral Philosophy, pp. 19–32. Oxford: Oxford University Press. Gert, J. (forthcoming). The Distinction Between Justifying and Requiring: Nothing to Fear. In: B. Maguire and E. Lord (eds), Weighing Reasons. Oxford: Oxford University Press. Hare, R.M. (1981). Moral Thinking. Oxford: Clarendon Press. Hare, R.M. (1989). Essays in Ethical Theory. Oxford: Oxford University Press. Harman, E. (2011). Does Moral Ignorance Exculpate? Ratio 24(4): 443–468. Harman, G. (1999). Moral Philosophy Meets Social Psychology: Virtue Ethics and the Fundamental Attribution Error. Proceedings of the Aristotelian Society 99: 315–331. Hawthorne, J. and Stanley, J. (2008). Knowledge and Action. Journal of Philosophy 105(10): 571–590. Heathwood, C. (2006). Desire Satisfactionism and Hedonism. Philosophical Studies 128(3): 539–563. Holton, R. (2002). Particularism and Moral Theory. Proceedings of the Aristotelian Society, Supplementary Volume 76: 191–209. Hooker, B. (2000). Ideal Code, Real World: A Rule Consequentialist Theory of Morality. Oxford: Oxford University Press. Horty, J.F. (2012). Reasons as Defaults. Oxford: Oxford University Press. Hurka, T. (2001). Virtue, Vice, and Value. Oxford: Oxford University Press. Hurka, T. (2006). Virtuous Act, Virtuous Dispositions. Analysis 66(1): 69–76. Hursthouse, R. (1999). On Virtue Ethics. Oxford: Oxford University Press. Ichikawa, J.J. (2012). Knowledge Norms and Acting Well. Thought 1(1): 49–55. Johnson, R. (2003). Virtue and Right. Ethics 113(4): 810–834. Kamm, F. (2007). Intricate Ethics: Rights, Responsibilities, and Permissible Harm. Oxford: Oxford University Press. Kawall, J. (2009). In Defense of the Primacy of the Virtues. Journal of Ethics and Social Philosophy 3(2): 1–21. Kearns, S. (forthcoming). Bearing the Weight of Reasons. In: B. Maguire and E. Lord (eds), Weighing Reasons. Oxford: Oxford University Press. Kearns, S. and Star, D. (2008). Reasons: Explanations or Evidence? Ethics 119(1): 31–56.

REFERENCES



Kearns, S. and Star, D. (2009). Reason as Evidence. Oxford Studies in Metaethics 4: 215–242. Kearns, S. and Star, D. (2011). On Good Advice: A Reply to McNaughton and Rawling. Analysis 71(3): 506–508. Kearns, S. and Star, D. (2013a). Reasons, Facts-About-Evidence, and Indirect Evidence. Analytic Philosophy 54(2): 237–243. Kearns, S. and Star, D. (2013b). Weighing Reasons. Journal of Moral Philosophy 10(1): 70–86. Kearns, S. and Star, D. (2015). Weighing Explanations. In: I. Hirose and A. Reisner (eds), Weighing and Reasoning: A Festschrift for John Broome, pp. 232–251. Oxford: Oxford University Press. Kolodny, N. (forthcoming). Instrumental Reasons. In: D. Star (ed.), The Oxford Handbook of Reasons and Normativity. Oxford: Oxford University Press. Littlejohn, C. (2012). Justification and the Truth-Connection. Cambridge: Cambridge University Press. Littlejohn, C. (2013). The Russellian Retreat. Proceedings of the Aristotelian Society 113(3): 293–320. Mackie, J.L. (1977). Ethics: Inventing Right and Wrong. Harmondsworth: Penguin Books. McBride, M. (2013). Kearns and Star on Reasons as Evidence. Analytic Philosophy 54(2): 229–236. McKeever, S. and Ridge, M. (2006). Principled Ethics. Oxford: Oxford University Press. McKeever, S. and Ridge, M. (2012). Elusive Reasons. Oxford Studies in Metaethics 7: 110–137. McKeever, S. and Ridge, M. (2013). Generalism vs. Particularism. In: H. LaFollette (ed.), International Encyclopedia of Ethics, pp. 2103–2111. Malden, MA: WileyBlackwell. McNaughton, D. and Rawling, P. (2011). The Making/Evidential Reason Distinction. Analysis 71(1): 100–102. Parfit, D. (1987). Reasons and Persons. Oxford: Oxford University Press. Parfit, D. (2011). On What Matters, Vol. 1. Oxford: Oxford University Press. Quinn, W. (1993). Morality and Action. Cambridge: Cambridge University Press. Rawls, J. (1999). A Theory of Justice, Revised Edition. Cambridge, MA: Harvard University Press. Raz, J. (1986). The Morality of Freedom. Oxford: Oxford University Press. Raz, J. (1990 [1975]). Practical Reason and Norms. Princeton: Princeton University Press. Raz, J. (1999). Engaging Reason: On the Theory of Value and Action. Oxford: Oxford University Press.



REFERENCES

Raz, J. (2005). The Myth of Instrumental Rationality. Journal of Ethics and Social Philosophy 1(1): 2–28. Raz, J. (2006). The Problem of Authority: Revisiting the Service Conception. Minnesota Law Review 90: 1003–1044. Raz, J. (2010). On Respect, Authority, and Neutrality: A Response. Ethics 120(2): 279–301. Rosen, G. (2003). Culpability and Ignorance. Proceedings of the Aristotelian Society 103(1): 61–84. Ross, W.D. (2002 [1930]). The Right and the Good. Oxford: Oxford University Press. Scanlon, T.M. (1998). What We Owe to Each Other. Cambridge, MA: Harvard University Press. Schroeder, M. (2007a). Slaves of the Passions. Oxford: Oxford University Press. Schroeder, M. (2007b). Teleology, Agent-Relative Value, and “Good”. Ethics 117 (2): 265–295. Setiya, K. (2012). Knowing Right from Wrong. Oxford: Oxford University Press. Setiya, K. (2014). What is a Reason to Act? Philosophical Studies 167(2): 221–235. Skorupski, J. (2004). Externalism and Self-governance. Utilitas 16(1): 12–21. Smith, M. (1994). The Moral Problem. Malden, MA and Oxford: Blackwell. Sorensen, R. (ms.). The Practical Dogmatist. Stanley, J. (2011). Know How. Oxford: Oxford University Press. Star, D. (2007). Sean McKeever and Michael Ridge, Principled Ethics. Notre Dame Philosophical Reviews. Star, D. (2008). Moral Knowledge, Epistemic Externalism, and Intuitionism. Ratio 21(3): 329–343. Star, D. (2010). Moral Skepticism for Foxes. Boston University Law Review 90: 497–508. Star, D. (2011). Two Levels of Moral Thinking. Oxford Studies in Normative Ethics 1: 75–96. Star, D. and Delmas, C. (2011). Three Conceptions of Practical Authority. Jurisprudence 2(1): 143–160. Swanton, C. (2005). Virtue Ethics: A Pluralistic View. Oxford: Oxford University Press. Thomson, J.J. (1976). Killing, Letting Die, and the Trolley Problem. The Monist 59(2): 204–217. Thomson, J.J. (1985). The Trolley Problem. Yale Law Journal 94(6): 1395–1415. Thomson, J.J. (2008). Normativity. Peru, IL: Open Court Publishing. Väyrynen, P. (2013). The Lewd, the Rude and the Nasty. Oxford: Oxford University Press. Way, J. (2013). Value and Reasons to Favor. Oxford Studies in Metaethics 8: 27–49.

REFERENCES



Weatherson, B. (2013). Running Risks Morally. Philosophical Studies 167(1): 1–23. Williams, B. (1988). The Structure of Hare’s Theory. In: D. Seanor and N. Fotion (eds), Hare and Critics, pp. 185–196. Oxford: Clarendon Press. Williams, B. (1993). Ethics and the Limits of Philosophy. London: Fontana Press. Williamson, T. (2000). Knowledge and its Limits. Oxford: Oxford University Press. Williamson, T. (2005). Contextualism, Subject-Sensitive Invariantism and Knowledge of Knowledge. The Philosophical Quarterly 55(219): 213–235. Williamson, T. (2007). The Philosophy of Philosophy. Malden, MA, and Oxford: Blackwell. Williamson, T. (forthcoming). Justifications, Excuses, and Sceptical Scenarios. In: J. Dutant and D. Dohrn (eds), The New Evil Demon. Oxford: Oxford University Press. Zagzebski, L. (2012). Epistemic Authority: A Theory of Trust, Authority, and Autonomy in Belief. Oxford: Oxford University Press. Zangwill, N. (2011). Cordelia’s Bond and Indirect Consequentialism. Oxford Studies in Normative Ethics 1: 143–165. Zimmerman, M.J. (1997). Moral Responsibility and Ignorance. Ethics 107(3): 410–426. Zimmerman, M.J. (2008). Living with Uncertainty: The Moral Significance of Ignorance. Cambridge: Cambridge University Press.

Index Adams, Robert M. 2n, 69n Annas, Julia 69n Anscombe, G.E.M. 69n Aristotle 2, 78n authoritativeness 39 authority concept of 41–3 de facto vs. de jure 40, 44–5 directives of 51–3 minimalist conception of x, 37, 42–6, 51–2, 56, 58 practical vs. epistemic 42–6 second-personal conception of 46–7 service conception of 37, 40–2, 56, 58 autonomy 23, 48–51 Barkhausen, Mark 126n Berker, Selim 115n blameworthiness 73, 80–2, 104 Bommarito, Nicolas 84n Bradley, Ben 6 Broome, John 7n, 17n, 37n, 40n, 58n, 106n, 126–7 Brown, Jessica 112–13 Brunero, John 17n, 37n Chalmers, David 126n commonsense morality 8n, 19, 27, 68–70 consequentialism vii, x, 131 act- 29 and reason 63–4 global 70n, 83 rule- 29–31 see also utilitarianism; virtue courage 86, 88–9, 100–1 Crisp, Roger 32n Dancy, Jonathan 3n, 31–2, 35, 129–30n Darwall, Stephen 42n, 43n, 46–8, 49n, 50–1 deliberation, practical 82, 94–7 Delmas, Candice 66n deontology 5, 101–2, 131 pluralist/Rossian 3n, 13n, 29, 35

disentanglement argument, the 86–7, 92 Doris, John 69n Dreyfus, Hubert 124 Driver, Julia 69, 70n, 74, 79n, 80–5, 100 Elstein, Daniel 70, 85–92, 93n, 94n, 95, 100–1, 103n embedded evaluations 89–91 error theory ix, 108 evidence and “ought to believe” (EOB*) 109–10, 117 defeasible vs. indefeasible 128–33 dependent vs. independent 33–4 misleading 66–7, 102–5 perceptual 129–30 possessed vs. unpossessed 14, 53–5, 109–10 that one ought to do an act 60–4 Williamson’s account of 54, 105 see also knowledge, as evidence; reasons (normative), as evidence evil demon cases 71–81 Feldman, Richard 55n Fletcher, Guy 37n Foot, Philippa 5 Gert, Joshua 15n Gibbard, Allan 92n Hare, R.M. 8–12, 25, 27, 29–30, 80 Harman, Elizabeth 104n Harman, Gilbert 69n Hawthorne, John 111, 113n, 114–15, 117n Heathwood, Chris 6 Holton, Richard 121, 123 Hooker, Brad 29–30, 69n Horty, John 123 Hume, David xii–xiii, 120 humility/modesty, epistemic 19, 22, 82, 85



INDEX

Hurka, Thomas 69n, 70, 85–92, 93n, 95, 100–1, 103n Hursthouse, Rosalind 21, 75n, 85

normative ethics/ethical theory vii, 2, 4–8, 12, 38, 131 Nye, Howard 24n

Ichikawa, Jonathan Jenkins 113, 117 ignorance and culpability xi, 80–2, 104 integrity 88–9, 101

ought, not deducible from “is” xii, 120

Johnson, Robert 75n, 79n justice, distributive 87–8, 101–2 Kamm, Frances 5 Kant, Immanuel 77, 78n, 99 Kantian ethics vii, 17n, 69, 77, 79 Kawall, Jason 69n Kearns, Stephen x, 13, 15n, 37, 55, 58, 95n, 97, 106, 117n, 127 knowledge as evidence 108, 117–18 as norm of reasoning/action (KA) 108–19 moral/ethical xii–iii, 11, 26–8, 104–5, 107–8, 120–5 non-inferential/basic 121, 132 through testimony 120 Kolodny, Niko 66 Littlejohn, Clayton 110n, 112n, 116n luck viii–ix, 2, 26–8, 63, 104 Mackie, J.L. 22n Manne, Kate 30n McBride, Mark 37n McDaniel, Kris 136n McDowell, John 87–8, 92 McKeever, Sean 18n, 31n, 32n, 33, 37n, 121, 123n McNaughton, David 37n modesty, as virtue of ignorance 83–5 moral fetishism 134 moral judgments/intuitions 9–10, 26–7, 30 moral thinking “intuitive” vs. “critical” 8–9 moral worth 77–9 non-identity problem 6–7 Normal Justification Thesis (NJT, NJT*) 40–1, 46–8, 55–6

Parfit, Derek 4n, 5–6, 7n, 21, 36n, 52, 69n particularism 31–5 Plato 10 prescriptivism 11, 27 principles 101, 121–5 defeasible generalizations 121–3, 125 epistemic 130–1 fundamental ethical vii–iii, 12, 27–8, 38 “That’s It!” 123–5 primacy, epistemic vs. practical xi, 99 quasi–realism ix, 11n, 135 Quinn, Warren 5 Rashdall, Hastings 100 rational appropriateness 110 Rawling, Piers 37n Rawls, John 10 Raz, Joseph x, 17n, 37, 40–51, 55–8, 64–5, 66n, 67n Razian Insight, the 55–9 realism, ethical ix, 135 reason (capacity) 37, 56–7, 63–4 “reason implies can” 95 reasoning, practical 18 reasons (normative) as evidence (R, RAE, RAE*) x, xiii, 13–15, 28, 37, 52–3, 58–9, 96n, 97–8, 109, 115–18 as (parts of) explanations 17–18, 126–7 as unanalyzable 52 atomism vs. holism 31–3 defeasible vs. indefeasible 128–33 desires, treated as 59–61 epistemic (R, RAE, RAEB*) 13, 37, 44, 52–3, 109–10 for emotions 97n fundamental vs. derivative viii, 15–19, 23–5, 31–5, 59–64, 125–6, 134 Humean theory of 39n instrumental 66–7

INDEX

justifying vs. requiring 15n moral vs. prudential 20–1, 38n possessed vs. unpossessed 14, 53–5, 109–10 pragmatic (for belief) 13, 109n preemptive and exclusionary 40–6, 56, 64–6 roles of/truisms concerning 18, 38, 56, 126–8 strength/weight of 108–10, 120–5 reflective equilibrium 10 responsibility, epistemic 24–5 Ridge, Michael 11n, 18n, 31n, 32n, 33, 37n, 121, 123n rightness and wrongness 71–2 subjective vs. objective 70, 74–6 Rosen, Gideon 104n Ross, W.D. 3n, 13n, 35, 92, 100, 123, 131 Scanlon, Thomas M. 36n, 52, 126 Schroeder, Mark 39n, 69n, 108n self-worth 83–5 Setiya, Kieran 29n, 37n, 125n Sidgwick, Henry 9, 20n, 27 Simple Example (SE) 13–15, 20, 38–9 Skorupski, John 79n Smith, Michael 61, 134 Sorensen, Roy 65 Stanley, Jason 111, 113n, 114–15, 117n, 119n Stevenson, C.L. 11n Swanton, Christine 69n thick concepts/terms 86–8, 90, 92 Thomson, Judith Jarvis 5, 37n trolley cases 5 utilitarianism viii, 8–12, 15–16, 26–8, 69, 132–3; see also consequentialism



value (goodness) 59, 62, 69, 90, 93–5, 101–2 Väyrynen, Pekka 103n virtue consequentialist accounts of xi, 68–70, 74, 79–83 degrees of 98 new account of x–xi, 21–5, 96–9 reasons of 91–5 reductive accounts of 85–99 Type I form of analysis 86–8, 90 Type II form of analysis 87–8, 90, 93n Type IIR form of analysis 93n Type III form of analysis 88–91, 93, 95–6 Type IIIR form of analysis 93, 95–6 Type IIIR* form of analysis 96, 98, 103, 105 Type IIIE form of analysis 98–9, 103, 105 virtue ethics 68–9 vicious people 72–6 virtuous acts vs. dispositions 90–1 virtuous people xi, 1–3, 19, 21–5, 31, 71–82 acting out of character 75–6 and the possession of virtues 76–7 Way, Jonathan 17n, 96n, 101n weakness of will 25n, 99 Weatherson, Brian 104n well-being, accounts of 5–6, 20–1, 48–9 Williams, Bernard 8–9, 12, 20, 25–6, 28–9, 71n, 80, 92 Williamson, Timothy xii, 53n, 54, 88n, 105, 108, 111, 112n, 117–18, 135n, 136n Zagzebski, Linda 66n Zangwill, Nick 19n Zimmerman, Michael 104n