Oxford Studies in Metaethics, Volume 14 0198841450, 9780198841456

Table of contents :
Contents
List of Contributors
Introduction
1. How Principles Ground • David Enoch
2. The Self-Undermining Arguments from Disagreement • Eric Sampson
3. Contextualism, Moral Disagreement, and Proposition Clouds • Jussi Suikkanen
4. Taking Prudence Seriously • Guy Fletcher
5. Internalism and Prudential Value • Jennifer Hawkins
6. Desiring under the Proper Guise • Michael Milona and Mark Schroeder
7. Infinitism about Cross-Domain Conflict • David Killoren
8. We Can Have Our Buck and Pass It, Too • Zoë Johnson King
9. Reason without Reasons For • Michael G. Titelbaum
10. The Fundamentality of Fit • Christopher Howard
Index

Citation preview

O X F O R D S TU DI E S IN M E TA E T H I C S

Oxford Studies in Metaethics Volume 14

Edited by RUSS SHAFER-LANDAU

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 © the several contributors 2019 The moral rights of the authors have been asserted First Edition published in 2019 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: 2019930897 ISBN 978–0–19–884144–9 (hbk.) 978–0–19–884145–6 (pbk.) 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.

Contents List of Contributors Introduction 1. How Principles Ground David Enoch 2. The Self-Undermining Arguments from Disagreement Eric Sampson 3. Contextualism, Moral Disagreement, and Proposition Clouds Jussi Suikkanen

vii ix 1 23

47

4. Taking Prudence Seriously Guy Fletcher

70

5. Internalism and Prudential Value Jennifer Hawkins

95

6. Desiring under the Proper Guise Michael Milona and Mark Schroeder

121

7. Infinitism about Cross-Domain Conflict David Killoren

144

8. We Can Have Our Buck and Pass It, Too Zoë Johnson King

167

9. Reason without Reasons For Michael G. Titelbaum

189

10. The Fundamentality of Fit Christopher Howard

216

Index

237

List of Contributors David Enoch is Rodney Blackman Chair in the Philosophy of Law and Professor of Philosophy, Hebrew University. Guy Fletcher is Senior Lecturer in Ethics and Political Philosophy, University of Edinburgh. Jennifer Hawkins is Associate Research Professor of Philosophy, Duke University. Christopher Howard is Research Assistant Professor of Philosophy, University of North Carolina at Chapel Hill. David Killoren is Post-Doctoral Fellow, Institute for Religion and Critical Inquiry, Australian Catholic University. Zoë Johnson King is Bersoff Faculty Fellow, New York University. Michael Milona is Assistant Professor of Philosophy, Ryerson University. Eric Sampson is graduate student, University of North Carolina at Chapel Hill. Mark Schroeder is Professor of Philosophy, University of Southern California. Jussi Suikkanen is Senior Lecturer, University of Birmingham. Michael G. Titelbaum is Professor of Philosophy, University of WisconsinMadison.

Introduction Russ Shafer-Landau This volume truly displays the breadth of work nowadays being done in metaethics, as its entries range broadly across so many of philosophy’s areas of specialization. We begin with a work in metaphysics, by David Enoch, who examines the nature of the grounding of particular moral facts. An intuitive picture has it that such facts are grounded in two things: a set of natural facts, and a moral bridge principle that links the natural facts to the moral one. An alternative view alleges that the natural facts alone serve as the ground of moral facts—moral principles are otiose when it comes to citing that in virtue of which moral facts hold. Enoch argues that we can preserve a grounding role for moral principles once we distinguish two kinds of grounding—normative and metaphysical. While natural facts alone can serve as the metaphysical grounds of particular moral facts, moral principles are needed to provide the normative grounds of moral facts. For those who might be suspicious of multiplying grounding relations beyond necessity, Enoch argues that we have independent reason to endorse both metaphysical and normative grounding. Next up: a pair of articles about moral disagreement, a phenomenon that can be approached from many different philosophical angles. Eric Sampson focuses on the metaphysics and epistemology of moral disagreement, while Jussi Suikkanen concentrates on issues from linguistics and the philosophy of language. Sampson surveys many arguments that start with a premise about actual or predicted fundamental moral disagreement and then draw anti-realist or skeptical conclusions. He argues that all such extant arguments are self-undermining, and that all such future arguments are likely to suffer the same fate. The central problem for all of these arguments is that they contain a premise that is either itself the subject of widespread disagreement or supportable only by means of at least one claim that is subject to such disagreement. Sampson is careful to note that he has not tried to show that all such arguments are unsound, but rather that no one can be justified in believing their conclusions. Suikkanen’s entry focuses issues that arise within the framework of contextualist moral semantics. Such views deny that moral predicates have fixed semantic values—rather, the properties referred to by moral predicates will vary with context. Different contextualist theories disagree about the

x

Introduction

relevant contextual features that determine which natural property is designated by a moral predicate. Suikkanen focuses on subjectivist and relativist versions of these views, according to which it is either the speaker’s own moral code or her moral community’s moral code that constitutes the reference-fixing context. A standard objection to these views is that they are unable to account for how we are able to disagree about moral matters. Another is that there is an arbitrariness worry about selecting the relevant context, especially for relativist views: which culture’s mores will determine the truth of a moral utterance when the speaker is a member of more than one culture? Suikkanen invokes the notion of a proposition cloud—a set of propositions, all of which a speaker puts into play in a given context by her utterance—as a way to defend against both objections. All the propositions in such clouds are available for rejection or acceptance on the part of listeners. The norms of conversation then guide interlocutors to make informative contributions to the conversation in ways that enable them to disagree and to non-arbitrarily determine which cultural elements will fix the reference of the moral predicates in use. In the next pair of articles, Guy Fletcher and Jennifer Hawkins invite us to reflect on metanormative issues regarding prudence. These are not efforts to specify what it is that makes our lives go well, but rather, in Fletcher’s case, an investigation into the nature of prudential reasons and prudential discourse. Fletcher first defends the claim that facts about an agent’s well-being generate, and so entail, reasons for her either to perform certain actions or to have certain attitudes. He then offers an argument from analogy (whose analogue is moral judgment) for the claim that prudential discourse is genuinely normative. According to Fletcher, the same features that reveal the normativity of moral judgments do so for prudential ones. Hawkins considers the question of whether existence internalism holds for prudential value. This form of internalism says that something can be good for an agent only if it is capable of motivating her. Hawkins draws our attention to a modal distinction between realized good (what is now or has been good for an agent) and merely potential good (what may contribute to her good at some future date). Hawkins argues that existence internalism is true when it comes to realized good: something is now good for an agent only if it elicits some kind of positive psychological response from her during the time at which it is good for her. However, she believes that philosophers should be existence externalists about merely possible good. Even if facts about one’s future good ground reasons to act in ways to bring it about, agents may fail to recognize such facts, and so may fail to be duly motivated.

Introduction

xi

Michael Milona and Mark Schroeder next take us into the realm of moral psychology. They direct their efforts to vindicating a specific version of the guise of the normative thesis: the view that all desires are associated with normative appearances or judgments. This thesis has two main versions. The first says that the relevant normative representations have to do with reasons; the second, with the good. Milona and Schroeder defend the former version as against the latter. The central idea is that we can use observations from the theory of content determination in order to constrain possible theories of the representational contents associated with desire. They argue that guise of the good views fail to meet those constraints, while a guise of reasons view can readily meet them. David Killoren’s article focuses on deep structural issues within ethics. His central claim is that if what he calls the overriding view is true, then infinitism about cross-domain conflict is true. The overriding view is the view that there are multiple normative domains (e.g., morality and prudence), whose requirements can conflict with one another, and that in some such conflicts, an obligation belonging to one normative domain overrides an obligation belonging to another. Infinitism about cross-domain conflict combines the overriding view with the following infinitist claim: for every case in which an obligation from one domain (call it DA) overrides an obligation from another, there is an infinitely long chain of domains D₁, D₂, D3, . . . , such that D₁ contains an obligation to act in line with the obligations of DA, and D₂ contains an obligation to act in line with the obligations of D₁, and D₃ contains an obligation to act in line with the obligations of D₂—and so on, indefinitely. The core idea is that resolution of cross-domain conflict requires a higher-order domain whose edicts possess the authority to resolve such conflict. If, for instance, a moral and a prudential requirement conflict, we cannot resolve the conflict just by letting morality assert its preeminence, so to speak. Some other, higherorder domain, rather than one of the domains involved in the dispute, must be introduced to resolve the conflict. But each such higher-order domain will be one among many at the same level. So, to possess the normative power needed to effect the conflict resolution, it needs to be invested with that authority by the edicts of a domain at a yet-higher level—and so on, and so on, ad infinitum. Killoren realizes that this view is going to strike most readers as unpalatable, much as infinitism in the epistemic domain has met with widespread doubt. Nor does he himself find the result very savory. But he sees no plausible way to avoid it. Zoë Johnson King’s article is a contribution to debates about what has become known as “buck-passing.” Parties to these debates are taking a stand on whether moral requirements are themselves reasons for action. So-called

xii

Introduction

buck-passers insist that such requirements cannot themselves be practical reasons, for if they were, then there would be too many reasons. Buckpassers think that the grounds of moral requirements are reasons, and that if the moral requirements themselves were additional reasons, then we’d get a pernicious form of normative double-counting. Johnson King is no buckpasser, and she offers a criticism of this oft-repeated redundancy argument. She emphasizes that this argument overgeneralizes, yielding implausible verdicts in all cases in which two or more reasons are arranged in relationships of metaphysical constitution. For all such cases, she proposes that all of these facts can be genuine reasons, whose normative weight is shared in virtue of the metaphysical relationships between them. Michael G. Titelbaum’s entry is also focused on structural relations among facts that count as reasons. He notes that metaethicists have recently devoted a great deal of attention to questions about when a fact counts as a reason for or against a particular conclusion, and how such reasons interact. But he asks a broader question: When a set of facts counts in favor of some conclusion, is that always because at least one of those facts is a reason for that conclusion? Titelbaum returns a negative answer. He offers a battery of examples in which a set of considerations supports a conclusion without any fact in that set’s being a reason for that conclusion. He then assesses the significance of such examples for philosophical methodology, metanormative realism, and the “reasons-first” program. Closing out the volume is Chris Howard’s award-winning entry in the 2017 Marc Sanders metaethics competition. As Howard notes, many authors favor a “reasons-first” ontology of normativity, which treats reasons as normatively fundamental. Others favor a “value-first” ontology, which treats value or goodness as normatively fundamental. Howard argues that both are mistaken, on the grounds that neither can account for all of the normative reasons there are. He develops an ontology of normativity, originally suggested by Franz Brentano and A. C. Ewing, according to which fittingness is normatively fundamental. The normative relation of fittingness is the relation in which a response stands to an object when the object merits—or is worthy of—that response. Howard argues that a “fittingness-first” ontology is no less parsimonious than either of its main competitors, and that it, unlike them, can plausibly accommodate the existence of all the normative reasons there are. Early drafts of most of the papers in this volume were delivered at the 12th annual Madison Metaethics Workshop. I’d like to thank John Brunero, Tristram McPherson, Andrew Sepielli, Daniel Star, Sigrún Svavarsdóttir, and Alex Worsnip for doing superlative work as the Program Committee for that event. I’d also like to thank Jamie Dreier, Sarah McGrath, and Jussi Suikkanen for having served as the review panel for the 2017 Marc Sanders

Introduction

xiii

Prize. Lastly, Richard Rowland and Derek Baker served as OUP’s reviewers for this volume. It’s a huge task; they offered dozens of pages of trenchant, constructive comments that have led to a great many improvements in the essays you have before you. From myself, and on behalf of the grateful contributors to this collection: thank you. And thanks, as always, to Peter Momtchiloff, editor par excellence and reliable steward of this series.

1 How Principles Ground David Enoch

It would be wrong of you not to give that grad student feedback on her paper in a timely manner. Why? Well, because she legitimately expects this feedback, and not getting it on time will harm her progress. But this is not all—more is needed. What makes it the case that it would be wrong of you not to give that grad student her feedback soon is all of that, together with the fact that you ought not to harm people by frustrating their legitimate expectations. I ought to send the workshop organizer my paper by the agreed deadline. I have this moral duty, it seems, not merely in virtue of having promised, but also in virtue of the fact that we ought to keep our promises. In these and in many other cases, it is very natural to refer to moral principles (in different levels of generality, perhaps) in offering grounding explanations of specific moral facts—that is, in answering “in virtue of what” or “what makes it the case that” questions about them. And as I proceed to argue in the first section, for Robust Realists like myself this may be more than just a natural thing to say—there may be theoretical reasons to think that Robust, non-naturalist, Realism needs moral principles to do serious grounding work. But a recent challenge from Selim Berker threatens to refute this natural picture of the grounding role of moral principles. Moral principles, Berker argues plausibly, are themselves already claims about what grounds specific moral facts, and so they leave no room for themselves to do such grounding work. I present this challenge in more detail in Section 1.2, and the rest of this paper is my attempt at responding to it. It turns out that in doing so, a detour through thoughts about the grounding role of legal rules comes in handy. I discuss how the law grounds legal facts in Section 1.3, offering there an analogue to Berker’s challenge, and coping with it. I then present my response to the challenge in the moral domain in Section 1.4. That solution relies on the distinction between different kinds of grounding relations, or on grounding-pluralism. Berker anticipates this line of response to his challenge, though, and in Section 1.5 I get into even

2

David Enoch

deeper metaphysical waters in order to defend the grounding pluralism line against Berker’s response. Sections 1.6 and 1.7 offer a somewhat broader view: In Section 1.6 I try to distinguish between domains in which principles do grounding work and domains about which we can settle for (roughly) Humean—metaphysically idle—principles. This discussion further supports the claim that morality is among the former. And in Section 1.7 I show that we—all of us, pretty much regardless of other metanormative commitments—have reason to accept grounding pluralism (and in particular, a notion of moral grounding that is distinct from metaphysical grounding), by highlighting (quickly) some of the other work grounding pluralism does in metaethics. I hope that this is of independent interest, but also, it serves to show that there’s nothing ad hoc in the apparatus I employ to respond to Berker’s Challenge.

1 . 1. T H E N A T U R AL R O B U ST R E AL I S T P I C T UR E According to Robust Metaethical Realism, there are response-independent moral facts and properties and truths. Furthermore, they are importantly different from natural facts, properties, and truths. This much is clear—even though it’s not as clear how to characterize the natural (and with it, presumably, the non-natural), or what “importantly different” precisely comes to. But because much has been said about such complications,¹ and because the precise details of ways of dealing with them won’t matter for our purposes here, let’s not spend time on them. Let me stipulate that the Robust Realism I’m interested in here is committed to the following: No moral fact is fully grounded in non-normative, natural facts. In other words, whenever in-virtue-of-what questions asked about moral facts (in virtue of what is that action wrong? In virtue of what is inequality bad, if it is?) have answers,² any full answer will have to invoke a normative, non-natural fact. This claim is meant to capture the idea that moral facts are as real, as ontologically respectable, and the most fundamental of them are as fundamental, as any others. When the gods finished creating the natural facts, as it ¹ Including, a little, by me. See my Taking Morality Seriously (2011a: 4) (hereafter TMS). ² Some in-virtue-of-what questions of this kind may not have substantive answers. Perhaps the fact that pain is (pro tanto) bad is ungrounded: there’s nothing in virtue of which it holds. Perhaps even similar claims about tokens of pain (the fact that this agony is bad) are ungrounded. If so, this will help accommodate the intuition (to which Tristram McPherson drew my attention) that there doesn’t seem to be any need for a generalization in grounding the badness of a specific token of agony.

How Principles Ground

3

is sometimes said, they still weren’t done—they still had to put in place the moral ones as well.³ Of course, this does not rule out moral facts being partly grounded in natural facts, which is a good thing, because obviously, they are—the fact that it would be wrong of you not to give your grad student comments on her paper is at least partly grounded in natural facts about the effects of such comments on her work. The Robust Realist merely insists that such stories—grounding particular moral facts in natural ones—will never give the full grounding story. The full grounds of particular moral facts will include more than just the relevant natural facts.⁴ Nor does this claim—that no moral fact is fully grounded in natural facts—entail that all moral facts are at least partly grounded in other moral facts. Some moral facts may be ungrounded.⁵ Such fundamental moral facts—principles, perhaps—then have a special role: they are not themselves grounded (in either natural or normative facts), and they play a partial grounding role—perhaps at the end of rather long grounding chains—in grounding non-fundamental moral facts. So according to such a picture, if utilitarianism is true, then the fact that you ought to maximize utility (or some such) is an ungrounded moral fact; and all other moral facts are partly grounded in it (and partly in the natural facts, facts about which line of action will have which consequences, and the like). And if some Rossian pluralism about prima facie duties is correct, then the moral fact that I ought to send the workshop organizer my paper by the deadline is partly grounded in the principle requiring that we keep our promises (together with natural facts about my having promised), and that principle itself is not grounded at all.

³ Metaphors are tricky. This one, for instance, may suggest that there was actually some time in which causing gratuitous pain was not wrong (until the gods got to putting in place moral properties). But this is a feature of the metaphor, not of my view. ⁴ Shafer Landau (2003: 76–7)—a self-proclaimed non-naturalist—seems willing to accept that normative facts are fully grounded in natural ones. As Leary (2017: 91) argues, though, this renders him vulnerable to the charge that at the end of the day he’s not a nonnaturalist after all. Perhaps Shafer-Landau can insist (as he has, in conversation; and I thank Jussi Suikkanen for a closely related point) that the philosophical motivations for nonnaturalism are consistent with seeing all normative facts as fully grounded in natural ones. Whether this is so will depend on the motivations for non-naturalism. In my case, this will take us to the just-too-different intuition, which I tie to grounding-talk below, in Section 1.6. ⁵ They may be ungrounded either because they are fundamental—that is, they are apt to be grounded, but standing at the end of a grounding chain, they are not—or perhaps because they are autonomous in Dasgupta’s (2014) sense, that is, they are not in the game of being grounded at all.

4

David Enoch

This, then, is a fairly natural picture even independently of Robust Realism. But there are reasons to think that for Robust Realism it’s more than just natural—it’s needed. First, there seems to be a very close relation between moral (and more generally normative) facts and natural ones. Actions are wrong because they are, or in virtue of being, harmful, or painful, or humiliating, or . . . ⁶ What is the nature of this relation between the natural and the moral facts? The Robust Realist cannot—on pain of deserting her non-naturalism—go for identity here. Grounding seems to work better, but not full grounding—for if the moral facts are fully grounded in natural ones, again naturalism (denied by Robust Realism, as characterized above) will have been endorsed. So the natural way of thinking of the relation between the natural and the moral facts is as a relation of partial grounding. Which raises the question—what else is needed for fully grounding moral facts except for the morally relevant natural ones? And then there don’t seem to be plausible alternatives to fundamental moral principles.⁷ Second, and relatedly, one of the main challenges facing Robust Realism is to explain the supervenience of the moral (or the normative) on the natural. Such supervenience—consisting, according to Robust Realism, in necessary connections between distinct entities—calls for explanation, and if Robust Realism can’t explain it (and if, as seems plausible, at least some alternative metaethical views have a much easier time explaining it), then this is very bad news for Robust Realism. When I suggested a way out—an explanation of supervenience—I relied on moral principles doing work.⁸ The details don’t matter right now, and neither do the reasons I’m no longer confident what I say there succeeds. The relevant point is that if anything like my story there is even just a part of the right story to tell here, principles must be doing grounding work roughly along the lines of the natural picture above.⁹ For these reasons, then, even if rejecting the natural picture is consistent with Robust Realism, still if the natural picture cannot be defended this would be especially bad news for Robust Realism.¹⁰ ⁶ See here also Väyrynen’s (2013: 155) related “dependence intuition.” ⁷ For this line of thought, see Rosen (2017a). The natural picture in the text is the one that Rosen (2017b) calls “bridge-law non-naturalism.” ⁸ TMS (136–50). ⁹ Leary (2017: 83 and on) is especially clear on putting my story on supervenience in terms of the grounding role of principles. ¹⁰ Moral particularists—some of whom want to defend a fairly robust realism, but without according any serious role to moral principles—are another story. Perhaps they will have to invoke a plethora of ungrounded specific moral facts. Or—going out on a limb here—perhaps they will want to say that specific moral facts are only partly grounded

How Principles Ground

5

1. 2 . BER K ER ’S CH A L LE N G E The utilitarian principle, it seems very plausible to assume, does not merely state an extensional equivalence—it is not merely the claim (somewhat roughly) that an action is wrong if and only if there’s an available alternative that generates higher utility.¹¹ At the very least, utilitarians think that such equivalence is no mere coincidence, so we need to add a “necessarily” before the equivalence. But even this is not enough. Utilitarianism is not indifferent regarding the Euthyphronic contrast between the claim that an action is wrong because it fails to maximize utility, and the claim that it fails to maximize utility because it is wrong. So utilitarianism is best understood as a grounding claim, stating, roughly, that an action is wrong if and only if, and because, it fails to maximize utility. Utilitarianism says, in other words, that utility-suboptimality grounds wrongness.¹² Furthermore, this seems to be a statement of wrongness’ full ground. The utilitarian doesn’t think that alongside utility-suboptimality something else is needed—in the same kind of way—to ground wrongness. Rather, she thinks that utilitysuboptimality is the full story underlying wrongness. We can understand, then, utilitarianism as the thought that utility-suboptimality fully grounds wrongness. But if one thing fully grounds another, then no other thing is needed in order to ground it. In particular, if whenever an action is wrong, utility-suboptimality fully grounds its wrongness, then the principle of utility plays no role—at the very least, no non-redundant role—in grounding its wrongness. So the best understanding of the principle of utility has it saying of itself, as it were, that it does not play a (non-redundant) role in grounding the wrongness of wrong actions.¹³ And this means that the

(in the relevant natural facts), but are never fully grounded by anything—perhaps, that is, they can insist that the natural facts can be, in Fine’s (2012) terms, the moral facts “strict partial grounds.” Other than this footnote, I will not be discussing particularism in this paper. For some discussion that ties moral particularism to the grounding literature—in particular, to rejecting grounding necessitarianism—see Wygoda-Cohen (2018). ¹¹ What I call “Berker’s Challenge” is taken from his “The Explanatory Ambitions of Moral Principles” (Berker forthcoming). ¹² What if a utilitarian insists that she’s only endorsing the modal extensional equivalence? Then it seems she has to dig deeper. For why is it, we can ask her, that, as a matter of necessity, all and only utility-suboptimal actions are wrong? It seems like she needs an answer, and in giving it she will be giving what really is her ultimate moral theory. Then we can run the line in the text on that theory. ¹³ There may be other possible ways of interpreting the utilitarian principle (some of them are discussed by Berker), but I don’t find them plausible.

6

David Enoch

natural picture from the previous section is false. Similarly, of course, for other moral principles.¹⁴ If Berker is right, then, the natural picture is false. The best understanding of moral principles does not allow them—on pain of inconsistency with their very content—to play a role in grounding particular moral facts. Moral principles, on this picture, are roughly Humean—metaphysically idle generalizations that do not themselves do any work (or anyway, not this work).¹⁵ And because of the naturalness of the natural picture, this is a result you may be worried about even if you’re not a Robust Realist.¹⁶

1. 3 . H O W L A W G R O UN D S Earlier today, when on my way from Tel Aviv to Jerusalem I drove at a speed of 120 kph, I acted illegally. The fact that I acted illegally (or—the illegality of my action)¹⁷ is partly grounded in the non-legal but legally relevant facts (my speed, for instance), but this is not its full ground. To get to the full ground we also need to refer to the relevant legal norm. The fact that I acted illegally holds in virtue of the fact that I was going 120 together with the legal fact that the speed limit there is 110 kph. This, you will recall, is precisely the legal analogue of the natural picture above, and going legal doesn’t take anything from its initial plausibility. How should we understand, though, the relevant legal norm? What is its content? Surely, it doesn’t merely say that all cases of driving over 110 (in the relevant area) amount to violation. The legal norm (perhaps together with the best understanding of its role) is not indifferent between the thought that the action is illegal in virtue of being a case of driving over 110, ¹⁴ Berker uses the examples of utilitarianism and of a Rossian prima facie duty against promise breaking. ¹⁵ Of course, this is not a point about Hume. All I mean by Humean is what follows in the text. Berker considers some other options—including the partial-grounding option, and the possibility that moral principles are self-referential, mentioning themselves among the full ground. I don’t find any of these ways out plausible. ¹⁶ Clearly, then, Berker doesn’t beg any question against the Robust Realist. Berker doesn’t assume that principles don’t play a grounding role (so assuming would amount to begging the question against the Robust Realist). Rather, he starts with theory-neutral observations about the content of moral principles, and tries to infer from them that principles don’t play a grounding role. (And, as I note below, he thinks that even that is consistent with Robust Realism.) I thank an anonymous reader for some related comments. ¹⁷ I take it the relata of the grounding relation are facts, but for ease of exposition I’ll allow myself to speak as if properties can also serve this role.

How Principles Ground

7

and going in the opposite Euthyphronic direction. The most plausible reading of the legal norm is as a grounding claim. This seems plausible with regard to the speed limit example: It says, roughly, that in the relevant cases, driving over 110 fully grounds illegality.¹⁸ But if you’re not convinced, other examples may work better: Think, for instance, of a legal rule stating the conditions needed for the validity of a will (a signature, a date, witnesses, etc.), and assume the rule says that these together are sufficient for the validity of a will. Such a rule seems to say of itself that it is not a part of the ground of the validity of wills. Berker’s reasoning applies, and therefore so does his conclusion—legal norms are Humean generalizations, they do no serious metaphysical work in grounding particular legal facts, like that I acted illegally when I drove 120 on the way to Jerusalem. The (legal version of ) the natural picture is false. But this can’t be right. Of course the legal norms play a grounding role here. If someone wants to understand what it is that makes it the case that I acted illegally, and you don’t mention the legal norm, you’re not giving them the full story they’re after. Think of the fact that had we been in another jurisdiction, or had someone changed the law in a relevant way, my action would not have been illegal. We’re now in a position, then, to draw the following interim conclusion: Berker’s reasoning applies in the legal case just as it does in the moral case; in the legal case, though, the analogue of Berker’s conclusion is clearly false; so unless Berker can point to a relevant disanalogy between the moral and the legal case, we’ve learned that something has gone wrong in Berker’s reasoning. The remaining task is to see what—first, in the rest of this section, vis-à-vis the law, and then, in Section 1.4, regarding morality. Consider, then, the tension between the following two propositions: L N G: The full ground of a particular legal fact (like that I acted illegally when driving 120 on the way to Jerusalem) includes reference to the legal norm (like the fact that the speed limit there is 110). N G W: The full ground of the illegality of my action is my speed (and perhaps other, legally relevant but non-legal facts).

N G W seems to follow from the content of the legal norm itself. L N G seems like an uncompromising aspect of our commonsensical theory of legal facts. But the two seem inconsistent. Can they be reconciled?

¹⁸ The legal case may be much more complicated, but in ways that needn’t worry us here.

8

David Enoch

They can, if we are allowed to distinguish between different kinds of grounding, and disambiguate accordingly. Let’s distinguish then between metaphysical grounding and legal grounding. When asking about legal grounding, one is asking about an in-virtue-of claim that is internal, as it were, to the legal story or the legal domain (more on this shortly). Now, N G W, restricted to legal grounding, sounds very plausible indeed: N L G W: The full legal ground of the illegality of my action is my speed (and perhaps other, legally relevant but nonlegal facts).

And it is very plausible that N L G W does follow form the content of legal norms (perhaps together with a plausible understanding of their role). For legal norms are plausibly understood as full legal grounding claims—they state the full in-virtue-of story internal to the law. But because restricted to legal grounding, this is consistent with allowing the legal norm to play a metaphysical grounding role: L N M G: The full metaphysical grounding of a particular legal fact includes reference to the legal norm.

Because L N M G and N L G W are not even in tension with each other, if we can help ourselves to the distinction between these two kinds of grounding, we can respect both the fact that the legal norms are themselves statements of full (legal) grounding, and the fact that legal norms are not Humean generalizations. Notice how nicely this captures what we intuitively want to say here. The legal norm is part of what creates or underlies the relevant legal facts, so of course it’s relevant to the particular legal facts. But within the framework created by the specific legal norms all that’s needed for my driving to be illegal is for it to be at 120 kph. The full metaphysical explanation of what makes it the case that I’m acting illegally has to refer to the legal norm. But the internally legal story does not.¹⁹ Within that story, the speed suffices. Can we help ourselves to these different kinds of grounding relation? What’s to stop us, really? I’ve made no further assumptions about them, except that they are distinct—I haven’t, for instance, said anything about

¹⁹ Things are a little more complicated than this, as we may need to refer to specific legal norms even within the legal story. If so, we should be able to tell a more complicated story that includes also the hierarchy of legal norms within a legal system, and something resembling a Kelsenian Grundnorm that is not itself a part of the legal picture. But nothing of importance here depends on this complication.

How Principles Ground

9

them being independent of each other or any such metaphysically committed thing (though I return to this below).²⁰ The two notions—metaphysical and legal grounding—seem to make sense, and to capture genuinely different things we are genuinely interested in (perhaps in different contexts). Think of games, for example: Federer won the point. Why? Because his forehand caught the line. Or should we add—and because in tennis, on-the-line counts as “in”? Well, whether we need this as a part of the answer depends on what the question precisely is. If the question is for the full metaphysical story, then yes. If the question arises in a context in which we are allowed to assume,²¹ as it were, the tennis-framework, and what is sought is a tennisinternal in-virtue-of, then the story is complete without referring to the relevant tennis norm. We can distinguish, then—rather unproblematically—between different grounding claims, in the minimal sense of the distinction that’s (so far) needed. Unless positing them can be shown to lead to incoherence or otherwise unacceptable results, then, we are entitled to use them. And this means that we have a clean way out of the analogue of Berker’s challenge that applies to the legal case. Before getting back to the moral case, I want to mention a closely related, but still different, possible grounding picture. On the story I’ve been telling, the legal norm (or the rules of tennis) played a role in directly grounding the relevant particular legal (or tennis) fact. But there’s an alternative way of fleshing out the intuition that the legal norm sets the stage or the framework for the legal story and plays no role in it. According to this alternative picture, the only (relevant) thing that grounds the illegality of my action is my speed. The legal norm never stands in the grounding relation— metaphysical or otherwise—with the illegality of my action. It does grounding work, alright, but not that grounding work. Rather, what it grounds is the fact that driving 120 grounds illegality. Similarly, perhaps the rules of tennis do not ground, in any way, the fact that Federer won the point; it’s just that these rules ground the fact that the ball catching the line grounds ²⁰ Rosen (2017b: 286): “no one will be tempted to posit sui generis relations of legal grounding, musical grounding, orthographic grounding, and so on.” Myself, I’m okay with multiplying grounding relations; I’m not sure about the sui generis part. I need to concede a point here (following a helpful objection from Nevin Climenhaga): Even I will have limits as to how liberal to be in multiplying grounding kinds. I don’t want it to be the case, say, that that grass is green is the grassy-ground of the conjunctive fact that grass is green and snow is white, and that that snow is white is its snowy-ground. And what I have to concede is that I don’t have a good answer for the question how liberal is too liberal. Still, intuitive ideas about domains or some such can help. There’s no grassy domain in a sense analogous to that in which there is a moral or a legal domain. ²¹ But I don’t think the matter can be accommodated purely pragmatically. I return to this later in the text.

10

David Enoch

the fact that it was a winner. On this picture, then, we get iterated grounding claims: LegalNormG(SpeedGIllegality) [Compare: (LegalNorm&²²Speed) GIllegality] TennisRulesG(OnTheLineGWinner) [Compare: (TennisRules& OnTheLine)GWinner]

Notice that the two accounts share the ability to accommodate both the thought that the relevant “external” fact does some serious grounding work, and the fact that “internally” it does not. Which is the better account may vary from case to case—in the legal case, for instance, it seems implausible that the legal norm does not stand in the (partial, metaphysical) grounding relation directly with the particular legal fact. Perhaps it’s even possible for one and the same thing to play both roles. For now, though, let us just bear this other option in mind. We’ll revisit it regarding the case we’re most interested in here, namely, that of morality.

1. 4. BA CK TO MORA LITY Berker’s challenge can now be seen as highlighting the need to reconcile the seemingly irreconcilable: M P G: The full ground of a particular moral fact includes reference to a moral principle. N G W: The full ground of (e.g.) the wrongness of wrong actions are their (morally relevant) natural characteristics.

And the way out is to distinguish between moral and metaphysical grounding, and disambiguate accordingly: M P M G: The full metaphysical ground of a particular moral fact includes reference to a moral principle. N M G W: The full moral ground of the wrongness of wrong actions are their relevant natural (but morally relevant) characteristics.

Notice, again, how nicely this captures what we want to say. Within the moral story, as it were, the fact that by failing to give your grad student comments in a timely fashion you will be harming her progress and ²² Or, if we’re going to be picky about these things, feel free to replace the conjunction with the set of the conjuncts.

How Principles Ground

11

frustrating her legitimate expectations suffices for wrongness. And what the promise-keeping principle says is that the fact that I promised the workshop organizer the paper by the end of August suffices for my moral duty to send him the paper by that date. But of course, that does not mean that the framework is somehow grounding-irrelevant. It’s relevant, because it’s a part of the fuller story, the more external story, of the grounding of these moral duties.²³ On this suggestion, then, Berker is right that the content of moral principles includes a grounding claim. He’s even right that it often includes a full grounding claim. On this suggestion, utilitarianism is understood (roughly) as the claim that utility-suboptimality morally fully grounds wrongness.²⁴ It’s just that the content of the moral principle, being a moral principle, is silent on metaphysical grounding. And why would you expect a moral principle to say something about that? We should now revisit the iterated grounding picture, according to which—now applied to the moral case—the moral principles do not themselves stand in any grounding relation to particular moral facts. Rather, they ground the fact that the relevant natural facts ground the particular moral fact. Semi-formally, again: MoralPrincipleG(NaturalFactGParticularMoralFact) [Compare: (MoralPrinciple&NaturalFact)GParticularMoralFact]

And in particular: OughtKeepPromisesG(PromisedGOught) [Compare: (OughtKeepPromises&Promised)GOught]

And here too, this story also nicely seems to capture what we want, such as the distinction between what goes on within the moral story (once the framework is in place, as it were), and what is a part of the external story that includes the internal one as a part of it.²⁵ How, then, are we to decide between the two?²⁶ And how bad is it if we cannot? ²³ Fine (2012: 77) notes that the plausible thought is that the normative is normatively, not metaphysically grounded in the natural. ²⁴ See Rosen (2017a: 155) where his last attempt at capturing the form of moral principles is a formal version of my point in the text here. ²⁵ At least one way of reading Shafer-Landau is as opting for this kind of a story. See Leary’s (2017: 90) discussion of “Lawfully Grounded Non-Naturalism.” Berker too considers this option, and rejects it for reasons closely related to those discussed in Section 1.5. ²⁶ Perhaps the iterated grounding view counts as naturalist according to the stipulated characterization above (for according to the iterated grounding view only natural facts stand in direct grounding relations with moral facts), but because of the grounding role moral principles play even on that view, it seems to fit only the letter of the

12

David Enoch

At times, we have direct intuitions about in-virtue-of claims, so it may be a good idea to consult these, and see if they can drive a wedge between the grounding-pluralist picture and the iterated-grounding one. But I’m not sure this can work: True, to the extent that we have intuitions about these things, I do have the intuition that “I have a duty to send in the paper on time partly in virtue of the promise-keeping principle” is true, and perhaps this can serve as some support for the grounding-pluralism picture. But I’m really not sure that whatever intuitions we do have about such in-virtue-of claims are sufficiently fine-grained to distinguish between the two stories. Perhaps a fair characterization of the intuition is just that the moral principle is somehow grounding-relevant to the particular moral fact. But that, of course, is accommodated by the iterated-grounding story as well. If I (somewhat hesitatingly) prefer the grounding-pluralist picture over the iterated-grounding one,²⁷ it’s for methodological reasons: As noted in Section 1.3 (and as further defended in Section 1.7), there is no significant theoretical price to positing many different kinds of grounding relations. Furthermore, in all likelihood even if you go for the iterated-grounding picture, you’ll still need also to go pluralist on grounding—either because of other places where such pluralism seems to be doing work (see Section 1.7), or perhaps even as you give more details regarding your MoralPrincipleG (NaturalFactGParticularMoralFact)—notice that the “G”s here too may have to be disambiguated. So it’s not as if there are clear advantages to going with the iterated-grounding picture. The iterated-grounding picture takes on board whatever apparatus the grounding-pluralism picture utilizes, and then adds the complexity of grounding iteration. Unless it can be seen to be more loyal to how we intuitively think of the grounding relations here—and above I suggested that we don’t have clear intuitive support for this—then perhaps it’s preferable to stick to the somewhat simpler grounding-pluralist picture. And this is what I will do in the rest of the paper—though I’m really not sure that anything significant hangs on this.

1. 5 . A N OTH ER R OU ND The way to save the natural picture of the grounding role of moral principles proceeds, I’ve suggested, via distinguishing different grounding relations. Berker anticipates this move, and offers reasons for rejecting it. characterization of naturalism, not its spirit. In other words, perhaps this shows a way in which the characterization above needs precisifying. ²⁷ Regarding morality, that is. As I noted in Section 1.3, which is the better way to go may vary between domains.

How Principles Ground

13

Berker offers a dilemma, having to do with the relation between the two posited grounding relations—metaphysical and moral grounding, in our case. The two are either in some way inter-definable, or they are not. If they are, then there’s some reason to suspect that no progress has been made—we can restate everything in terms of just one grounding relation (having replaced each definiendum with its definians), reaching undesirable circularity or regress territory. If, on the other hand, they are not inter-definable, then the grounding pluralist relies, in his attempt to save the natural picture we started with, on a mysterious, unexplained connection between these two metaphysically distinct grounding relations. Why should the two be related in such a way, asks Berker, to secure the natural picture of the role of principles? Is this supposed to be just some coincidence, perhaps an especially fortunate one for the Robust Realist? Wouldn’t such coincidence be too much to believe? I have some doubts regarding the details of Berker’s discussion of circularity in the first horn of the dilemma, but let me put them to one side (for now), because I want to focus on the second. Accept, then, that moral and metaphysical grounding are not inter-definable. Are we stuck with an unacceptable—or even just theoretically expensive—mystery? Lurking in the background here is Kit Fine’s pluralism about both modality and grounding.²⁸ Fine believes that there are three kinds of necessity—metaphysical, natural, and normative—and that neither of these is reducible to any other; all three are needed for a full account of the modal universe. Somewhat derivatively, he believes that there are three kinds of grounding relation—the same three kinds—of which none is reducible to the other. When Berker thinks of what he calls “the grounding pluralism gambit,” this is the kind of view he has in mind. But recall that I was being, at least initially, much less ambitious. The crucial thing for me was nothing about non-inter-definability or irreducibility. I merely insisted on the phenomena, at a pre-theoretical stage. At that stage, I insisted, we can and do distinguish between the within-tennis in-virtue-of relations and those that are somehow external to tennis (even if the relevant grounded fact is a tennis fact); and between the intra-legal grounding relations and the extra-legal ones (even when the grounded fact is a legal one). I could afford to be so liberal on the number of different grounding relations (even more than three!) partly because—so far—I could afford to remain noncommittal on the precise relations between them. Perhaps some really are independent of some others; perhaps some—further down the theoretical road, as it

²⁸ Fine (2002; 2012).

14

David Enoch

were—can be reduced to others.²⁹ All of this remains to be seen. At the outset, though, a rather liberal grounding pluralism seems unobjectionable, and it is this kind of pluralism that got us out of Berker’s challenge. Berker may insist, though, that his point is precisely that if such response is to succeed, there must be metaphysical constraints further down the road. And I want to acknowledge that—but to insist that such further constraints need not be in terms of inter-definability, or any terms that would lend us in a problematic way on the first horn of Berker’s dilemma. Consider the law again. I suggested that particular legal facts are fully legally grounded in the legally relevant non-legal facts, whereas they are partly metaphysically grounded also in the relevant legal norms. Is there some deep metaphysical mystery about the relation between legal grounding and metaphysical grounding here? The kind of mystery that can only be avoided by insisting that, say, legal grounding can be defined in terms of metaphysical grounding? I don’t see why. It’s quite sensible to ask for more by way of explanation here—a further explanatory layer, as it were—but not necessarily in terms of inter-definability. Perhaps an explanation in terms of essence can do the work:³⁰ For it’s a very natural thought that it lies in the nature of the legal that particular legal facts are grounded in the conjunction of legal norms and non-legal facts.³¹ Furthermore, it seems to lie in the nature of the legal grounding relation that it is related to the metaphysical grounding relation in something like the following way: When a legal fact is metaphysically grounded in a legal norm together with legally relevant nonlegal facts, it is fully legally grounded in the legally relevant non-legal facts alone. This is just what the legal grounding relation is, one is tempted to say. And none of this seems at all mysterious. Perhaps the relation between legal and metaphysical grounding on this picture is not one of inter-definability (though I get back to this shortly), but it is not brute either: It is given a plausible metaphysical explanation in terms of the nature of the relevant kinds of fact and the relevant grounding relations. Similarly for the moral, I would say. It lies in the nature of the moral that (particular) moral facts are metaphysically grounded in the conjunction of moral principles and morally relevant natural facts.³² And it lies in the nature of the moral grounding relation that when a moral fact is ²⁹ I think that a fairly strong legal positivist view that accepts a reduction of the legal to the social can generate optimism about the possibility, in the legal case, of a reduction of the kind mentioned in the text. ³⁰ Berker criticizes the attempt to offer essentialist replies to his challenge, but ones that are importantly different from the ones in the text. ³¹ For similar claims, see Rosen (2017b: 7). ³² “What may be plausible, though, is that it should lie in the nature of goodness, say, that it should have some ground in what is natural . . . ” (Fine 2012: 77).

How Principles Ground

15

metaphysically grounded in the conjunction of a moral principle and a set of morally relevant natural facts, it is fully morally grounded in the relevant set of natural facts alone. And here too, I don’t see inter-definability, but I don’t see bruteness or unacceptable mystery either. Two objections come to mind. The first is that there’s always something suspicious—and rarely something explanatory—about references to essences.³³ I confess some sympathy to this general suspicion—“essence,” after all, is not some magic problem-solving word (nor are “lies in the nature of”). I want to respond, then, in a somewhat concessive spirit: To the extent that there’s a mystery in the first place about the relation between metaphysical and moral (or legal) grounding, such plausible thoughts about essence and what lies in the nature of the moral and of moral grounding solve it. To the extent that they amount merely to a rewording of the things already said before introducing essence-talk and lies-in-the-nature-of-talk, this just shows that no further explanatory layer was called for to begin with here.³⁴ The second objection is that while I did not use the words “definition” or “inter-definability,” the picture I’ve been painting is precisely one of interdefinability. For haven’t I in effect offered a definition of the moral (or legal) grounding relation in terms of metaphysical grounding? The point cannot be about these words, of course; the question must be whether the kind of essential relation between moral grounding and metaphysical grounding here invoked lands us on the other horn of Berker’s dilemma. And it does not. For Berker’s worry about an inter-definable plurality of grounding relations is that such inter-definability will introduce metaphysical circularity into the account: We will have, Berker worries, moral principles that are supposed to do serious metaphysical grounding work, the content of which refers to moral grounding, itself defined in terms of metaphysical grounding—and all circular hell will break loose. But for this problem to arise, the relation between moral and metaphysical grounding has to be one of reduction, or identity, or definition. And I have made no such assumptions. I do think it lies in the nature of the moral grounding relation that it’s related to metaphysical grounding in the way I suggested—but this in no way allows a replacement of the reference to moral grounding in the content of moral principles with something about metaphysical grounding. That relation between the two kinds of grounding—that it lies in the nature of moral grounding that, roughly, the moral principles needed to do ³³ Berker expresses a similar worry at the point where he gets closest to considering my line in this section. ³⁴ For positing essence-facts as explaining grounding-facts, see Rosen (2017a: 149; 156). Rosen also suggests that the relations to metaphysical grounding may lie in the nature of the normative grounding relation.

16

David Enoch

metaphysical grounding work don’t do moral grounding work—is one that I’m happy to accept as brute.³⁵

1. 6. W H E N A R E H U M E A N LA W S EN O U G H? If I am right, then, Berker’s case for going Humean regarding moral principles fails. For anything thus far said, at least, the natural picture, with the grounding role it assigns to moral principles, can stand. But it may be worth our while to reconsider the Humean option. How bad on reflection is it if moral principles turn out to be merely Humean generalizations? One way of thinking about things here comes from more examples.³⁶ In some domains, it seems very hard to accept a Humean picture—think of the law, again. In others it seems like the obvious way to go. Here’s an example from metaphysics: Conjunctions, it seems pretty safe to assume, are grounded in their conjuncts. This is a generalization, but it seems Humean: The fact it’s summer and I’m writing a metaethics paper is grounded in the fact that it’s summer and in the fact that I’m writing a metaethics paper, but not also in the generalization that conjunctions are grounded in their conjuncts. So the question about morality may be thought of as the question whether moral principles, in the relevant way, are more like legal ones or like metaphysical ones.³⁷ And it may be helpful to think of further examples as well: Do epistemological principles do grounding work, or are they best seen as Humean generalizations? How about esthetic ones? Logical ones?³⁸ The reason this discussion may be relevant in our context is that I let the implausibility of Berker’s conclusion—or the naturalness of the natural picture I started with—play a role in how well all I’ve been saying fits together. If there’s no intuitive pressure to recognize a grounding role for moral principles, there’s less justification to go for grounding pluralism ³⁵ Nor will it follow from this picture that moral principles are themselves—as Berker says when fleshing out the circularity horn of his dilemma against grounding pluralism— grounded in facts about metaphysical grounding. Recall that on the natural picture I am trying to vindicate, the basic moral principles are ungrounded. ³⁶ See Rosen (2017a: 139) again (he uses disjunction as an example). In his (2017b) Rosen mentions other examples where Humeanism doesn’t seem at all plausible— perhaps mostly, cases of merely formal normativity, a generalization of the legal example. ³⁷ It’s possible, of course, that only some generalizations in metaphysics are best understood as Humean. ³⁸ I am avoiding laws of nature. The problem is that their nature and status are, obviously, very controversial, and I just don’t know these discussions well enough. But let me note that a fuller discussion would have to include them as well. For some such discussion (in a related context), see Elliott (2014).

How Principles Ground

17

(though see Section 1.7), and perhaps less justification for the entire Robust Realism picture that includes the natural picture as a part.³⁹ And of course, if we can come up with a criterion—when do generalizations do serious grounding work, and when are they best understood as Humean?—this will be of independent interest. What’s the criterion, then, such that the legal satisfies it and the metaphysical does not? And does morality? (And do these other domains?). I don’t think it’s plausible to think that the relevant phenomenon is entirely explainable in pragmatic terms. True, the conversational context does affect which principles have to be mentioned in an explanation or justification or argument; but the intuition that some rules or principles (as in the legal case) do serious grounding work survives even in a context in which the audience may take the relevant rules or principles for granted. One possible suggestion is modal. Recall that in the legal case, when pushing the intuition that the legally relevant non-legal facts (that I was going 120 kph) do not suffice as the grounds for the legal fact (that I acted illegally), I drew attention to other jurisdictions, or to the possibility of changing the law in the same jurisdiction. That is, I referred to the law’s contingency. Now, if the generalizations or principles or norms in a given domain are (metaphysically) contingent, then it is very natural to think that they’re playing a role in grounding the particular facts that are sensitive to this contingency (as in the legal case).⁴⁰ Metaphysical laws—like the one about the grounding of conjunctions—are presumably absolutely necessary, and so there’s no need for them to do grounding work. So the modal distinction gets right at least these two paradigmatic cases of Humean and of non-Humean rules.⁴¹ What does this criterion have to say about moral principles? The answer will depend on whether moral principles are (metaphysically) necessary. And this is a matter of rather heated debate. Myself, I used to think moral principles are metaphysically necessary; if this is so, and if we have here the ³⁹ Berker insists that his conclusion does not refute Robust (meta-normative) Realism, and I agree. But Berker suggests that almost any metanormative view is an instance of Robust Realism. This sounds to me—if true—not as good news for Robust Realism, but as bad news for the metanormative debate (or, perhaps, for the reasoning that leads Berker to this conclusion). To me it seems that the rejection of the natural picture—called for by Berker’s reasoning, if no reply works—renders Robust Realism significantly weaker. ⁴⁰ Perhaps this doesn’t have to be so—perhaps one can still think of the principles as indicating which world is relevant, not as making it the case that one world (or jurisdiction) is relevant. But the picture in the text—even if not strictly speaking required—is still a very natural one. ⁴¹ Also—I think that in the natural laws context, there is such a close connection between the discussion of Humeanism and the discussion of natural laws’ modal status. See Carroll (2016: 9 and the references there).

18

David Enoch

right criterion that distinguishes grounding principles from Humean ones, it turns out that Berker is right after all. But I am no longer sure about the metaphysical necessity of moral principles,⁴² and regardless, I want to offer a somewhat different criterion for our distinction, one that subsumes the modal one as a particular instance. In arguing—to the extent possible—against naturalist reductions of the normative, I end up relying on what I call “the just too different” intuition—normative properties and facts, I claim, seem too different from natural ones to be identical to some of them.⁴³ At least, I argue, this intuition raises the bar that a specific naturalist reduction will have to pass in order to be acceptable. I want to return to this intuition here. Principles (or some such) have to play a grounding role when the other candidate grounds are just too different from the grounded. This is why no such role is needed in the metaphysical case. The fact that it’s summer and I’m writing a metatehics paper is not of an entirely different kind from the fact that it’s summer, and the conjunctive fact is (almost⁴⁴) nothing over and above the conjunct facts. Nothing like the just-too-different intuition applies, and so the generalizations may be Humean. In the legal case, things are a bit tricky—whether the fact that I acted illegally is just too different from the fact that I drove 120 kph is presumably a matter of jurisprudential controversy, one that fortunately we can bypass here.⁴⁵ For we can note that even if legal facts are natural facts, they are different natural facts, and that anyway—because of the contingency noted above—more needs to be in place for me to have driven illegally than merely my having driven at 120 kph. So here too it would be false to say that the legal fact is nothing over and above the non-legal (but legally relevant) one. So we get the legal case— where going Humean seems unacceptable—because here too, in a somewhat different way, the underlying facts do not suffice for the full needed grounding work. And this is the criterion I am offering, then: Principles (rules, generalizations) do grounding work when the rest is not enough to do all the grounding work that needs doing, either because of contingency, or because of a just-too-different intuition (or perhaps for other reasons). And so, perhaps we are back with the gods metaphor: When the gods created the

⁴² Of course, they are morally or perhaps normatively necessary, but this is a different matter. See Fine (2012: 38–40). ⁴³ TMS (100–5). ⁴⁴ Rosen (2017b) repeatedly characterizes the thought that one thing is grounded in another as the thought that it’s almost nothing over and above it. (Grounding’s irreflexivity precludes the grounded being strictly speaking nothing over and above what grounds it.) ⁴⁵ For my view in this debate, see my “Reason-Giving and the Law” (2011b), and for my view of this debate see my “Is General Jurisprudence Interesting?” (2019).

How Principles Ground

19

conjunct-facts, they were already done with creating the conjunction-fact as well. Not so in the legal and moral cases. With this (imprecise) criterion in mind, whether Berker’s conclusion about moral principles can be accepted depends on such things as whether moral facts are just too different from natural ones. And, well, you know where I stand on this. (It would be an interesting exercise to check regarding the other mentioned domains.)

1 .7 . O T H E R M E T A E T H I C AL P A Y O F F S OF GROUNDING PLURALISM I want to conclude by hinting at some other payoffs that grounding pluralism has in moral philosophy. Showing such payoffs is one way to guard against the accusation that invoking pluralism is objectionably ad hoc, a desperate step whose only motivation is to save Robust Realism and the natural picture of principles from Berker’s challenge. It is also, quite simply, a way of arguing for grounding pluralism:⁴⁶ For, to paraphrase David Lewis (1986: vii), if there are many ways in which systematic philosophy, and in particular, metaethics, goes more easily if we may presuppose grounding pluralism, I take this to be a good reason to think that grounding pluralism is true. Obviously, then, this is going to be very quick and so merely suggestive. Anyway, here goes. Do our reasons depend on our desires, preferences, and the like? Well, it’s very hard to deny that sometimes, what one has reason to do at least partly depends on what one wants. I have a reason to order the pizza at least partly in virtue of my wanting to eat something, well, pizza-ish. You have a reason to go to the party at least partly in virtue of how much you love dancing (and the fact that there’s going to be dancing at the party).⁴⁷ Do such mundane observations suffice to refute all response-independence meta-normative views, views according to which normative facts are not grounded in our attitudes and responses? That would be awkward. A lot can ⁴⁶ There are other ways—most notably, perhaps, Fine’s. But because I don’t agree with all of the details in Fine’s arguments, and because discussing them in detail will take too much time and space, I will not rely on them here. In the opposite direction, a full account would have to include a refutation of arguments against pluralism, most notably, perhaps, Berker 2018. But I cannot here discuss these arguments. Let me just note that the suggestion above according to which normative and metaphysical grounding, though distinct, are closely related (for it lies in the nature of moral grounding . . . ) is the first step to accommodating Berker’s examples of cross-groundings transitivity cases. ⁴⁷ Schroeder’s (2007) lead example.

20

David Enoch

be said here, of course, but one very natural thing to say is that if we can help ourselves to grounding pluralism, the problem disappears. A responseindependent theorist will insist that normative facts are not metaphysically grounded in our responses (not fully, that is, and when partially, only together with or in virtue of a normative principle; and no fundamental moral principle is grounded in them at all). This leaves open the possibility that some normative facts are normatively grounded in our desires.⁴⁸ This also nicely saves a difference between two types of view that may seem similar: The first is a subjectivist theory according to which, roughly, to have a reason to do something just is to have some relevant desire; and the second is a Robust Realist theory about the nature of reasons, except one that on normative grounds accepts that the only thing we have reason to do is (roughly) to satisfy our own desires. We can now see how the two theories differ—both ground all your reasons in your desires, but the former does this metaphysically, and the latter normatively.⁴⁹ Discussions of intrinsic value are sometimes confused because of a failure to distinguish the different grounding relations. It is sometimes said that something is of intrinsic value iff its value is not grounded in its relation to anything else. Then, if you are a response-dependence theorist, you find yourself committed to the view that nothing has intrinsic value. And you may find it an unwelcome result. At least, you may think—as do I—that the two discussions (the metaethical one about response-dependence and the value-theoretical one about which things are good, as it were, in their own right) are distinct discussions. Grounding pluralism gives us the resources to distinguish between them in a helpful way. One discussion—the metaethical one—is about the metaphysical grounding of (for instance) value. Another discussion is a first-order one, about the normative or evaluative grounding of values—when we ask, say, whether autonomy is of intrinsic (or sometimes ultimate or final) value, we ask whether it’s normatively grounded in other values, or whether it’s normatively ungrounded (even if metaphysically it is).⁵⁰

⁴⁸ This doesn’t mean that the mundane examples are irrelevant. The responseindependence theorist may still owe us an explanation of such examples. This is the spirit with which Schroeder introduces his example. ⁴⁹ So I agree with Berker 2018 when he claims that the unity of grounding means the collapse of the distinction between ethics and metaethics. But I save the latter by rejecting the former. I do a little more with the observation in the text—in the context of criticizing David Sobel’s subjectivism—in my (2018). ⁵⁰ For a discussion along these lines, and for an attempt to discipline the use of the relevant terms (intrinsic value, final value) in a way that will preempt further confusion, see Rabinowicz and Rønnow-Rasmussen (2000).

How Principles Ground

21

Lastly—and in a very sketchy way: Quasi-realists have been insisting that they are not committed to response-dependence in any way. They reject such counterfactuals as “Had we been in favor of bullfighting, bullfighting would have no longer been wrong,” by insisting—correctly—that what on their view (and any other sensible view) makes bullfighting wrong is the suffering of the bulls, not anything about how we view it. This line always seemed to me like cheating, though it has proved remarkably hard to show exactly why. But now we can: Sure, what normatively makes bullfighting wrong (or what normatively grounds its wrongness) is something about the bull’s suffering, and the quasi-realist can accept this with the rest of us. But what if we ask about the metaphysical grounds of such wrongness?⁵¹ Here, the fact that the quasi-realist can (purportedly) do first-order ethics with the rest of us won’t help. Here he must either say what metaphysically grounds such normative truths (and if not our responses, on his account, what does?), or he has to claim that nothing does (are basic normative facts ungrounded? Or are they autonomous?⁵² Something else?); or perhaps he can find a flaw in the question he’s being asked now; and so on. The point is that he has to say something in response to the question about metaphysical grounding. Merely repeating sensible things about normative grounding won’t do. And, let me predict, there are going to be theoretical costs whichever way the quasi-realist goes here.⁵³

References Berker, S. 2018. “The Unity of Grounding,” Mind 127, 729–77. Berker, S. Forthcoming. “The Explanatory Ambitions of Moral Principles,” Noûs. Carroll, J. W. 2016. “Laws of Nature,” The Stanford Encyclopedia of Philosophy (Fall 2016 edition), Edward N. Zalta (ed.), https://plato.stanford.edu/archives/ fall2016/entries/laws-of-nature/. ⁵¹ Quasi-realists insist that they don’t hear any other understanding of the counterfactual than the one they can reject on normative considerations. The point about grounding pluralism in the text can be seen as saying more about the sense of such counterfactuals they insist on not hearing. ⁵² In Dasgupta’s (2014) sense. ⁵³ Previous versions of this paper have been given as talks in Frankfurt, Uppsala, Zurich, Sheffield, Notre Dame, Jerusalem, Prague, and the Madison Metaethics Workshop. I thank the audience members for the helpful discussions. For comments and relevant conversations, I thank Dan Baras, Jonathan Barker, Selim Berker, Paul Bloomfield, Annette Bryson, Daniel Fogal, Nils Franzén, Ori Herstein, Andrew Kernohan, Stephanie Leary, Tristram McPherson, Ofer Malcai, Kate Nolfi, Michael Pendlebury, Henrik Rydéhn, Geoff Sayre-McCord, Aaron Segal, Jonathan Shaheen, Alex Silk, Pekka Väyrynen, Preston Werner, and anonymous readers for OUP. My research was supported by the Israel Science Foundation, Grant 439/15.

22

David Enoch

Dasgupta, S. 2014. “The Possibility of Physicalism,” The Journal of Philosophy 111 (9–10): 557–92. Elliott, A. 2014. “Can Moral Principles Explain Supervenience?” Res Philosophica 91 (4): 629–59. Enoch, D. 2011a. Taking Morality Seriously: A Defense of Robust Realism. Oxford: Oxford University Press. Enoch, D. 2011b. “Reason-Giving and the Law,” Oxford Studies in the Philosophy of Law, vol. 1: 1–38. Enoch, D. 2018. “Review of David Sobel, From Valuing to Value: A Defense of Subjectivism,” Ethics 128 (3): 672–7. Enoch, D. 2019. “Is General Jurisprudence Interesting?” in David Plunkett, Scott Shapiro, and Kevin Toh (eds.), Dimensions of Normativity: New Essays on Metaethics and Jurisprudence. Oxford: Oxford University Press; 65–86. Fine, K. 2002. “Varieties of Necessity,” in John Hawthorne and Tamar Szabó Gendler (eds.), Conceivability and Possibility. Oxford: Oxford University Press, 253–82. Fine, K. 2012. “Guide to Ground,” in Fabrice Correia and Benjamin Schnieder (eds.) Metaphysical Grounding: Understanding the Structure of Reality. Cambridge: Cambridge University Press, 37–80. Leary, S. 2017. “Non-Naturalism and Natural Necessities,” Oxford Studies in Metaethics, vol. 12: 76–105. Lewis, D. 1986. On The Plurality of Worlds. Oxford: Blackwell. Rabinowicz, W. and Rønnow-Rasmussen, T. 2000. “A Distinction in Value: Intrinsic and for its own Sake,” Proceedings of the Aristotelian Society 100 (1): 33–51. Rosen, G. 2017a. “What Is a Moral Law?” Oxford Studies in Metaethics, vol. 12: 135–59. Rosen, G. 2017b. “Ground by Law,” Philosophical Issues 27 (1): 279–301. Schroeder, M. 2007. Slaves of the Passions. Oxford: Oxford University Press. Shafer-Landau, R. 2003. Moral Realism: A Defence. Oxford and New York: Oxford University Press. Väyrynen P. 2013. “Normativity and Grounding,” Proceedings of the Aristotelian Society Supplementary Volume 87: 155–78. Wygoda-Cohen, S. 2018. “Not All Partial Grounds Partly Ground,” Philosophy and Phenomenological Research, https://onlinelibrary.wiley.com/doi/pdf/10.1111/phpr. 12524.

2 The Self-Undermining Arguments from Disagreement Eric Sampson 2.1. INTRODUCTION Moral disagreement is pervasive among both philosophers and nonphilosophers. Perhaps it’s no surprise that non-philosophers disagree about ethics, but widespread, intractable moral disagreement among philosophers is more striking (and more troubling). Aren’t moral philosophers supposed to be the experts—maybe not with respect to being moral but at least with respect to knowing the moral truths? What do ethicists do at their jobs all day if they’re not discovering, or making progress toward discovering, the moral truths? And why, after several thousand years of ethical inquiry, have they not been able to make any progress toward consensus? It seems philosophers are no closer to reaching agreement about ethics today than they were when Socrates walked the streets of Athens. Indeed, dissensus seems to have grown since then. These reflections can cause one to have doubts about morality. One might doubt, first, that many of our moral beliefs are epistemically justified: If the people who devote their lives to ethical inquiry can’t come to agreement about ethics, then how can we (or they) be confident that our (or their) moral beliefs are true? One might even doubt that there are any moral truths at all. After all, one possible explanation for the pervasiveness and intractability of moral disagreement among moral philosophers is that there are no moral truths about which to agree. Perhaps disputes about how we ought to live are no better than those disputes in ancient times about whether the moon is male or female—there’s just no fact of the matter. These kinds of doubts find their expression in arguments from disagreement in metaethics.

24

Eric Sampson

Arguments from disagreement are typically aimed at non-skeptical moral realism.¹ As I will understand this view, it comprises four claims: C: moral judgments express beliefs that aim to represent moral reality, N-: some of these positive moral judgments are true,² S-: the moral truths do not constitutively depend on the attitudes of any actual or hypothetical agent(s), and N-: we are justified in believing some moral propositions.

While arguments from disagreement might be employed to target other views, such as constructivism or expressivism, I will only be interested in considering the challenge these arguments pose for non-skeptical realism.³ Arguments from disagreement begin by calling attention to (or supposing) widespread, fundamental moral disagreement among a certain group of people (e.g., the folk, moral philosophers, idealized agents). Then, some skeptical or anti-realist-friendly conclusion is drawn. These conclusions come in three varieties: epistemological (e.g., we don’t know any moral propositions), metaphysical (e.g., there are no moral truths), or semantic (e.g., moral judgments express non-cognitive attitudes). This paper focuses only on the first two kinds of arguments. Moreover, it focuses on arguments that take as their starting point a particularly troubling kind of disagreement: disagreement among excellent moral inquirers. For many years now, moral realists have argued that we shouldn’t be terribly concerned about moral disagreement in society, or among “the folk.” Moral disagreements in society (or across societies), realists often say, are the product either of ignorance about the non-moral facts, genuine moral indeterminacy, failures of imagination and sympathy, the distorting influences of self-interest, or some other kind of cognitive bias or shortcoming. The standard realist line, then, is that moral disagreement in society persists because most people are bad (or less-than-ideal) moral inquirers.⁴ If this is correct, then the more troubling kind of disagreement for nonskeptical realists is the kind we see among moral philosophers or the kind we might expect to see between perfectly informed, perfectly rational agents. ¹ I’ll call it “non-skeptical realism” or just “realism” from now on. ² A positive moral judgment is one whose truth entails either that it would be good (or bad) for an agent to do something, that an agent morally ought (or ought not) to do something, that an agent has moral reason to do (or not to do) something, or something similarly unacceptable to moral error theorists. ³ Much of what I say in defense of non-skeptical realism can, however, be adopted by expressivists and constructivists. ⁴ This popular line is taken by Brink (1989: ch. 7), Enoch (2011: ch. 8), and ShaferLandau (2003: ch. 9).

The Self-Undermining Arguments from Disagreement

25

After all, moral philosophers have more time than almost anyone to investigate the morally relevant empirical facts, try to rid themselves of bias, try to imagine the plight of others, and so on. And ideally informed, ideally rational agents wouldn’t be subject to any of the cognitive shortcomings to which realists often appeal to explain disagreement in society. So, if disagreement in these ideal (or nearly ideal) conditions persists, then realists cannot respond to concerns about moral disagreement in the standard way. When realists do respond to arguments from disagreement, those responses tend to be piecemeal.⁵ They take each argument from disagreement one at a time and try to respond accordingly. In this paper, however, I argue that arguments from disagreement share a structure that makes them vulnerable to a single, powerful objection: they self-undermine. For each formulation of the argument from disagreement, at least one of its premises casts doubt either on itself or on one of the other premises employed on the way to the argument’s skeptical or anti-realist conclusion. Upon reflection, this should not be surprising. Arguments from disagreement are philosophical arguments. They therefore employ philosophical— specifically metaphysical and epistemological—premises to support their conclusions. But deep, widespread, intractable disagreement about philosophy (especially metaphysics and epistemology) is pervasive, both among philosophers and non-philosophers.⁶ Thus, any non-trivial metaphysical or epistemological premise an argument from disagreement may employ is likely to be the subject of deep, widespread, intractable disagreement. And if the existence of such disagreement about a moral proposition casts doubt on that proposition or on our moral beliefs more generally, as arguments from disagreement purport to show, then, by those arguments’ own standards, such disagreements cast doubt on those arguments’ own controversial premises. By those arguments’ own standards, then, they fail.⁷

⁵ Again, this is the strategy employed by Brink, Enoch, and Shafer-Landau in their defenses of realism. I think that their responses are largely correct. Thus, everything I say in this paper is meant to be compatible with those defenses. ⁶ Shafer-Landau makes this point forcefully in (2003: 220). He then argues that arguments from disagreement for anti-realism in ethics overgeneralize, committing their proponents (implausibly) to anti-realism about philosophy. I suspect that some arguments from disagreement do fall to this objection, but not the ones I consider in this paper. ⁷ Enoch (2011: 215–16) raises a self-defeat worry, too. He contends that arguments from disagreement seek to establish a metaethical position. But, given the level of disagreement in metaethics, arguments attempting to show that there is no fact of the matter about ethics would also show that there’s no fact of the matter about metaethics. And one cannot coherently defend an argument for a metaethical position if that argument also supports the conclusion that there is no fact of the matter about metaethics. Again, there are likely some arguments that fall to this objection, but not the ones I consider here.

26

Eric Sampson

I don’t have space to consider every (metaphysical and epistemological) argument from disagreement. I will therefore show how the selfundermining objection applies to some of the most recent and most powerful arguments, which I take to be representative of many others. I’ll consider two kinds of epistemological arguments and two kinds of metaphysical arguments. In each case, I argue, these arguments employ premises that are highly controversial, and in precisely the same ways that the moral propositions they purport to undermine are controversial. The result is that these arguments undermine themselves. Once I’ve shown how the selfundermining objection applies to the arguments I consider in this paper, it should be clear how the objection will apply to other arguments from disagreement not discussed here. If the argument presented in this paper is sound, it provides realists with a single, unified strategy for responding to arguments from disagreement. Moreover, it provides a challenge for any future arguments from disagreement: any future argument must rely on premises that are not themselves controversial in the way the moral propositions they seek to undermine are. But, again, given the pervasiveness of disagreement among both philosophers and non-philosophers about metaphysics and epistemology, this will be an incredibly difficult task. Thus, I conclude, realists should not be concerned about existing arguments from disagreement, and they should rest assured that any future arguments will have an extraordinarily high bar to meet.

2 .2 . AR G UM E N T S F R O M C O N C I L I A T I O N I S M The first kind of argument I’ll consider draws on the peer disagreement literature. The central question in that literature is this: How should we revise our doxastic attitude(s) toward a proposition p when we discover that we disagree about p with an epistemic peer—i.e., someone we judge to be just as intelligent, well-informed, open-minded, and otherwise epistemically virtuous with respect to p as we are? One popular view is C: If an agent A learns that an apparent epistemic peer, B, disagrees with her about p, then A is rationally required either to suspend judgment about p, adjust her credence in p significantly in the direction of B’s, or both.⁸, ⁹ ⁸ For simplicity, I will mostly speak in terms of full (dis)belief, rather than credences. ⁹ Defenders of Conciliationism include Bogardus (2009), Christensen (2007; 2009; 2013), Elga (2007, 2010), Feldman (2007), Frances (2010), Kornblith (2010), Matheson (2009; 2016), Pittard (2015), and Vavova (2014).

The Self-Undermining Arguments from Disagreement

27

If Conciliationism is the correct view about how to respond to peer disagreement, then it has clear implications for our moral views. Given that many of our moral beliefs are the subject of disagreement among excellent moral philosophers (some of them our epistemic peers or superiors), Conciliationism implies that we are rationally required to suspend judgment about a large proportion of our moral views. There are, after all, excellent philosophers on virtually every side of every moral controversy. If so, then here is a very incomplete list of the moral issues about which Conciliationism would require us to suspend judgment: consequentialism vs. non-consequentialism; whether the doctrine of double effect is true; whether the killing/letting die distinction is morally significant; the moral (im)permissibility of eating meat, abortion, capital punishment, genetic enhancement, prostitution, and many, many other debates in moral and political philosophy. The complete list would be enormous.¹⁰ Armed with Conciliationism, one can advance The Argument from Conciliationism¹¹ (1) If Conciliationism is true, then we’re rationally required to suspend judgment about any moral claim that is the subject of disagreement among our epistemic peers. (2) Conciliationism is true. (3) So, we’re rationally required to suspend judgment about any moral claim that is the subject of disagreement among our epistemic peers. (4) A very large proportion of our moral beliefs are the subject of disagreement among our epistemic peers. (5) Therefore, we are rationally required to suspend judgment about a very large proportion of our moral beliefs.

If this argument succeeds, it would not entail that an extreme version of moral skepticism, according to which peer disagreement requires us to suspend judgment about all our moral beliefs, is correct. Moral philosophers don’t disagree about everything. There is very little disagreement about, for instance, whether pain is bad, whether a person’s hair color alone counts as a reason to harm them, whether a crying infant deserves to be punished for waking you up from your nap, and so on. These beliefs would not be threatened by Conciliationism. Nevertheless, the truth of Conciliationism ¹⁰ You might, like me, think it’s plausible that we ought to suspend judgment about some of these issues. But I think we ought to suspend judgment because the first-order evidence is inconclusive, not because philosophers disagree about these issues. ¹¹ A version of this argument is advanced by McGrath (2008), Matheson (2016), Rowland (2017), and Sinnott-Armstrong (2007).

28

Eric Sampson

would entail that we are required to suspend judgment about most of our interesting moral and political views. So the argument entails a kind of skepticism (or agnosticism) that is troubling enough to warrant a response. Some have tried to resist The Argument from Conciliationism by attacking (1)—the claim that Conciliationism entails that we ought to suspend judgment about a large proportion of our moral views.¹² Others have attacked (4)—the claim that we have many epistemic peers who disagree with us about moral issues—by suggesting that, as a matter of fact, we have very few epistemic peers about moral matters.¹³ I don’t find either strategy compelling, but I will not argue here that they fail. Instead, I’ll argue that no one, not even conciliationists, should accept this argument because (2)—the claim that Conciliationism is true—is self-undermining. If I am correct, then our justification for resisting the Argument from Conciliationism doesn’t depend on the cases others have made for rejecting (1) and (4). Conciliationism is a popular view in the epistemology of peer disagreement, but it hasn’t won universal acceptance. Many epistemologists hold a version of a competing view, known widely as the Steadfast View (see, e.g., Bergmann (2009), Decker (2014), Enoch (2010), Kelly (2005), Titelbaum (2014), van Inwagen (2010), Weatherson (2013), Wedgwood (2010), Weintraub (2013)).¹⁴ According to steadfasters, disagreement about p between epistemic peers is rarely sufficient, all by itself, to rationally require either peer to suspend judgment about p. Thus, Conciliationism is itself the subject of disagreement among excellent philosophers. This fact poses an obvious problem for conciliationists. If, as virtually all conciliationists admit, conciliationists have many epistemic peers who disagree with them about the truth of Conciliationism, then it seems to be irrational, by conciliationists’ lights, for conciliationists to believe (or be confident in) their own view. Call this the self-undermining problem (SUP).¹⁵ Conciliationists are, of course, aware of this problem. Some of them accept without reservation the obvious implication that we cannot rationally believe Conciliationism. For instance, David Christensen, a prominent conciliationist, writes: [I]t seems to me those of us who find ourselves strongly drawn toward Conciliationism in these contentious times should not be confident that Conciliationism is

¹² See, e.g., Vavova (2014). ¹³ See, e.g., King (2012). ¹⁴ As with Conciliationism, the Steadfast View is best described as a family of views. ¹⁵ I’m not the first to point out that arguments from Conciliationism face the SUP. Decker and Groll (2013) and Horn (2016) do this, too. But they don’t engage conciliationist attempts to justify confidence in Conciliationism even in the face of peer disagreement, as I do here.

The Self-Undermining Arguments from Disagreement

29

correct. (Of course, we may still work hard in producing and disseminating arguments for the view, hoping to hasten thereby the day when epistemic conditions will brighten, consensus will blossom, and all will rationally and whole-heartedly embrace Conciliationism.) (2009: 763)

And later he writes: [Recognition of the SUP] puts the advocate of CV [conciliatory views] in a situation that’s puzzling in a number of ways. For one thing, it would seem that, in the present epistemological climate, at least, CV has the property that one cannot rationally believe it (at least very strongly), even if it’s true. (2013: 78)

Christensen later notes, correctly, that the fact that conciliationists cannot rationally believe their view, given the current controversy about it, does not entail that Conciliationism is false. Nevertheless, he does concede that Conciliationism cannot now be rationally believed, and this is what concerns me now. I will not (and need not) argue that Conciliationism is false to show that we should not accept the Argument from Conciliationism. It’s sufficient, for my purposes, to show that the Argument from Conciliationism employs a premise we cannot rationally believe. After all, no one should accept the conclusion of an argument on the basis of that argument if, by one’s own lights, one cannot rationally believe one of the premises. While Christensen doubts that we can rationally believe Conciliationism, other conciliationists are more optimistic. They argue that we are rationally permitted to believe Conciliationism, even in the face of peer disagreement about it. If they’re correct, then conciliationists can escape the force of the SUP and Conciliationism wouldn’t self-undermine in the way I’ve described. But, as I’ll now argue, these defenses fail for roughly the same reason. Each defense depends for its success on philosophical claims that are themselves the subject of disagreement among conciliationists’ epistemic peers. It is therefore irrational, by conciliationists’ own lights, to believe the claims they employ in their defenses. It is thus irrational to believe that these defenses succeed and, given the SUP, irrational to continue believing Conciliationism in the face of peer disagreement about it.¹⁶ Begin by considering Tomas Bogardus’s (2009) defense of Conciliationism against the SUP. He argues that the truth of Conciliationism is obvious and can be known by direct acquaintance on the basis of rational intuition. Thus, any apparent epistemic peer who denies Conciliationism can be demoted from the status of epistemic peer on the grounds that they deny an obviously true proposition known by direct acquaintance. Bogardus writes,

¹⁶ I argue for this conclusion in much more detail in (Sampson, ms.). What follows is a rough sketch of that argument.

30

Eric Sampson

[T]he antecedent of the Equal-Weight View [that some agent is an apparent epistemic peer] might not be satisfied in cases involving knowledge from that unmediated access to the truth of propositions sometimes afforded by rational intuition. And it’s plausible that the Equal-Weight View is itself a deliverance of rational intuition . . . With further reflection, I think, one can come to just see the truth of the View—not only does it seem obvious, but upon further reflection it just is obvious. (2009: 333)

Clearly, this defense succeeds only if the following claim is both true and can be rationally believed. O: The Equal Weight View—one version of Conciliationism— is obviously true and is known by direct acquaintance on the basis of rational intuition.

On Bogardus’s view, Obvious serves as the justification for demoting any steadfaster from the status of epistemic peer. But one can rationally demote a peer on the basis of a belief in Obvious only if one can rationally believe Obvious. Thus, the success of Bogardus’s defense depends on the claim that Obvious is both true and can be rationally believed. Unfortunately for conciliationists, however, Obvious cannot, by conciliationists’ standards, be rationally believed since it is itself highly controversial. Not only do all steadfasters deny Obvious, but so also do many conciliationists. One might suggest that Obvious, too, is obvious or known on the basis of rational intuition. If this were true and we could rationally believe it, then conciliationists could demote any apparent epistemic peer who disagrees with them about Obvious on the grounds that they deny a proposition that is obviously true. But the claim that Obvious is obvious or known on the basis of rational intuition is even more controversial than Obvious itself. Again, every steadfaster will deny this and so also will many conciliationists. Conciliationism thus entails that it’s not rational to believe that Obvious is obvious or known on the basis of rational intuition. So one cannot rationally demote an apparent epistemic peer on the basis of a belief that Obvious is obvious. One might attempt to make the same move yet again—i.e., insist that it’s obvious that Obvious is obvious. But I take it that no conciliationist would want to pursue this desperate strategy. The same difficulty afflicts Adam Elga’s (2010) famous defense of Conciliationism against the SUP. Elga employs a self-exempting strategy to defend Conciliationism: He argues that we rationally ought to conciliate in the face of peer disagreement, except when the disputed proposition is Conciliationism itself. The resulting view is

The Self-Undermining Arguments from Disagreement

31

E’ S-E C (ESEC): (1) Conciliationism: If A learns that an apparent epistemic peer, B, disagrees with her about p, then A is rationally required to adjust her credence in p significantly in the direction of B’s, (2) unless p is Conciliationism. If p is Conciliationism, then A ought to remain steadfast in her confidence—indeed have a credence of 1—in p.

Obviously, Elga needs to explain how ESEC’s second clause isn’t objectionably ad hoc. Elga’s answer is that Conciliationism—ESEC’s first clause—is an inductive method. And all correct inductive methods must be dogmatic—i.e., recommend credence 1—about their own correctness. If an inductive method were not dogmatic about its own correctness, Elga argues, then there would be possible circumstances in which the method calls for its own rejection. But that would render the method incoherent. It would be as if the method were to say, “Take the following advice: do not take this advice.” That’s incoherent. Elga thinks that any method that could possibly give such incoherent advice is a false (or bad, or incorrect) method. So, on Elga’s view, for any correct inductive method M, M will never advise an agent to stop believing or taking advice from M. That includes whatever method ought to govern our response to peer disagreement. Thus, Elga argues, his self-exempting Conciliationism is not objectionably ad hoc. It treats Conciliationism like any other inductive method. If this defense is correct, and we can rationally believe it, then he’s successfully rebutted the SUP. Unfortunately for Elga, whether his defense succeeds is the subject of disagreement, not only among steadfasters (who obviously believe that his defense fails), but among his fellow conciliationists, too. For example, David Christensen has criticized Elga’s self-exempting Conciliationism at length (2013: 88–9). Christensen’s criticism, in short, is that, to endorse Elga’s defense by adopting credence 1 in Conciliationism, one would be committed to believing either that one is infallible when it comes to reasoning about peer disagreement or that one is just incredibly lucky to have hit upon the correct view. But neither suggestion is plausible. So, according to Christensen, we cannot rationally endorse ESEC. If Christensen is correct, then Elga’s response cannot insulate Conciliationism from the force of the SUP. The point of the preceding paragraph is not to argue that Christensen’s criticism is correct (though I’m inclined to think that it is). Rather, the point is that the question of whether Elga’s response succeeds in defending Conciliationism from the SUP, is itself the subject of disagreement among excellent philosophers—not just among steadfasters, but among conciliationists, too. By Elga’s own lights, then, it is irrational to believe that his

32

Eric Sampson

defense succeeds. It would therefore be irrational to believe Conciliationism in the face of the SUP on the basis of Elga’s defense. So, unless some other defense of Conciliationism against the SUP comes to the rescue, conciliationists cannot rationally believe their view. Perhaps such a defense will come, but it doesn’t exist now.¹⁷ The upshot, then, is that no one should believe the conclusion of the Argument from Conciliationism on the basis of that argument, since, given the actual distribution of opinion about Conciliationism, its second premise is self-undermining. No one can rationally believe it. And if no one can rationally believe that premise, then no one can rationally accept the conclusion of the Argument from Conciliationism on the basis of that argument.

2 .3 . A R GU M E N T S F R O M T H E U N R E L I A B I L I T Y OF THE METHOD OF MORAL INQUIRY Another popular form of epistemological argument against non-skeptical realism exploits moral disagreement to cast doubt on the reliability of the method used in moral inquiry. These arguments begin by calling attention to the troubling fact that moral disagreement among philosophers persists despite philosophers’ use of (what is widely regarded as) the best method for moral inquiry: reflective equilibrium. Reflective equilibrium is, roughly, the process of making intuitive judgments about moral cases and general moral principles, and then bringing those judgments into coherence with one another and with our beliefs about the non-moral facts. While moral philosophers don’t agree about much, it’s remarkable how much they agree that reflective equilibrium is the correct way to proceed with ethical inquiry. This makes it all the more discouraging that this method has failed to lead many, perhaps most, philosophers to the moral truth. After all, moral disagreement is pervasive among philosophers, and it can’t be that all their logically incompatible moral judgments are correct—not on non-skeptical realism, at least. This suggests that even our best method of moral inquiry is unreliable. And it seems clear to many that if one is using a method of inquiry that one knows is unreliable, then one is not epistemically justified in believing the outputs of that method. In the case of moral inquiry, then, it seems that we are not epistemically justified in holding our moral beliefs— the products of reflective equilibrium. ¹⁷ I argue elsewhere (Sampson, ms.), however, that it is highly unlikely that any such defense will emerge in our lifetimes, given the extent of disagreement in epistemology, especially about the standards of rational belief.

The Self-Undermining Arguments from Disagreement

33

Perhaps the best version of this argument comes from Justin Horn (2016). Horn argues that there are plausible cases of fundamental moral disagreement. A moral disagreement is fundamental if it would persist even if the disputants had all the morally relevant non-moral information, were making no logical mistakes, and were not subject to any other sort of bias. One example of such a disagreement, Horn suggests, is the disagreement about whether certain items of non-sentient nature (e.g., rivers, rock formations, mountain ranges) are non-instrumentally valuable. Some environmentalists suggest that trees, for instance, in addition to being instrumentally valuable (because they give off oxygen and can be made into paper) are also non-instrumentally valuable. Others insist that trees are valuable only insofar as they help contribute to the flourishing of sentient life. It’s not clear that either group is misinformed about some non-moral matter. Nor is it clear that either has made a mistake in reasoning or is subject to some bias. If that’s correct, then the disagreement among philosophers about the noninstrumental value of non-sentient nature is fundamental. We might wonder what explains fundamental moral disagreement when it occurs. Why do excellent philosophers reach different moral judgments? Horn suggests that such disagreements are plausibly explained by the hypothesis that the two parties have different starting points—namely, conflicting moral intuitions. If so, then the difference in intuitive judgments is systemic such that not even flawless coherentist reasoning could bring the two parties into agreement. Cases of systemic differences in moral intuitions, Horn thinks, raise a serious epistemological challenge for non-skeptical realists. He writes: If I have a disagreement with an equally rational and (non-morally) informed interlocutor and this disagreement is ultimately traceable to a systemic difference in our intuitions, then at least one of us must have systemically misleading intuitions. Given that we are equally rational and non-morally informed, why should I think that my intuitions happen to reflect the truth while my opponent’s do not? And if there’s reason to think that even our most careful moral judgments fail to reliably track any stance-independent moral reality, is there any remaining reason to believe in such a reality at all? (2016: 4)

Horn then proceeds to advance what he calls the Reliability Argument from Disagreement (RAD): (6) If there is widespread fundamental moral disagreement, then many people would have a significant number of false moral beliefs even if they were to flawlessly employ our best method of moral inquiry. (7) If many people would have a significant number of false moral beliefs even if they were to flawlessly apply our best method of moral inquiry, then that method is not reliable for many people. (8) If one is aware that our best method of moral inquiry is not reliable for many people, then one cannot be justified in believing the

34

Eric Sampson outputs of such a method unless one has special reason to believe that it is reliable in one’s own case. (9) It is not the case that any of us has special reason to believe that the best method of moral inquiry is reliable in our own case. (10) Therefore, if we become aware of widespread fundamental moral disagreement, then we cannot be justified in believing the outputs of our best method of moral inquiry (2016: 12).

Three features of the RAD are worth noting. First, the conclusion is conditional. Horn is reluctant to speculate about the extent to which there is actual fundamental moral disagreement. He correctly notes that that’s an empirical question that cannot be settled from the armchair. He is more interested in exploring the implications of such widespread disagreement if it exists. I will, however, grant for the sake of argument that there is such widespread, fundamental disagreement and that we are aware of it. I will therefore proceed as if the conclusion of the RAD is not conditional— that the conclusion is that we cannot be justified in believing the outputs of our best method of moral inquiry. Second, the RAD purports to undermine the justification we might have for any of our moral beliefs, not just our controversial moral beliefs. This is what distinguishes the RAD from the Argument from Conciliationism. On Horn’s view, reflective equilibrium is our best (perhaps only) method of moral inquiry. Widespread, fundamental moral disagreement reveals that there is something deeply flawed about that method. We therefore ought not to trust that any of the outputs of that method are correct. Third and finally, the RAD is specifically about moral disagreement. There is, however, no reason to restrict its scope to moral but not epistemic disagreement. Anyone who accepts the RAD as an argument for moral skepticism should accept it as an argument for epistemic skepticism as well, provided there is widespread, fundamental epistemic disagreement. After all, reflective equilibrium is the best (perhaps only) method we have for epistemological inquiry, too. We should ask, then: Is there widespread, fundamental epistemic disagreement? It seems that there are disagreements in epistemology, even among the experts, that bear all the same marks of supposed fundamental moral disagreement. Consider the disagreement among epistemologists about the conditions under which a belief is epistemically justified. There have, for decades now, been clearly defined camps, among which are foundationalists, coherentists, reliabilists, infinitists, and others. It doesn’t seem that the members of these camps are lacking some crucial nonnormative information, or are subject to some bias, or are making some obvious mistake in reasoning. As with fundamental moral disagreement, it seems that disagreements about epistemic justification are explained by a

The Self-Undermining Arguments from Disagreement

35

difference in epistemic intuition between the members of the opposing sides. If so, then the debate about which theory of epistemic justification is correct is an example of fundamental epistemic disagreement. There are other examples. There are long-lasting debates about, among many other issues, internalism vs. externalism about epistemic justification, whether knowledge is closed under known entailment, whether skepticism can be refuted, whether there is a priori knowledge, and more recently, whether one ought always to conciliate in the face of peer disagreement, and whether a body of evidence permits differing doxastic attitudes toward a proposition for an agent (or between different agents). There’s no more reason, that I can see, to think that these epistemological debates could be settled by providing the disputing parties with more non-normative information, removing biases, or correcting mistakes in reasoning than there is in the case of moral disagreement. If anything, epistemological disagreements seem less likely than moral disagreements to be the product of mistaken nonnormative information or bias. After all, the practical stakes seem much lower in these epistemological debates than in the moral debates. Thus, epistemological disagreements aren’t as easily explained away by the distorting influence of self-interest, or a failure of imagination or empathy, or mistaken empirical beliefs. If this is correct—if, that is, there is at least as much fundamental disagreement in epistemology as there is in ethics—then the RAD succeeds in undermining the justification we have for our moral beliefs only if it also succeeds in undermining the justification we have for our epistemological beliefs. But premises (8) and (9) of the RAD are epistemological claims. So, if the RAD undermines our moral beliefs, then it undermines our justification for believing (8) and (9), too. But if we shouldn’t believe (8) and (9), then we shouldn’t believe the conclusion of the RAD on the basis of the argument, since it employs premises that, even by the argument’s standards, we’re not justified in believing. Now we have a clear dilemma for proponents of the RAD. Either the RAD is sound or it isn’t. If the RAD is not sound, then we shouldn’t accept the RAD’s conclusion on the basis of the RAD (since the argument is unsound). If the RAD is sound, then we shouldn’t accept the RAD’s conclusion on the basis of the RAD (since the RAD would undermine our justification for believing premises (8) and (9)). Either way, we shouldn’t accept the RAD’s conclusion on the basis of the RAD. As with the Argument from Conciliationism, I clearly haven’t shown that the RAD is unsound. I’ve shown only that no one can rationally believe the conclusion of the RAD on the basis of the RAD. Some might find this unsatisfying, but all I can say is that I don’t. After all, neither the RAD nor the Argument from Conciliationism purport to show that our moral beliefs

36

Eric Sampson

are false, only that we cannot rationally hold them. I’ve shown that precisely the same criticism applies to our beliefs about at least one of the premises in the RAD and the Argument from Conciliationism.¹⁸ Thus, by those arguments’ own standards, we cannot rationally endorse either argument as one that supports moral skepticism (or agnosticism).

2.4. IBE ARGUMENTS FROM INTRACTABLE D IS A G R E E M E N T I’ll now turn to arguments that draw some troubling metaphysical conclusion about morality. One popular form of metaphysical argument calls attention to disagreement among a certain group of people and suggests that the best explanation for the existence and character of that disagreement is that there are no moral truths about which to agree.¹⁹ The best version of this argument, in my view, identifies moral philosophers, past and present— the supposed experts vis-à-vis moral inquiry—as the relevant class of disputants. Brian Leiter (2014) has recently advanced an argument of this kind that takes its inspiration from Nietzsche.²⁰ Leiter begins by describing the kind of disagreement that his metaphysical hypothesis—namely, that there are no moral truths—is meant to explain better than non-skeptical moral realism. He writes: For what [Nietzsche] calls attention to is not “ordinary” or “folk” moral disagreement, but rather what seems to me the single most important and embarrassing fact about the history of moral theorizing by philosophers over the last two millennia: namely, that no rational consensus has been secured on any substantive, foundational proposition about morality. By a “foundational” moral proposition about morality, I am thinking of, for example, deontological or utilitarian theories which specify the criteria in virtue of which concrete or particular moral judgments are thought warranted: so e.g. “it is wrong to break this promise” is a concrete moral judgment, while “the wrong-making feature of an action is its effect on utility” is a foundational proposition. With regard to such foundational propositions, the history of moral ¹⁸ The strength of my critique of the RAD and the Argument from Conciliationism is thus exactly proportional to the (supposed) strength of those arguments’ critique of our moral beliefs. This should not be surprising since my claim is that whatever defect proponents of the arguments think attaches to our moral beliefs also attaches to at least one of the premises in those arguments. ¹⁹ Mackie’s (1977) “argument from relativity” is the most famous version of this kind of argument. ²⁰ In Leiter’s article, he is doing both Nietzsche interpretation and defending an argument from disagreement. Leiter attributes the argument I’ll consider in this section to Nietzsche, but I will attribute it and its supporting claims to Leiter, since Leiter is the one who does the most to defend them.

The Self-Undermining Arguments from Disagreement

37

philosophy is the history of intractable disagreement. Is the criterion of right action the reasons for which it is performed or the consequences it brings about? If the former, is it a matter of the reasons being universalizable, or that they arise from respect for duty, or something else? If the latter, is it the utility it produces or the perfection it makes possible? If the former, is utility a matter of preference satisfaction (as the economists almost uniformly believe) or preference satisfaction under idealized circumstances . . . ? (2014: 131)

The idea, then, is that moral philosophers have had roughly twenty-five hundred years to construct, and come to agreement about, a systematic account of ethics. Over that time, they have produced a host of incompatible philosophical theories: Aristotelianism, Kantianism, Humeanism, Utilitarianism, contractualism, and so on. Each purports to answer foundational questions in ethics, but none has been able to secure the endorsement of more than a fraction of moral philosophers. Thus, there is very little agreement and, moreover, very little in the way of rational persuasion among them. You almost never see a Kantian converting all at once (or even gradually) to Utilitarianism, or a Libertarian being rationally persuaded to embrace socialism, or any similar kind of conversion. For the most part, philosophers endorse (or incline toward) a certain ethical theory at the beginning of their careers and remain in that corner, as it were, forever.²¹ Their disagreements are thus amazingly intractable. Though philosophers often agree about particular moral judgments, that agreement ends as soon as they’re called on to explain why their particular judgments, and not their opponents’, are correct. And it is this fact, Leiter thinks—the fact that there is such intractable moral disagreement about foundational moral issues, not particular moral judgments—that calls out for explanation. The explanation Leiter (inspired by Nietzsche) provides has two components: a psychological and a metaphysical component. He writes: [T]he “best explanation” argument asks: what is the best explanation for the fact that philosophical theories, in the sense just noted, reach different and quite intractable conclusions about foundational matters? Nietzsche’s skeptical answer will be that the best explanation is that the psychological needs of philosophers lead them to find compelling dialectical justifications for very different basic moral claims, and there are no objective moral facts to stand in the way of satisfying those psychological needs. (2014: 134)

The psychological component, then, is that philosophers have psychological needs that are satisfied when they believe their preferred moral theory. The reason there are so many different moral theories is that moral philosophers have many different psychological needs. Some moral theories will satisfy ²¹ This is, at any rate, how Leiter sees things. For the sake of argument, I’m happy to accept his description of the sociology of moral philosophy.

38

Eric Sampson

some philosophers’ psychological needs while other moral theories will satisfy others’. The metaphysical component of Leiter’s explanation is that there are no moral truths to “stand in the way” of philosophers satisfying their psychological needs by believing their preferred moral theory. The thought seems to be that, if there were the kind of moral truths that non-skeptical realists posit, we would expect to see moral philosophers slowly but surely coming to believe them, despite the distorting influence of their psychological needs. We would therefore expect to see a slow but steady increase in the amount of agreement among moral philosophers as they discover the moral truths. Since we don’t see that kind of slow and steady increase in agreement, we can be confident that there are no such truths influencing their moral beliefs. We might represent Leiter’s central argument as follows: The IBE Argument from the History of Moral Philosophy (11) If there is long-standing, intractable disagreement among the (supposed) experts about a subject matter S, then the best explanation for the existence and character of that disagreement is that there are no facts about S. (12) There is long-standing, intractable disagreement among the (supposed) experts about ethics. (13) Therefore, the best explanation for the long-standing, intractable disagreement among moral philosophers is that there are no facts about ethics.

There are, of course, dangers associated with representing an inference to the best explanation as a deductive argument. For instance, Leiter might not accept (11) in full generality. But since Leiter accepts (11) as it pertains to the subjects that I will discuss here (i.e., ethics, epistemology, metaphysics), I can afford to simplify in this way. Premise (12) says that ethics is characterized by long-standing, intractable disagreement among its supposed experts. But, of course, ethics is not unique in this respect. The same is true of, at least, metaphysics, epistemology, political philosophy, aesthetics, philosophy of language, philosophy of logic, and philosophy of science. If such disagreement in ethics is best explained by Leiter’s hypothesis that there are no moral facts, then a similar hypothesis should apply to the other subjects. That would entail that there are no facts about metaphysics, epistemology, philosophy of logic, and so on. Leiter anticipates this response—for metaphysics and epistemology, at least. He understands the challenge as an appeal to “companions in guilt.” Philosophers appeal to companions in guilt when they want to rebut an argument that has undesirable implications for their view. They do this by suggesting that the argument leveled against their preferred view

The Self-Undermining Arguments from Disagreement

39

overgeneralizes, having undesirable implications for views that their opponents do not wish to undermine. Leiter, however, is unmoved by appeals to companions in guilt. He writes: Some recent writers (such as Bloomfield 2004 and Shafer-Landau 2005) think this kind of “companions in guilt” consideration counts in favor of moral realism, notwithstanding the disagreement among moral philosophers. It is not entirely clear why they rule out, however, the other natural conclusion. Nietzsche, as far as I can see, has no reason to resist it, since he believes that, as an explanatory matter, the moral commitments of the philosopher—at least the great philosopher—are primary when it comes to his metaphysics and epistemology. (2014: 146)

Leiter’s (surprising) reply, then, is to admit that metaphysics and epistemology are just as “guilty” as ethics. On the Leiter/Nietzsche view, philosophers adopt their metaphysical and epistemological views because they support, or cohere with, their antecedently held ethical views. If so, then, on Leiter’s view, there are no facts of the matter when it comes to foundational questions in metaphysics and epistemology—just as in ethics. But if there are no facts about metaphysics and epistemology, then what are we to make of Leiter’s premise (11)? (11) is either a metaphysical or an epistemological claim. Either way, if Leiter is correct that there are no facts of the matter when it comes to metaphysics and epistemology, then by endorsing (11) and its implications, he has thereby undermined (11). He is committed to the claim that there is no fact of the matter about whether (11) is correct. But if that is so, then we clearly should not accept his IBE argument from the history of moral philosophy. For, by its own standards, there is no fact of the matter about whether its first premise, (11), is correct.

2 . 5 . A R G U M E N T S F R O M D IS A G R E E M E N T IN IDEAL CONDITIONS The final popular form of argument I’ll consider begins with the supposition that there is significant fundamental moral disagreement among people generally—both philosophers and non-philosophers. Fundamental moral disagreement, as it’s meant here, is moral disagreement that would persist even between agents who are fully informed about non-moral matters and are ideally-rational. The argument then proceeds to draw a troubling metaphysical conclusion about morality, namely, that, for many debates in ethics, there is no fact of the matter about which view is correct. And if so, then our understanding of morality is deeply flawed and non-skeptical realism is false. After all, on non-skeptical realism, we do not have a deeply

40

Eric Sampson

flawed understanding of morality. The best version of this argument comes from Richard Rowland (2016). Rowland begins by arguing for an accessibility condition on moral truths. He thinks that, if there are moral truths, they must be epistemically accessible to (i.e., discoverable by) some possible agent. He reasons as follows. If there is a fact of the matter about the moral status of some action Φ, then that fact makes a difference to how some possible agent ought to act. For example, if it’s (im)permissible to eat meat, then that fact makes a difference to how we ought to live. But if it is impossible for any possible agent to discover the truth about the moral (im)permissibility of eating meat, then that fact wouldn’t make a difference to how any possible agent ought to live. So, there wouldn’t be any fact of the matter about whether it’s permissible to eat meat. Thus, the only way that an action can have a moral status is if some possible agent could discover the truth about its moral status. I’ll call this the accessibility condition. Rowland then argues that, for many debates in moral philosophy, it’s not unlikely that the putative moral status of the relevant act is inaccessible to any possible agent. For example, he thinks that it’s not unlikely that the putative facts about the moral status of vegetarianism, bombing Syria, torture, the death penalty, breaking promises, pushing the fat man off the bridge, and lying are inaccessible to any possible agent because disagreement about the moral status of these actions would persist even among agents in epistemically ideal conditions (2016: 2). That is, even if the disputants in these debates were given full non-moral information and were endowed with perfect reasoning capacities, they would still disagree about the moral status of these actions. If so, Rowland argues, then the supposed facts about the moral status of these actions would be inaccessible to any possible agent. So, given the accessibility condition, there would be no fact of the matter about the moral status of these actions. Our understanding of morality would thus be deeply flawed and non-skeptical moral realism would be false. We could represent the argument as follows: The Argument from Disagreement in Ideal Conditions (14) For any action Φ, if Φ-ing has a moral status, then the moral status of Φ-ing is accessible to some possible agent. (15) For many debates in ethics, the moral status of Φ-ing is inaccessible to any possible agent. (16) So, for many debates in ethics, Φ-ing has no moral status. (17) If (16), then our understanding of morality is deeply flawed and non-skeptical realism is false. (18) Therefore, our understanding of morality is deeply flawed and non-skeptical realism is false.

The Self-Undermining Arguments from Disagreement

41

I’ll focus on (14)–(16), since I accept (17). (14) is the accessibility condition, which I’ll leave unchallenged. The support for (15) includes two suppositions. The first is that there is fundamental disagreement about many debates in moral philosophy. I’m happy to grant that. The second is that, if there is fundamental disagreement about the moral status of Φ-ing, then the moral status of Φ-ing is inaccessible to any possible agent. This is where I will focus my criticism. Rowland doesn’t give an account of what epistemic accessibility consists in, but he does offer a necessary condition for a proposition’s being epistemically accessible. It’s a condition that he believes is not met for many propositions that are the subject of fundamental disagreement in ethics. Here’s the condition: (2.1) A has epistemic access to p at T1 only if there is a justification R that A can justifiably believe p on the basis of at T1 that is such that A’s belief that p on the basis of R would not be defeated in more ideal or idealized conditions. (2016: 12)

In other words, an agent has epistemic access to a proposition p only if she can justifiably believe p in epistemically ideal conditions. The key claim in Rowland’s case for the claim that ideally informed, ideally rational agents engaged in a moral disagreement do not meet this condition with respect to many moral propositions is (2.3) If idealized reasoners A and B hold conflicting beliefs about the moral status of Φ-ing, know this, and know that they are idealized reasoners, then neither A nor B are justified in holding conflicting beliefs about the moral status of Φ-ing; A and B are only justified in suspending belief about the moral status of Φ-ing. (2016: 12)

The idea is that, if A and B are idealized reasoners and know this about one another, then they each know of the other that they have the same evidence (that is, all the non-moral facts) and have made all and only correct steps in reasoning. So, neither can claim to have more (or different) information, and neither can claim to have followed a better line of reasoning than the other. Rowland thus concludes that neither can claim to have the “better” view, and neither would be justified in holding their own view in the face of this disagreement. They’re both rationally required to suspend judgment about the disputed proposition. Though Rowland formulates (2.3) in terms of moral disagreement, there’s no motivation for restricting it to moral, but not normative, disagreement more generally. Extending (2.3) in this way yields (2.3*) If idealized reasoners A and B hold conflicting beliefs about the normative status of Φ-ing, know this, and know that they are idealized

42

Eric Sampson reasoners, then neither A nor B are justified in holding conflicting beliefs about the normative status of Φ-ing; A and B are only justified in suspending belief about the normative status of Φ-ing.

Φ-ing would now refer, not only to acting in various ways, but also believing in various ways. It’s clear that (2.3*) is itself a normative claim, since it’s about the conditions under which holding a certain belief is not epistemically justified. The first question we should ask, then, is: Would (2.3*) itself be the subject of disagreement among agents in epistemically ideal conditions? If so, then, by (2.3*), even agents in ideal conditions would not be justified in believing (2.3*). So, by (2.1), no possible agent would have epistemic access to (2.3*). Combined with premise (14), this would entail that (2.3*) has no normative status—i.e., there would be no fact of the matter about whether (2.3*) is correct. Obviously, then, Rowland must insist that (2.3*) would not be the subject of disagreement in ideal conditions. But what could justify this insistence? Why should anyone believe that agents in epistemically ideal conditions would come to agreement about (2.3*) and, furthermore, agree to accept, rather than to reject or suspend judgment about, (2.3*)? I can imagine two strategies for justifying this claim. The first is to endorse Conciliationism. The second is to endorse the Interpersonal Uniqueness thesis. I’ll consider each strategy in turn. If Conciliationism is correct, then that would explain why ideally informed, ideally rational agents would suspend judgment about controversial moral propositions once they become aware that their ideally informed, ideally rational interlocutor disagrees with them. Plausibly, such agents are epistemic peers and Conciliationism requires disagreeing epistemic peers to suspend judgment about any proposition that is under dispute between them. There are, however, two problems with appealing to Conciliationism to rule out the possibility that (2.3*) would itself be the subject of disagreement in ideal conditions. The first, as I argued in Section 2.2, is that, even by conciliationists’ lights, we cannot now rationally believe that Conciliationism is correct, given the current controversy about it among excellent philosophers. Thus, we cannot now rationally accept Rowland’s argument if it depends on Conciliationism. I’ll discuss the second problem with appealing to Conciliationism to justify (2.3*) in a moment. The second way to justify the claim that (2.3*) would not be the subject of disagreement in idealized conditions is to appeal to T I U T (IUT): Given a body of evidence E, a proposition p, and any two agents A and B, there is precisely one doxastic attitude toward p that is rationally permissible for A and B relative to E.

The Self-Undermining Arguments from Disagreement

43

If this were correct, then ideally rational agents with the same evidence would never disagree about anything. But, of course, the IUT is highly controversial, too.²² Even now, a debate is raging between permissivists who deny the IUT and non-permissivists who accept it.²³ This debate is at an early stage in its development, so it’s not clear how it will turn out. So, it’s certainly not clear that the consensus, should it ever emerge, will settle on acceptance (rather than agnosticism about or rejection) of the IUT. Indeed, the IUT is at least as controversial as any of the moral issues about which Rowland suggests there might be fundamental disagreement. Speaking of the IUT, Matt Kopec and Michael Titelbaum write: [T]he question of whether Uniqueness is true has many of the trappings of a classic philosophical puzzle. The thesis seems obviously false to many philosophers, and obviously true to many others. The arguments in favor of Uniqueness are at first rather compelling, but a closer inspection reveals various flaws. Nevertheless, those who think Uniqueness is false often find it difficult to devise any counter examples that are able to convert the Uniqueness defenders. (2016: 189)

If this is correct, then the IUT, like Conciliationism, shares all the relevant features of debates about vegetarianism, bombing Syria, torture, and so on—debates about which Rowland suggests there might be fundamental disagreement. Thus, any reason to believe that moral disagreement would persist among ideal reasoners about these issues favors believing that epistemic disagreement would persist among them, too—especially about the IUT and Conciliationism. But if such disagreement would persist, then, by Rowland’s own lights, these principles are not epistemically accessible to any possible agent. Combined with (14)—the accessibility condition—this entails that there is no fact of the matter about whether these epistemic principles are correct. If so, then Rowland has undermined the most promising support for (2.3*)—his central claim in his argument from disagreement. Moreover, since the support for (2.3*) likely depends on the IUT or Conciliationism, and these principles would be the subject of disagreement in ideal conditions, there is excellent reason to believe that (2.3*) would itself be the subject of disagreement in ideal conditions. By Rowland’s own lights, that entails that there is no fact of the matter about whether (2.3*) itself is correct. We therefore shouldn’t accept his argument from disagreement, which crucially depends on (2.3*).

²² Enoch makes this point in (2011: 208). ²³ Among the non-permissivists are Christensen (2007), Feldman (2007), Greco and Hedden (2017), Horowitz (2014), Matheson (2011), and White (2013). The permissivists include Decker (2012), Kelly (2013), Meacham (2014), Peels and Booth (2014), Schoenfield (2018), Sharadin (2015), and Titelbaum (2014).

44

Eric Sampson 2 . 6 . C ON C L U S I O N

I’ve now considered four different popular forms of argument from disagreement and argued that each fails in its own way. It’s clear, however, that the flaw in these arguments is structural—each fails for roughly the same underlying reason. Each argument attempts to support a very strong conclusion about the metaphysics or epistemology of morality. To do this, these arguments must employ very strong metaphysical or epistemological premises. But given the pervasiveness of disagreement in philosophy—especially about metaphysics and epistemology—such strong premises are always, or nearly always, the subject of deep, widespread, intractable disagreement. If such disagreement is sufficient to support troubling conclusions about morality, as these arguments purport to show, then such disagreement is sufficient to support troubling conclusions about metaphysics and epistemology, too. By these argument’s own standards, then, such disagreement undermines the strong metaphysical and epistemological premises these arguments employ to derive their radical conclusions about morality. In short, then, given the pervasiveness of disagreement in philosophy, these arguments are destined to undermine themselves.

References Bergmann, M. 2009. “Rational Disagreement After Full Disclosure,” Episteme 6 (3): 336–53. Bogardus, T. 2009. “A Vindication of the Equal-Weight View,” Episteme 6 (3): 324–35. Brink, D. 1989. Moral Realism and the Foundations of Ethics. Cambridge: Cambridge University Press. Christensen, D. 2007. “Epistemology of Disagreement: The Good News,” Philosophical Review 116 (2): 187–217. Christensen, D. 2009. “Disagreement as Evidence: The Epistemology of Controversy,” Philosophy Compass 4 (5): 756–67. Christensen, D. 2013. “Epistemic Modesty Defended,” in D. Christensen and J. Lackey (eds.), The Epistemology of Disagreement: New Essays. Oxford: Oxford University Press, 77–97. Decker, J. 2012. “Disagreement, Evidence, and Agnosticism,” Synthese 187 (2): 753–83. Decker, J. 2014. “Conciliation and Self-Incrimination,” Erkenntnis 79 (5): 1099–134. Decker, J. and Groll, D. 2013. “The (In)significance of Moral Disagreement for Moral Knowledge,” in R. Shafer-Landau (ed.), Oxford Studies in Metaethics, Volume 8. Oxford: Oxford University Press, 140–67.

The Self-Undermining Arguments from Disagreement

45

Elga, A. 2007. “Reflection and Disagreement,” Noûs 41 (3): 478–502. Elga, A. 2010. “How to Disagree About How to Disagree,” in T. Warfield and R. Feldman (eds.), Disagreement. New York: Oxford University Press. Enoch, D. 2010. “Not Just a Truthometer: Taking Oneself Seriously (but not Too Seriously) in Cases of Peer Disagreement,” Mind 119 (476): 953–97. Enoch, D. 2011. Taking Morality Seriously: A Defense of Robust Realism. Oxford: Oxford University Press. Feldman, R. 2007. “Reasonable Religious Disagreements,” in L. Antony (ed.), Philosophers Without Gods. Oxford: Oxford University Press, 194–214. Frances, B. 2010. “The Reflective Epistemic Renegade,” Philosophy and Phenomenological Research 81 (2): 419–63. Greco D. and Hedden, B. 2017. “Uniqueness and Metaepistemology,”Journal of Philosophy 113 (8): 365–95. Horn. J. 2016. “Moral Realism, Fundamental Moral Disagreement, and Moral Reliability,” The Journal of Value Inquiry 51 (3): 363–81. Horowitz, S. 2014. “Immoderately Rational,” Philosophical Studies 167 (1): 41–56. Kelly, T. 2005. “The Epistemic Significance of Disagreement,” in J. Hawthorne and T. Gendler (eds.), Oxford Studies in Epistemology, Volume 1. Oxford: Oxford University Press, 167–96. Kelly, T. 2013. “Evidence Can Be Permissive,” in M. Steup, J. Turri, and E. Sosa (eds.), Contemporary Debates in Epistemology, 2nd edn. Hoboken, NJ: John Wiley & Sons. King, N. 2012. “Disagreement: What’s the Problem? Or A Good Peer is Hard to Find,” Philosophy and Phenomenological Research 85 (2): 249–72. Kopec, M. and Titelbaum, M. (2016). “The Uniqueness Thesis,” Philosophy Compass 11 (4): 189–200. Kornblith, H. 2010. “Belief in the Face of Controversy,” in R. Feldman and T. Warfield (eds.), Disagreement. New York: Oxford University Press. Leiter, B. 2014. “Moral Skepticism and Moral Disagreement in Nietzsche,” in R. Shafer-Landau (ed.), Oxford Studies in Metaethics, Volume 9. Oxford: Oxford University Press, 126–49. McGrath, S. 2008. “Moral Disagreement and Moral Expertise,” in R. Shafer-Landau (ed.), Oxford Studies in Metaethics, Volume 3. Oxford: Oxford University Press, 87–108. Mackie, J. 1977. Ethics: Inventing Right and Wrong. London: Penguin Books. Matheson, J. 2009. “Conciliatory Views of Disagreement and Higher Order Evidence,” Episteme: A Journal of Social Epistemology 6 (3): 269–79. Matheson, J. 2011. “The Case for Rational Uniqueness,” Logos and Episteme 2 (3): 359–73. Matheson, J. 2016. “Moral Caution and the Epistemology of Disagreement,” Journal of Social Philosophy, 47 (2): 120–41. Meacham, C. 2014. “Impermissive Bayesianism,” Erkenntnis 79 (6): 1185–217. Peels, R. and Booth, A. (2014). “Why Responsible Belief is Permissive Belief,” Analytic Philosophy 55 (1): 75–88.

46

Eric Sampson

Pittard, J. 2015. “Resolute Conciliationism,”Philosophical Quarterly 65 (260): 442–63. Rowland, R. 2016. “The Significance of Fundamental Moral Disagreement,” Noûs 50 (4): 1–30. Rowland, R. 2017. “The Epistemology of Moral Disagreement,” Philosophy Compass 12 (2): 1–16. Sampson, E. ms. “Can We Rationally Believe Conciliationism?” Schoenfield, M. 2018. “Permissivism and the Value of Rationality: A Challenge to the Uniqueness Thesis,” Philosophy and Phenomenological Research, online first, https://doi.org/10.1111/phpr.12490. Shafer-Landau, R. 2003. Moral Realism: A Defence. Oxford: Oxford University Press. Sharadin, N. 2015. “A Partial Defense of Permissivism,” Ratio 28 (2): 57–71. Sinnott-Armstrong, W. 2007. Moral Skepticisms. New York: Oxford University Press. Titelbaum, M. 2014. “Rationality’s Fixed Point,” in T. Szabo Gendler and J. Hawthorne (eds.), Oxford Studies in Epistemology, Volume 5. Oxford: Oxford University Press, 253–94. van Inwagen, P. 2010. “We’re Right. They’re Wrong,” in R. Feldman and T. Warfield (eds.), Disagreement. New York: Oxford University Press, 10–28. Vavova, K. 2014. “Moral Disagreement and Moral Skepticism,” Philosophical Perspectives 28 (1): 302–33. Weatherson, B. 2013. “Disagreements, Philosophical and Otherwise,” in D. Christensen and J. Lackey (eds.), The Epistemology of Disagreement. Oxford: Oxford University Press, 54–75. Wedgwood, R. 2010. “The Moral Evil Demons,” in R. Feldman and T. Warfield (eds.), Disagreement. New York: Oxford University Press, 216–46. Weintraub, R. 2013. “Can Steadfast Peer Disagreement Be Rational?” Philosophical Quarterly 63 (253): 740–59. White, R. 2013. “Evidence Cannot Be Permissive,” in M. Steup, J. Turri, and E. Sosa (eds.), Contemporary Debates in Epistemology, 2nd edn. Hoboken, NJ: John Wiley & Sons.

3 Contextualism, Moral Disagreement, and Proposition Clouds Jussi Suikkanen 3.1. INTRODUCTION I understand Contextualism to be the following semantic framework for understanding the meaning of moral terms.¹ Consider the sentence ‘eating meat is wrong’. According to Contextualism, this sentence does not have truth-conditional content because ‘ . . . is wrong’ is not a predicate that has a fixed property as its semantic value. Rather, on this view, its meaning consists merely of a ‘character’ which can be understood as a function from contexts of use to fixed contents (Kaplan 1989).² Contextualism thus claims that, due to what the predicate means, only in a context of use ‘ . . . is wrong’ will have a specific property as its semantic value—the context plays an ineliminable role in determining what that property is. There are two familiar versions of Contextualism according to which the characters of moral predicates contain an essential indexical element.³ According to these views, the contexts on which the characters of moral predicates operate include moral codes that are accepted by either the speaker (indexically, from the speaker’s perspective, my moral system) or her moral community (our moral system). Thus, according to the subjectivist versions of Contextualism (hereafter the SC-views), when I utter the

¹ My formulation of Contextualism is inspired by Wong (1984: ch. 5), Dreier (1990: 8; 1999; 2009: 79–81), Brogaard (2008: 385–6, 400), and Björnsson and Finlay (2010: 7–8). ² Character represents what one must understand in order to grasp the meaning of the term whereas the truth-evaluable content is what an expression says (Dreier 2009: 81). ³ Historically, these views are more often understood as versions of subjectivism and relativism. See Sumner (1906), Montaigne (1958 [1572–4]), Harman (1975), and Hobbes (2012 [1651]). ‘Character’ here refers to the Kaplanian characters defined above: the functions from contexts of use to fixed properties as semantic values.

48

Jussi Suikkanen

sentence ‘eating meat is wrong’, in this context the utterance ascribes a certain natural property N₁ (say that of not maximizing the amount of happiness) to eating meat because my moral code—as an essential element of the context—is against actions that are N₁. Likewise, when you utter that same sentence, your utterance ascribes a different natural property N₂ to eating meat because your moral code—as an essential element of your context—is against actions that are N₂. Thus, on this view, the character of ‘ . . . is wrong’ is a function from contexts of use—essentially consisting of the speakers’ moral codes—to certain natural properties determined those codes. Relativist versions of Contextualism (hereafter the RC-views) work in a similar way except here the moral code that determines the semantic value of ‘ . . . is wrong’ in a context is not the speaker’s personal code but rather the code of her moral community. Thus, according to RC-views, when I utter the sentence ‘eating meat is wrong’ that utterance will ascribe a certain natural property to actions of eating meat because the moral code internalized by the majority of people in my moral community is against the actions that have that property. Both subjectivist and relativist versions of Contextualism (hereafter the SRC-views) have important theoretical advantages. These views do not require positing any mysterious non-natural properties and yet they are able to rely on the standard truth-conditional, compositional semantics in explaining the meaning of more complex moral sentences. These views furthermore promise to solve many epistemic challenges facing other metaethical views and even help us to give an account of the internal connection between moral judgements and motivation.⁴ However, because there are many compelling objections to the SRCviews, until quite recently most metaethicists have taken these views to be merely good for philosophical target practice. Yet, more recently, several metaethicists have tried to develop the SRC-views in ways that could enable these views to avoid the standard objections to them (see footnote 1). This chapter takes part in that project by focusing on two objections. Section 3.2 first explains how the SRC-views seem unable to accommodate intuitive moral disagreements. It also considers how these views seem to lead to an unfortunate form of arbitrariness because we live in multicultural societies. This chapter then argues that these two problems can be solved together: the second problem provides a solution to the first one. Section 3.3 first explains how the popular Kratzer-styled contextualist accounts of epistemic modals suffer from a corresponding disagreement problem. It also outlines

⁴ For the latter advantage, see Dreier (1990); for the former, see Capps, Lynch, and Massey (2009).

Contextualism, Moral Disagreement, and Proposition Clouds

49

how Kai von Fintel and Anthony Gillies (2011) rely on proposition clouds to address that challenge. Section 3.4 then develops a similar response to the disagreement objection to the SRC-views by relying on an insight extracted from the arbitrariness objection. The basic idea is that, given that we live in multicultural societies, the contexts in which we utter moral sentences fail to provide unique truthevaluable contents for our moral assertions. When we make moral utterances, we can hence only put into play proposition clouds. Furthermore, because our moral discourse has a fundamental action coordinating function, we, as audience members, should react with disagreement to those propositions in the relevant clouds the rejection of which will best enable us to coordinate choices and actions. Section 3.5 concludes by outlining this proposal’s advantages over its main rivals.

3 . 2. D I S A G R E E M E N T A N D A R BI T R A R I N E S S Consider the following exchange: (1) (2)

Ann: It is not wrong to eat meat. Ben: No, eating meat is wrong!

When Ann and Ben utter (1) and (2), they intuitively disagree. The fundamental problem with the SRC-views is that they cannot guarantee this. Let’s stipulate that Ann and Ben have internalized different moral codes. In this case, the SC-views entail that according to Ann’s utterance eating meat lacks a certain natural property whereas according to Ben eating meat has a different natural property. In this situation, Ann and Ben would not be disagreeing but rather talking past one another—their claims could, after all, be both true at the same time.⁵ The RC-views do enable Ann and Ben to disagree when they belong to the same moral community.⁶ However, if Ann and Ben belong to different moral communities with different moral codes, these views again make them talk about different natural properties of actions. Consider also the following response from Ben (Björnsson and Finlay 2010: 19): (3) Ben: #When you utter the sentence ‘eating meat is not wrong’, what you say is true. Nevertheless, eating meat is wrong.

⁵ This objection was introduced by Moore (1912: ch. 3) and Hare (1952: 148–9). Since them, it has become ubiquitous. ⁶ See, for example, Shafer-Landau (2010: 286).

50

Jussi Suikkanen

This response makes no sense and yet the SRC-views seem to make it appropriate. On these views, what Ann asserts when she utters (1) is that eating meat does not have the relevant natural property determined by her moral code. According to the SRC-views, by uttering the first part of (3), Ben could agree that it is true that eating meat lacks that property. By uttering the second part of (3), he could then claim that eating meat is still wrong (i.e. eating meat has the natural property which his moral code picks out as salient). But, given that (3) makes no sense, the SRC-views arguably fail (see also Section 3.5). The second arbitrariness objection first observes that it is a philosophical myth that, in every area, there is a single moral community such that every person in that area is a member of that moral community and it only (Shafer-Landau 2004: ch. 10).⁷ Instead our societies are multicultural and we belong to many moral communities simultaneously. Consider ShaferLandau’s (2004: 44) real-life example in which, years ago in France, a young woman was about to be forced to undergo a clitoridectomy by her family. She was, fortunately, able to escape and she even won a lawsuit against her family. Consider then the young woman’s utterance ‘What they are doing to me is wrong!’ According to the SRC-views, in order to know whether this utterance is true, we must first know which moral code the context of the utterance picks out as the one that determines which natural property her assertion ascribes to clitoridectomy. The problem is that the utterance is made in a context in which several different communities are salient. Given that the young woman identified herself as a member of the French society, we might think that the code relevant in the context is the conventional French morality. Yet, the young woman also belonged to the Algerian immigrant community. Thus, the relevant moral code in the context could equally well be this community’s code. The problem is that, whichever community’s moral code is salient, that code determines which natural property the previous assertion ascribes to clitoridectomy and therefore also whether the assertion is true. If the French moral code is salient, then the young woman’s assertion is true whereas, if the Algerian immigrant culture’s code is salient, then that assertion might turn out to be false. In response, the defenders of the SRC-views could try to stipulate that the characters of moral predicates are such that they uniquely pick out a certain kind of communities and codes as salient in every context. The idea could be ⁷ Shafer-Landau (2004: ch. 8) calls a different objection to the SRC-views the arbitrariness objection. According to this objection, the concerns on which the moral standards are based cannot be supported by reasons in the SRC-framework and hence they must be arbitrary. For a response, see Blackburn (1998).

Contextualism, Moral Disagreement, and Proposition Clouds

51

that, given the meaning the moral terms have, it is always the larger society whose moral code is salient. Or, perhaps it is always the smaller, more tightknit community. Or, perhaps we should let the speaker herself choose which moral community and code is relevant. The fundamental problem with all these solutions is arbitrariness. As Shafer-Landau (2004: 46) points out, whether what the woman’s family was trying to do was wrong is a substantial moral question. It seems objectionable that such a question could be solved merely by considering whether the character of moral terms is such that the larger community’s moral code is always salient (or the smaller community’s or . . . ). That would be solving a substantial moral problem on arbitrary semantic grounds.

3 .3 . E P I S T E M I C M O D A LS , D I S A GR E E M E N T , A N D P R O P O S I T I O N C LO U D S This chapter responds to the previous objections by drawing from the recent debates concerning epistemic modals.⁸ The following two utterances are examples of how modal terms can be used in an ‘epistemic flavour’ (i.e. as epistemic modals): (4) (5)

Ed: My keys might be in the car. Fred: Your keys must be in the car.

The standard semantic model for understanding these statements was developed by Angelika Kratzer (1977; 1981; 1991) on the basis of David Lewis’s (1973) seminal work. It is known as the Lewis–Kratzer Semantics or the Standard Ordering Semantics. According to this model, epistemic modals are propositional operators. The propositions that embed within their scope (in (4) and (5), the proposition that Ed’s keys are in the car) are called ‘prejacents’. The semantic role of these operators is then to quantify over a certain contextually determined set of worlds. The strict necessity modal ‘must’ functions as a universal quantifier whereas the weak possibility modal ‘may’ functions as an existential quantifier. Must (p) is thus true when p is true in all the relevant worlds whereas May (p) is true when p is true in some of those worlds. The model then claims that the context of use determines in two stages the relevant sets of worlds over which the epistemic modals quantify. ⁸ For representative essays, see Egan and Weatherson (2011).

52

Jussi Suikkanen

The context first picks out a ‘modal base’, which consists of all the worlds that are consistent with a body of information that has certain properties salient in the conversational context (such as being what the speaker knows, for example). The conversational context then also provides an ordering source that ranks the worlds in the modal base with a certain standard. In the case of epistemic modals, this ordering source is based on what is stereotypical. The worlds in the modal base are ranked in terms of how well they match with what is the normal course of events relative to the conversational background. Must (p) is then true when p is the case in all the highest ranked worlds in the modal base whereas May (p) is true when p is true in some of them. The Standard Ordering Semantics too faces a disagreement objection (von Fintel and Gillies 2011: 114–17). Imagine that Ed and Greg are trying to find Ed’s keys. Ed utters (4) and Greg responds: (6) Greg: No, they cannot be in the car. You still had them when we came in.

Here Ed and Greg disagree about whether Ed’s keys might be in the car. When we apply the standard model, we must first consider which informational state is salient in the context as that state determines which worlds belong to the modal base. One alternative is to think that, in any context of use, what the speaker knows always constitutes the relevant information state. Yet, this assumption would make Ed and Greg talk past one another. Ed would be claiming that in some of the worlds compatible with what he knows the keys are in the car, but Greg could not reject that proposition by claiming that the keys are not in the car in the worlds compatible with what he knows. The information states that are salient in the relevant contexts cannot therefore be what the speakers know as individuals. The alternative is then to think that, when Ed utters (4), he is claiming that the keys are in the car in some of the worlds compatible with the aggregate of what he and Greg know. Given that it is possible to understand Greg’s claim to be about the fact that the keys are not in the car in those same worlds, this view can make Ed and Greg disagree. This proposal, however, leads to a different problem. Given how little we usually know about what others know, the previous response to the disagreement problem makes it mysterious how we could ever use epistemic modals to make warranted assertions. If Ed does not know what Greg knows about the keys, he cannot be justified in asserting that the keys are in the car in the worlds that are compatible with his own and Greg’s knowledge. Yet, surely, we often are justified in using epistemic modals even when we know little about what others know.

Contextualism, Moral Disagreement, and Proposition Clouds

53

Von Fintel and Gillies (2011: 117–29) provide an ingenious solution to the previous dilemma.⁹ They first observe that many contexts fail to ‘successfully resolve all indeterminacies and ambiguities’ (2011: 117). Imagine that you are at a conference where a stranger wants to know where you are from. In these types of cases, there are often not enough contextual cues that would help you to interpret what the stranger is asking (2011: 118). Does she want to know where you were born, where you grew up, where you did your PhD, where you work, or what? The context is just too sparse to resolve the ambiguity in question. Von Fintel and Gillies suggest that, similarly, when Ed uses the epistemic modal ‘might’ in (4), the context leaves open the knowledge of which group constitutes the information state that generates the modal base. The conclusion they draw is that utterances of epistemic modals take ‘place against a cloud of admissible contexts—one for each resolution of the relevant group that is compatible with the facts as they are’ when the utterance is made (2011: 118). They thus believe that, when we use epistemic modals, the best we can do is to put into play a proposition cloud. Each proposition put into play corresponds to the prejacent being true in either all or some of the worlds (depending on the epistemic modal) that are compatible with a body of information that is an aggregate of what a different group knows. For the audience, all the propositions put into play are then available for rejection and acceptance. In the previous example, by uttering (4), Ed puts into play at least two propositions: (i) that the keys are in the car at least in some of the worlds compatible with what he knows and (ii) that they are in the car in some of the worlds compatible with what he and Greg know in aggregate. In order to be justified in putting into play these propositions, Ed must be justified in asserting at least one of them. In this case, that proposition naturally is the one according to which the keys are in the car in the worlds compatible with what he knows (von Fintel and Gillies 2011: 120). In order to put the relevant propositions into play, he thus need not be justified in asserting the other proposition the content of which also depends on what Greg knows. This avoids the second horn of the dilemma above. To which propositions put into play by Ed should Greg then react? Von Fintel and Gillies (2011: 121) claim that, given the general standard norms of conversation, this is determined by the denial or acceptance of which proposition would make a suitably informative contribution to the conversation given its purpose. When Ed puts the previous propositions into play, he is asserting the proposition that the keys are in the car in the worlds compatible with what he knows and, in a way, simultaneously asking Greg ⁹ For objections, see, e.g. MacFarlane (2010).

54

Jussi Suikkanen

whether the other proposition is true too. Knowing whether (ii) is also true would, after all, be informative given that this would enable Ed to know whether the keys are in the car in the more restricted set of worlds also compatible with whatever additional evidence Greg has. Knowing whether those propositions are true would hence help Ed to find his keys, which is the purpose of the conversation. In order to make an informative contribution, Greg should therefore either confirm or reject the proposition (ii). This proposition is something Greg will have a reasonable opinion on assuming that he has some idea of what he knows about the location of the keys. If part of what Greg knows is that Ed came in with them, then he can justifiably rule out that the keys are in the car in the worlds compatible with what he and Ed know in aggregate. As a consequence, we can understand the utterances (4) and (6) to express a genuine first-order factual disagreement. Ed puts into a play a cloud of propositions by uttering (4) and by uttering (6) Greg denies that some of those very propositions are true. The disagreement between Ed and Greg can thus be understood in the traditional way as one person asserting and the other rejecting the very same proposition. Section 3.4 investigates whether the previous account can provide a new response to the disagreement objection to the SRC-views in metaethics. Before that, one further observation is needed. It has been argued that the truth of moral utterances is information-sensitive (Prichard 1932), which suggests that moral concepts contain an element of epistemic modality. Consider the miners case in which a group of ten miners are trapped in one of two shafts (Kolodny and MacFarlane 2010). We do not know in which shafts the miners are and rising waters threaten their lives. We have time to block only one shaft—if we block the wrong one everyone dies whereas if we block neither shaft only one miner dies. Many have the intuition that it would be wrong to block a shaft because doing so risks nine additional lives. It has then been suggested that, when we say that it would be wrong to block a shaft, we mean that in all the worlds both (i) compatible with what we know and (ii) ranked highest by morality we do not block a shaft.¹⁰ This interpretation applies the Standard Ordering Semantics to moral expressions by assuming that (i) what is morally best ¹⁰ The Standard Ordering Semantics itself is problematic here. Of the worlds that are compatible with what we know, many ordering sources would rank blocking A first in the worlds in which the miners are in A and blocking B first in the rest of the worlds. According to the Standard Ordering Semantics, this entails that it would not be wrong to block a shaft. For this reason, it has been suggested that we additionally need to rely on information-sensitive ordering sources that rank the relevant worlds in the epistemic modal base in terms of expected value (Dowell 2013; Silk 2014). For an objection, see Finlay (2016: 186–7).

Contextualism, Moral Disagreement, and Proposition Clouds

55

provides the relevant ordering source and (ii) what we know provides the epistemic modal base of the worlds that are ordered. The latter feature of the proposal means that this view builds epistemic modality into the semantics of ‘wrong’. This proposal enables us to apply the previous idea of proposition clouds directly to some moral disagreements. Consider the following conversation between Charles who does not know where the miners are and David who knows they are in A: (7) Charles: It would be wrong to block a shaft. (8) David: No, we are definitely required to block shaft A—the miners are in there.

In the current framework, Charles’s utterance puts into play a cloud of propositions because the context fails to determine a unique group whose aggregated knowledge generates the epistemic modal base for the evaluation of the truth of (7).¹¹ Thus, according to one proposition Charles puts into play, neither shaft is blocked in the worlds that morality ranks highest of those that are compatible with what he knows.¹² Similarly, according to another proposition he puts into play, neither shaft is blocked in the worlds that morality ranks highest of the worlds compatible with what he and David know in aggregate. David’s utterance (8) should then be interpreted as making an informative contribution to the conversation. Given that his evidence rules out the worlds in which the miners are in B, he can react with rejection to the proposition according to which no shaft is blocked in the worlds that are ranked highest by morality of the worlds that are compatible with what he and Charles know in aggregate. This contribution is highly informative because rejecting those propositions helps Charles to save all the miners.¹³ Thus, insofar as the truth of moral assertions is information-sensitive and epistemic modality is built into the semantics of moral terms, already the previous account of epistemic modals in its original form helps us to make sense of some moral disagreements in the contextualist framework. In this form, however, the view only helps us to accommodate moral disagreements

¹¹ According to Dowell (2013), Charles’s own referential intentions are sufficient to determine which proposition he asserts. It’s not clear, however, how Charles’s audience could access those intentions so as to know to which proposition they are to react. ¹² We only get this consequence if the ordering source provided by morality is itself information-sensitive. See footnote 9. ¹³ On this view, moral utterances made by informed agents can often play an instrumental role in helping the agents to choose the most moral course of action. This does not entail, however, that the agents’ interest in morality needs to be merely instrumental.

56

Jussi Suikkanen

between people who share the same moral principles but have different amounts of empirical information. Charles and David do not really disagree about whether it is right to save people but rather David only disagrees with Charles because he knows the location of the miners and thus is in a position to help save more lives. Yet, when Ann and Ben disagree when they utter (1) and (2), they can be in a context in which they know all the same empirical facts and they also know that they know the same facts. Because of this, their disagreement cannot be explained with the idea that one of them has some additional empirical information and thus is in a position to rule out additional worlds from the modal base. Rather, their disagreement is a more fundamental moral disagreement about what is right based on accepting different moral codes. The idea that speakers put different moral propositions into play with their moral utterances because epistemic modals are built into the meaning of moral terms cannot thus explain what is going on in these more fundamental moral disagreements. This is why Section 3.4 develops a contextualist account of disagreements that could be applied more broadly also to the more fundamental moral disagreements.¹⁴

3 . 4 . M O R A L D I S A G R E E M E N T AN D PROPOSITION CLOUDS My proposal begins from an observation concerning the traditional forms of relativism in metaethics: these views classified moral communities in a maximally coarse-grained way. They assumed that we all belong to just one moral community. The idea was that, within the borders of a state, the citizens of that state always form the relevant moral community whose moral standards determine which actions are right and wrong within that state. In some cases, the relevant moral community was assumed to be an even larger unit: the Western moral community, the Asian moral community, or the like. These views just did not seem to recognize any smaller moral communities. Shafer-Landau’s (2004: ch. 10) arbitrariness objection shows that these assumptions are untenable. He is surely right in pointing out that, in addition to large national moral communities, there are also many smaller ¹⁴ For reasons of space, the proposal below will not include epistemic informationsensitivity within the contextualist semantics of moral terms. The proposal is, however, fully compatible with adding that aspect of the character of moral terms to the account. For SRC-views formulated explicitly in that way, see Björnsson and Finlay (2010) and Dowell (2013).

Contextualism, Moral Disagreement, and Proposition Clouds

57

moral communities (such as the Algerian immigrant community in France) to which many of us belong. When we recognize these smaller moral communities and the fact that we can belong to many of them simultaneously, there is no reason why we could not classify moral communities in a more fine-grained way. We could think, plausibly enough, that there exists a plenitude of overlapping moral communities so that we all belong to a vast number of different ones. Some of these moral communities are formed by people who live in the same area (the Europeans, the Brits, the Brummies, etc.), whereas others are formed through class membership, age, general interests, shared political views, immigration, ethnicity, profession, Facebook groups, and the like. We can even stipulate that, whatever relation ties a certain group of individuals together, these individuals thereby form a moral community. Let us then return to the original disagreement. According to the RC-views, when Ann utters (1), given the character of ‘wrong’ and the context of use, her utterance asserts that the actions of eating meat lack a certain natural property determined by the moral code of her community. The immediate problem is that, as we have just seen, Ann belongs to a number of different moral communities and so there is no such thing as her unique moral community. Furthermore, if the context of utterance is an ordinary conversational context, it is hard to see what contextual cues would suffice to help Ann’s audience to interpret against the moral code of which community the truth of her utterance should be evaluated. The context in question will help to rule out some moral communities as irrelevant. Given the constraints provided by charity, presumably she cannot be talking about the type of actions that are forbidden by the moral code of the majority of the group consisting of herself, the oldest citizen of Iceland, and the fastest runner in Switzerland. Yet, even so, the context seems too poor to be able to disambiguate between all the things she could reasonably be interpreted to mean. The natural idea is then to think that, as a consequence, the best Ann can do by uttering (1) is to put into play a cloud of all the propositions that are not ruled out by the contextual cues as reasonable interpretations. Each one of these propositions asserts that the action of eating meat lacks a different natural property because the content of each proposition in the cloud is determined by a different moral community’s moral code. Ann is justified in putting into play this proposition cloud if she is in a position to assert at least one of the propositions in it. Given that Ann herself already forms a moral community of one person, Ann must at least be justified in putting into play the relevant proposition cloud. She has to be in a position to assert that eating meat lacks a certain natural property determined by her own personal moral code.

58

Jussi Suikkanen

To which proposition put into play by Ann should Ben then react? Above, it was suggested that the response to this question is provided by the general Gricean norms of conversation. Grice (1975: 45–6) suggested that the Cooperative Principle that captures what we can be expected to do in a conversation contains two maxims of quantity. We are to make our conversational contributions as informative as is required by the purposes of the conversation (we should not say too little) and we should also not make our contributions more informative than required by those purposes (we should not say too much). To which proposition should Ben then react so as to make his conversational contribution appropriately informative? Grice’s own formulation of the maxim of quantity answers this question. In order to evaluate the informativeness of a conversational contribution, we must first consider the purpose of the conversation in question. In the example concerning Ed’s keys, the purpose was to find Ed’s keys. This is why it was sufficiently informative for Greg to reject the propositions according to which the keys are in the car in the worlds that are compatible with what he knows. After all, ruling those worlds out was enough to enable Ed to find his keys. This means that, in order to find out to which propositions it would be appropriate for Ben to react, we need to consider the fundamental purpose of moral conversations. Here I will rely on a widely held assumption according to which one fundamental purpose of moral discourse is to coordinate choice and action in a mutually beneficial manner.¹⁵ The starting point of this assumption is that, because we all have a vast number of different needs and desires and we live in circumstances in which resources are not unrestrictedly abundant, there are bound to be conflicts between us concerning resource allocation.¹⁶ Yet, if such conflicts were frequent and escalated into violence, our lives would be in Hobbes’s (2012 [1651]: pt. 1, ch. 13, para. 9) words ‘solitary, poor, nasty, brutish, and short’. My assumption will be that one raison d’être of moral discourse and thought is to help us to avoid the previous type of escalating vicious conflicts.¹⁷ When individuals share a moral code (including its actionguiding principles and the corresponding reactive attitudes towards norm violations), they are able to avoid the previous kind of conflicts and their

¹⁵ This assumption thus grants that moral discourse generally serves an instrumental purpose. This again need not entail that the individuals who take part in the discourse are similarly motivated by instrumental reasons. ¹⁶ In John Rawls’s (1971: §22) words, we thus live in the ‘circumstances of justice’. ¹⁷ In fact, it can be argued that moral thought and talk evolved in human beings because of their ability to coordinate the actions of communities in a way that was an evolutionary advantage ( Joyce 2006: ch. 4).

Contextualism, Moral Disagreement, and Proposition Clouds

59

escalation in a way that also enables mutually beneficial cooperation. Because the adoption of a moral code requires certain practical attitudes, the individuals who adopt a shared moral code will thus come to have a more cooperative and altruistic stance towards interaction. They will constrain the selfish pursuit of their own interests so as to comply with the shared code, which prevents harm to others. They will also have certain shared expectations concerning the execution of contracts and the distribution of the benefits of social cooperation. All these shared attitudes then mean fewer conflicts and more of the kinds of social interaction that benefit everyone. Shared moral standards also prevent the escalation of conflicts by coordinating third parties to side with one of the two parties in a conflict and thus by putting social pressure on the parties in the conflict to solve their conflict peacefully.¹⁸ In the following, I thus assume that one important purpose of moral conversations is that having such conversations serves the fundamental action coordinating function of the moral discourse. Through moral conversations we can put pressure on the choices and actions of others in a way that is required for meeting our ‘competing needs in a social, cooperative setting’ (Blackburn 1993: 168). This assumption then helps us to answer the question of to which of the propositions put into play by Ann Ben should react so as to make his contribution suitably informative. I propose that, in the context of moral conversations, we should understand the informativeness of conversational contributions in terms of the contribution’s ability to coordinate behaviour in a way that serves the fundamental action coordinating purpose of the moral discourse. Recall that, when Ann utters (1), she puts into play a proposition cloud. According to every proposition in this cloud, the actions of eating meat lack a certain natural property determined by the moral codes of different moral communities. Hence, when Ben disagrees with Ann by uttering (2), he must be rejecting at least one of these propositions. By uttering (2), he must be asserting that eating meat actually has at least one of the salient natural properties determined by the moral codes of the relevant moral communities.¹⁹ Above, I acknowledged that there exists a plenitude of different moral communities. Yet, the mere fact that certain moral communities are against the type of actions to which eating meat belongs will put very little pressure on Ann’s actions. For example, Ann will not be moved by the fact that the ¹⁸ See DeScioli and Kurzban (2013). ¹⁹ In Finlay’s (2017: 188–90) terminology, my strategy to accommodate intuitive moral disagreements is ‘content-based’ rather than non-content-based (see Section 3.5 in this chapter).

60

Jussi Suikkanen

majority of those people who, like her, were born at 4.24 a.m. are against certain type of actions.²⁰ Firstly, it can well be that, even if most of those people are against the actions of the type in question, Ann does not belong to the majority in this respect. This means that the mere fact that a moral community to which she belongs is against eating meat need not in itself provide an internal reason for her not to eat meat—the kind of reason that would motivate her in practical reasoning that begins from her own cares and concerns.²¹ Secondly, Ann need not identify strongly with the other members of this group. She does not need to think that who she really is as a person is in some way dependent on being a member of this group. Finally, the members of this group also seem too unconnected to one another so as to be able to coordinate their reactive attitudes and actions in a way that would be required for putting social pressure on the choices and actions of the other members of the group. For these reasons, if Ben were to reject the proposition according to which actions of eating meat do not have a certain natural property determined by the moral code of the majority of people born at 4.24 a.m. are, this would not be an appropriately informative contribution to his conversation with Ann. It would not be sufficiently informative because it would fail to serve the fundamental action coordinating function of the moral discourse. The previous example also helps us see to which propositions Ben should react so as to make an informative, action coordinating contribution. Ann also belongs to at least some moral communities such that: i. she shares the same core cares and concerns that are reflected in the moral code that is accepted by the majority of the community’s members, ii. she considers her own personal identity to be at least in part grounded in being a member of the community, and iii. the community in question is sufficiently close-knit both to be able to put significant social pressure on the choices and actions of its members and to be able to effectively side with one party in conflict cases. According to at least some of the propositions put into play by Ann, eating meat lacks a natural property determined by what types of actions are ²⁰ Because it is so obvious that no agent will be moved by this fact, the proposition that this community is not against eating meat did probably not belong to the proposition cloud put into play by Ann. The contextual cues present in all contexts should be sufficient for Ben to know that this could have been what Ann meant. I am merely using the example for illustrative purposes to provide a contrast to the communities that matter. ²¹ See Williams (1979).

Contextualism, Moral Disagreement, and Proposition Clouds

61

opposed by the previous special kind of moral communities of which Ann is a member. If Ben can reject these propositions (i.e. if he is in a position to assert that in fact eating meat has those very natural properties), then presumably this will put pressure on Ann’s actions in a way that can coordinate her and Ben’s behaviour. This is why rejecting those propositions would be informative (assuming that we understand the informativeness of moral utterances in terms of the moral discourse’s fundamental action coordinating purpose). As a consequence, we know to which propositions Ben should react in the cloud put into play by Ann: to those that put significant pressure on Ann’s choices and actions. These are the propositions in which the semantic value of ‘ . . . is wrong’ is determined by the moral codes of those moral communities that have the features i.–iii. introduced above. Let me finish by making a few further observations. Firstly, even if I have described to which propositions Ben should react, this does not mean that he will respond to those propositions. Ben can fail to make an informative, action coordinating contribution. He can, for example, react to the proposition according to which eating meat does not have a certain natural property determined by his own personal moral code.²² Yet, rejecting this proposition will have only a limited effect on Ann’s behaviour—it will move her only to the extent that she cares about her relationship to Ben. How can Ann know to which proposition Ben is actually reacting? Ben’s own utterance (2) itself is not very helpful in this respect—it will not enable Ann to understand what Ben means. After all, just like Ann’s own utterance, it too can only put into play a proposition cloud. There are, however, two methods that Ann can use to home in on the meaning of Ben’s response. Firstly, she too has internalized the same conversational maxims that she should assume guide Ben’s conversational contributions. Thus, she can initially assume that Ben is reacting to the propositions that matter in terms of coordination. As a consequence, she is warranted in assuming that Ben is talking about the natural properties that are determined by the moral codes of the communities that are important to her are.²³

²² Assuming that this proposition was one of the propositions put into play by Ann, in this case Ben would still be disagreeing with her. It’s just that his disagreeing contribution to their conversation would not be very informative. ²³ One potential objection at this point concerns disagreement cases in which the participants share no common cares, concerns, or moral standards. Imagine a disagreement case like this between a vegan and a carnivore, where the participants are also members of distinct vegan and carnivore communities. In this case, it might be argued that these individuals could not disagree according to my proposal because these individuals do not belong to the same communities that could coordinate their actions effectively. Here I want to suggest that one of the propositions in the cloud put into play by the

62

Jussi Suikkanen

Secondly, moral conversations rarely consist only of a short exchange of blunt utterances such as (1) and (2). Rather, the conversations that start in this way usually continue with a game of giving and receiving reasons. Ben would presumably next talk about the considerations that he thinks make eating meat wrong—he might argue that factory farming causes suffering or that meat production is a significant contributor to climate change. One way to understand these further contributions is to think of them as narrowing down the context of utterance for (2). They provide contextual cues for Ann that enable her to interpret to which proposition Ben is reacting. The previous claims about reasons, for example, should tell her that Ben’s utterance in (2) must in some way concern the moral communities to which she belongs, the majorities of which are against actions that cause suffering or climate change. That is, by referring to these reasons, Ben can help Ann to understand that he rejected the propositions in the relevant cloud according to which eating meat does not have the natural properties determined by what type of actions are disapproved of by the moral communities that care about pain, suffering, and climate change and of which Ann is a member. Let me then conclude by summarizing how the previous proposals enable the SRC-views to make sense of the genuine disagreement between Ann and Ben: (i) given the context of utterance and the lack of contextual cues, when Ann utters (1) she can only put into play a cloud of propositions; (ii) by uttering (2) Ben reacts to some of the propositions put into play by outright rejection and so the disagreement is an ordinary first-order factual disagreement; (iii) we can explain to which propositions Ben should react to by considering the action coordinating function of the moral discourse; and (iv) these norms and further exchanges about reasons can help Ann to understand what Ben in fact means.

3.5. THE ALTERNATIVES This final section will briefly outline two alternative theories, which the defenders of the SRC-views have already tried to use for making sense of moral disagreements. I will also introduce some of the problems of these proposals and consider how the previous section’s proposal can avoid them. vegan could be that the use of animal products has a certain natural property determined by the moral code accepted by the majority of mankind. The carnivore could then reject at least this proposition. This contribution to the conversation would, of course, be only weakly coordinating but sometimes, given the conversational context, the most informative utterances we can make are not very informative.

Contextualism, Moral Disagreement, and Proposition Clouds

63

Traditionally, the defenders of the SRC-views have often endorsed the expressivists’ disagreement in attitude account of moral disagreement.²⁴ This proposal grants that Ann’s and Ben’s assertions (1) and (2) can both be true at the same time because they are about different natural properties. This proposal thus accepts that there need not be a factual disagreement between Ann and Ben. It then, however, notes that Ann and Ben presumably have conflicting attitudes of approval and disapproval towards eating meat.²⁵ Ann is, after all, implying that actions of eating meat are not the type of actions against which her own moral code is whereas Ben implies that those actions are the kind of actions against which his own moral code is. Ann and Ben can thus be claimed to have a disagreement in attitude: they have a ‘clash of practical attitudes’ in the same way in which we can disagree when we have different plans concerning where we are to go for dinner together. To see why this view is problematic, we must first notice that Ben could express his disagreement with Ann also in the following way (Schroeder 2008: 17): (9)

Ben: What Ann just said is false.

One problem is that, according to the traditional interpretations of the SRC-views, Ben could not make this assertion because the proposition Ann asserted—what she said—is true according to these views. The actions of eating meat can really lack the natural property determined by Ann’s own moral code. In response, Brogaard (2008: 400), Björnsson and Finlay (2010: 20–1), and Finlay (2017: 201) have suggested that, due to the conversational context, the expression ‘What Ann just said . . . ’ does not here pick out the proposition Ann asserted but rather the proposition which Ben could assert by uttering ‘Eating meat is wrong’ himself in the context.²⁶ However, even if there exists linguistic devices that pick out unasserted propositions in the previous radically context sensitive way, ‘What X just said . . . ’ does not seem to one of them—at least not in ordinary non-moral contexts.²⁷ Rather, in these contexts, this expression seems to function uniformly as a device for indirect quotation. ²⁴ Stevenson introduced the account first in his prescriptivist framework (1944: 2–8). In the debates concerning the SRC-views, it has been defended, for example, by Wong (1984; 2011), Harman (1996), Dreier (1999: 569; 2009: 106), Jackson (2008: 81–2), Björnsson and Finlay (2010: 27–8), and Finlay (2014). ²⁵ Finlay (2017: 191) calls this account ‘quasi-expressivist’ because many defenders of the view suggest that moral utterances pragmatically express the conflicting attitudes. ²⁶ As Finlay (2017: 200–1) notes, defenders of the disagreement in attitude view could turn metalinguistic at this point. For problems of this response, see below. ²⁷ For examples of such devices, see Björnsson and Finlay (2010: 19).

64

Jussi Suikkanen

Consider a situation in which Ann utters: (10) Ann: I was told that Sally stole the money.²⁸

In this example, I could use the phrase ‘Yes, that’s true’ to agree with the proposition that Sally stole the money (which was not asserted by Ann). Yet, by saying that ‘What she just said is false!’ I would only be able to deny that Ann was told that Sally stole the money (i.e. the asserted proposition). In this non-moral context, the reference of this expression just does not switch from the asserted proposition about what Ann was told to the unasserted proposition concerning what Sally has done. This means that defending the SRC-views with the disagreement in attitude strategy requires thinking that expressions such as ‘What she just said . . . ’ pick out the propositions that are not asserted in the moral contexts and yet asserted propositions in the nonmoral contexts. The problem is that this assumption seems ad hoc—it seems that the only motivation for making it would be theory-driven, motivated by an attempt to save the SRC-views from the objection.²⁹ In contrast, the proposal described above does not require the previous assumption. It is compatible with the view that ‘What she just said . . . ’ functions also in the moral contexts in the standard way as a device for indirect quotation that uniformly picks out the proposition asserted by the previous speaker. It’s just that, according to this proposal, at least in the moral contexts, the previous speaker puts into a play a whole cloud of propositions and so the new speaker can, in principle, use a sentence beginning with ‘What she just said . . . ’ to express her agreement or disagreement with any one of these propositions. Thus, one reason for preferring the proposition cloud proposal is that it does not require making any ad hoc assumptions about how certain expressions used to refer to other propositions function. The second suggestion is that we should understand moral disagreements such as the one between Ann and Ben as metalinguistic negotiations concerning which moral concepts should be adopted.³⁰ Many formulations

²⁸ I borrow this example from Björnsson and Finlay (2010: 19). ²⁹ Perhaps due to this kind of concerns, more recently Finlay (2017: 202) has suggested that we can explain at least some of our intuitions about disagreements away on the basis of semantic opacity—the idea that we, as ordinary speakers, are not aware of the fact that in some cases we can end up talking past one another. For further problems of the view, see Ridge (2013) and Björnsson (2017). ³⁰ For different variations of the view, see Robinson (2010), Plunkett and Sundell (2013), Khoo and Knobe (2016), Silk (2016: 128–9; 2017: 214, 222–3), and Bolinger (ms.), even if Silk (2017: 225) himself does not classify his view as metalinguistic (for a discussion, see Finlay 2017: 196 fn. 26).

Contextualism, Moral Disagreement, and Proposition Clouds

65

of this proposal rely on the idea that different speakers and communities have different moral concepts, which is not compatible with the kind of Contextualism explored in this chapter. For this reason, I need to reformulate the proposal in a slightly different way. Let’s assume that the character of the moral terms is such that when Ann asserts (1) her assertion is about a certain natural property determined by the moral code of her own moral community. Likewise, when Ben asserts (2) his assertion is about a different natural property determined by the moral code of his moral community. As a consequence, let’s assume that there is no first-order factual disagreement between Ann and Ben as they are talking past one another. Yet, intuitively we would think that they disagreed. Here’s how the metalinguistic proposal promises to accommodate that disagreement intuition. One way to understand the character of a moral term is that it is a function from the contexts of use to different natural properties as semantic values. This means that, in the previous example, even if Ann and Ben are talking to one another and so in a loose sense belong to the same conversational context, we should properly speaking think that their utterances are made in different contexts of use. The context of use for Ann’s utterance is the moral code of her community, which explains why the semantic value of ‘ . . . is wrong’ in Ann’s mouth is a certain natural property. Likewise, the context of use for Ben’s utterance is a different moral code—his community’s code, which explains why the semantic value of ‘ . . . is wrong’ in his mouth is a different natural property. The metalinguistic proposal then is that, even if Ann and Ben are literally talking past one another and there is no factual first-order disagreement between them, they can be understood to be disagreeing in a different, nonstandard way. This is because, in the conversation, they are both trying to get their conversational partner to move to the same context of use as they are in themselves. Thus, by asserting (2), Ben is trying to get Ann to join his moral community that is against eating meat so that the context of use for Ann’s moral utterances too would become the moral code of his moral community. If Ben were successful, not only would Ann’s previous assertion become false but she would also come to adopt the vegetarian moral code of Ben’s community, which would effectively coordinate her and Ben’s actions to not eating meat. As we saw above, this type of action coordination can be understood to be the reason why the conversation is taking place in the first place. Likewise, Ann too is, according to this proposal, trying to get Ben to switch over to the context of the moral code of her own moral community, which too would help her to get Ben to coordinate his actions with her meat eating habits.

66

Jussi Suikkanen

The problem with this proposal is that it fails to make sense of disagreements in eavesdropping cases.³¹ Consider the following exchange between Isabel and James who are listening to the conversation between Ann and Ben from a distance: (11) Isabel: No, Ann is mistaken—eating meat is wrong! (12) James: You are right—she is mistaken.

Intuitively, here too, Isabel disagrees with Ann. Yet, this disagreement between Isabel and Ann cannot be construed as a metalinguistic negotiation concerning which moral code should be adopted as the shared context of use. The purpose of Isabel’s utterance cannot be to get Ann to join the context of Isabel’s moral code given that Ann is not even aware of Isabel’s existence and so her utterance cannot have any influence on which moral code Ann adopts as the relevant context for her moral utterances. The fact that Isabel disagrees with Ann also cannot be understood as consisting of a metalinguistic negotiation between Isabel and James given that they already know that they both share the same moral code and thus belong to the same context. The metalinguistic proposal thus seems to lack the resources required for dealing with disagreements such as the one between Ann and eavesdropper Isabel. In contrast, the proposal outlined in Section 3.4 can do better. According to it, Ann’s assertion (1) puts into play a proposition cloud. Her audience can then react to these propositions by either accepting or rejecting them. This is why Isabel can react with rejection to those propositions in play, the rejection of which would make a suitably informative contribution to the conversation she is having with her fellow eavesdropper James (see von Fintel and Gillies 2011: 127–9). Thus, Isabel’s disagreement with Ann can again be understood as an ordinary first-order factual disagreement— she is flat out rejecting one of the things Ann said. Yet, her rejection of that proposition put into play by Ann can be informative in the conversion she has with James, as it can coordinate action and choice. By uttering (11), Isabel can confirm whether she and James are on the same page with regard to eating meat and also check whether they would side with the same person in disagreements over meat-eating on future occasions. ³¹ For a similar objection, see Finlay (2017: 194). One strategy to respond to this concern is to limit the scope of the disagreements to which the account is supposed to apply (see Khoo and Knobe (2016: 3) and Silk (2017)). Yet, this entails that my proposal should be preferred on the grounds of simplicity. According to Finlay, different versions of the metalinguistic proposal also threaten to mislocate the disagreements (2017: 192–3), fail to accommodate disagreements between people who are not speaking to one another (195), and/or entail both too many disagreements and too few instances of agreement (197).

Contextualism, Moral Disagreement, and Proposition Clouds

67

This means that one advantage of my proposal is that it can accommodate a wider range of disagreement cases than the metalinguistic proposal. In addition to enabling the SRC-views to understand moral disagreements as ordinary first-order factual disagreements and thus in addition to addressing the most common objection to these views, the proposal outlined in Section 3.4 also therefore does better than the previous proposals. It does not require making any ad hoc assumptions about how certain devices of indirect quotation function in moral contexts and it can also deal with the disagreement cases, which the metalinguistic proposal fails to accommodate. I thus conclude that the proposition cloud account is the most promising alternative for the contextualists.

References Björnsson, Gunnar (2017): ‘The Significance of Ethical Disagreement for Theories of Ethical Thought and Talk’. In T. McPherson and D. Plunkett (eds): The Routledge Handbook of Metaethics (New York: Routledge), 275–91. Björnsson, Gunnar and Finlay, Stephen (2010): ‘Metaethical Contextualism Defended’. Ethics 121/1: 7–36. Blackburn, Simon (1993): ‘How to Be an Ethical Anti-Realist’. In his Essays in Quasi-Realism (Oxford: Oxford University Press), 166–81. Blackburn, Simon (1998): ‘Moral Relativism and Moral Objectivity’. Philosophy and Phenomenological Research 58/1: 195–8. Bolinger, Renee (ms.): ‘Metalinguistic Negotiations in Moral Disagreement’. Brogaard, Berit (2008): ‘Moral Contextualism and Moral Relativism’. Philosophical Quarterly 58/232: 385–409. Capps, David, Lynch, Michael, and Massey, Daniel (2009): ‘A Coherent Moral Relativism’. Synthese 166/2: 413–30. DeScioli, Peter and Kurzban, Robert (2013): ‘A Solution to the Mysteries of Morality’. Psychological Bulletin 139/2: 477–96. Dowell, Janice (2013): ‘Flexible Contextualism about Deontic Modals’. Inquiry 56/2–3: 149–78. Dreier, James (1990): ‘Internalism and Speaker Relativism’. Ethics 101/1: 6–26. Dreier, James (1999): ‘Transforming Expressivism’. Noûs 33/4: 558–72. Dreier, James (2009): ‘Relativism and the Problem of Disagreement’. Philosophical Perspectives 23: 79–110. Egan, Andy and Weatherson, Brian (eds) (2011): Epistemic Modality (Oxford: Oxford University Press). Finlay, Stephen (2014): Confusion of Tongues: A Theory of Normative Language (Oxford: Oxford University Press) Finlay, Stephen (2016): ‘ “Ought”: Out of Order’. In N. Charlow and M. Chrisman (eds): Deontic Modality (Oxford: Oxford University Press), 169–99.

68

Jussi Suikkanen

Finlay, Stephen (2017): ‘Disagreement Lost and Found’. In R. Shafer-Landau (ed.): Oxford Studies in Metaethics, Volume 12. Oxford: Oxford University Press, 187–205. Fintel, Kai von and Gillies, Anthony (2011): ‘ “Might” Made Right’. In A. Egan and B. Weatherson (eds): Epistemic Modality (Oxford: Oxford University Press), 108–30. Grice, H. Paul. (1975): ‘Logic and Conversation’. In P. Cole and J. L. Morgan (eds): Speech Acts (New York: Academic Press), 41–58. Hare, R. M. (1952): The Language of Morals (Oxford: Oxford University Press). Harman, Gilbert (1975): ‘Moral Relativism Defended’. Philosophical Review 84/1: 3–22. Harman, Gilbert (1996): ‘Part 1: Moral Relativism’. In G. Harman and J. J. Thomson’s Moral Relativism and Moral Objectivity (Oxford: Blackwell). Hobbes (2012 [1651]): Leviathan, vols. I–III, ed. by N. Malcolm (Oxford: Oxford University Press). Jackson, Frank (2008): ‘The Argument from the Persistence of Moral Disagreement’. In R. Shafer-Landau (ed.), Oxford Studies in Metaethics, Volume 3. Oxford: Oxford University Press, 75–86. Joyce, Richard (2006): The Evolution of Morality (Cambridge, MA: MIT Press). Kaplan, David (1989): ‘Demonstratives’. In J. Almog, J. Perry, and H. Wettstein (eds): Themes from Kaplan (New York: Oxford University Press), 481–563. Khoo, Justin and Knobe, Joshua (2016): ‘Moral Disagreement and Moral Semantics’. Noûs, https://doi.org/10.1111/nous.12151. Kolodny, Niko and MacFarlane, John (2010): ‘Ifs and Oughts’. Journal of Philosophy 107/3: 115–43. Kratzer, Angelika (1977): ‘What “Must” and “Can” Must and Can Mean’. Linguistics and Philosophy 1/3: 337–55. Kratzer, Angelika (1981): ‘The Notational Category of Modality’. In H. Eikmeyer and H. Rieser (eds): Words, Worlds, and Contexts (Berlin: de Gruyter), 38–74. Kratzer, Angelika (1991): ‘Modality’. In A. Stechow and D. Wunderlich (eds.): Semantics: An International Handbook of Contemporary Research (Berlin: de Gruyter), 639–50. Lewis, David (1973): Counterfactuals (Oxford: Blackwell). MacFarlane, John (2010): ‘Epistemic Modals: Relativism vs. Cloudy Contextualism’. MS, Chambers Philosophy Conference on Epistemic Modals, University of Nebraska. Montaigne, Michel de (1958 [1572–4]): ‘Of Custom, and not Easily Changing an Accepted Law’. In M. Frame (ed.): The Complete Essays of Montaigne (Stanford, CA: Stanford University Press), 77–89. Moore, G. E. (1912): Ethics (Oxford: Oxford University Press). Plunkett, David and Sundell, Tim (2013): ‘Disagreement and the Semantics of Normative and Evaluative Terms’. Philosophers’ Imprint 13/23: 1–37. Prichard, H. A. (1932): ‘Duty and Ignorance of Fact’. Proceedings of the British Academy 18: 67–92. Rawls, John (1971): A Theory of Justice (Cambridge, MA: Harvard University Press).

Contextualism, Moral Disagreement, and Proposition Clouds

69

Ridge, Michael (2013): ‘Disagreement’. Philosophy and Phenomenological Research 86/1: 41–63. Robinson, Denis (2010): ‘Reflections on Moral Disagreement, Relativism, and Skepticism about Rules’. Philosophical Topics 38/2: 131–56. Schroeder, Mark (2008): Being For (Oxford: Oxford University Press). Shafer-Landau, Russ (2004): Whatever Happened to Good and Evil? (Oxford: Oxford University Press). Shafer-Landau, Russ (2010): The Fundamentals of Ethics (Oxford: Oxford University Press). Silk, Alex (2014): ‘Evidence Sensitivity in Weak Necessity Modals’. Journal of Philosophical Logic 43/4: 691–723. Silk, Alex (2016): Discourse Contextualism: A Framework for Contextualist Semantics and Pragmatics (Oxford: Oxford University Press). Silk, Alex (2017): ‘Normative Language in Context’. In R. Shafer-Landau (ed.), Oxford Studies in Metaethics, Volume 12. Oxford: Oxford University Press, 206–43. Stevenson, Charles (1944): Ethics and Language (New Haven, CT: Yale University Press). Sumner, William (1906): Folkways (Boston, MA: Ginn and Company). Williams, Bernard (1979): ‘Internal and External Reasons’. In R. Harrison (ed.): Rational Action (Cambridge: Cambridge University Press), 101–13. Wong, David (1984): Moral Relativity (Los Angeles: University of California Press). Wong, David (2011): ‘Relativist Explanation of Interpersonal and Group Disagreement’. In S. Hales (ed.): A Companion to Relativism (Malden, MA: WileyBlackwell), 411–29.

4 Taking Prudence Seriously Guy Fletcher 4 . 1. I N T R O D U C T I O N Philosophers have long theorized about which things make people’s lives go well, and why, and the extent to which morality and self-interest can be reconciled. By contrast, we have spent little time on meta-prudential questions, questions about prudential discourse.¹ This is surprising given that prudence is, prima facie, a normative form of discourse and, as such, cries out for further investigation of how exactly it functions and whether it has problematic commitments.² It also marks a stark contrast from moral discourse, about which there has been extensive theorizing. By ‘prudential discourse’ I mean two things. Firstly, thought and talk about what’s good/better/best for, or bad/worse/worst for, harms, benefits (etc.) someone. Secondly, thought and talk about what someone prudentially ought to do, has reason to do, must do, or needs to do, along with analogous claims about attitudes.³ Prudential discourse thus encompasses evaluative prudential discourse (e.g. ‘good for’) and deontic prudential discourse (e.g. ‘oughtprudential’).⁴ Here are some paradigmatic instances: ‘You’ll be better off if you have the surgery.’ ‘Her spouse’s death was terribly bad for her.’

¹ Some exceptions to this are Railton (1989), Hooker (1991), Rosati (1995a; 1995b; 1996), and Darwall (2004). ² Much of the history of moral philosophy presumes that prudential normativity is genuine and more easily justifiable than moral normativity. c.f. Darwall (2016). ³ A complication: ‘good for’ (and so on) talk extends beyond welfare subjects (beings with levels of well-being). For example: ‘oil is good for engines’, ‘sand is bad for watches’. How to draw the boundary between these uses of ‘good for’ and those connected to wellbeing is an issue that I bracket here. See Kraut (2007), Fletcher (2012; ms.), Finlay (2014), Rosati (ms.). ⁴ ‘Deontic’ is an imperfect label, I grant.

Taking Prudence Seriously

71

‘Sarah should take the job.’ ‘You need friends in order to live well.’

The big picture meta-prudential questions that prudential discourse gives rise to include (at least): (1) Are prudential judgements normative judgements? (2) Does prudential discourse involve the ascription of normative properties (in either a deflationary sense or a robust sense) and, if robust, what is the nature of prudential properties? (3) Do evaluative prudential properties generate prudential reasons? (4) Is prudential talk partly (or wholly) expressive? Herein, I explicitly take up questions 1 and 3. Some of my discussion bears on question 2 but I leave question 4 (and others) for another day. Here is the plan. I first examine (Section 4.2) whether there is a distinct set of prudential reasons, generated by evaluative prudential properties. I do so because this is an important issue in its own right but also because points that arise in that section bear on the later discussion. I argue that there are prudential reasons (and that prudential reasons of one sort are compatible with Humeanism about reasons for action). More precisely, my thesis is: P V M (PVM): Evaluative well-being facts generate agent-relative reasons (for action or for attitudes) for the relevant agent.

In the second half of the chapter (Sections 4.3–4.5) I take up question (1), asking whether prudential discourse is normative.⁵ To motivate doing so, note that many think it obvious whether prudential discourse (etc.) is normative.⁶ Unfortunately, they disagree. Moreover, their discussion of the question tends to be in passing. On the positive side we have Steve Finlay, who writes: There is close isomorphism between morality and self-interest. For one thing, each is a normative domain: there is a moral “ought” and an “ought” of self-interest.⁷

⁵ For analogous inquiries into other prima facie normative properties and domains, see Schroeder (2015) and Sylvan (2018). ⁶ I think it is fair to attribute to them the implicit view that the answer is obvious by the limited amount of space they spend arguing for their respective views. The ‘etc.’ here is because some of these claims are specifically about concepts, some about properties, some about domains. Nonetheless, I think that they are all fairly interpreted as about prudential discourse. ⁷ Finlay (2007: 138).

Guy Fletcher

72 And David Enoch who writes:

Paradigmatic examples may however help clarify the kind of things I refer to as normative[:] that I should go on a diet, that you have a reason to read Kant [ . . . ] that pain is pro-tanto bad for the person whose pain it is[.]⁸

By contrast, others contend that prudential discourse is non-normative. Bart Streumer comments: [T]hat a painting is beautiful can be a reason to look at it and can make it wrong to destroy it, and being a reason and being wrong are normative properties. But that does not mean that beauty is itself a normative property. [S]imilar claims apply to disgust, funniness, harm, benefit[.]⁹

And Jonas Olson writes: The second consideration that makes me hesitant to accept the idea that welfare should be seen as a normative concept is purely economical. We have a plethora of normative concepts already[.] Why introduce yet another one if it seems that we can do without it?¹⁰

To simplify things, I will assume that some region of discourse is normative if the judgements essential to it are normative judgements. Thus, I will argue that prudential discourse is normative by arguing for the following thesis: P J  N (PJN): ments are normative judgements.

Prudential judge-

A natural retort at this point: What makes a judgement a normative judgement? Before outlining my strategy for establishing PJN, let me introduce and set aside three alternate proposals whose weaknesses are instructive. First, I do not answer the question of what makes a judgement a normative judgement via some meta-normative theory. For instance, if one characterizes normative judgements as essentially involving ascription of non-natural, irreducibly normative, properties then one’s characterization has immediately assumed a form of non-naturalism or robust realism (or error theory).¹¹ By contrast, if one characterizes normative judgements as essentially desire-like states then one has immediately assumed a form of non-cognitivism (and the same goes for other parochial proposals). It is not

⁸ Enoch (2011: 1). The reading example could be read differently—as a moral reason to read—but the other examples are plausibly prudential. ⁹ Streumer (2017: 105) (my italics). ¹⁰ Olson (2006: 176). Granted, Olson’s claim is specifically about the concept . ¹¹ For discussion of various options for characterizing what makes a property normative, see Streumer (2017: §45).

Taking Prudence Seriously

73

my ambition here to argue for a specific meta-prudential theory, only the more minimal thesis that prudential discourse is a normative form of discourse. I thus need a less parochial way of characterizing normative judgements. Second, I do not want to treat the normativity of prudential judgements as settled by the fact that such judgements are made using evaluative and deontic terms. The trouble with this test alone is that there are many kinds of judgements that deploy evaluative and deontic vocabulary that seem not to belong in the same theoretical category as e.g. moral judgements, the paradigmatic instance of normative judgement.¹² Familiar examples here are judgements about attributive goodness—good kettles and good toasters, etc.—the rules of clubs and games, legal judgements, and etiquette judgements. Though these judgements are normative in some sense of ‘normative’ they are not normative in the way moral judgements are. Third, one might suggest that prudential judgements are normative if they ascribe properties that entail normative reasons. Unfortunately this proposal collapses the distinction between normative properties and normatively-relevant properties (and between normative judgements and normatively-relevant judgements).¹³ Many judgements ascribe properties that entail normative reasons but not all of these are plausibly normative judgements themselves. For example: ‘a phone is ringing in the cinema’, ‘the fridge is empty’, ‘that’s a snake’ are judgements that ascribe properties that entail normative reasons without themselves being normative judgements. I thus don’t want to characterize normative judgements in terms of reasonsentailment.¹⁴ The alternate strategy that I will pursue I call the ‘parity strategy’. This is to argue for PJN by analogy with moral discourse. It is widely held that morality is a normative domain. Thus if we can find relevant similarities between moral judgements and prudential judgements, and no significant disanalogy, that gives us grounds for holding that prudential judgements are normative. I will pursue such a strategy, arguing, first (Section 4.3), that the features of moral judgements that metaethicists appeal to when articulating, explaining, and justifying the claim that moral judgements are normative are also possessed by prudential judgements. I then consider (Section 4.4) various objections to the analogy. ¹² I use ‘normative’ in the broad(er) sense that encompasses the evaluative rather than contrasting with it. ¹³ This distinction (in terms of facts) comes from Parfit (1997). ¹⁴ What about necessary reason-entailment? I think not. Pleasure and pain facts are plausible candidates for being necessarily reason-entailing without, themselves, being normative facts.

74

Guy Fletcher 4. 2 . PR UD ENTI AL R EA S ONS

We are accustomed to thinking that well-being at least sometimes generates normative reasons because at least some moral duties (etc.) are explained by facts about making lives go better or worse. Without denying that, the question here is not whether well-being generates normative reasons at all. Rather, it is whether there are specifically prudential reasons (which are distinct from moral reasons and the like). What must be true for there to be normative prudential reasons (I henceforth sometimes suppress ‘normative’ for brevity)? There are prudential reasons if and only if the following is true: P V M (PVM): Evaluative well-being facts generate agent-relative reasons (for action or for attitudes) for the relevant agent.¹⁵

To better understand PVM it is useful to look at two general ways of denying it (I consider instances of each strategy shortly).¹⁶ One way to deny PVM is to allow that things are (e.g.) good for people but deny that such facts generate distinctly prudential reasons.¹⁷ Thus one might reject the existence of prudential reasons by holding that all reasons for action, even if generated by well-being, are agent-neutral. A second way to deny PVM is to hold that there are agent-relative reasons for an agent (to e.g. promote their own well-being) but deny that these reasons are fundamentally generated by evaluative well-being facts. This would be true if one thought that all reasons were generated by some other property P and that sometimes an agent’s promoting their own well-being had property P. Thus, on this view, an agent sometimes has reasons to act in ways that will, in fact, promote their own well-being but this reason does not stem from the fact that by so acting they will promote their own well-being but instead from the P property of the outcome. We now know what must be true for there to be prudential reasons. Are there any? It seems like there are. Across many decision-making contexts, one assesses options at least partly with an eye to which is in one’s best interests. ¹⁵ I leave aside the buck-passing proposal that there are prudential reasons but generated (only) by the grounds of evaluative well-being properties. Though I think this incorrect, I lack the space to address it in full. I don’t deny that the grounds of evaluative well-being properties can generate reasons. ¹⁶ Another way to deny PVM is via the view that nothing is (e.g.) good for anyone. This is not a view that I expect many to hold. For this reason, and constraints of space, I pass over discussing it. ¹⁷ For example, Nagel (1970), Worsnip (2018), Hayward (2017).

Taking Prudence Seriously

75

When deciding which career to take up, one considers among other things how the options will impact one’s well-being. One might choose to be a lecturer rather than an aid worker because, although better impartially, the latter option will be worse for oneself. How to weigh these competing considerations is, of course, difficult. But it seems plausible that, in such decisions, one factor we weigh against the moral reasons we have is a set of reasons to promote and secure our own good. The same seems true for smaller decisions (e.g. joining a gym, going on holiday). Suppose you are asked why you did something. Just as ‘it was the right thing to do’ serves as a justification for performing some action in most, perhaps all, contexts, so too ‘it was best for me’ serves as a justification across many circumstances.¹⁸ So evaluative prudential properties seem to generate prudential reasons for action. Why, then, would someone deny such prudential reasons?

4.2.1. Objection 1: All Reasons are Agent Neutral One ground for denying PVM is the view that all reasons are agent-neutral. This would be the view, for example, of someone who accepted a very broad form of utilitarianism (that well-being is the sole fundamental generator of all reasons for action) who denied that a person’s well-being generates distinctive reasons for themselves, over and above the moral reasons that everyone’s well-being generates for everyone.¹⁹ What is the difference between agent-relative and agent-neutral reasons?²⁰ Whilst there is no consensus on how precisely to draw the distinction,²¹ we get the general idea via cases such as: Case 1 Option 1 Option 2

V’ W-B 500 600

J’ W-B 1000 901

Suppose you must decide which option to bring about. Vikram and Julia are strangers (to you and each other). One reason that you have for choosing

¹⁸ I take the plausibility of this—that something’s being your best option gives you prudential reason to choose it—to undermine Scanlon’s (1998) scepticism about the normative importance of well-being. For discussion see Fletcher (ms.). ¹⁹ This would be a very broad view because it would be utilitarianism about not only moral reasons but all practical reasons. ²⁰ The terminology of agent-neutral vs agent-relative was introduced by Parfit (1984). The modern debate about such reasons stems from Nagel (1970). ²¹ For detailed discussion of the extant proposals, see Ridge (2017).

Guy Fletcher

76

Option 2 over Option 1—the greater total well-being—is plausibly fully general (or agent-neutral). Now, suppose that Julia is making the decision (Vikram and Julia are still strangers to each other). If there are agent-relative reasons then the situation is now different in the following way. Whilst Julia has a reason to choose Option 2 that is exactly like one you had, there is also a reason for Julia to choose Option 1, namely that her own well-being is higher under that scenario. Those who claim that all reasons are agent-neutral must deny the existence of this reason for Julia. On their view, there is no difference in the reasons that you or Julia would have in making these decisions. I think that such a view is implausible for reasons for action and we can see that by considering this type of case: Case 2 Option 1 Option 2

V’ W-B 500 1

J’ W-B 1 500

Suppose I am making the decision between these options and that there are no further relevant differences between these two outcomes and these two people are perfect strangers. In such a situation, I need some way of resolving the tie between these outcomes (e.g. flipping a coin) because there is no reason to choose one option over another in terms of total well-being or any other fully general consideration. If all reasons are agent-neutral then it follows that, if Julia were in the position of choosing in this case, then she too would need to (e.g.) flip a coin in order to decide between these options. But that seems highly implausible. Presumably her own well-being can be a tie-breaker in this case, giving her some reason to choose Option 2. This suggests that there are agent-relative prudential reasons, even if they only serve to break ties. However, note that PVM is compatible with the denial of agent-relative reasons for action as long as there can be agent-relative reasons for attitudes. And this seems especially plausible. Looking again at Case 2, would it not be mistaken for Vikram and Julia to be completely indifferent to which option I choose, given the huge difference in their well-being on each option?²² Don’t they each have a reason to prefer the outcome in which they do much better? I take this to be evidence that there are prudential reasons for attitudes at the very least.

²² Reminder: they are strangers to each other.

Taking Prudence Seriously

77

4.2.2. Objection 2: Humeanism about Reasons Let me now return to the second general objection to PVM. This objection allows for agent-relative reasons for an agent to e.g. promote their own wellbeing but, crucially, denies that these reasons are generated by evaluative well-being facts themselves. The most likely species of this genus assumes the truth of a Humean or ‘internalist’ theory of reasons so I will examine that one in particular. On a Humean theory of reasons all reasons for action stem from, and are explained by, an agent’s desires. Whenever someone has a reason to do something this is because the action is conducive to the fulfilment of a desire of theirs. Put more precisely: H  R  A: X has a reason to Ф iff and because Ф-ing promotes the fulfilment of a desire of X’s.²³

One might wonder precisely why Humeanism would lead one to deny PVM. Remember, PVM is: P V M (PVM): Evaluative well-being facts generate agent-relative reasons (for action or for attitudes) for the relevant agent.

In many cases Humeanism and PVM deliver identical verdicts about the existence of reasons; both say that the relevant agent has a reason to perform some action. That will be true in any case where an agent has a desire that is fulfilled by getting what is good for them. However, Humeanism and PVM offer competing fundamental explanations of why the agent has the reason. On the Humean view, the fundamental explainer of the reason—that in virtue of which the agent has the reason—is the fact that the relevant action promotes the fulfilment of the agent’s desire. For PVM, by contrast, the explainer of the reason is the fact that the option is good for the agent (and this reason remains in place even if the desire disappears). Thus the challenge to PVM posed by the Humean theory of reasons is not the view that there are no reasons to promote

²³ For canonical recent defences of the view, see Schroeder (2007) and Markovits (2014). For the sake of space I focus on this, standard, Humean view but many Humeans, and commentators, sophisticate the theory in different ways. Such ways include: allowing the reasons that an agent has to be explained by desires that they do not have but would have under some counterfactual condition (e.g. Williams 1981); allowing desires that the agent does not yet have but will have to be relevant (e.g. Lin 2015); allowing the desires of other agents to be relevant (e.g. Manne 2016). For dissent see, for example, Enoch (2005).

78

Guy Fletcher

well-being. Rather, it is that there are no distinctively prudential reasons, reasons that are fundamentally explained by well-being.²⁴, ²⁵ Having explained this challenge to PVM from the Humean view of practical reasons let us assess the prospects for prudential reasons. Of course, the viability of Humeanism about practical reasons is a major issue, one that would take rather more than one chapter to settle. I do not pretend to offer a decisive case against it. I aim only to point out, firstly, precisely how counterintuitive it is when we consider prudence. I do this because it is easy to think that, unlike morality, Humeanism is not all that counterintuitive with respect to prudence. I also aim to show that Humeanism about reasons for action is compatible with the plausible claim that there are prudential reasons for attitudes, such that one can combine Humeanism and PVM. Here are some cases to bring out how counterintuitive Humeanism is with respect to well-being: C C: Colin is raised in a cult and believes his torment and suffering to be necessary for the sake of humanity. He has many desires including the desire to suffer for the sake of humanity. None of his desires would be fulfilled by escaping the cult. However, were he to escape, he would lead a life with a much higher level of well-being. D D: Dave is severely depressed. His condition leaves him with no desire that would be fulfilled by his continued existence. However, if he were to get medical help, he would go on to lead a life with a much higher level of well-being. E E: Emily is a child who desires only to watch cartoons and eat ice cream. None of her desires would be fulfilled by going to school. However, were she to do so, she would lead a life with a much higher level of well-being.

In each case it is plausible that the relevant agent has a reason to perform some action, generated by evaluative well-being facts.²⁶ These scenarios do not assume any first-order theory of well-being and we can generate these kinds of cases even assuming a desire theory of well-being. Take the simplest desire theory of well-being:

²⁴ For detailed discussion of these points, see Worsnip (2018). ²⁵ The issues here are structurally similar to those concerning Humeanism and moral reasons. Humeans hold that morality and its demands cannot fundamentally generate reasons, only desires can. They then typically argue that on their view we still have reasons to do what morality requires by giving explanations of how conforming to the demands of morality in fact promotes the fulfilment of our desires. See, for example, Schroeder (2007). ²⁶ For discussion of these issues, see Lin (2015).

Taking Prudence Seriously

79

D T  W-B: something is good for someone if and only if and because it fulfils one of their actual desires.

Even on the assumption of this theory, PVM and Humeanism give different verdicts in the cases above. In the case of Colin, PVM plus the desire theory entails that Colin has reason to leave the cult so long as he would thereby end up with a larger set of fulfilled desires. Similarly, we get the verdict that Dave has (prudential) reason to seek help, stemming from the prudential value of that option (where that value comes from the desire-fulfilments within it). Cases like these reveal one way in which Humeanism is revisionary about reasons and well-being. For in these scenarios, Humeanism entails (i) that the well-being that would be gained by performing some action is not, itself, a reason to do it and (ii) that the absence of a desire that would be fulfilled by performing the beneficial action entails that there is in fact no reason to perform the action. But this is extremely counterintuitive, especially in the case of Dave. Our sense that Dave has reason to get help is not undercut, I assume, by discovering that his depression is so bad that he has no desires served by getting help. Rather, this is evidence of just how strong are his reasons to get help. Cases like Colin, Dave, and Emily put pressure on Humeanism by bringing out how revisionary it is when it comes to well-being in a way that is, I suggest, even starker than Humeanism’s verdicts with respect to morality.²⁷ We can accentuate it by considering a different case:²⁸ B B: Belinda has a choice between two different career paths. Her desires are such that each option will promote the fulfilment of the same number of them (which are all of equal intensity) so her desires leave the options weighted exactly equally. However, one option will give her a significantly higher level of wellbeing than the other.

Humeanism regards the two options as identical, in terms of reasons for action, given that each option promotes the fulfilment of the same number of desires (of equal intensity). But that is to think that no amount of extra well-being for one option can provide any normative weight, even as a

²⁷ Starker, I think, because it is easy to think of internal reasons as being closely connected to one’s own well-being (especially if one holds a desire-based view of wellbeing) and to fail to appreciate how often Humeanism will fail to deliver the reasons that we think agents have to promote their own well-being. ²⁸ Unlike the earlier cases, this case does require assuming the falsity of at least some forms of the desire fulfilment theory of well-being.

80

Guy Fletcher

tie-breaker. And that seems extremely counterintuitive.²⁹ Couldn’t someone recommend that Belinda take one of the options on precisely this prudential ground? And couldn’t Belinda choose an option on this ground? If so, then it looks like prudential properties at least generate prudential reasons to break ties. I have pointed out that Humeanism gives counterintuitive results in a range of cases. But reasons for action are not the only candidate here. Even if the foregoing fails, and Humeanism holds for reasons for action, PVM would still be true if there are prudential reasons for attitudes.³⁰ We saw one example earlier (Case 2) which demonstrated the plausibility of prudential reasons for attitudes. Here is another: U’ U: Ursula is checking in for her flight. The person at the desk tells Ursula that she might get a free upgrade and that this will be decided between now and when she reaches the gate. Ursula will be better off if her ticket is upgraded.

In this case, Ursula does not have reason to do anything to get an upgrade, given that she cannot affect her chances. Presumably, though, she has reason to hope that she is upgraded, stemming from the fact that she would be better off. Further, this does not seem dependent upon Ursula having some relevant desire which the hope would promote the fulfilment of. Thus even if Humeanism is true for reasons for action, PVM could still be true because of prudential reasons for attitudes. Let me recapitulate. In this section I have argued that there are prudential normative reasons and tried to undermine two main challenges to them. In each case I pointed out the counterintuitive results that come from these denials of prudential reasons. I then showed, more concessively, how each is compatible with the existence of prudential reasons for attitudes. I now move on to the second part of the chapter: whether prudential judgements are normative.³¹ ²⁹ See Hayward (2017) for a recent defence of the most stark dissenting view—that there are no normative requirements of prudence, even the sort of structural requirements people often think of as constitutive of rationality. See also Worsnip (2018). ³⁰ Is it peculiar to examine the possibility of Humeanism about reasons for action but not for attitudes? I do not deny that some will be Humeans about both. But it is hard to know what the orthodox Humean position is on this issue. In A Treatise of Human Nature (1738: 2.3.3.6), Hume famously declared there to be no reasons for preference: ‘Tis as little contrary to reason to prefer even my own acknowledg’d lesser good to my greater, and have a more ardent affection for the former than for the latter’ (my italics). By contrast, contemporary Humeans typically leave unclear the extent to which their views extend beyond reasons for action. Two exceptions: Markovits (2014)—who is explicit in rejecting Humeanism for reasons for belief—and Cowie (2014; 2016) who explicitly extends his Humeanism to reasons for belief. ³¹ It might be tempting to connect these two issues thus: if there are normative prudential reasons this means (at least some) prudential judgements are normative in

Taking Prudence Seriously

81

4 . 3 . T H E P A R I T Y S T R A T E G Y (I ) : T H E P O S I T I V E C A S E As I outlined at the beginning of the chapter, I will argue for the claim that P J  N (PJN): ments are normative judgements,

Prudential judge-

by looking at some of the grounds that are offered for the claim that moral judgements are normative (which I will call ‘markers’ of normativity) and arguing that they carry over to prudential judgement. My claim is not that any of these alone is a sufficient condition of some judgement type being normative. It is easy to find false positives—namely, forms of nonnormative judgements that have at least one of these features. However, it seems plausible that there are no forms of non-normative discourse that share all of them. Finally, let me emphasize that these markers serve as evidence that some type of judgement is normative. I do not take them to constitute the nature of normative judgement. Normative types of judgements have these features (or at least some subset of them) because they are normative, whatever precisely that comes to, rather than being normative because they have these features.³²

4.3.1. Markers of Normativity 4.3.1.1. Evaluative and Prescriptive One feature of moral judgements commonly cited in connection with their normativity is that they are evaluative, critical, or prescriptive. Moral judgements do not merely represent how the world, agents, and their actions are. Rather they are about how the world or its constituents ought to be, how we have reason to make it, how it would be good, bad, better, best, or worst for it to be. This feature of moral judgements is reflected in the vocabulary in which they are made and expressed. We use deontic terms ‘rights’, ‘wrong’, ‘permitted’; evaluative terms ‘good’, ‘bad’, ‘value’, ‘best’; and modal terms ‘ought’, ‘should’, ‘must’. virtue of their content. However, we do not need there actually to be any prudential reasons for these judgements to be normative in virtue of their content. Thus the truth of PVM, though sufficient, is not necessary to show that (some) prudential judgements are normative. ³² To expand: I take the points I am about to develop to be very general articulations of the data that we use when formulating and assessing metaethical views, data that must be explained, or explained away, by such views.

82

Guy Fletcher

Here we find a perfect fit. Prudential judgements are equally evaluative, critical, or prescriptive. They do not merely represent how the world, agents, and their actions are but instead how the world or its constituents ought to be, how we have reason to make it, how it would be good, bad, better, best, or worst for it to be. We also find the same evaluative (‘good for’, ‘better for’), deontic (‘reason’), and modal (‘ought’, ‘should’, ‘must’, ‘needs’) vocabulary used to make prudential judgements.³³

4.3.1.2. Authority Moral judgements are typically held up as having a kind of authority which in turn gives them a special place in deliberation. It is often taken to be a feature of moral discourse that morality makes a difference to the reasons an agent has such that if an action is morally wrong this entails reasons for the agent not to perform that action and reasons for the agent to feel shame or guilt and to be blamed by others.³⁴ This authoritative nature of morality is also reflected in the kind of centrality with which it is treated in deliberation where deciding what to do all-thingsconsidered is closely connected with deciding what one morally ought to do.³⁵ It is an everyday feature of conversation and deliberation that accepting that it would be morally wrong to perform some action is typically treated as settling whether one ought to do it all-things-considered. Furthermore, even where people allow that morality does not settle the question of what an agent ought to do all-things-considered, it is typically common ground that morality makes a difference to what an agent ought to do, by entailing reasons. Here we have a close fit between moral and prudential judgements. As argued above, prudential value plausibly makes a difference to the reasons an agent has; if an option is best for them then this gives them (at least) prudential reason to prefer that option. One might be sceptical as to whether prudence has such a central role in deliberation as that which we find with morality. I agree that there is a less close connection between what one ought to do all-things-considered and what one prudentially ought to do. But this, I suggest, is specifically because moral demands plausibly trump prudential demands across many (though not all) contexts. If we focus on contexts where moral demands are screened off, the normative authority of prudence is clearer to see. In these contexts ³³ On the use of ‘needs’ in prudential discourse, see Fletcher (2018a). ³⁴ That is not to deny the existence of scepticism on this point (e.g. the Why Be Moral? challenge). ³⁵ I here assume the coherence of ought all-things-considered. c.f. Tiffany (2007), Baker (2017), Copp (ms.).

Taking Prudence Seriously

83

there is a close connection in between what one ought to do and what one prudentially ought to do; that some option would be (sufficiently) bad for you is treated as settling whether you ought to bring about that option (and vice versa for what would be best for you). Furthermore, in contexts where moral demands are in play, we treat prudential considerations as to-beweighed against moral requirements and one of the most plausible sets of cases where morality fails to settle what one ought to do all-thingsconsidered is the set in which prudence (specifically, rather than any other normative domain) trumps morality because morality’s demands are too costly. This suggests that prudence enjoys the same kind of authority as morality.³⁶

4.3.1.3. Affective Connection For at least a central subset of moral judgements, there is a close connection between them and certain affective responses (including reactive attitudes).³⁷ For example, judging that morally I ought not to perform some action is closely connected to having some sorts of negative attitude towards it. We typically expect someone who makes the judgement that some action is morally wrong to blame and feel anger towards others who perform the action, and guilt towards themselves if they perform such an action. And if an agent judges that, morally, they ought to perform some action we expect them (even if defeasibly) to be motivated to perform that action, in two senses of ‘expect’. We make predictions about affective responses and motivation in light of moral judgements and we make these predictions on the basis that that an agent who falsified them would be criticizable in some way.³⁸ Here, again, there is a perfect parallel with prudential judgment.³⁹ In the case of prudential judgements, or at least the analogous central subset of them, we find a close connection between those judgements and affective responses (including reactive attitudes). Judging that an action is what one prudentially ought to do is closely connected to being motivated to perform

³⁶ Though, again, this depends on one’s view about cases where prudence can override morality in determining what one ought to do all-things-considered. ³⁷ ‘Close connection’ is deliberately vague. The most hardcore of externalists about moral motivation can accept this thesis as long as they think it just notes a correlation. ³⁸ For representative discussion of these issues, see e.g. Smith (1994), Svavarsdóttir (1999). ³⁹ Perhaps it is imperfect in this way: the connection is plausibly closer for prudential judgements than for moral judgements because every form of internalism about moral judgement and motivation is less plausible than the analogous claim about prudential judgement.

84

Guy Fletcher

that action and judging that some option would be best for oneself is closely connected to preferring that option. We similarly predict that agents have responses that line up with their prudential judgements and we make these predictions on the grounds that to lack such responses would be a failing.

4.3.1.4. Disagreement Moral discourse is striking in the way that it permits of radical disagreement without manifestation of conceptual incompetence or loss of common subject-matter.⁴⁰ People of radically divergent general world-views and across large differences in social and cultural milieu seem equally to make moral judgements and to be in the condition of agreeing and disagreeing morally, despite widespread differences in their wider set of beliefs. If there are limits to what one can think is morally wrong without thereby showing oneself to be incompetent with moral concepts, these limits are lax, permitting a large range of fundamental moral disagreement.⁴¹ Here too we get a perfect fit. Prudential discourse also permits extremely radical disagreement, without manifestation of conceptual incompetence or loss of a common subject matter. In Welfare and Rational Care, Stephen Darwall provides an argument on precisely these lines: [I]t seems possible for two people who care about someone, S, to coherently disagree about whether something, X, is good for S, even though they agree completely about all the non-normative facts concerning X and S. Suppose, for example, that X is a pleasant illusory belief of S’s, say, that S’s novel has sold 10,000 copies (when in fact it has sold only 12). It would seem that two people could be agreed about everything else, but simply disagree about whether this pleasant illusory belief is good for S or makes some contribution to his welfare, other things being equal . . . If this is right, then welfare . . . must be an explicitly normative concept.⁴² ⁴⁰ See Hare (1952), Horgan and Timmons (1992). For scepticism about the evidentiary value of such cases, see Dowell (2016). ⁴¹ Foot (1958: 512) gives the example of thinking it morally wrong to look at hedgehogs in the light of the moon or run around trees right-handed. Perhaps this is a limit on what can count as a moral judgement. For the attempt to motivate a wide(r) range of conceptual restrictions on moral judgements, see Cuneo and Shafer-Landau (2014). I think that the same applies to prudence. Perhaps there are such fixed points when it comes to prudence (e.g. one cannot coherently hold that the only thing with prudential value is looking at hedgehogs in the light of the moon). Morality and prudence thus seem on a par here. ⁴² Darwall (2004: 10) gives this as a direct argument for the claim that prudential judgements are normative (along with his particular account of their content). This argument is criticized by Olson (2006) on the grounds that this disagreement phenomenon is found in non-normative domains, such as genre judgements. Olson’s objection succeeds, I think, against the argument that this feature suffices for a type of judgement to be normative.

Taking Prudence Seriously

85

We can broaden out from illusory pleasant belief and note that people of radically divergent general world-views and across large differences in social and cultural milieu seem equally to make prudential judgements and to agree and disagree prudentially, despite widespread differences in their wider set of beliefs. Close to home, we have Nozick’s experience machine example where some philosophers hold views of well-being where there is nothing prudentially deficient at all about life in the experience machine (assuming a large enough supply of hedons).⁴³ By contrast, many philosophers think that such a life would hold little to no prudential value, because what matters prudentially is the exercise and development of our human capacities. Some philosophers think that getting what you want is what matters fundamentally; others think that some things have prudential value independently of whether we desire those things. Moreover, it is easy to imagine radically different theories of prudential value across different communities (which thus constitute prudential ‘twin earth’ scenarios).⁴⁴ We can, for instance, imagine a pair of communities where one has a set of practices that suggests a hedonistic conception of prudential value whilst the practices in the other community suggest a perfectionist conception of prudential value. In the first community, people advise their loved ones to only develop their capacities if by doing so they thereby maximize enjoyment, whereas in the second community they advise their loved ones to only seek pleasures where this is connected to the maximal exercise and development of their capacities. Similarly, in the first community those that are envied are the ones who experience most pleasure whilst in the second community it is those who develop and exercise their capacities most even if their lives are known to be less pleasurable. From cases like these it seems safe to say that prudential discourse, like moral discourse, permits a wide range of fundamental disagreement without talking past each other or manifesting conceptual incompetence. To sum up this section, I have introduced four markers of the normativity of some set of judgements—four (sets of) ideas typically appealed to when articulating the normativity (and theoretical interest and importance) of moral judgements. I then argued that we find sufficiently close, often perfect, parallels in the case of prudential judgements. This gives us grounds for holding that prudential judgements are normative, on a par with moral judgements. As I noted earlier, to show that prudential judgements are normative by analogy with moral judgements I need to establish relevant parallels between them and show that there is no significant disanalogy. In the next section ⁴³ Nozick (1974: 42).

⁴⁴ On this, see Fletcher (2018b).

86

Guy Fletcher

I thus consider objections to the effect that there is a significant disanalogy between moral and prudential judgements and that PJN is false.

4.4. OBJECTIONS TO PJN How might one argue that prudential and moral judgements are significantly disanalogous such that prudential judgements are not normative? In this section I consider three such objections. The first of these simply fails, as I shall explain. In response to the second and third I point out the considerable argumentative burdens which their proponent takes on.

4.4.1. Objection 1: Prudential Value is Desire-Based The first such ground for rejecting PJN that one might offer is that prudential value is desire-based and that this means that prudential judgements are non-normative. Spelt out further, the argument is that a desirebased theory of well-being is true, that this entails that well-being properties are non-normative properties, which means that prudential judgements are non-normative.⁴⁵ Because this objection to PJN assumes the truth of a particular first-order theory, I will talk in terms of prudential properties and facts (but this should be construed in a sufficiently minimal way so as to beg no questions between e.g. cognitivism vs non-cognitivism about prudential discourse). To fairly examine this objection, let us assume the following theory: D T  W-B: Something is good for someone if and only if and because it fulfils one of their actual desires.

Would the truth of this theory entail that prudential properties and judgements are non-normative? No, it would not. The argument assumes that the fact that something fulfils a person’s desire is a non-normative fact, which we can grant for the sake of argument. Would that suffice to show that the left-hand side, the prudential fact, is a non-normative fact? No, that’s too quick. Just because a prudential fact is explained by a non-normative fact it doesn’t follow that it is a non-normative fact. ⁴⁵ As I read Mackie (1977: 29) he endorses this inference when considering the scope of his error theory: ‘But if we think rather of a counsel of prudence as being related to the agent’s future welfare, to the satisfaction of desires that he does not yet have—not even a present desire that his future desires should be satisfied—then a counsel of prudence is a categorical imperative, different indeed from a moral one, but analogous to it.’

Taking Prudence Seriously

87

To see why, note the failure of the analogous argument for the non-normativity of morality, on the assumption of the dependence of the moral on the non-moral. Suppose this moral theory is true: M T: It’s morally wrong for X to PHI iff and because X’s PHI-ing doesn’t maximize pleasure.

On this kind of view, a fact about total pleasure that would be generated by an action—a paradigmatic non-normative fact—explains a moral fact. Does the truth of this explanatory claim, coupled with the right-hand side being non-normative, mean that the moral fact is also a non-normative fact? Plainly not. We do not simply infer from the non-normative nature of the facts that make actions right and wrong that facts about rightness and wrongness are, themselves, non-normative. We find many cases where non-normative facts ground normative facts and even if one side of an explanatory claim is non-normative it is a further question whether the fact on the left-hand side is thereby non-normative. Just to be clear about the dialectic here, my claim is that prudential value being desire-based would not entail that prudential properties and judgements are non-normative. For even if the facts that ground prudential facts are non-normative this does not, itself, settle the nature of prudential facts.⁴⁶ And we can see that this inference is dubious by looking at other domains. The way in which this argument fails extends to other theories of well-being. Just as the truth of a desire theory of well-being is insufficient to show that prudential properties and judgements are non-normative, the truth of hedonism (or any other theory) would be similarly insufficient. Thus we cannot point to the non-normativity of the grounds of prudential facts to show that prudential properties and judgements are non-normative.

4.4.2. Objection 2: Reductionism about Prudential Properties Let us now consider a sophisticated successor to the objection that we just rejected. This objection holds that prudential properties are identical to desire-fulfilment properties and that is why prudential discourse is nonnormative. To make maximally clear how this differs from the previous objection, on this new view it is not merely that desire-fulfilment has prudential value. Rather, what it is to be good for someone just is to fulfil their desire. This view thus suggests a reductive account of the evaluative ⁴⁶ On this see Hampton (1998), Bedke (2009).

88

Guy Fletcher

prudential property of (e.g.) being good for someone.⁴⁷ It combines this with the claim that such a reductive account entails that prudential judgements are non-normative. This objection need not be run with the specific claim that it is desirefulfilment that the property of being good for someone is identical to. Another natural candidate would be that evaluative prudential properties are identical to hedonic properties—such that to be good for someone just is to be pleasurable for them (and to be bad for someone just is to be painful for them). What, then, to make of this objection? There are two weaknesses to it, in the form of heavy argumentative burdens which the objection incurs (burdens which make it difficult to provide a direct reply to the objection, without it being worked out in a lot more detail). First, notice that for the objection to work, the objector must supply the reductive account of the property of being good for someone. This is, for reasons familiar from metaethics, no mean feat. As argued above, one of the normative markers of prudential discourse which it shares with moral discourse is the possibility of radical prudential disagreement. Such disagreement provides one obstacle to identifying the property of being good for someone with (e.g.) the property of fulfilling the person’s desire (or, similarly, some hedonic property).⁴⁸ The second argumentative burden that this strategy takes on is that of showing that the reductive account of the property of being good for someone entails that this property, and judgements ascribing it, are non-normative. Put another way, suppose that the property of being good for someone just is some hedonic property (and so on for the other evaluative prudential properties). The question now is: why would that entail that prudential properties and judgements are non-normative? This strategy must therefore show why a form of reductive naturalism about prudential properties is incompatible with prudential judgements’ being normative. I do not claim to have shown that this kind of reductionist move against PJN cannot succeed. Space constraints preclude a comprehensive examination of this proposal. I have simply made clear the task for anyone who pursues this strategy against PJN. Those inclined to reject PJN on the grounds of a reduction of prudential properties have much to do.

⁴⁷ For discussion of the issues addressed here, see Schroeder (2005; 2007). ⁴⁸ See Heathwood (2009) for discussion of this issue.

Taking Prudence Seriously

89

4.4.3. Objection 3: Prudential Reasons are Different from Moral Reasons An analogous objection to the previous one, this time focused on deontic prudential properties, is the view that prudential reasons are not normative reasons (and so on for other deontic prudential properties). That is, someone might grant that there are prudential reasons, and oughts, and so on, but hold that these are somehow not normative or, at least, not normative in the way that moral reasons are. One model that might be appealed to is the idea that prudential reasons are like reasons (or ‘reasons’) of etiquette or that what it is for something to be prudentially required is similar to what it is for something to be required by law, or by a company code of conduct, or by etiquette.⁴⁹ Another model would be that prudential reasons are hypothetical, rather than categorical, reasons. We find instances of this position in the debate about what to do with moral discourse if error theory is true. In that debate, it is treated as utterly unproblematic to invoke prudential reasons within a dialectical context that supposes that there are no moral reasons (and perhaps no normative reasons whatsoever). For example, Richard Joyce writes: But it is important to remind ourselves that even the eliminativist error theorist will still have plenty of good and strong reasons—many of them self- interested reasons— for being nice to her fellows[.]⁵⁰

This passage, like many others in that debate, embodies a view of prudential reasons as non-normative (or, at least, as normative in some different, lesser, way than moral reasons). It is important to distinguish between the claim that there are no prudential reasons (examined above) and this position, namely that there are prudential reasons but that they are not bona fide normative reasons of the sort that could make prudential judgements normative judgements on a par with moral judgements. What could the difference between prudential and moral reasons consist in, such that prudential reasons are non-normative? As we saw above, it cannot be merely that prudential reasons are grounded in desires because a desire-fulfilment theory of well-being is true. That is compatible with their being normative. Rather, the property of being a prudential reason must be

⁴⁹ For discussion, see e.g. Stroud (1998: 172), Olson (2011a, 2011b), Dorsey (2016: ch. 1). ⁵⁰ Joyce (2008: 75).

90

Guy Fletcher

radically different from the property of being a moral reason, such that the prudential one is non-normative. We also saw above that there is a genuine clash between prudential reasons and Humeanism about reasons. This means we cannot treat prudential reasons as distinct from moral reasons by treating them as hypothetical, rather than categorical, reasons. This is because, spelled out, this just is the view that there are no prudential reasons. If we have reasons to promote well-being only because that promotes the fulfilment of our desires, then that is to think that evaluative prudential facts themselves do not, fundamentally, generate reasons (which is, ipso facto, to deny that there are prudential reasons). There is thus no way of holding that there are prudential reasons and that they are non-normative by being internal/ hypothetical. As with the objection we considered in Section 4.4.2, the best version of the idea is that being a prudential reason just is some relation such as that of promoting fulfilment of the agent’s desire, or generating pleasure for the agent. And, again, as with the previous objection, the person making this objection has two jobs to do. They must do the hard work of supplying a plausible reduction of the relation of being a prudential reason and show that this relation (and judgements ascribing it) are thereby non-normative. Furthermore, they must explain why prudential discourse (like moral discourse) possesses the markers of normativity, in stark contrast to the kinds of domains that might be appealed to as models for this reduction of prudential discourse (namely law, codes of conduct, and etiquette).⁵¹ Rhetorically: if prudential reasons are so different from moral reasons, why do they seem so similar? Before moving on, let me note that whilst I considered the objections in Sections 4.4.2 and 4.4.3 individually, neither alone can be a fully general objection to PJN for the following reason—PJN encompasses evaluative and deontic judgements and each of these objections targets one of evaluative or deontic judgements. This means that these objections, individually, could only establish that one part of prudential discourse is non-normative. Someone who wanted to argue that prudential discourse is wholly nonnormative must either (a) combine these objections (or similar ones) to show that neither evaluative nor deontic prudential judgements are normative or (b) collapse these tasks into one by treating all prudential normative

⁵¹ These domains have some of the markers but do not possess them all. In each case the connection to affect and motivation is implausible. It is also difficult (especially for codes of conduct and etiquette) to hold that there could be the same kind of radical disagreement.

Taking Prudence Seriously

91

properties as reducing to one and showing how that property (and judgements ascribing it) are reduced to something non-normative.⁵² These tasks are significant and it behoves those who wish to treat prudential normativity as second-class normativity to undertake them.

4.5. CONCLUSION In this chapter I have argued for the following two claims: P V M (PVM): Evaluative well-being facts generate agent-relative reasons (for action or for attitudes) for the relevant agent. P J  N (PJN): Prudential judgements are normative judgements.

I argued for PVM by showing how counterintuitive it is to deny that there are prudential reasons for action (or, at least, for attitudes). I argued for PJN by outlining the markers of normativity—features of moral discourse appealed to in defence of the claim that moral discourse is normative—and then showing that prudential discourse shares those features. I then considered objections to PJN, showing that these objections either fail or incur significant, hitherto undischarged, argumentative burdens. If prudential judgement is normative, like moral judgement, this has a number of important implications throughout moral philosophy and metaethics.⁵³ It means that we should examine prudential discourse directly, both for its own sake but also in order to assess various meta-normative theories. It’s time to take prudence seriously.⁵⁴

⁵² For example, a buck-passing view of all prudential properties would only need to provide one such reduction—of the property of being a prudential reason. ⁵³ On this, see Fletcher (ms.) ⁵⁴ Acknowledgements: I am indebted to (at least) the following for various forms of help with this chapter: James Brown, Krister Bykvist, Steve Campbell, Richard Chappell, Chris Cowie, Alex Gregory, Daniel Groll, Louise Hanson, Anandi Hattiangadi, Daniel Haybron, Uriah Kriegel, Marcy Lascano, Eden Lin, Kathryn Lindeman, Elinor Mason, Jonas Olson, Jason Raibley, Mike Ridge, Debbie Roberts, Connie Rosati, Richard Rowland, Justin Snedegar, Bart Streumer, Sarah Stroud, Teemu Toppinen, Joey Van Weelden, Pekka Väyrynen, Eric Wiland, Daniel Wodak, Jack Woods; audiences in Edinburgh (*2), California Long Beach, Dublin, Geneva, Montreal, Paris, Rome, St Louis; and The Madison Metaethics Workshop. Sincere thanks also to two anonymous referees, whose comments were invaluable.

92

Guy Fletcher

References Baker, D. (2017). ‘Skepticism about “Ought” Simpliciter’, in R. Shafer-Landau (ed.), Oxford Studies in Meta-Ethics, vol. 13 (Oxford: Oxford University Press), 230–52. Bedke, M. (2009). ‘Might all Normativity be Queer?’ Australasian Journal of Philosophy 88 (1): 41–58. Copp, D. (ms.). ‘Normative Pluralism and Skepticism about Ought Simpliciter’. Cowie, C. (2014). ‘In Defence of Instrumentalism about Epistemic Normativity’, Synthese 191 (16): 4003–17. Cowie, C. (2016). ‘Revisionist Responses to the Amoralism Objection: A Reply to Julia Markovits’, Ethical Theory and Moral Practice 19 (3): 1–13. Cuneo, T. and Shafer-Landau, R. (2014). ‘The Moral Fixed Points: New Directions for Moral Nonnaturalism’, Philosophical Studies 171 (3): 399–443. Darwall, S. (2004). Welfare and Rational Care (Princeton, NJ: Princeton University Press). Darwall, S. (2016). ‘Making the “Hard” Problem of Moral Normativity Easier’, in E. Lord and B. Maguire (eds), Weighing Reasons (Oxford: Oxford University Press), 257–78. Dorsey, D. (2016). The Limits of Moral Authority (Oxford: Oxford University Press). Dowell, J. (2016). ‘The Metaethical Insignificance of Moral Twin Earth’, in Russ Shafer-Landau (ed.), Oxford Studies in Metaethics, vol. 11 (Oxford: Oxford University Press), 1–27. Enoch, D. (2005). ‘Why Idealize?’ Ethics 115 (4): 759–87. Enoch, D. (2011). Taking Morality Seriously (Oxford: Oxford University Press). Finlay, S. (2007). ‘Too Much Morality’, in Paul Bloomfield (ed.), Morality and SelfInterest (Oxford: Oxford University Press), 136–54. Finlay, S. (2014). Confusion of Tongues (Oxford: Oxford University Press). Fletcher, G. (2012). ‘The Locative Analysis of Good for Formulated and Defended’, Journal of Ethics & Social Philosophy 6 (1): 1–26. Fletcher, G. (2018a). ‘Needing and Necessity’, in Mark Timmons (ed.), Oxford Studies in Normative Ethics, vol. 8 (Oxford: Oxford University Press), 170–92. Fletcher, G. (2018b). ‘Pain for the Moral Error Theory: A New Prudential Companions-in-Guilt Argument’, Australasian Journal of Philosophy 96 (3): 474–82. Fletcher, G. (ms.). ‘Dear Prudence’. Foot, P. (1958). ‘Moral Arguments’, Mind 67 (268): 502–13. Hampton, J. (1998). The Authority of Reason (Cambridge: Cambridge University Press). Hare, R. M. (1952). The Language of Morals (Oxford: Oxford University Press). Hayward, M. K. (2017). ‘Practical Reason, Sympathy and Reactive Attitudes’, Noûs, https://doi.org/10.1111/nous.12203. Heathwood, C. (2009). ‘Moral and Epistemic Open Question Arguments’, Philosophical Books 50 (2): 83–98.

Taking Prudence Seriously

93

Hooker, B. (1991). ‘Theories of Welfare, Theories of Good Reasons for Action, And Ontological Naturalism’, Philosophical Papers 20 (1): 25–36. Horgan, T. and Timmons, M. (1992). ‘Troubles for New Wave Moral Semantics: The “Open Question Argument” Revived’, Philosophical Papers 21 (3): 153–75. Joyce, R. (2008). ‘Morality, Schmorality’, in Paul Bloomfield (ed.), Morality and Self-Interest (Oxford: Oxford University Press), 51–76. Kraut, R. (2007). What is Good and Why: The Ethics of Well-Being (Cambridge, MA: Harvard University Press). Lin, E. (2015). ‘Prudence, Morality, and the Humean Theory of Reasons’, Philosophical Quarterly 65 (259): 220–40. Mackie, J. L. (1977). Ethics: Inventing Right and Wrong (London: Penguin). Manne, K. (2016). ‘Democratizing Humeanism’, in E. Lord and B. Maguire (eds), Weighing Reasons (Oxford: Oxford University Press), 123–40. Markovits, J. (2014). Moral Reason (Oxford: Oxford University Press). Nagel, T. (1970). The Possibility of Altruism (Princeton, NJ: Princeton University Press). Nozick, R. (1974). Anarchy, State, and Utopia (Malden, MA: Basic Books). Olson, J. (2006). ‘Review of Stephen Darwall’s Welfare and Rational Care’, Economics and Philosophy 22 (1): 171–7. Olson, J. (2011a). ‘In Defense of Moral Error Theory’, in Michael Brady (ed.), New Waves in Metaethics (Basingstoke: Palgrave Macmillan), 62–84. Olson, J. (2011b). ‘Error Theory and Reasons for Belief ’, in A. Reisner and A. Steglich-Petersen (eds), Reasons for Belief (Cambridge: Cambridge University Press), 75–93. Parfit, D. (1984). Reasons and Persons (Oxford: Oxford University Press). Parfit, D. (1997). ‘Reasons and Motivation’, Proceedings of the Aristotelian Society, Supplementary Volume 71: 99–130. Railton, P. (1989). ‘Naturalism and Prescriptivity’, Social Philosophy and Policy 7 (1): 151–74. Ridge, M. (2017). ‘Reasons for Action: Agent-Neutral vs. Agent-Relative’, The Stanford Encyclopedia of Philosophy, ed. Edward N. Zalta, https://plato.stanford. edu/entries/reasons-agent/. Rosati, C. (1995a). ‘Naturalism, Normativity, and the Open Question Argument’, Noûs 29 (1): 46–70. Rosati, C. (1995b). ‘Persons, Perspectives, and Full Information Accounts of the Good’, Ethics 105 (2): 296–325. Rosati, C. (1996). ‘Internalism and the Good for a Person’, Ethics 106 (2): 297–326. Rosati, C. (ms.). ‘Welfare and Rational Fit’. Scanlon, T. M. (1998). What We Owe to Each Other (Cambridge, MA: Harvard University Press). Schroeder, M. (2005). ‘Realism and Reduction: The Quest for Robustness’, Philosophers’ Imprint 5 (1): 1–18. Schroeder, M. (2007). Slaves of the Passions (Oxford: Oxford University Press). Schroeder, M. (2015). ‘Is Knowledge Normative?’ Philosophical Issues 25 (1): 379–95. Smith, M. (1994). The Moral Problem (Oxford: Blackwell).

94

Guy Fletcher

Streumer, B. (2017). Unbelievable Errors (Oxford: Oxford University Press). Stroud, S. (1998). ‘Moral Overridingness and Moral Theory’, Pacific Philosophical Quarterly 79 (1): 170–89. Svavarsdóttir, S. (1999). ‘Moral Cognitivism and Motivation’, The Philosophical Review, 108 (2): 161–219. Sylvan, K. (2018). ‘Knowledge as a Non-Normative Relation’, Philosophy and Phenomenological Research 97 (1): 190–222. Tiffany, E. (2007). ‘Deflationary Normative Pluralism’, Canadian Journal of Philosophy 37 (5): 231–62. Williams, B. (1981). ‘Internal and External Reasons’, in his Moral Luck: Philosophical Papers 1973–1980 (Cambridge: Cambridge University Press), 101–13. Worsnip, A. (2018). ‘Eliminating Prudential Reasons’, in M. Timmons (ed.), Oxford Studies in Normative Ethics, vol. 8 (Oxford: Oxford University Press), 236–57.

5 Internalism and Prudential Value Jennifer Hawkins

Many philosophers accept that prudential value is a distinct type of value, different both from moral value and also from general non-moral goodness.¹ Those who (like myself ) accept the category of prudential value think of it as the type of value at stake when we say that something is non-morally good for a particular person. It is the kind of value theories of welfare or well-being are about. It is what we seek for ourselves on those occasions when we seek our own self-interest. And it is the kind of value we seek to realize for others when we try to benefit them. There are many interesting and (relatively) underexplored foundational questions about prudential value. This paper is concerned with whether or not some form of existence internalism (defined below) holds for prudential value, and if so what it amounts to.² Existence internalism is the idea that there is a necessary connection between something’s having a certain positive normative status (its being right or its being intrinsically prudentially good) and its ability to elicit positive psychological responses from human subjects.³ Existence internalism is typically contrasted with judgment internalism, which is a claim about what follows, as a matter of conceptual content, from sincere

¹ Philosophers differ in how they carve up value. For example, Moore (1903) and Scanlon (1998) are skeptical of the very idea of prudential value that is relative to individuals. However, they are not skeptical of the idea of general, non-relational nonmoral goodness. Others who discuss general, non-relational goodness take the notion of intrinsic goodness to be primarily a moral notion (Zimmerman 2015). ² The idea that some form of internalism must hold for prudential welfare is commonly expressed, but there are relatively few explicit discussions of what such a constraint might amount to. At the time this essay was written in 2014, the main discussions of prudential internalism were Rosati (1996), Velleman (2000), and Sarch (2011). Since then there have been several new additions to the literature. See, in particular, Dorsey (2017) and Heathwood (ms.). ³ The term ‘existence internalism’ and its close relative ‘judgment internalism’ come from Darwall (1983: 54). A similar distinction with different labels occurs in Brink (1989: 40).

96

Jennifer Hawkins

judgments about value.⁴ Existence internalism, however, is a metaphysical thesis, not a conceptual one. Put another way, it is the claim that facts about human psychological responsiveness constrain the metaphysics of value in particular ways. Attempts to specify existence internalism for a particular domain of value are attempts to say what the metaphysical constraints are for that type of value. Peter Railton characterizes the task in terms of avoiding “alienness.” He writes, “What is intrinsicially valuable for a person must have a connection with what he would find to some degree compelling or attractive, at least if he were rational or aware. It would be an intolerably alienated conception of someone’s good to imagine that it might fail in any such way to engage him.”⁵ However, there remain different views about what would suffice to avoid alienness. My aim is to introduce and defend as viable a new account of what such constraints might look like. Moral existence internalism is typically formulated as a claim about motivation. This reveals an important sense in which discussions of prudential value differ significantly from discussions of moral value. In the moral case, we are typically invited to imagine an agent who has the opportunity to act rightly or wrongly. The question is whether the moral facts are such that her mere awareness of them would give her (at least some, perhaps defeasible) motivation to do the right thing. But this scenario is prospective. Talk of motivation makes sense only when action is required, and action is only sometimes required in relation to prudential value. Sometimes we are in pursuit of the good. But at other times we talk about goodness that is realized, and here talk of motivation seems out of place. When we talk about goods that a person ‘has’ or ‘possesses’ we need different language. Rather than speak of motivation we speak of caring about the good or appreciating the good or some other contemporaneous positive response. Several theorists have recognized that when we theorize about prudential value we need a broader psychological notion than just motivation.⁶ However, this issue of terminology points to something with more than merely terminological significance. It points to a modal distinction between types of prudential facts—a distinction between goodness that is currently realized and goodness that is merely possible; although no one has made much of this modal distinction, I will do so here. In what follows, I defend the idea that ⁴ Smith (1994) defends moral judgment internalism. For detailed surveys of the many forms metaethical internalism can take and the relevance for particular debates, see Van Roojen (2015) and Rosati (2016). ⁵ Railton (1986a). ⁶ In addition to Rosati (1996), see Velleman (2000), Dorsey (2017), and Heathwood (ms.).

Internalism and Prudential Value

97

prudential existence internalism should be understood simply as a constraint on facts about realized goodness. In those cases where talk of motivation might make sense—cases involving possible good—I argue that we should be motivational externalists. In Section 5.1, I first briefly describe what I call “the structure of good.” The aim is simply to make explicit the important relationships between different kinds of goodness claims. In Section 5.2, I then introduce and defend the coherence of my particular version of internalism. I don’t pretend to offer a direct argument for the plausibility of prudential existence internalism itself.⁷ Instead, since many theorists of well-being accept internalism, I aim to demonstrate the coherence and plausibility of a particular way of formulating existence internalism for prudential value. In Section 5.3, I examine the most influential alternative formulation of the internalism constraint, Rosati’s two-tier internalism, and argue that my account is preferable to hers.

5.1. THE STRUCTURE OF GOOD Ordinary talk about what is or would be good for a person is usually talk about prudential goods. We seek to realize various kinds of intrinsic prudential goods in our lives. I adopt an understanding of intrinsic value as final value, such that its natural opposite is instrumental value.⁸ However, it is worth noting that the word ‘intrinsic,’ when used to signal final value, could also apply to the kind of value had by a good life.⁹ Here, I only use the word for prudentially valuable items within a life. Furthermore, I take no stance here on what kinds of things may count as basic intrinsic prudential goods. ⁷ Some theorists do argue for it. See Rosati (1996), Velleman (2000), and Dorsey (2017). Sarch (2011) offers important criticisms of the arguments in Velleman and Rosati. Dorsey (2017) argues that we should not accept prudential existence internalism simply because we find it highly intuitive. There are, he claims, far too many intuitions that pull in the opposite direction as well, such that we must have some substantive reason for holding onto internalism. I think he overstates his point, however. Those who find internalism compelling rarely find the alleged counterexamples to internalism compelling as well. Rather this is one of the places in philosophy where (sadly) intuitions seem to fall on different sides of a seemingly unbridgeable divide. ⁸ See Korsgaard (1983) and Kagan (1998). Intrinsic value can refer either to the value something has based only on its intrinsic properties (the contrast is extrinsic value), or it can refer to the kind of value something has when we value it for itself, as a final end (the contrast being instrumental value). Kagan defends the coherence of retaining the label “intrinsic value” for the concept of final value, which is the approach I adopt. For more about the concept of intrinsic value, see Zimmerman (2015). ⁹ This point is made by Feldman (2000).

98

Jennifer Hawkins

Everything I say can be adapted (sometimes with a little tweaking of my examples) to the full range of views about prudential goods. Basic, intrinsic, prudential goods are good for us at a specific time or times.¹⁰ If riding a bicycle is intrinsically good for me, then, plausibly, it is good for me while I am riding. Some goods are temporally brief; others last most of a lifetime. It is important to mark a distinction between goods that are currently realized for a person (currently adding value to her life) and goods that could be realized (depending on how things go). We must constantly be aware of the modal distinction between realized good and possible good. Because I maintain that prudential goods are the building blocks of life value, my view is unabashedly additive. What makes a life good are the goods in it. Of course, some theorists reject additive views.¹¹ However, I suspect that many people are suspicious of additive accounts because they focus on rather simple versions of additivism. For additivists like myself, talk of a good life depends on the notion of overall value, or net value. Importantly, a life can be said to be good in two distinct ways.¹² In one sense, a life is good if it has net positive value, if the intrinsic goods in that life outweigh the intrinsic bads, even if only by a little. Something like this seems to be what (some) philosophers have in mind when they talk about “a life worth living.” However, a different sense of good captures the sense in which we all aspire to have a good life. A good life in this sense is one that satisfies a high threshold. It is a life that has a fair amount of net positive value. It is something to aim at. Lives like this are “better than most.” Insofar as we exercise agency in pursuit of our good, what concerns us most are good choices. To understand choices, we must consider the structure of possible future good. For any given individual there are many possible continuations of her life. What counts as the best continuation depends on what the set of possibilities are. The best continuation is the one with the greatest net value. However, it is important to remember that the best continuation of a ¹⁰ Not everyone will agree that goods must be realized at determinate times (e.g. Nagel 1970). I argue for the idea that prudential goods must be realized at particular times in Hawkins (2014). ¹¹ A certain sub-set of theorists reject additive views because they take seriously the idea that the shape of a life matters prudentially. They tend to assume that this rules out additivity (or as some say “intra-life aggregation”). See e.g. Slote (1984), Velleman (1991), Temkin (2012), Glasgow (2013). Dorsey (2015) demonstrates that, depending on why you think shape matters, taking shape seriously may be compatible with aggregation. I am more willing than many to simply downplay the importance of shape of life considerations. On this see Hawkins (2014). ¹² I find it intuitive to distinguish (as I do in the text) between a minimal sense of good and a more robust, aspirational sense that depends on a high threshold. However, sometimes the two are collapsed. For example, Sidgwick (1907) equates welfare with happiness and defines happiness as a “surplus of pleasure over pain.”

Internalism and Prudential Value

99

life (which means ‘best’ relative to the options) may not be good in either of the senses sketched above: it may not have positive net value, and it may not be good in the aspirational sense of a life “better than most.” The best choice for an individual at a time is the one that moves her in the direction of realizing the best continuation of her life. A good choice is one that moves her towards one of the better continuations in the set of possible continuations. In ordinary speech, evaluations of choices and outcomes often get blurred. For example, we are often reluctant to call a choice good if the outcome it produces is not (even though it leads to the least worst outcome). And sometimes we call a choice good if it helps us to realize something good immediately, even though the further consequences of the choice are not good. In what follows, however, I stick to the use of “good choice” defined here. Sometimes realizing a good continuation of a life requires the subject to make dramatic changes—changes of direction or goal. It may require new skills or training. It may even require changing aspects of herself. This is important, because goodness facts that lie on the other side of great change are the hardest for us to “see.” It could thus be true of an individual that the best choice for her is one she does not recognize as good at all. Like intrinsic prudential goods, choices have value at a time. But their value is instrumental, depending on how the choice relates to the possibilities the subject confronts.

5 . 2 . E X I S T E N CE IN T E R N A LI S M A N D P R U D E N T I A L V AL U E Let us now consider potential internalist constraints on goodness facts. Recall that existence internalism is the claim that there is a necessary connection between, on the one hand, normative facts (in this case prudential facts) and on the other hand, human psychological responses. What might this connection be? In the case of prudential value there is a rather simple suggestion one can make. Unsurprisingly, being simple it has problems. However, these are more easily dealt with than typically supposed. I shall consider and respond to these problems along the way, some in this section and some later. But for the moment, reflection on the simple suggestion is a good way to start. What I call the simple suggestion is best brought out through examples.¹³ Imagine a woman named Samia who is a newspaper writer. Various people ¹³ The following example presupposes a non-mental state theory of welfare since the activity of writing is being considered as a candidate for intrinsic goodness. Nothing much turns on this as I hope will be clear by the end. But a concrete example has to make some sort of assumption about welfare.

100

Jennifer Hawkins

claim that being a writer is intrinsically good for Samia. They mean that it is good for her now, not that it will be good for her at some point in the future, nor that it would be good for her if only she were different in some way. Because they mean it is good for her now—as she is—they are making a claim about realized intrinsic prudential goodness. As we learn more about her case, the facts cast doubt on their claim. First, we discover that writing and the writing life do not give Samia any pleasure.¹⁴ Next, we consider whether it makes her happy. After all, people can sometimes be made happy by activities or projects that they would not exactly describe as pleasurable. But the evidence suggests that Samia is not happy. Writing leaves her feeling depressed and drained. Sometimes people pursue things they value even though the pursuit doesn’t fit their affective profile very well. In such cases, despite how the pursuit makes them feel, they think the thing they are pursuing has value. It is because they value it that they persevere. But Samia doesn’t value writing either. She doesn’t think it is important in its own right. In fact, she doesn’t see any value in it at all. For whatever reason, it is just the job she has right now and she has to do it to make a living. At this point it becomes incredibly hard to believe that writing is intrinsically good for Samia. This suggests a general type of necessary condition on realized intrinsic goodness. Call this the   (SC): SC: A necessary condition on something X being a basic intrinsic prudential good for a subject, S, at a time T, is that S must have some kind of positive response to X at T.

Here, I intend positive response to be construed quite broadly to include positive sensations, feelings, affective states, and positive evaluative attitudes.¹⁵ However, positive responses must be psychological and at least partly conscious. It is also important to note that the condition is framed as necessary, but it says nothing about the sufficiency conditions for good. One problem with SC emerges quickly. It has to do with our understanding of ‘response’ and whether it is flexible enough to cover cases of temporally extended value and the attitudes that ground such value. Suppose (unlike a hedonist) I think that relationships can have intrinsic prudential value. A relationship might last for a long time and have value for me during all of that time. But presumably I am not having a positive response

¹⁴ Even non-hedonists may think that pleasure is a sign of the presence of value. ¹⁵ My intention is to be as broad as possible within the range of positively valenced psychological reactions. Heathwood (ms.) similarly argues for the need to define the psychological side of an internalist constraint broadly. We thus differ from Dorsey (2017) who defines the constraint in terms of valuing attitudes.

Internalism and Prudential Value

101

to this relationship (or the person I am related to) at each of the times when (intuitively) it is good for me. What seems likely is that the relationship becomes good when I first develop a certain attitude of valuing towards it. But this attitude will not be at the forefront of awareness from then on. Of course there is a sense in which, all through that time, I ‘have’ the valuing attitude. This is a dispositional sense. Most beliefs and evaluative attitudes require a dispositional analysis. For years I have had a belief about where I was born. If you ask me if I was born there I will say yes. But I hardly ever think about this fact, despite believing it. Only rarely does this content rise to the level of occurrent thought. Evaluative attitudes are more complex than ordinary non-evaluative beliefs. It might call into doubt my valuing of X if I rarely thought about X. If I value X, occurrent positive thoughts about X must occur, at least periodically. Moreover, if X is something (or someone) I interact with regularly, one would expect me to have various kinds of positive response to X on a regular basis. It was (at least in part) the absence of such regular, positive responses to writing that raised doubts about Samia. It is, of course, entirely unclear how many responses are required or how often they must occur. But the details need not concern us here as long as we can agree that it is too strong to say that positive response must be constant, and too weak to require that positive response only occur once. I suggest we talk about valuing and other ongoing positive attitudes as ‘positive engagements,’ in recognition of the fact that such attitudes, though dispositional, will with some regularity become manifest in occurrent positive responses. Then we can make the following amendment: SC2: A necessary requirement on X’s being a basic, intrinsic prudential good for S at T is that S must either have a positive response to X at T, or S must have an ongoing positive engagement with X that still holds at T.

However, a second problem follows quickly in the wake of the first. For SC2 seems to require that all good things enter the awareness of a subject. Most contemporary theorists of welfare reject what James Griffin labeled “the experience requirement,” the idea that prudential goodness (or badness) must enter our experience or our awareness.¹⁶ They insist that intrinsic prudential value can be realized even when the subject is ignorant of this. The most straightforward example of theories that allow this are traditional versions of desire fulfillment theory that allow that the satisfaction of a desire (an event in the world) is good for a person even if she doesn’t ever know it has been satisfied. ¹⁶ Griffin (1986).

102

Jennifer Hawkins

This is a problem; however, it is important to keep certain goals in mind. The aim is to articulate a very general internalism requirement that focuses on realized intrinsic prudential good. Any candidate must minimally explain our intuitions about cases like Samia. But we can accomplish these dual goals without at the same time building in an experience requirement. We just change SC2 to SC3: SC3: A necessary condition on something, X, being a basic, intrinsic prudential good for a subject, S, at a time T, is that: If S is aware of X at T, S must either have a positive response to X at T, or S must have an ongoing positive engagement with X that still holds at T, or If S is unaware of X at T, it must be true that S would have had a positive response to X at T if she had been aware of X at T.

This formulation leaves room for the possibility that something, X, might be good for S at T, even though S does not respond positively to it then because S knows nothing about it. It holds onto the idea (which seems key) that a certain kind of psychological relation must obtain between S and X at the very time X is good for S. What is relinquished is the idea that this relation must involve a direct experience of the good thing. The right kind of psychological relations can obtain at a time as long as certain psychological dispositions for response obtain at that time.¹⁷ It is also important that the counterfactual element here is minimal. No appeal is made to how S would respond to X in conditions extremely different from the ones she is currently in. The counterfactual appeals to a situation exactly like the real one but for one difference: S’s awareness of X. Finally, such an account can still easily handle the more ordinary cases in which someone is aware of a putative good. Writing is not good for Samia, since Samia clearly is aware of writing yet has no positive responses to it and no ongoing positive engagement with it. Some theorists may still find SC3 problematic. Even though there is no requirement in SC3 that S actually respond positively to X, there is a requirement that S have certain response dispositions toward X at T. Some theorists think that once you give up the experience requirement there is no reason to place restrictions on the subject’s psychology at the time when goodness is realized.¹⁸ But this is mistaken. Without some such requirement ¹⁷ The current framing of the constraint assumes that the intrinsically good thing must be the object of a psychological response. As will become clear in later discussion, I also allow that a theorist could modify SC slightly to include cases in which the psychological response is itself the intrinsic good. Heathwood (ms.) also discusses this. ¹⁸ Derek Parfit (1984: 495) seems to suggest this.

Internalism and Prudential Value

103

it is far too easy to generate bizarre cases in which a person is benefited at a time by something she would (if she actually knew about it) view as deeply bad. The constraint we are articulating requires (as a necessary condition only) a kind of fit between current psychological outlook and goodness facts. No doubt different theorists will be tempted to explain in different ways why such fit is important. But whatever explanation one gives, it is hard to see why it will not apply just as well in the case where the subject is ignorant of the putative good. In what follows, I shall continue to use the phrase “simple constraint” but I will now take it to refer to SC3.

5.2.1. Internalism and Theories of Welfare In addition to whatever surface plausibility the simple constraint has, it gains some indirect support from the fact that many theorists of welfare, particularly those of a subjectivist bent, seem to be—at least implicitly—concerned with preserving something like it.

5.2.1.1. Hedonism Consider first classical hedonism according to which pleasure is the only basic intrinsic good and pain the only basic intrinsic bad. One natural way of understanding the allure of hedonism is to note that pleasure is a mental state that is, when experienced, always welcomed, at least to some degree, by the person whose pleasure it is. It is, of course, possible to have mixed reactions to pleasure. I might feel pleasure, but because I think my feeling pleasure is inappropriate, I might also feel shame, or embarrassment. In short, I can both feel pleasure at a time (and when feeling it welcome it) and yet at a different, higher level of my psychology, adopt a disapproving attitude towards my own pleasure and so desire not to be experiencing it. I suggest we think of this ‘welcoming’ response as an involuntary affective response, one that operates at a different level from more cognitively sophisticated attitudes. It is this immediate, low-level reaction to pleasure—what I am calling the welcoming response—that seems to underwrite our sense that experiences of pleasure satisfy SC3. Moreover, the welcoming response is something we reliably have whenever we experience pleasure, and it lasts only for as long as the pleasure lasts. Thus pleasure naturally and easily satisfies the simple constraint: for the hedonist, pleasure is good whenever S experiences it, and whenever S experiences it, S has a positive (welcoming) response to it at that time.

104

Jennifer Hawkins

Someone might object that whether or not pleasure satisfies the simple constraint depends on the nature of pleasure.¹⁹ There is something to this. To see why, consider two views—distinguished by L. W. Sumner—about the nature of pleasure.²⁰ On internalist views, pleasure is simply a particular type of sensation. What various pleasures have in common is “their positive feeling tone: an intrinsic, unanalyzable quality of pleasantness that is present to a greater or lesser degree in all of them.”²¹ Externalist views, on the other hand, define pleasure in terms of some attitude or response that an agent has towards a sensation. The various sensations might be quite diverse, but what makes them pleasures is the subject’s attitude towards them.²² I find it helpful to think of such theories as bundle theories, because on such views pleasure is a bundle with two elements: a sensation and a certain kind of positive response to that sensation. Both internalists and externalists typically acknowledge the existence of positive response in cases of pleasure. They differ merely in whether they view that response as a reaction to pleasure, or as a part of pleasure. For internalists the welcoming response is a response to a sensation, and the sensation is pleasure. Since the response is a response to pleasure and it is pleasure that is claimed to be good, this neatly satisfies the simple constraint. But for externalists the welcoming response is a part of pleasure. If you combine this view with the claim that pleasure is good, the view in question will not satisfy the simple constraint. The simple constraint claims that something can only be good when the subject has a positive response to it, and that would require a welcoming response (or some other positive response) to the bundle that already contains a welcoming response. In short, if we accept the simple constraint, then in order for externalist pleasure to count as good it would have to be true that we routinely welcome the thing that contains welcoming. This is implausible. How problematic is this? First, it is important to note that it doesn’t threaten my central claim which is this: at least part of the attraction of hedonism historically has come from the fact that theorists intuitively wanted to capture something like the simple constraint and thought that

¹⁹ Guy Fletcher raised this objection when this paper was presented at the 2014 Metaethics Conference. Namely, he pointed out that hedonism plus an externalist account of pleasure would violate SC3. He has also made a similar point in Fletcher (2013). ²⁰ Sumner (1996). It is unfortunate that so many different debates in philosophy are framed in terms of the labels internalism/externalism. Here the terms are currently being used in the service of a debate about the nature of pleasure—a debate entirely unrelated to the metaethical debate that is the main focus of the paper. ²¹ Sumner (1996: 88). ²² Heathwood (2007) is a contemporary defense of an externalist view of pleasure.

Internalism and Prudential Value

105

equating goodness facts with pleasure was a way to do this. Even if externalist theories are currently more popular with philosophers, historically many hedonists were internalists and others appear to have been confused or undecided about which view to accept.²³ Moreover, I suspect internalism about pleasure is the naïve view, in the sense that it is the one most of us begin with. But then, since internalism about pleasure is compatible with the simple constraint, my claim about the historical allure of hedonism may still be correct. Second, the problems here may be pretty minor. Clearly it is the welcoming response (whether thought of as part of pleasure or as a reaction to pleasure) that underwrites many people’s sense that pleasure is good for the one who experiences it (whether or not it is an all-things-considered good is a different matter). Given that this is so, an externalist about pleasure could make several moves. She could modify the simple constraint slightly. For example a hedonist who is externalist about pleasure could adopt SC4, which, in spirit, is not so different from the earlier versions of SC: SC4: A necessary condition of X’s being a basic, intrinsic prudential good for a subject, S, at T is that X be the kind of mental state that has a positive welcoming response built into it.²⁴

Alternatively, the externalist could simply say that we misspeak when we say that pleasure is what has intrinsic value. Rather it is the central element of pleasure—the sensation at its core—that has value, and it has this in virtue of the fact that we naturally welcome it. Finally, one might simply adopt an internalist account of pleasure. After all, it does seem much more natural to say that we welcome pleasure (as opposed to saying that pleasure is partly a welcoming).²⁵ Though I do not have space to consider other theories besides hedonism that focus on occurrent attitudes or experiences, it should be obvious that many of them will also have the resources to satisfy SC3.

5.2.1.2. Desire Fulfillment Theories Although desire theories are different from hedonism, desire theorists also seem to be concerned with something like the simple constraint.²⁶ The ²³ For a discussion of the different strands of thought running through the tradition see Sumner (1996: 87–91). ²⁴ A value pluralist who counts pleasure as one value among others could combine SC4 with the earlier SC3 to get a hybrid principle. ²⁵ For recent, more sophisticated internalist views of pleasure, see Crisp (2006), Smuts (2011). ²⁶ It is sometimes said that one of the most intuitive features of desire theories is that they (better than other theories) capture the internalist constraint. See e.g. Arneson

106

Jennifer Hawkins

evidence for this is indirect, but nonetheless powerful. Without trying to name all the different desire theories or go into their details, we can point to a particular direction that the development of such views has taken over time, one that is suggestive of concerns about positive response to realized basic intrinsic good. The most basic form of desire theory claims that what is good for a person is for her actual desires to be satisfied. It is the desire satisfaction—a state of affairs in the world—that is the bearer of intrinsic value. The mental state ‘desire’ is simply relevant as the device that tells us which states of affairs benefit which people. Only those states of affairs that are the satisfaction of my desires benefit me. What matters for our purposes is the way these views have developed over time. Most theorists take ‘desire’ to be a prospective attitude.²⁷ I desire now what I shall (perhaps) get later. It is this feature of desire that creates problems for basic desire theories. It is usually assumed that the satisfaction of a desire benefits a person at the time the desire is satisfied. However, a person might desire X at T, but when her desire is satisfied at T2 and she is aware of this, it might be that she is disappointed in X. The problem isn’t just that she is unhappy. Desire theorists pride themselves on being able to accommodate the fact that not everyone cares about happiness. Some individuals do quite well in life despite having little happiness. The issue is not happiness per se, but rather lack of any kind of positive response to the putative good. Not only does it not make her happy or give her pleasure, but now that she ‘has’ it she sees no value in it. This is just the same point made earlier: if something is supposedly intrinsically good for a person at a time (and she is aware of this thing) we expect there to be some kind of positive response to the good thing at that time. Theorists can differ about the type of response that matters, but if (1999), Noggle (1999), Brink (2008), Dorsey (2017). However, what is meant is not that desire theories preserve something like the constraint I defend here. Rather such comments presuppose an entirely different vision of what prudential internalism amounts to, a vision I think we should reject (see my arguments against Rosati in Section 5.3). Although the features of desire theory commonly thought to preserve internalism are not my focus, I think one can trace in the development of desire theory an implicit commitment to something like my constraint. So far as I know, such a commitment has never been explicitly articulated by desire theorists or labeled by them as internalism. But that the commitment is there at all strikes me as highly significant and telling. ²⁷ Sumner (1996: 129). Heathwood (2006) bucks this trend, using ‘desire’ to name a concurrent attitude, a pro-attitude towards something at the time you have it. He thereby becomes immune to the particular problem for desire theories outlined here. However, it is doubtful that his theory (a species of what Parfit (1984) called “preference hedonism”) would be classified as a form of desire theory. Most simple desire theories do face the problem.

Internalism and Prudential Value

107

there is no positive response at all we cannot sustain the thought that the thing really is good. Because there are so many ways that prospective desires can be defective and thus so many ways that the satisfaction of prospective desires can disappoint us, simple desire theories have very few advocates. However, it is precisely the recognition of this problem as a problem that I want to highlight, because theorists with otherwise quite diverse approaches have shared this intuition. Desire theorists themselves have tried to adjust their view to avoid this problem, which suggests they accept that the counterexamples point to a real issue. This reflects an implicit acceptance of the simple constraint. In response, desire theorists have moved to some type of informed desire theory. Some of these views are like the simple version in holding that the satisfaction of informed desires is of intrinsic value. But others depart from this somewhat. Although still called desire theories, here the idea is that under certain improved epistemic conditions a person would be able to act as her own best advisor.²⁸ From the better vantage point she would be able to ‘see’ the prudential goodness facts, facts about goods but also facts about choices and lives. The individual’s good is defined as what from this special epistemically advantaged position she would want for her actual self if she were to contemplate her actual self ’s position as someone about to assume her actual self ’s position.²⁹ The advantaged self might want particular intrinsic goods for her actual self, but she might also want her actual self to make certain choices that will lead (perhaps only in the long run) to intrinsic goods. The problem with all such accounts is that it still seems possible that someone might get the thing her informed self wants for her and yet not find any value in it. Alternatively, she might make the choice that her informed self wants her to make and persevere in this direction, yet never come to respond positively to or see value in the things she thereby acquires. In the simpler case we find it implausible to suppose that a thing could be intrinsically good at a time when a person (who is aware of it) sees no good in it. In the slightly more complex case, we find it impossible to believe that a choice is good, unless we are confident that it will at least eventually lead to encounters with things we respond positively to. But when we have no confidence that continued pursuit of a particular path will eventually lead to positive response of some form, we find it impossible to view the case as one where an individual is proceeding down a path towards something good for her. Even when the ideal epistemic conditions are described as

²⁸ Railton (1986a; 1986b).

²⁹ Railton (1986b: 174).

108

Jennifer Hawkins

incorporating full information and full rationality, philosophers continue to have doubts about whether such cases (cases in which the actual self finds no value in the things selected for her by the ideal self ) will be eliminated.³⁰ In other words, the difference that remains between theorists who accept full information accounts of prudential good and theorists who don’t does not appear to be a disagreement about whether a theory should satisfy SC3 but instead looks like a disagreement about whether that particular theory in fact satisfies SC3. Both parties seem committed to the idea that the theory would fail if it failed to satisfy the simple constraint.

5.2.1.3. Objective List Theories Objective list theories are traditionally thought of as those theories that reject the necessity of any kind of psychological response. Although this means that most of them fail to satisfy SC3, this is hardly damning since most of them explicitly reject any form of existence internalism. However, it is worth noting that a number of theorists have recently tried to amend this very feature of objective list theories, recognizing it as the most problematic part of the view. As Shelly Kagan puts it, “Friends of an objective account of well-being seem forced to accept the unappealing claim that I could be extremely well-off, provided that I have the right objective goods in my life, even though these things hold no appeal for me, and I am, in fact, utterly miserable.”³¹ It is precisely the unattractiveness of such a claim that leads Kagan (as well as a number of other theorists interested in objective views)³² to consider a modified version according to which objective goods can improve a person’s life only if he or she has positive attitudes towards them. On Kagan’s view a person must enjoy the objective goods. Of course it is not at all clear whether these theorists would be willing to distinguish between realized good and merely potential good and so not clear whether they would embrace the simple constraint. But the quote from Kagan certainly does seem to suggest the simple constraint. For it emphasizes the extreme implausibility of “having” or “possessing” an intrinsic good at the very time it makes you miserable.

5.2.2. Further Objections Despite its initial plausibility, many theorists continue to worry about certain putative counterexamples to SC3. These challenge the necessity of ³⁰ Sobel (1994), Rosati (1995). ³¹ Kagan (2009: 254). ³² e.g., Adams (1999); Darwall (2002).

Internalism and Prudential Value

109

positive response (or positive engagement) in cases where the subject clearly has knowledge or awareness of the putative intrinsic good. A common case is one where a person is in the process of acquiring some new skill or transitioning into some new set of circumstances. For example, consider Maria who, unlike Samia, has the talent to become an excellent writer and would thrive as one if she could just get through the excruciating first years. Surely (says the objector) it is good for Maria to keep going despite the fact that currently she has no positive response? Yes. However, the sense in which it is good for her to do this is instrumental. It is a good choice to persevere because genuine intrinsic goods lie in wait for her if she sticks with it and acquires the skill. A different case raises more serious concerns. Consider someone in a severe clinical depression. He is so depressed he does not currently have the capacity to respond positively to anything. He feels no pleasure. He is not happy. He values nothing. His family is loving and supportive and is trying to get him into a good treatment program. Surely (says the objector) it is good for him to go into treatment? Surely the love and support of his family are good for him even now, while he cannot appreciate their efforts? The simple constraint seems to imply that nothing is intrinsically good for this person right now. But (so the objection goes) that can’t be right. First, it is important to notice that the case goes beyond clinical reality (and so may never actually be instantiated). People who are severely depressed often cannot feel much pleasure. And they often value far less than they did beforehand. But it is not clear that there really are people who remain selfaware and yet have no capacity to respond positively to anything at all. Yet there are people whose capacities for positive response are severely limited. So the question remains, what is good for them? The answer depends on thinking about the structure of good. Certain kinds of claims involving ‘good’ are true of this individual. If the person is treatable, then there exist possible continuations of his life in which he could realize many basic, intrinsic prudential goods that he cannot realize now. Assuming that some of these possible life continuations are ones that contain net positive intrinsic value, he has a normative prudential reason now to try and bring about one of these. Whether or not he realizes it, he has a reason now to try and end his depression. Mostly likely he doesn’t realize it. But it is also true that those who love him have reason now to help him out of depression and should do so precisely because he can’t appreciate his own situation. My account does say that very few things count as basic intrinsic prudential goods for him at this point in time. But this seems right. Indeed, it is part of the explanation of the badness of depression that it narrows our evaluative connections to the world in this way. It undermines us as valuers and as responders.

Jennifer Hawkins

110

Some people, however, are still not satisfied. Surely, they argue, the love and support of the family is intrinsically good now? It is not clear to me why we should want to say this. If he currently has no ability to respond positively to support and cannot maintain positive engagement with those who love him, it seems that currently the love and support is quite literally not benefiting him. And because of this, it fails to be intrinsically good for him now. This is compatible with the claim that there are strong reasons now for providing love and support. It may lead to the individual getting the right help. The love and support may also be good in the sense that it is good to keep trying to engage someone who is depressed. You never know when your expressions of love might “get through.” Finally, some theorists might wish to say that loving another person is a (morally) good thing for the one who loves or that it is just non-relationally good—good simpliciter—that there be more love in the world. All of this is compatible with recognizing that the love and support is not intrinsically good for the depressed person while he is so deeply depressed.

5 . 3 . M O T I V A T I O N T O P U R S U E T H E GO O D ? So far I have focused entirely on one type of constraint, a constraint on truths about realized good of the form: “X is a basic, intrinsic prudential good for S now.”

But what about potential good? What about truths of the form: “Y would be good for S”?

Should such truths motivate agents in the way that some philosophers think moral truths should motivate individuals to act rightly? We need several clarifications before we can answer. To begin with, the sentence “Y would be good for S” is ambiguous. ‘Y’ could refer to some putative intrinsic good or to a choice. Moreover, it could be true that something—perhaps a particular relationship—would be intrinsically good for a subject S, but still false that she ought to pursue it. For it might also be true that the relationship (which is a good) is part of a path through life that is not as good overall. Even if the relationship is intrinsically good, if it is not part of one of the good continuations of S’s life, then it is not a good choice. So if we want to consider issues of motivation, we should focus on true claims about good choices. If any prudential facts should motivate individuals to act, surely it would be these. Are all subjects such that if they are minimally rational and confronted with such facts they will be (at least somewhat) motivated? The right answer,

Internalism and Prudential Value

111

I think, is: no.³³ When it comes to merely potential good, we should be motivational externalists. To see why, we first need some clarity about what is at stake. We are interested in existence internalism, which is a metaphysical thesis. It is easy to confuse intuitions in this realm with intuitions about judgment internalism. Judgment internalism holds that there is a necessary connection between sincere judgment involving evaluative concepts and motivation. For all I know, judgment internalism may be true for prudential good. Perhaps the notion of prudential good is such that if you genuinely believe Y is the best choice for you, you will necessarily have some motivation to choose Y. I take no stand on that. But notice that if judgment internalism is true, people will be motivated by anything they believe to be good, whether or not it really is good. However, existence internalism is concerned with the relation between normative facts naturalistically construed, on the one hand, and human responsiveness, on the other. Theorists who want to capture some form of existence internalism tend to equate prudential value facts with facts they think most people care about. Most people care about their happiness, even if that is not all they care about. Because of this some theorists have hoped to secure the right kind of internalist connection by equating prudential value facts with facts about what would make someone happy. Indeed, the constraint would work if we could be sure that everyone is always, at least to some degree, motivated by facts about what would make them happy. But this is not the case. Nor am I aware of any other candidate set of goodness facts that people are reliably motivated to pursue, independently of their evaluative beliefs about prudential good. In short, to test existence internalism we have to imagine a subject who is considering a choice that, according to our favored theory, is good for her. She need not think of the choice or its outcome as good, but she must have correct descriptive beliefs about what her life would have in it if she made the choice and what her life would be like for her at various points in the future if she made it. The question is whether we can identify any plausible potential goodness facts that we can be certain she will motivated by. There is no reason to think that we can. If I am right, then we should be motivational externalists when it comes to unrealized, merely potential good. Of course, even externalists can allow that people are sometimes motivated by such facts. What they deny is that such ³³ Many theorists incline towards ‘yes.’ Rosati (1996) defends an indirect form of motivational internalism. Hall and Tiberius (2016) assume the truth of motivational internalism, and argue that one point in favor of subject-dependent theories is that they are well placed to explain this truth. However, it should now be clear that not all subjectdependent theories actually preserve motivational internalism about potential good.

112

Jennifer Hawkins

motivational connections are necessary. And that seems right, particularly once we get clear about the structure of good. Recall that sometimes it can be good for a person to change her circumstances or herself dramatically. It is highly plausible that, at least once in a while, the best path forward through life for an individual will be one of the paths that requires dramatic self-change. But then our theory will say dramatic change is the best choice. And in these kinds of cases, even if we have great confidence in our theory, we simply cannot guarantee that such facts will have motivational force for the agents. To help make clear that the problem is not specific to a particular theory, I will consider two examples. For the first I will assume the truth of hedonism, and for the second the truth of some form of value pluralism. Consider Miserable Maud, a religious ascetic.³⁴ She believes pleasures come from the devil and must be avoided as much as possible. She lives her life in a way that pretty successfully minimizes pleasure. But Maud is certainly capable of experiencing pleasure and sometimes despite herself she experiences some. According to hedonism the best path forward for Maud is one that leads her to experience a great deal of pleasure. However, Maud is unlikely to embark on such a path without a dramatic change in her outlook, since her current attitudes lead her to avoid pleasurable things and also serve to dampen her experience of the few pleasures she unwittingly stumbles upon. Not only is the best life for Maud one in which she has more pleasure, the best life is one in which she changes herself in ways that make it possible for her to experience much more pleasure than her current attitudes allow. Thus if hedonism is true it may well be that the best choice for Maud now is to begin the process of altering her religious beliefs. She need not abandon religion. But she needs a religious view that (unlike her current one) encourages pleasure. The internalist question is whether we can expect Maud to be motivated now by knowledge of the descriptive facts about how much pleasure she could experience if only she changed her beliefs. Is it likely that she would be motivated to make such a choice? Absolutely not! From her current position she has zero motivation to change. Of course, for all I have said so far, the problem could be a problem with hedonism rather than motivational internalism about potential good. If hedonism is false, then it will not be surprising if someone is not motivated by knowledge of facts about how she could change to maximize her experience of pleasure. However, the problem arises for other views as well. To see this, consider the next example. Consider Miserable Martha. Martha is not always miserable, but she does suffer from regular bouts of serious unipolar depression, and when in the grips of these spells she is miserable. When she is not depressed she is quite

³⁴ The example is borrowed from Fletcher (2016: 67–8).

Internalism and Prudential Value

113

artistic. She strongly suspects that part of what contributes to the quality of her art is something she gains from her depressive spells. Suppose she is right. And suppose she has the following option: she could take a medication that would relieve her of her depressions, but it would also alter slightly her personality. She would be capable of living a wonderful life as the person she would become, but she would not be nearly as creative and the quality of her artwork would be significantly less. She could, however, have healthy relationships that are currently impossible given her recurrent depressions. And she could succeed in many other, equally gratifying, though nonartistic pursuits. Here I assume a pluralistic view of welfare that allows for many different basic, intrinsic prudential goods, including creative production of art, other kinds of achievement, and relationships. Suppose the best possible continuation of Martha’s life—the one with the greatest net positive intrinsic prudential value—is one of the lives she could lead only with the medication. This means she has most reason right now to give up her artistic pursuits, take the medication, and embark on a highly different path through life. Should we assume that the person she is now— during one of her non-depressed moments—would be motivated by knowledge of these facts? Again, recall that what she knows are just the descriptive facts about what her life will have in it and what it will be like for her once she becomes the post-medication person. Will knowledge of such facts necessarily motivate her to choose the medication? It seems clear to me that we cannot expect this. Indeed, it is quite likely that the facts won’t motivate her at all, particularly if from her current position she can’t personally identify with the individual she would become if she pursued that path. I don’t think this is a problem. We shouldn’t really expect her to be motivated by facts about her potential good that are so removed from her current circumstances. The prudential goodness facts sometimes point in directions that, from our limited, current perspectives, seem odd. It may just be that the complexity of future prudential value ensures that we cannot always care about our future good from afar. Motivational externalism about potential good begins to seem like the sensible option.

5.3.1. Two-Tier Internalism? Before we settle for SC3, we should examine carefully the view of Connie Rosati, labeled “two-tier internalism.”³⁵ To date this is the most welldeveloped alternative to my view. ³⁵ Rosati (1996: 307). Alexander Sarch (2011) argues against Rosati’s two-tier internalism. However, he does not contemplate the possibility of an internalist constraint on realized good like the one defended here.

114

Jennifer Hawkins

To grasp two-tier internalism we need a couple of definitions. First, we need the notion of “ordinary optimal conditions.” These are just the conditions of judgment most of us occupy most of the time. They exclude the obvious sorts of conditions that distort judgment (or make it impossible)—things like being asleep, being drunk, being delusional, being in a state of intense emotion, or being clinically depressed.³⁶ Second we need the notion of “ideally authoritative conditions.” This is the idea of conditions that improve judgment to the point where it can’t go wrong and which are authoritative for that reason. Most of us realize our ordinary judgments are hampered by various limitations. We lack knowledge. We are prone to many forms of irrationality. Thus, it is natural to suppose that if we had full knowledge, or could be always rational, our judgments would presumably be much better. If we push such ideas to the limit we arrive at the notion of ideally authoritative conditions—the set of epistemic conditions that are ideally authoritative for making judgments about prudential value.³⁷ I shall refer to these conditions as C*. We can now explain two-tier internalism as follows. Consider a case of potential good (as we shall see it works the same way for realized good). Facts about potential good can typically be translated into claims about good choices in the present. So, suppose it is true that Y is the best choice for S now. Speaking roughly, two-tier internalism requires first that S herself under ideally authoritative conditions recognize this fact about her good (that Y is the best choice). For various reasons the actual requirement needs to be formulated in a slightly more complicated way as follows: it must be true that in C* if S contemplates the situation of her actual self as someone about to assume the actual self ’s position, she would want her actual self to choose Y.³⁸ The second requirement of two-tier internalism is that under ordinary optimal conditions a person, S, cares about what she would want in C*. This means that even if in ordinary circumstances she can’t recognize facts about her good as facts about her good (and so can’t be motivated by them), she still cares at that time about what her ideal self would want in C*. Of course, presently she can’t make the connection between what is really good for her (Y) and what the ideal self would want in C*. But the fact that right now she cares about what her ideal self would want in C* means that if she could

³⁶ Rosati (1996: 304–5). ³⁷ Rosati (1996: 304). ³⁸ For ease of exposition, I often leave out these details, but they should always be understood to be present. I sometimes use short phrases like “what S would want in C*” to stand in for longer phrases of the sort: “what S would want in C* for her actual self if she were contemplating the circumstances of her actual self as someone about to assume the actual self ’s position.”

Internalism and Prudential Value

115

make the connection—if she could come to see that Y is what the ideal self would want in C*, she would right now have some motivation to pursue Y. Rosati clearly intends two-tier internalism to serve as a constraint on both facts about realized prudential good and facts about potential prudential good. In other words, she does not make the modal distinction that has been central to my discussion so far. For that reason, let us briefly consider in turn, two-tier internalism as a constraint on realized good, and then as a constraint on possible good. Rosati and I agree that there is an internalist constraint on facts about current realized intrinsic good—that there must be some kind of psychological link between a subject and facts about what is good for her now. We simply offer two different accounts of what that constraint is. Like SC3, Rosati’s constraint is best understood as a necessary condition only. Although two-tier internalism and SC3 are naturally understood as competing accounts of a single constraint on facts about realized good, nothing rules out in advance the possibility that there are two constraints. This turns out to be relevant in the following way. It is incredibly difficult to undermine the two-tier internalist requirement for realized good directly. To do so we would need to have a clear case in which we could all agree that something, X, is currently good for a subject, S. And we would then need to establish in that case two-tier internalism fails to hold. That would decisively show that the two-tier constraint is not necessary, as claimed. However, for various reasons, it is incredibly difficult to know in any given case whether or not two-tier internalism is satisfied. So rather than undermine it directly, my aim is to show its irrelevance. First, it seems to me that no matter what the status of two-tier internalism, we still need SC3 in the case of realized goodness facts. Suppose, for example, a theory were to say that writing is intrinsically good for Samia at the very time when she is so utterly indifferent to it. I confess, I simply can’t imagine anything that would convince me to view something as intrinsically good for her at that time given the psychological facts about her at that time. So SC3 remains necessary. Second, it seems important to note that one of the prime motivations for two-tier internalism fails in the case of realized goodness. Two-tier internalism was formulated in response to worries about the ability of a subject to recognize and positively respond to facts about her own good. It is designed to provide a kind of indirect psychological link in cases where a direct link is lacking. But once we make the modal distinction, and recognize the importance of SC3 for facts about realized goodness, we also see that indirect links are not needed, because direct links (currently existing links) are required. SC3 rejects any theory where they are absent. In other words, once we focus narrowly on facts about realized good, we see that there are no cases in which X is currently good for S but S as she actually

116

Jennifer Hawkins

is fails to respond positively to X in any way. But since no such facts exist, there is no need to salvage internalism for these cases by forging an indirect link through facts about what S would want in C*. Nonetheless, one might think that we need something like the two-tier requirement when we focus on facts about possible good. It is unlikely that any theory of welfare will be able to ensure that agents are always reliably motivated to pursue what would be best for them. But one might hope that two-tier internalism could forge an indirect link between the facts about potential individual good and the subject’s current motivations. Consider again Miserable Martha for whom (let us assume) the best choice is medication and self-transformation. If the two-tier requirement is satisfied, this means both (a) that in C* Martha would want her actual self to adopt the life that medication makes possible, and (b) right now (assuming she is not currently in the grips of depression) Martha cares about what her ideal self would want in C*. The question is whether the best theory of welfare can ensure both that Martha in C* wants the right things and that actual Martha cares about the desires of her ideal self. To put the point more dramatically, should we reject a theory of welfare if it fails to ensure these two claims? I think not. There are several reasons for this, though I grant that the first reason is not widely shared. I am skeptical of the idea that we can describe a set of ideal epistemic conditions that could be occupied by a real human being and would reliably reveal to this human occupant all the truths about her good.³⁹ In other words even if we make sense of the idea of an agent who is fully informed and fully rational, I have no faith that the pronouncements of such an idealized self would really track the goodness facts. I am not sure we should trust the pronouncements of such a self. Nor do I think there is some way to amend the description of ideal conditions to fix the problem. This is the basis for my skepticism about full information theories of prudential good. But it is relevant to the question of whether to accept two-tier internalism, for although it is not a theory of prudential value, it is a proposed constraint on such theories. If there is no C* such that the pronouncements from C* are always correct, then we cannot require a theory of welfare to conform to a restriction that appeals to C*. However, even if one does not share my skepticism about C*, it is possible to raise worries about two-tier internalism. Two-tier internalism assumes that there is some description of ideal conditions C* such that C* satisfies two requirements. On the one hand, C* is such that all welfare agents view C* as authoritative (this is what underwrites the confidence that

³⁹ Similar skepticism is expressed by Sobel (1994) and Rosati (1995).

Internalism and Prudential Value

117

under ordinary optimal conditions agents will care about what their ideal selves want in C*). And on the other hand, C* is such that agents in C* reliably perceive the goodness facts (or at least reliably want for their nonideal selves the things that are really good). The problem, however, is that there may be no C* that can satisfy both conditions. Suppose for the sake of argument that there is some set of ideal epistemic conditions C* such that an agent in these conditions reliably judges her own good. Still, actual agents do not all agree about what these conditions are. Probably it is true for the vast majority of people that there is some description or other of better epistemic conditions they would accept as authoritative. Let us call the conditions accepted by Martha C+Martha, and the ones accepted by Maud C+Maud, and for each person the conditions she would view as authoritative are C+hername. In ordinary optimal conditions Martha cares what she would want in C+Martha and Maud cares what she would want in C+Maud. But unless C+Martha is the same as C* this will not forge a motivational link with the true goodness facts. In short, if people do not generally agree about ideally authoritative conditions, then even if there is a C* that offers a reliable epistemic perspective on the prudential goodness facts, it will be impossible to establish an indirect motivational link from individuals to those facts. It is important to remember that many people, even when they are free of violent emotion, are not delusional, and so on, are willing to grant authority to less than ideal judgments—judgments made without full information or that are irrational in one or another sense. Nor can they all easily be brought to see the error of their epistemic ways. Even if very many people can, we have to remember that internalism is supposedly a necessary condition. My point is just that there is no value for C* such that C* offers the right kind of perspective on the goodness facts and C* is viewed as authoritative by all people in ordinary optimal conditions. In the very best case the link remains highly contingent. Therefore we should not require a theory of welfare to satisfy the two-tier internalism requirement. Accepting motivational externalism about potential good might be more distressing if we were also externalists about realized good. To embrace a dual form of externalism would be to accept that the psychology of individual subjects places no necessary constraint on prudential goodness facts. That does seem deeply problematic. Since most theorists until now have not distinguished realized good and potential good and so have not considered whether the constraints for each might be different, it is easy to see why a claim like two-tier internalism would be popular. But after reflection it seems that it just isn’t necessary. The right link between the psychology of agents and goodness facts is a link that holds when basic, intrinsic goodness is realized. And once we see that, we see we have all we need.

118

Jennifer Hawkins 5 . 4 . C ON C L U S I O N

We have covered a lot of ground. I first reviewed my understanding of the “structure of good,” emphasizing a modal distinction between realized intrinsic prudential goods and merely potential good. Against this backdrop I argued that we should understand existence internalism for prudential value as a constraint on realized intrinsic good. In an attempt—at least for now—to keep the formulation of the constraint broad so that many theories might satisfy it, I proposed the following formulation: SC3: A necessary condition on something, X, being a basic, intrinsic prudential good for a subject, S, at a time T, is that: If S is aware of X at T, S must either have a positive response to X at T, or S must have an on-going positive engagement with X that still holds at T, or If S is unaware of X at T, it must be true that S would have had a positive response to X at T if she had been aware of X at T.

This is a merely necessary constraint. Positive response should be taken to include a wide range of positive psychological states. Having defended SC3, I considered whether we should also want some form of internalism constraint that would guarantee that subjects are motivated, at least to some degree, by truths about their own potential good. We should not. We should accept motivational externalism about potential good. Individuals will not necessarily be motivated to seek their good, but if they seek it, and get it, they will respond positively to it. Finally, I considered and rejected two-tier internalism, a highly sophisticated attempt to forge an indirect link between prudential goodness facts and an individual’s motivations. SC3 is superior to two-tier internalism when we focus on facts about realized goodness. And in the case of potential good it is ultimately wiser to accept that no internalist constraint, even an indirect one, can be had.

References Adams, Robert Merrihew. 1999. Finite and Infinite Goods. Oxford: Oxford University Press. Arneson, Richard. 1999. “Human Flourishing versus Desire Satisfaction,” Social Policy and Philosophy 16 (1): 113–42.

Internalism and Prudential Value

119

Brink, David O. 1989. Moral Realism and the Foundations of Ethics. New York: Cambridge University Press. Brink, David O. 2008. “The Significance of Desire,” in Russ Shafer-Landau (ed.), Oxford Studies in Metaethics, Volume 3. Oxford: Oxford University Press, 5–45. Crisp, Roger. 2006. “Hedonism Reconsidered,” Philosophy and Phenomenological Research 73 (3): 619–45. Darwall, Stephen. 1983. Impartial Reason. Ithaca, NY: Cornell University Press. Darwall, Stephen. 2002. Welfare and Rational Care. Princeton, NJ: Princeton University Press. Dorsey, Dale. 2015. “The Significance of a Life’s Shape,” Ethics 125 (2): 303–30. Dorsey, Dale. 2017. “Why Should Welfare ‘Fit’?” Philosophical Quarterly 67 (269): 685–708. Feldman, Fred. 2000. “Basic Intrinsic Value,” Philosophical Studies 99 (3): 319–46. Fletcher, Guy. 2013. “A Fresh Start for the Objective List Theory of Well-Being,” Utilitas 25 (2): 206–20. Fletcher, Guy. 2016. The Philosophy of Well-Being: An Introduction. New York: Routledge. Glasgow, Joshua. 2013. “The Shape of a Life and the Value of Loss and Gain,” Philosophical Studies 162 (3): 665–82. Griffin, James. 1986. Well-Being: Its Meaning, Measurement, and Moral Importance. Oxford: Oxford University Press. Hall, Alicia and Valerie Tiberius. 2016. “Well-Being and Subject Dependence,” in Guy Fletcher (ed.), The Routledge Handbook of Philosophy of Well-Being. New York: Routledge, 175–86. Hawkins, Jennifer. 2014. “Well-Being, Time, and Dementia,” Ethics 124 (3): 507–42. Heathwood, Chris. 2006. “Desire Satisfactionism and Hedonism,” Philosophical Studies 128 (3): 539–63. Heathwood, Chris. 2007. “The Reduction of Sensory Pleasure to Desire,” Philosophical Studies 133 (3): 23–44. Heathwood, Chris. ms. “What Is the Resonance Constraint?” Kagan, Shelly. 1998. “Rethinking Intrinsic Value,” The Journal of Ethics 2 (4): 277–97. Kagan, Shelly. 2009. “Well-Being As Enjoying the Good,” Philosophical Perspectives 23 (1): 253–72. Korsgaard, Christine. 1983. “Two Distinctions in Goodness,” The Philosophical Review 92 (2): 169–95. Moore, G. E. 1903. Principia Ethica. Cambridge: Cambridge University Press. Nagel, Thomas. 1970. “Death,” Noûs 4 (1): 73–80. Noggle, Robert. 1999. “Integrity, the Self, and Desire-Based Accounts of the Good,” Philosophical Studies 96 (3): 303–31. Parfit, Derek. 1984. Reasons and Persons. Oxford: Oxford University Press. Railton, Peter. 1986a. “Facts and Values,” Philosophical Topics 14 (2): 5–31. Railton, Peter. 1986b. “Moral Realism,” The Philosophical Review 95 (2): 163–207.

120

Jennifer Hawkins

Rosati, Connie. 1995. “Persons, Perspectives and Full Information Accounts of the Good,” Ethics 105 (2): 296–325. Rosati, Connie. 1996. “Internalism and the Good for a Person,” Ethics, 106 (2): 297–326. Rosati, Connie. 2016. “Moral Motivation,” The Stanford Encyclopedia of Philosophy (Winter 2016 edition), Edward N. Zalta (ed.), https://plato.stanford.edu/arch ives/win2016/entries/moral-motivation/. Sarch, Alexander. 2011. “Internalism About a Person’s Good: Don’t Believe It,” Philosophical Studies 154 (2): 161–84. Scanlon, T. M. 1998. What We Owe to Each Other. Cambridge, MA: Belknap Press of Harvard University Press. Sidgwick, Henry. 1907. The Methods of Ethics, 7th edn. New York: Hackett. Slote, Michael. 1984. “Goods and Lives,” in Goods and Virtues. Oxford: Oxford University Press, 9–37. Smith, Michael. 1994. The Moral Problem. Malden, MA: Blackwell. Smuts, Aaron. 2011. “The Feels Good Theory of Pleasure,” Philosophical Studies 155 (2): 241–65. Sobel, David. 1994. “Full Information Accounts of Well-Being,” Ethics 104 (4): 784–810. Sumner, L. W. 1996. Welfare, Happiness and Ethics. Oxford: Clarendon Press. Temkin, Larry. 2012. Rethinking the Good. Oxford: Oxford University Press. Van Roojen, Mark. 2015. Metaethics: A Contemporary Introduction. New York: Routledge. Velleman, J. David. 1991. “Well-Being and Time,” Pacific Philosophical Quarterly 72 (1): 48–77. Velleman, J. David. 2000. “Is Motivation Internal to Value? in The Possibility of Practical Reason. Oxford: Clarendon Press, 85–98. Zimmerman, Michael J. 2015. “Intrinsic vs. Extrinsic Value,” The Stanford Encyclopedia of Philosophy (Spring 2015 edition), Edward N. Zalta (ed.), https://plato. stanford.edu/archives/spr2015/entries/value-intrinsic-extrinsic/.

6 Desiring under the Proper Guise Michael Milona and Mark Schroeder 6.1. INTRODUCTION According to the thesis of the guise of the normative, all desires are associated with normative appearances or judgments. This thesis has held promise, according to many, to explain why motivation by desire rationalizes or makes intelligible, and doesn’t merely explain, action.¹ And it holds promise to yield a non-mysterious source for normative epistemology.² The guise of the normative can take different forms. Most familiarly, it is the thesis that a desire that p represents it as being good if p. This thesis is a version of the guise of the good. But another version of the guise of the normative holds that when you desire to φ, you see yourself as having reasons to take means to φ. This thesis is a version of the guise of reasons. Both of these versions of the guise of the normative hold promise to make good on the key motivations for that thesis. In this paper we will be arguing for the comparative thesis that the guise of reasons is more promising than the guise of the good. Many authors have discussed the guise of the good or the guise of reasons without explicitly taking on the comparative question of which view is to be preferred. For example, Anscombe (1963/2000), Davidson (1969/2000), Quinn (1993), Oddie (2005), and Tenenbaum (2007) all defend forms of the guise of the good with little if any explicit commentary on why it is to be preferred to the guise of reasons. In contrast, Scanlon (1998) and Bedke (2009) defend forms of the guise of reasons, with little if any explicit commentary on why it is to be preferred to the guise of the good. But the fact that this question has mostly not been tackled explicitly does not mean that it is not important. In particular, the guise of the normative cannot be true unless some particular thesis about what normative features ¹ See particularly Anscombe (1963/2000), Quinn (1993), Tenenbaum (2007). ² Oddie (2005), Schroeder (2007).

122

Michael Milona and Mark Schroeder

are represented is true. And it matters which form of the guise of the normative is true. For example, some objections might apply to one but not the other.³ Moreover, for many of the philosophical applications of the guise of the normative, it makes a great difference which form the guise of the normative takes. For example, we are particularly interested in the epistemological applications of the guise of the normative. If desires involve perceptions of the normative, then normative epistemology could have a perceptual basis, analogous to the justificatory story for ordinary empirical beliefs. Such a perceptual basis would not be debunked by the empirical observation that our moral judgments are highly dependent on affective states, which is of course exactly what this view predicts.⁴ But it makes an important difference whether we think that our basic epistemic access to normative facts runs through reasons, or through the good. The structure of our argument is simple. It is that each form of the guise of the good turns out to be incompatible with an important constraint from the theory of how perceptual states come to have contents at all. We diagnose the source of the problem as analogous to a familiar puzzle in moral philosophy, that of trying to explain agent-centered constraints in terms of the value of outcomes. Both structural problems can be resolved by appeal to reasons, in place of the good.

6.2. THE GUISE OF THE NORMATIVE Traditionally, many distinct theses have fallen under the heading of ‘the guise of the good’ or, generalizing, ‘the guise of the normative.’ But we have in mind a very particular target:  : There is some particular normative property or relation N and some other relation R, such that every time an agent has a desire with object o, she perceives things that are R-related to o as being N.

This formulation is neutral about a number of things. It is neutral about which normative property N is represented by desire. This makes it common ground between the guise of the good and the guise of reasons. It is likewise neutral about the relationship of priority between perceiving things as satisfying this normative property, and having the desire. So it can be common ground between those who hold that having a desire consists or ³ Compare Gregory (2013). ⁴ Compare especially Haidt (2001), Greene (2008).

Desiring under the Proper Guise

123

partly consists in perceiving these normative things, and those who hold instead that there are independent facts about the nature of desire, of which it is a consequence that desires are associated with representations of these normative things.⁵ Furthermore, our statement of Generic Guise is neutral about both the object of desire, o, and the relation, R, between the objects of desire and the normative property, N. One form of the guise of the good, for instance, maintains that the objects of desire are propositions, that R is the relation of identity, and that N is the property of being impersonally good—the property that we express by saying things like ‘it would be good if there were fewer conflicts,’ and which is appealed to by traditional, agent-neutral act consequentialism. On this view, whenever you have a desire, you desire that p, and you perceive it as being impersonally good that p. But Generic Guise is flexible. You might hold, for instance, that the objects of desire are individuals (e.g., a coffee) yet still maintain that desire is under the guise of the impersonal good (whereby impersonal goodness is a property of a proposition). This is because R can be a relation other than that of identity. Such a view might say that whenever you desire, you desire some individual x, and you perceive it as being good that you possess x. On this view, the objects of desire are individuals, N is the property of being impersonally good, and R is the relation of being the proposition that you possess o. One dimension along which our formulation of Generic Guise is definitely not neutral, is that we specify that desire is associated with the perception that something has the relevant normative property. It is possible to hold similar views according to which desire is instead associated with something more akin to a judgment that something has a normative property. Our position is that the perceptual version of the view is more promising, and that it is better positioned to play an important role in normative epistemology. But in any case, Alex Gregory (2013) has already argued for a similar conclusion on the assumption of a judgmentalist form of the guise of the normative. Generic Guise is not committed, however, to the claim that you must in any literal sense perceive that something has the relevant normative property; the state it is committed to calling a ‘perception’ earns the title through its analogous phenomenological and epistemological role to ordinary perceptions. A second dimension along which our formulation of Generic Guise is not neutral, is that it assumes that there is uniform relationship between desire and the representation of some normative property or relation. This is a

⁵ This is something we ourselves don’t agree about—Scanlon (1998) defends the former approach and Schroeder (2007) the latter.

124

Michael Milona and Mark Schroeder

substantive assumption, and one that it is possible to relax. Later we will see that one of the best hopes for defending the guise of the good in response to our argument will be to distinguish between different good properties and generalize in this way. Our question, then, is not whether Generic Guise is true, but rather, if it is, what the appropriate values are for N and R, and hence, in addition, what the object of desire is, in general. In the next section, we will narrow this question by defending a general answer about the nature of the objects of desire.

6.3. PROPERTITIONAL DESIRE Ascriptions of desire can be divided grammatically into three classes: determiner-phrase (DP) desire, non-finite desire, and complementizer phrase (CP) desire. These three classes correspond to the following examples: DP  -  CP 

Sue desires a coke. Sue desires to drink a coke. Sue desires that she drinks a coke.

Each of these grammatical distinctions corresponds to a distinct possible object of desire. In DP desire ascriptions, the verb ‘desire’ is followed by a determiner phrase such as ‘a coke.’ These ascriptions correspond to the idea that when you have a desire, you desire a thing. In non-finite desire ascriptions, ‘desire’ is followed by a non-finite clause, such as ‘to drink a coke.’ These ascriptions correspond to the idea that when you have a desire, it is a desire to do some thing or to be some way. And in CP desire ascriptions, the verb ‘desire’ is followed by a complementizer phrase such as ‘that Sue drinks a coke.’ These ascriptions correspond to the idea that when you have a desire, it is a desire for something to be the case. Despite the fact that these three forms of desire ascription appear to make ‘desire’ out to take three different kinds of object, they cannot plausibly be maintained to be independent. In general, whenever a DP ascription such as ‘Sue desires a coke’ is true, there is a corresponding true non-finite ascription, which says that Sue desires to bear some important relation to a coke. For example, ‘Sue desires a coke’ might be true because Sue desires to drink a coke, or because Sue desires to own a coke. But it can’t be true unless there is some way in which Sue desires to be related to a coke. So every DP desire ascription has a corresponding non-finite desire ascription, with which it is equivalent. But the converse is not true. ‘Sue desires to fly’ does not correspond straightforwardly to any DP desire ascription.

Desiring under the Proper Guise

125

This asymmetry, and the fact that non-finite desire ascriptions make explicit structure that is only understood from context in DP desire ascriptions, show that DP desire ascriptions should be understood in terms of nonfinite desire ascriptions. This helps us to avoid positing two kinds of desire—desires for things and desires to be some way.⁶ Advocates of the view that desires are propositional attitudes hold that non-finite desire ascriptions can be similarly reduced to CP-desire ascriptions, which seem to relate the agent to a proposition denoted by the CP clause. However, even if desire is, in some sense, a propositional attitude, CP-desire ascriptions are not the most general form for ascriptions of desire. This is because though in general, if Sue desires to drink a coke, Sue desires that she drinks a coke, the entailment does not go the other way around. Suppose, for example, that Sue is in a busy shopping center, looks across the crowd, sees someone glossy-eyed and grumpy-looking, and forms the desire that she—the person who Sue sees looking glossy-eyed and grumpy-looking—drinks a coke (so that she perks up). If it turns out that Sue is seeing herself in a mirror,⁷ then though it is true that Sue desires that she drinks a coke, it is not true that Sue desires to drink a coke. In general, non-finite desire ascriptions truth-conditionally entail that their subjects conceive of themselves in the normal, first-personal, way in the content of the desire, but propositional desire ascriptions do not. What this shows is that nonfinite desire ascriptions are more fine-grained than propositional desire ascriptions.⁸ So insofar as we are trying to formulate a general thesis about what is represented by desires, we should focus on the case of non-finite desire ascriptions. But non-finite clauses, we believe, do not express propositions. Some verbs, such as ‘neglect,’ take non-finite clauses as their arguments, as in ‘Jake neglected to come to the talk.’ But ‘neglects’ does not express a relation between agents and propositions. Jake cannot neglect an arbitrary proposition; he can only neglect to do something. The proper object of ‘neglects’ is something that you can do, which you have in common with everyone else who is doing it. That is, it is a property. Therefore, we can infer that ‘to come to the talk’ expresses a property.⁹ So we may infer that ‘Jake desires to come to the talk’ expresses a relation between Jake and this same ⁶ Brewer (2009: 62–3) appears to formulate the guise of the normative in terms of objectual desire. But for reasons we point out below, we think that this is a mistake (see also Sinhababu 2015 for detailed criticism). ⁷ Compare the classic cases discussed by Mach (1897) and Perry (1979). ⁸ It is important for this point that Sue’s desire that the woman over there drinks a coke can be understood as Sue’s desire to be such that the woman over there drinks a coke. Lewis (1979) calls such vacuous properties “properties that are shared by worldmates.” ⁹ See Schroeder (2011), especially footnote 31.

Michael Milona and Mark Schroeder

126

property—the property of coming to the talk. Desire, in the most general case, is a relation toward properties—a propertitional attitude, we might say, rather than a propositional attitude.¹⁰ The resulting view of desire resembles David Lewis’s (1979) claim that desire is an attitude toward properties, rather than propositions. We don’t, however, accept all of Lewis’s commitments. What is important for us in this paper is not the proper metaphysics of desire nor the proper semantics of ‘desire’ ascriptions, but just fixing on the most general case of desire. Because DP and CP desire ascriptions can always be rephrased as non-finite desire ascriptions, they are what we must fix on in order to evaluate general claims such as Generic Guise. Consequently, from here forward, we will assume that in the fully general case, the objects of desire are properties, and we will evaluate each implementation of the Generic Guise thesis in that light. So our question is, when an agent desires to F, what is the normative property or relation N and the relation R such that she perceives things that are R-related to F as being N ?

6.4. THE STRUCTURE OF GOOD What makes the guise of the good distinctive is that it claims that the relevant normative property N is the property of being good, or some closely related property. So in order to formulate a version of the guise of the good precisely, we need to know what the relation R is. Characterizing R, however, requires knowing what sort of thing the property good applies to—R must be a relation that relates that kind of thing to properties (the objects of desire). As with ‘desire,’ attributions of ‘good’ can be grammatically distinguished as falling into several different classes. In fact, with ‘good,’ there are many more possible surface forms of attributions, but they can be naturally classified as falling under three categories: CP     

It is good that you came./It would be good for you to come. It is good for you to eat your vegetables. Those are good vegetables.

In attributive ‘good’ attributions, ‘good’ functions as a predicate modifier, transforming some predicate such as ‘vegetables’ into a complex predicate, ¹⁰ Compare Chierchia (1989).

Desiring under the Proper Guise

127

‘good vegetables.’ Such uses of ‘good’ can be applied to individuals, but strictly speaking ‘good’ does not pick out a property of individuals, but only transforms kind terms like ‘vegetables’ into corresponding properties of individuals, good vegetable. We take it that attributive goodness is the least promising alternative for the content associated with desire. In general, any kind of thing can be said to be a good K, for some value of ‘K’. But we have already argued that desires are oriented toward properties, rather than individuals. More generally, it’s not clear which kind it is such that desires should be associated with the perception that things are good things of that kind. So it’s no surprise that attributive goodness is not usually appealed to by philosophers who wish to make sense of the guise of the good.¹¹ Some other sentences attributing ‘good’ do seem to predicate ‘good’ directly of individuals. For example, consider the sentence, ‘vegetables are good,’ which appears to predicate ‘good’ of vegetables. But this is a mistake; this sentence is systematically related to sentences like ‘vegetables are good for you to eat,’ ‘it is good for you to eat vegetables,’ and ‘vegetables are good to eat.’ Robert Shanklin has shown that as used in all of these sentences, ‘good’ belongs to the syntactic category of what are known as experiencer adjectives.¹² Another example is ‘fun’ for which we have ‘vegetables are fun,’ ‘vegetables are fun to eat,’ ‘vegetables are fun for you to eat,’ and ‘it is fun for you to eat vegetables.’ The basic form of experiencer-adjective sentences is ‘it is good for you to eat vegetables,’ in which ‘good’ expresses a relation between you and the proposition that you eat vegetables. The word ‘it,’ in this sentence, does not refer to anything; it is simply included because English grammatically demands each sentence to have a subject. This same grammatical demand can be satisfied through movement: the word ‘vegetables’ can be moved up to the beginning of the sentence, to get the transparently equivalent, ‘vegetables are good for you to eat.’¹³ The sentences ‘vegetables are good to eat,’ ‘vegetables are good for you,’ and ‘vegetables are good’ are just elisions of this sentence. So though they all appear to predicate ‘good’ of vegetables, this is a grammatical illusion. Really, they all predicate ‘good’ of a relation between someone and some proposition. They belong to the category of good for sentences—what we might call personal good. The final category of ‘good’ attributions is CP attributions. In these sentences, ‘good’ is grammatically predicated of some complementizer phrase, ¹¹ Although there are many possibilities to try, we believe that the main problems that we point out for the other forms of the guise of the good apply here as well. ¹² For arguments and discussion, see Shanklin (2011). ¹³ Radford (2004) is a good introduction to the relevant issues in syntax.

128

Michael Milona and Mark Schroeder

such as ‘that you came’ or ‘for you to come’ or ‘if you come,’ to get sentences like ‘it was good that you came,’ ‘it would be good for you to come,’ or ‘it will be good if you come.’ It is important to distinguish the role of ‘for you’ in these sentences from its role in good for attributions, and this distinction can be brought out by contrasting sentences like ‘it would be better for Jeffrey Dahmer to be punished for him not to be punished’ with sentences like ‘it would be better for Jeffrey Dahmer to not be punished than to be punished,’ both of which intuitively express truths. In the former sentence, ‘for’ is a complementizer, and in the latter, it is a preposition. The latter sentence says that Dahmer and not being punished stand in the good for relation; the former says that the outcome of Dahmer’s being punished is good. CP ‘good’ ascriptions ascribe ‘good’ to propositions—the referents of complementizer clauses.¹⁴ It is this reading of ‘good’ in which traditional consequentialists are interested when they claim that each agent ought always to perform that action such that it will be best for them to perform it. On this view, CP ‘good’ ascriptions ascribe a monadic property, good, to propositions—the property of what we might call being impersonally good. It is important to note that on some views—most famously, that of Foot (1985) and also Thomson (2003)—CP ‘good’ ascriptions do not ascribe a monadic property because they are really just a special case of good for ascriptions, with the person for whom the outcome is good elided or quantified out. But if ‘good’ denotes a simple property at all, CP attributions are the best case for making that out. We have ruled out the idea that a plausible version of the guise of the good thesis will be formulated using attributive goodness. So that leaves the possibility of formulating it with personal good, or with impersonal good. The former is a relation between a person or subject and a proposition, and the latter is a property of a proposition (both, perhaps, relative to a standard, since ‘good’ is gradable). If an agent desires F, then if the impersonal view is true, there must be some related proposition P such that she has a perception as of it being good for it to be the case that P. In order to get P from F, we need to supply some individual, and the only obvious individual is the agent—the subject of the attitude. So the easiest way to make sense of this view is as the view that when an agent desires F, she has an associated perception as of its being good for it to be the case that she is F. G I: If X desires F, then she has an associated perception as of its being impersonally good that she is F.

¹⁴ See especially Finlay (2014) for discussion and defense of this point.

Desiring under the Proper Guise

129

And this is intuitively what impersonal guise of the good theories hold. If Lauren desires to drink water, then she perceives it as being good that she drinks water. Similarly, if an agent desires F, then if the personal view is true, there must be some related proposition or outcome, P, such that she has a perception as of it being good for someone for it to be the case that P. In order to make sense of this view, we need to fill in for whom she perceives it as being good for it to be the case that P, and in order to get P from F, we need to supply some individual who might have F. Again, the most obvious answer to each of these questions is the agent herself—the subject of the attitude. So the easiest way to make sense of this view is as the view that when an agent desires F, she has an associated perception as of its being good for her for it to be the case that she is F—that it is good for her to be F. G P: If X desires F, then she has an associated perception as of its being personally good for X that X is F.

And this is intuitively what personal guise of the good theories hold. If Lauren desires to drink water, then she perceives it as being good for her to drink water. It is possible, of course, that a guise of the good theory could claim that personal good is represented, without it being the personal good of an individual, or without it being the personal good of the desirer. For example, she might instead perceive only that it is good for someone to drink water, or that it is good for someone that Lauren drinks water, or even that it is good for her that someone drinks water. Or she might even perceive that it is good for (for example) Larry that Lauren drinks water. All of these would count as versions of Generic Guise. But we think it is not surprising that such views have never to our knowledge been defended. The quantified version, for example, places limits on how useful desires can be in moral epistemology, and generally seems too weak. Furthermore, the version that appeals to another agent’s personal good makes unclear what determines whose personal good is perceived. Similarly, the guise of the good theorist could claim that impersonal good is represented, without it representing that it is good for the subject to have the property. For example, Lauren could perceive that it is good for it to be the case that someone drinks water. But again, this is intuitively much too weak. None of these weaker views help to make sense, for example, of why desires rationalize the actions that they do. In consequence, we will assume in what follows, at least to get our argument started, that the best formulations of each Guise theory are Guise Impersonal and Guise Personal. These formulations will bring our challenge into particularly sharp focus. Although Generic Guise provides a framework for developing new versions

130

Michael Milona and Mark Schroeder

of the guise of the good, we anticipate that novel theories will face obstacles in the same kinds of cases that constitute our challenge, in Section 6.6, or else answer the challenge at the expense of undermining some of the initial action-theoretic and epistemological motivations for adopting a guise of the good theory in the first place.

6 .5 . T H E A C C U R A C Y C O N S T R A I N T According to Generic Guise, desires are systematically associated with perceptual states with normative contents. We argue that one can get a handle on the question of which normative contents are represented by thinking more closely about the question of how any normative content could come to be represented at all. Philosophers of mind have written extensively about the question of how it is that ordinary perceptual states come to the contents that they do. So theorists who accept Generic Guise owe a similar explanation of how it is that states associated with desires come to represent normative contents.¹⁵ Indeed, there is a greater debt of explanation, because it is much more controversial whether we have perceptions with normative contents than it is whether our ordinary perceptual states have contents about the world. In the philosophy of mind, theories differ sharply over exactly what is required in order for a perceptual state to come to have a particular content. According to some, a covariation condition suffices; other theories are causal or have a causal component, and yet other theories require teleological conditions to be satisfied.¹⁶ Often combinations of various forms of each of these conditions are packaged together. Against the background of such disagreement, it is still possible to formulate a constraint on content determination that is broadly acceptable to theorists of very different persuasions. Such a constraint will pose a necessary condition on a perceptual state coming to have a particular content, and can therefore be used to tell against possible normative contents that cannot plausibly be construed as satisfying that condition. The intuitive idea behind the constraint that we will defend is that the presence (or absence) of a systematic covariation relation between an experience and what it is hypothesized to be about at least provides evidence for (or against) the hypothesis, whether or not covariation is ultimately what

¹⁵ Schroeder (2008) and Schafer (2013) note that this question has received very little attention. ¹⁶ Compare, for example, Millikan (1989), Dretske (1995), Neander (2013).

Desiring under the Proper Guise

131

explains content.¹⁷ To give an illustration, consider the hypothesis that squarish perceptual experiences (i.e., perceptual experiences with the phenomenology as of squares) are about triangularity. This, of course, is an implausible view. On the assumption of this hypothesis, the squarish perceptual experiences of human agents with normal, well-functioning psychologies would be systematically in error, since as a matter of fact such agents tend to have such experiences in the presence of square objects, not triangular ones. That there would be such systematic inaccuracy tells against the initial hypothesis, and it tells against it because they would be inaccurate in this systematic way. Of course, sometimes there are systematic sources of error in our perceptual experiences. Perceptual illusions, including famous cases such as the Müller-Lyer illusion, occur under systematic conditions. It is too strong, then, to say that any systematic source of error tells against the hypothesis that the perceptual state has a content that makes it inaccurate. But we can get some insight into when systematic error is possible by considering such ordinary cases of perceptual error or illusion. It turns out that we know quite a lot about how perceptual illusion works in ordinary cases. Take the Müller-Lyer illusion, for example (Figure 6.1). Most humans who have been raised in an ordinary built environment perceive one of two lines as longer than another, despite the fact that they are exactly the same length. This can be explained by the fact that in a normal built environment, the line segment that visually appears to be longer actually is longer. Even though the images projected on the retina are identical in size, the line flanked by outward facing arrowheads corresponds in normal settings to something farther away; and if two objects

Figure 6.1

¹⁷ We take this constraint (which will be refined below) to be amenable to the explanations of perceptual content offered by Millikan (1989), Dretske (1995), Burge (2010), and Neander (2013), among others. Neander, for instance, has a causal-teleological theory of content, but nevertheless she accepts that covariation is evidential. For example, see Neander (2013: 29).

132

Michael Milona and Mark Schroeder

project the same size retinal image, but one is perceived as farther away, we perceive the one that is apparently at a greater distance as larger.¹⁸ Other illusions cut across the senses in interesting ways. To take a familiar example, ventriloquists make a living creating auditory illusions that are triggered by visual cues. Specifically, a ventriloquist moves a puppet in a way that corresponds to how we would expect someone to move who is speaking in the way that the ventriloquist is speaking, but the ventriloquist herself does not move her mouth at all. This helps to create the auditory illusion that the sounds are coming from a different direction than they in fact are.¹⁹ In both cases, then, the dispositions that lead to the illusory perceptions are such as to produce accurate perceptions in typical environments. Similar explanations go for other visual and auditory illusions. In fact, it should not be a surprise that this is so, because one of the standard methods of research into perceptual cognition presupposes that it is so. According to this methodology, one of the principle sources of insight into how perceptual cognition works is to pay special attention to perceptual illusions. We can learn a lot about how a perceptual process works by assuming that the process in an illusion is an extension of a process that is accurate in its core cases.²⁰ Here, then, is the constraint that we are looking for, formulated in a general way: A C: A hypothesis about the content of a perception is plausible only if perceptions of that sort arising out of normal, well-functioning psychologies are accurate, or else can plausibly be understood on the model of an illusion.

In the next section, we will use this Accuracy Constraint to argue against Guise Impersonal and Guise Personal. This is because the desires of agents with normal, well-functioning psychologies are systematically associated with states that fail in typical contexts to covary with what is impersonally good and with what is good for the desirer.

6 .6 . P R O B LE M C AS E S We can now precisely formulate the problem for Guise Impersonal and Guise Personal—a problem to which we will later argue that the guise of ¹⁸ See Gregory (1997). Rock (1995) offers an alternative explanation. The key, however, is the shared assumption that the dispositions that lead to the illusions are such as to be accurate in more typical cases. ¹⁹ Howard and Templeton (1966). ²⁰ Gregory (1997) is a good example of this approach.

Desiring under the Proper Guise

133

reasons offers an elegant and natural solution. The easiest way to begin to see the trouble is by thinking about several of the ‘flavors’ that normal, wellfunctioning desires come in: egoistic, impersonal, and other-regarding but not purely impersonal. Consider Larry. Larry wants to help the homeless in his neighborhood. He doesn’t want to do this merely because doing so is a way of, say, impressing his peers or creating pleasure for himself. Of course, he’s not completely impersonal in his desires. He really does want to impress his peers, get pleasure for himself, and so on; and he doesn’t have those desires merely because fulfilling them would be instrumental to some impersonal desire. Others of Larry’s desires are not egoistic but also not completely impersonal; a prime example is his desire for his child to win a prestigious fellowship. Most of us are like Larry. We have a mixture of genuinely egoistic, purely impersonal, and other-regarding but not purely impersonal desires. The difficulty for the guise of the good is that Guise Personal runs into trouble with purely impersonal desires, Guise Impersonal struggles with egoistic desires, and both views struggle with other-regarding but not purely impersonal desires. It’s easy to see how purely impersonal desires make trouble for Guise Personal. According to this theory, when an agent desires F, she has an associated perception as of it being good for her that she is F. So if Kunal sees that a friend of a friend is up for a promotion, and desires that she gets it, then, according to Guise Personal, he has an associated perception as of its being good for him that she gets a promotion. These kinds of cases create trouble in light of the Accuracy Constraint. To begin, unless there is some coincidence, in general it is not good for Kunal that friends of his friends get promotions. So the vast majority of such cases turn out to be inaccurate, on this view. But such impersonal, or more specifically, altruistic desires seem to arise out of normal, well-functioning psychologies.²¹ There is nothing defective about the desiderative system of a person who wants others to get promotions, for example. And, finally, it is not attractive to think of such desires as illusory. If under normal evolutionary conditions, our ancestors would have seen someone up for a promotion only under conditions under which it would have been accurate to conclude that it was better for the viewer for that person to get a promotion, then that would be the right kind of thing to explain why it is that we experience an illusion when we see someone up for a promotion. But the assumptions required by this explanation seem to be straightforwardly false. Under normal conditions—and even under normal evolutionary conditions—not everything that is good for

²¹ Altruistic desires are a paradigm case of impersonal desires, but not all impersonal desires are altruistic.

134

Michael Milona and Mark Schroeder

the people (or even friends of friends) one encounters is good for oneself. And similar problems seem to arise for similar styles of explanation. Guise Personal, then, falls afoul of the Accuracy Constraint. It’s equally straightforward why egoistic desires make trouble for Guise Impersonal. According to this theory, when an agent desires F, she has an associated perception as of its being good for it to be the case that she is F. We can begin to see trouble by considering egoistic desires in competitive contexts: R: An Olympic runner, Shelly, is about to compete for the 400 meter gold. She desires to win the race and is averse to any of the other runners winning. The other runners, of course, also want to win the race and their opponents not to win.²²

It makes no difference (or no significant difference) to the total goodness of different possible outcomes who wins the race. So if proponents of Guise Impersonal want it to turn out that Shelly’s desires are associated with accurate perceptions, they will be forced to say that the other runners’ desires are associated with inaccurate perceptions. This is for the simple reason that, according to the theory, Shelly represents it as good that she wins the race, and the other runners represent it as not good that Shelly wins. But it cannot be both impersonally good and not good that Shelly wins the race. Cases like Race are quite typical and could be multiplied. Furthermore, competitive, egoistic desires seem to arise out of normal, well-functioning psychologies. There is nothing defective about the desiderative system of a runner who wants to win a race and is averse to other runners winning, for example. It is also not plausible that competitive desires are illusory. Such desires do not respond to considerations that tend to covary closely with the impersonal good. A competitor’s desires will normally be highly and systematically sensitive to thoughts of her glory, her status, her future endorsements and so on; and these egoistic considerations just aren’t relevant to the impersonal good. At this point, a proponent of the guise of the good may make the following maneuver. Instead of adopting either Guise Impersonal or Guise Personal, she may insist that somehow both are correct. The easiest way to make sense of this idea is to suppose that there are actually two types of desire, egoistic desires which are associated with perceptions as of personal goodness and impersonal desires which are associated with perceptions as of impersonal goodness. The problem with this view is that it fares no better with other-regarding but not purely impersonal desires. In fact, the vast majority of our ²² This example is adapted from Oddie (2018).

Desiring under the Proper Guise

135

impersonal desires are not purely impersonal at all but are rather what we call competitively altruistic. Consider the following, more harrowing kind of competition: H: Jasmine and Jacqueline are members of the same tribal community, but they’re not close family members. During one especially harsh winter, food is limited and survival is in question. On occasions when there is not enough food for everyone, Jasmine desires that her family get the food. But on those occasions, Jacqueline desires that her family get the food.

In Hungry, as in Race, it makes no difference to the impersonal good who ‘wins’ the competition, and so the guise of the good theorist, even on the disunified version of the theory, cannot have it that both Jasmine’s and Jacqueline’s desires are accurate. But at the same time, both of their desiderative systems seem to be functioning well. And, moreover, it is hard to see why we should expect illusions. Jacqueline’s desire is sensitive to the fact that a certain family is hers, and similarly for Jasmine. But these considerations are irrelevant to the impersonal good, and yet their desires are highly and systematically sensitive to such considerations. So once we have the Accuracy Constraint on board, it becomes very implausible not only that desires in general are systematically associated with perceptions of the impersonal good but even that altruistic desires are. The trouble is, we might say, with the desirer-centeredness of desire; plenty of altruistic desires are centered on the desirer in a way that is difficult even for the disunified theory of desire to accommodate. So far, we have not ruled out every possible version of the guise of the good. Instead of distinguishing between two kinds of desire, which represent different good properties, we might instead distinguish between three. Or we might relax the assumption that desire must be associated with a single normative property or relation, and instead assume that the way that desire represents is context-sensitive—there is some function which tells us, in any given situation s, which form of the guise of the normative is correct for that context. There are too many possible variations on this theme to refute each one in detail, but we believe that every such variation fails either by running into the same sorts of trouble in variants on cases of other-regarding but not purely impersonal desires, by being too weak, or by failing to play the same role in rationalizing action that we expect of a form of the guise of the normative. For example, it is worth paying attention to just how hard it is to characterize a form of the guise of the good that yields plausible results about cases of competitive altruistic desires, let alone fits into a general account that makes sense of all cases of desire. The problem is that

136

Michael Milona and Mark Schroeder

competitive altruistic desires don’t track the good of any individual or group—rather, at best they track a weighted balance of the goods of individuals. For example, Jasmine and Jacqueline may desire that each other’s families get food more than they desire that some other family outside of their tribe gets food. So their desires track the individual good of the same group of individuals, but weight those goods differently. So if you want a version of the guise of the good to make Jasmine’s and Jacqueline’s competitive altruistic desires to come out as associated with approximately true representations of some facts about what is good for some people, these facts cannot just be facts about what is good for some group of people; they must be complex facts about the weighted balance of the good for some group of people—and the weighting may need to turn out to be different for different people’s desires, and for the desires of the same person at different times! It is very hard to see how to give a non-ad hoc account of this kind, and it is even more puzzling how it could give us anything helpful for a perceptual moral epistemology. If our basic justification for normative judgments is for a complicated weighing of facts about what is good for individuals, and the weighing itself is different in different situations in ways that are not transparent to us, then it is extremely hard to see how this is going to help us to support the apparently basic normative judgments that we actually go on to make.

6 . 7 . C O N S T R AI N T S IN M O R A L P H I L O S O PH Y We have just been arguing that the guise of the good thesis has trouble making sense of competitive desires. This is because good and better are properly predicated of propositions, so getting one agent’s competitive desires to come out as accurate in how they represent the good forces other agents’ desires to come out as inaccurate. But the problem is not solved, either, by the form of the guise of the good theory which appeals to personal value, because psychological egoism is false. We will now argue that this very same structural problem is familiar from consideration of the paradox of deontology, which is a philosophical puzzle about the possibility of agent-centered constraints in moral philosophy. And there, we believe, the puzzles have a familiar and straightforward answer: what makes actions right is the reasons that support them, rather than the goodness of the state of affairs that results if they are performed. A similar moral shows that what desires represent is what we have reasons to do. An agent-centered constraint is an action that an agent ought not to perform, even if her performing it would result in a better outcome. According to common-sense, most forbidden actions are like this. It is not

Desiring under the Proper Guise

137

okay to kill an innocent person just to prevent two deaths, and that is not just because killings are worse than deaths—because under ordinary circumstances it would be wrong to kill an innocent person even to prevent two killings of innocent persons. It is impersonally better for there to be only one innocent person who is killed rather than two, but it is not, intuitively, okay for you to kill one innocent person who would not otherwise be killed, just to prevent the worse outcome in which two innocent people are killed. The puzzle of how this could be so is sometimes called the ‘paradox of deontology.’ The reasoning behind the puzzle is simple: any explanation of what is wrong with killing the innocent person must appeal to something that is bad about your killing that innocent person. But by parity of reasoning, the same thing must be bad about the killings of innocent people that you prevent. And so unless you are special, we must conclude that there is even more of whatever tells against your killing telling in favor of it. But you are not special. And even if you were, it would not be okay for other people to kill, just to prevent you from killing. So it is hard, on this view, to understand how it could be wrong to kill, in this situation. Hence the idea that deontological intuitions are paradoxical. Cases involving constraints are analogous to cases of competitive desires. There is no single account of what is impersonally good that makes each of a set of interlocking competitive desires accurate. Similarly, there is no single account of what is impersonally good that explains the possible agentcentered constraints facing each agent. And just as there is nothing special about you, such that we should explain the constraints facing you but not those facing other people, similarly there is nothing special about Shelly, such that we should explain why her perceptions about impersonal goodness are accurate, but not those of her competitors. So the reasons why we can’t explain the wrongness of actions in terms of impersonal goodness, if we want to make sense of constraints, are closely analogous to the reasons why we can’t make competitive desires turn out to be accurate as representations of impersonal good. In the case of competitive desires, one apparent alternative was to suppose that they are instead associated with perceptions as of what is good for the agent. But just as it is implausible that the objects of altruistic desires systematically happen to be good for the agents who desire them, it is implausible that it is always bad for an agent for it to be the case that she kills an innocent person. The fact that these are wrong things to do does not depend on whether it is plausible that doing them detracts from the agent’s well-being.²³ ²³ One strategy for resolving the paradox of deontology appeals to agent-relative value, which is supposed to be distinct from impersonal and personal goodness. Similarly, one might propose that we can salvage the guise of the good by taking desires to involve

138

Michael Milona and Mark Schroeder 6 .8 . R E A S O N S S U P P O R T P R O P E R T I E S

What we’ve just argued is that the paradox of deontology derives from the assumption that what explains why you should not do something is some fact about what is good or bad. Because such facts are always facts about goodness or badness as ascribed to propositions or outcomes, such explanations of why you should not steal carry over to commitments that equally tell against your not preventing the killings of others. We conclude that the wrongness of actions is not explained by facts about what is good or bad. It is explained, in contrast, by reasons against performing those actions. In general, a reason is always a reason for someone to do something. The reason relation holds between considerations or facts (which are what we identify as ‘the reason’), agents (who are the people for whom it is a reason), and actions (as expressed by phrases like ‘to not kill’). But actions in this sense are just a special case of properties. The action of riding a bike, for example, is the property that is shared by everyone who is riding a bike. And as properties, actions are the right sort of thing to be denoted by non-finite clauses like ‘to ride a bike,’ and hence to be reported by reason ascriptions like ‘the fact that it is good exercise is a reason for you to ride a bike.’ Since reasons tell in favor of actions, rather than outcomes, and actions are properties, rather than propositions, we get a very simple explanation of why the reasoning behind the paradox of deontology fails. The explanation of why you ought not to kill appeals to a reason for you not to kill. This reason is a reason for you not to kill, but it does not support the proposition that you do not kill; it just supports the action of not killing. So parity of reasoning, based on the observation that you are not special, leads us to conclude only that there are also reasons for other people in support of the very same action—not killing. Their reasons don’t concern your killing any more than your reasons concern theirs. In both cases, your reasons are just reasons not to kill. So the idea that reasons count in favor of properties is not only wellmotivated in its own right, but it offers a clean and elegant explanation of where the reasoning behind the paradox of deontology gets into trouble.

perceptions as of agent-relative value. But we are not optimistic. For one, despite the long history of the guise of the good theory, no philosophers that we are aware of have ever developed a version of it in terms of agent-relative value. Furthermore, there is a serious question of what agent-relative value is even supposed to be, and the most promising ways of making sense of it analyze it in terms of reasons (e.g., what an agent has reason to desire). See Portmore (2011: 125). The resulting theory, then, would best be conceived as a version of the guise of reasons.

Desiring under the Proper Guise

139

Organizing the explanation of wrongness around propositions, rather than around properties, distorts the phenomena. This motivates the thought that the thesis of the guise of reasons might offer similarly better prospects for satisfying the Accuracy Constraint. We will now argue that this is right.

6. 9 . D E S I R E A N D T H E G UI S E O F R E A S O N S In order to see how the guise of reasons does better in light of the Accuracy Constraint, we first need to get clear about what the view is. According to any version of the guise of reasons, desires are systematically associated with perceptions as of reasons. On the version that we favor, these reasons will be considerations that favor actions which are means to the satisfaction of the desire. For example, if you want to drink a coke and there are cokes in the fridge, then you will have a perception as of a reason to go to the fridge. Of course, if you don’t believe that there are cokes in the fridge, then it won’t seem to you that you have a reason to go to the fridge. So we need to be careful in formulating the theory. It says when you desire to drink a coke, what appears to you to be a reason will depend on your beliefs; if given your beliefs, some action would help bring it about that you drink a coke, then that will appear to you to be a reason. Put generally, G R: If X desires F she perceives herself as having reasons to do actions which, given her beliefs, would help to bring it about that she is F.

Now let’s work through how the guise of reasons helps with competitive desires, focusing on the case which made trouble for Guise Impersonal, Race. In this scenario, each of the competitors wants to win the race. And so, according to Guise Reasons, each of the runners will see herself as having reasons do actions which, given her beliefs, would help to bring it about that she wins the race. For example, if the race is two weeks away, each runner will see herself as having reasons to eat well, train hard, and avoid distractions, among other things. And these perceptions are entirely accurate; the runners really do have reasons to take such means to help bring about the object of their desire. Guise Reasons also does well when our competitive desires are altruistic rather than egoistic. In Hungry, for example, there is not enough food for all of the children and Jacqueline wants her children to get the food. She will thus see herself as having reasons to take steps which ensure that her children, rather than the other children, get the food. And to allow that her desires involve accurate normative perceptions, we are not forced to say that there is anything objectionably special about Jacqueline’s children,

140

Michael Milona and Mark Schroeder

for other parents in the tribe have desiderative profiles analogous to Jacqueline’s, and accurately perceive themselves as having reasons to bring it about that their children get the food. Guise Reasons also does well with non-competitive altruistic desires. To take a simple case, suppose that Alejandro sees someone in pain and desires for her pain to stop. According to the theory, Alejandro will see himself as having reasons to bring about the cessation of her pain (e.g., a reason to call an ambulance). And, of course, Alejandro does have such reasons. Other cases of normal, well-functioning impersonal desires will receive similar treatment. Guise Reasons, then, apparently does quite well in light of the Accuracy Constraint, resolving all the problem cases that arose for the two major formulations of the guise of the good.²⁴ In Section 6.6, we argued that in order to satisfy the accuracy constraint across the full range of possible desires ranging between the purely altruistic and the purely impersonal, desires would have to represent a kind of weighted sum of personal goodness. Guise Reasons again fares comparably quite well. Whatever its explanation, we do have reasons to act in ways that are partial. So whereas any viable version of the guise of the good will require the contents of normative perceptions to be quite complex, it follows from Guise Reasons that they can be very simple. That Guise Reasons has these advantages shouldn’t be surprising, given what we have already observed about how explaining obligations in terms of reasons offers an elegant solution to the paradox of deontology. To reiterate, what we’re obligated to do (ought to do, should do, etc.) is always a property (an action), and reasons tell in favor of properties. By contrast, goodness and badness are ultimately ascribed to propositions, and so trying to explain obligations in terms of goodness and badness is much less natural. Matters are structurally similar with desire, since desires, like obligations, are property-oriented. In many cases, desires will have as their object properties which are also actions. For example, we might desire to win a race, ride a bike, or drink a smoothie. But even when a desire is oriented to a property which is not an action—say, a desire to be such that there is world peace— the desire still motivates us (all else being equal) to behave in ways which, given our beliefs, would satisfy the desire. Guise Reasons adds that her desire involves perceptions as of reasons to do these actions; and so the normative perception is neatly integrated with the motivational character of desire. ²⁴ Sometimes an agent may wish for something yet not believe that there are any possible means she could take to fulfil her wish. But remember that, according to Guise Reasons, an agent who lacks such a belief will not perceive any reasons. One may also doubt that certain such unmotivating desires are really desires, after all, but are instead mere wishes (sharing certain features with desire but requiring a different analysis).

Desiring under the Proper Guise

141

Finally, one of the concerns that has interested us throughout is whether the thesis of the guise of the normative will have any sort of payoff in moral epistemology. One of the important points that we made in response to the increasingly complicated sorts of variants on the guise of the good that one might propose in response to the Accuracy Constraint was that such views make it increasingly puzzling how what they yield could be the basis for a plausible moral epistemology—particularly if the content represented by desires varies between situations in ways that are not transparent or are difficult to unpack. In contrast, if the contents that are represented by desires just concern the agent’s own reasons, then this fits smoothly into a plausible perceptual moral epistemology, as it is highly plausible that an agent’s judgments about what it is okay or not okay to do in different situations are sensitive to her perception of her reasons in those situations. So the failures of the Accuracy Constraint for the guise of the good simply look worse, if we are trying to get an epistemological payoff.

6. 10 . CONC L U S ION The guise of the normative is an important theory that promises exciting results in action-theory and moral epistemology. But it is underappreciated how many different forms the theory can take, and it is an important task to determine which version is the most promising. In this paper, we have argued that the guise of reasons approach to the guise of the normative is more promising than the guise of the good approach. We first argued against two of the most attractive versions of the guise of the good—Guise Impersonal and Guise Personal. In particular, we argued that both theories run afoul of the Accuracy Constraint, which is a useful tool for determining whether a state is really associated with certain contents. By contrast, Guise Reasons does well in light of the Accuracy Constraint. A brief detour through constraints in moral philosophy allowed us to unpack the structural issue plaguing familiar guise of the good theories, and to explain why it is not surprising that Guise Reasons does better. What makes us so confident in our conclusion, then, is that it proceeds so immediately from what we have argued is a deep diagnosis of the structural challenge facing the guise of the good.²⁵

²⁵ Special thanks to Stephen Finlay, Janet Levin, Ralph Wedgwood, the attendees of the 12th Annual Wisconsin Metaethics Workshop, the DEthiX workshop at UC Davis in April 2017, and the Schroeder advisee workshop for helpful comments and suggestions. Thanks also to two helpful anonymous referees.

142

Michael Milona and Mark Schroeder

References Anscombe, Elizabeth (1963/2000). Intention, 2nd edn (Cambridge, MA: Harvard University Press). Bedke, Matt (2009). “The Iffiest Oughts: A Guise of Reasons Account of End-Given Conditionals,” Ethics 119 (4): 672–98. Brewer, Talbot (2009). The Retrieval of Ethics (New York: Oxford University Press). Burge, Tyler (2010). The Origins of Objectivity (New York: Oxford University Press). Chierchia, Gennaro (1989). “Anaphora and Attitudes ‘De Se’,” in R. Bartsch, J. F. A. K. van Benthem, and P. van Emde Boas (eds.), Semantics and Contextual Expressions (Dordrecht: Reidel), 1–31. Davidson, Donald (1969/2000). “How Is Weakness of the Will Possible?” in Essays on Actions and Events (New York: Oxford University Press), 21–42. Dretske, Fred (1995). Naturalizing the Mind (New York: Cambridge University Press). Finlay, Stephen (2014). A Confusion of Tongues (New York: Oxford University Press). Foot, Philippa (1985). “Utilitarianism and the Virtues,” Mind 94 (374): 196–209. Greene, Joshua (2008). “The Secret Joke of Kant’s Soul,” in W. Sinnott-Armstrong (ed.), Moral Psychology, vol. 3: The Neuroscience of Morality: Emotion, Disease, and Development (Cambridge, MA: MIT Press), 35–79. Gregory, Richard (1997). Eye and Brain, 5th edn (Princeton, NJ: Princeton University Press). Gregory, Alex (2013). “The Guise of Reasons,” American Philosophical Quarterly 51 (1): 63–72. Haidt, Jonathan (2001). “The Emotional Dog and Its Rational Tail: A Social Intuitionist Approach to Moral Judgment,” Psychological Review 108 (4): 814–34. Howard, Ian and William Templeton (1966). Human Spatial Orientation (New York: John Wiley and Sons). Lewis, David (1979). “Attitudes De Dicto and De Se,” Philosophical Review 88 (4): 513–43. Mach, Ernst (1897). Analysis of Sensation (Chicago, IL: Open Court Press). Millikan, Ruth (1989). “Biosemantics,” Journal of Philosophy 86 (6): 281–7. Neander, Karen (2013). “Toward an Informational Teleosemantics,” in D. Ryder, J. Kingsbury, and K. Williford (eds.), Millikan and Her Critics (Malden, MA: Blackwell), 21–40. Oddie, Graham (2005). Value, Reality, and Desire (New York: Oxford University Press). Oddie, Graham (2018). “Value Perception, Properties and the Primary Bearers of Value,” in A. Bergqvist and R. Cowan (eds.), Evaluative Perception: Aesthetic, Ethical, and Normative (Oxford: Oxford University Press), 239–57. Perry, John (1979). “The Problem of the Essential Indexical,” Noûs 13 (1): 3–21. Portmore, Douglas (2011). “The Teleological Conception of Practical Reasons,” Mind 120 (477): 117–53.

Desiring under the Proper Guise

143

Quinn, Warren (1993). Morality and Action (Cambridge, MA: Harvard University Press). Radford, Andrew (2004). Minimalist Syntax (Cambridge: Cambridge University Press). Rock, Irvin (1995). Perception (New York: Scientific American Library). Scanlon, T. M. (1998). What We Owe to Each Other (Cambridge, MA: Harvard University Press). Schafer, K. (2013). “Perception and the Rational Force of Desire,” Journal of Philosophy 110 (5): 258–81. Schroeder, Mark (2007). Slaves of the Passions (Oxford: Oxford University Press). Schroeder, Mark (2008). “How Does the Good Appear to Us?” Social Theory and Practice 34 (1): 119–30. Schroeder, Mark (2011). “Ought, Agents, and Actions,” Philosophical Review 120 (1): 1–41. Shanklin, Robert (2011). “On Good and ‘Good,’ ” PhD Dissertation, University of Southern California: http://digitallibrary.usc.edu/cdm/ref/collection/ p15799coll127/id/632791. Sinhababu, Neil (2015). “The Advantages of Propositionalism,” Pacific Philosophical Quarterly 96 (2): 165–80. Tenenbaum, Sergio (2007). Appearances of the Good: An Essay on the Nature of Practical Reason (New York: Cambridge University Press). Thomson, Judith (2003). Goodness and Advice (Princeton, NJ: Princeton University Press).

7 Infinitism about Cross-Domain Conflict David Killoren 7 . 1. I N T R O D U C T I O N According to the overriding view, (i) there are multiple normative domains, including the domains of morality and prudence, and perhaps also including an epistemic domain, an esthetic domain, a domain of etiquette, and so on. (ii) Normative domains can come into conflict with one another. Such conflicts occur when an obligation from one domain is incompatible with an obligation from another domain. For example, an agent may be morally obligated to ϕ₁ and prudentially obligated to ϕ₂ in a case where the agent cannot do both ϕ₁ and ϕ₂. And finally, (iii) in some such conflicts, an obligation belonging to one normative domain overrides an obligation belonging to another normative domain. For example, a moral obligation may override a prudential obligation or vice versa. In such cases, we may say, the first domain trumps the second domain. My purpose in this paper isn’t to defend the overriding view, but is instead to tease out some of its implications. So I won’t say much in defense of the overriding view. I’ll avoid discussion of the merits of (i) entirely. But I will say a bit about (ii) and (iii), before I explain the implications of the overriding view that will be my main focus in this paper. There are theories that undermine (ii). For example, ethical egoism twists morality into something resembling prudence; therefore ethical egoists will not see any conflict between morality and prudence and thus will reject (ii) (Copp 1997: 89). Similarly, some moral theories, including certain kinds of Aristotelian eudaimonism, hold that it is in one’s own self-interest always and everywhere to be morally excellent. Such theories twist prudence into something resembling morality, and so defenders of such theories will naturally resist (ii). And, as Parfit (2011: 142) observes, certain religious ideas rule out (ii) by positing a divine force that rewards right action and thus ensures that it’s always prudent to do what’s moral.

Infinitism about Cross-Domain Conflict

145

But, setting religious ideas aside for the present purpose, I think all of the theories that lead to the denial of (ii) are implausible, precisely because they erase a distinction and a conflict that is an obvious and pervasive part of everyday normative experience. Morality seems very often to require us to behave imprudently. For example, we seem often to be morally obliged to make ourselves worse off (or to forego benefits for ourselves) in order to help (or to avoid doing harm to) those who are disadvantaged, oppressed, or helpless. The conflict between morality and prudence is a version of what Sidgwick (1874) calls “the profoundest problem” in moral philosophy. I don’t intend to solve the problem posed by such conflict in this paper, but I will offer a sketch of what I believe a solution ought to look like. My assumption, first, is that we would have a solution to the conflict between morality and prudence—or between the epistemic domain and the esthetic domain, or between any two normative domains one wants to consider—if we could show that (iii) is true. That is, my assumption is that a conflict between (say) morality and prudence in a given case would be resolved if we could show that morality wins out over prudence, or that prudence wins out over morality.¹ So my assumption is that if the overriding view, including (iii), were true, then we would have a solution to the problem posed by conflicting normative domains. My aim in this paper is to explore what follows from the overriding view. The thesis of this paper is: If we endorse the overriding view, then we ought to endorse the view that I call infinitism about crossdomain conflict. Infinitism about cross-domain conflict is a direct analog of epistemic infinitism, a view that has been given its seminal defense by Peter Klein (1998). Epistemic infinitism is an alternative to epistemic coherentism and to epistemic foundationalism. Roughly, according to epistemic infinitism, a given belief that p is epistemically justified iff there is an infinitely long chain of reasons q, r, s, . . . , such that q supports p, r supports q, s supports r, and

¹ Sidgwick seems to have thought that a satisfactory solution to the present problem requires that morality wins and prudence loses; see Crisp (ms.). But many philosophers, e.g., Susan Wolf (2015), are sanguine about the possibility that prudence sometimes wins out over morality. In any case, this issue need not concern us here; an unsatisfying solution to a given problem is still a solution to that problem. Cf. McPherson (2018). A related point is that (iii) isn’t the only possible solution to the present problem. For instance, David Copp (ms.) argues for a kind of pluralism according to which, in the event of conflicts between morality and prudence or between any other pair of domains, there is simply no winner and no loser; see also Tiffany (2007) and Enoch (2019). If such a position can be given a good defense—a matter I won’t address in this paper—then it would count as a solution of sorts to the problem here.

146

David Killoren

so on, indefinitely. Infinitism about cross-domain conflict can be defined analogously, as follows: Infinitism about cross-domain conflict There is at least one case in which an agent has an obligation OA that belongs to domain DA, and an obligation OB that belongs to domain DB (where DA and DB are different domains), and OA overrides OB. For every such case, there is an infinitely long chain of domains D₁, D₂, D3, . . . , such that D₁ contains an obligation to act in line with the obligations of DA, and D₂ contains an obligation to act in line with the obligations of D₁, and D₃ contains an obligation to act in line with the obligations of D₂, and so on, indefinitely.

There is a deep literature on epistemic infinitism, but no one has given a careful argument for the view that I’m calling infinitism about cross-domain conflict. When this sort of view is mentioned, it is usually treated with casual disregard; e.g., Dale Dorsey (2016: 23) considers a view in the neighborhood and simply dismisses it as “a wildly unattractive picture of the normative enterprise.” My argument from the overriding view for infinitism about cross-domain conflict is a regress argument and as such it formally resembles many other regress arguments that have appeared in philosophy. The argument has an important precedent in Copp (1997). It is fruitful to compare similarities and differences between my argument and Copp’s argument but unfortunately I do not have the space to provide that comparison here. My argument is as follows. The argument for infinitism about cross-domain conflict (1) OA overrides OB iff OA belongs to DA and OB belongs to DB and DA trumps DB. (This is just a stipulative definition of trumping.) (2) If DA trumps DB, this is because there is a novel higher-order domain, D₁, that explains why DA trumps DB. (3) D₁ explains why DA trumps DB only if D₁ has the requisite authority. (4) D₁ has the requisite authority only if D₁ trumps at least some of its same-order competitors (call them D₁*, D₁**, D₁***, . . . ). (5) If D₁ trumps any of its same-order competitors, this is because there is a novel higher-order domain, D₂, that explains why D₁ trumps those same-order competitors.

Step (5) marks the start of a regress. From this point, the argument proceeds to a step analogous with (3), and then to a step analogous to (4), and then to a step analogous with (5), and so on, continuing in this way

Infinitism about Cross-Domain Conflict

147

indefinitely. This, as I’ll explain in due course, leads to infinitism about cross-domain conflict. Section 7.2 will develop the argument. Section 7.2.1 will lay groundwork for the argument; Section 7.2.2 will argue for each of the premises. Section 7.3 will address objections. 7 . 2. T H E A R G U M E N T F O R I N F I N I T I S M ABOUT CROSS-DOMAIN CONFLICT In this section I will assume the overriding view and argue for infinitism about cross-domain conflict. In Section 7.2.1, I will argue for position concerning what I’ll call novel higher-order domains. Specifically, I will argue that novel higher-order domains exist and that standpoints associated with them play a crucial epistemic role. Then, in Section 7.2.2, I will complete the argument for infinitism about cross-domain conflict. First I need to make some preliminary stipulations and definitions. I’ll first explain the notion of a normative standpoint. Suppose Monica has a moral reason to donate a sum to charity. In virtue of this reason, suppose, Monica is (all-things-considered) morally obligated to donate to charity. Further, Monica has a prudential reason to invest the sum in her retirement plan. In virtue of this reason, Monica is (all-things-considered) prudentially obligated to invest. Monica can’t do both. Finally, suppose that it intuitively appears that Monica’s moral obligation to donate to charity overrides her prudential obligation to invest. Then, in this case, it appears that cross-domain overriding occurs: Monica’s moral obligation overrides her prudential obligation; morality trumps prudence. From the moral standpoint, it seems Monica ought to donate, whereas from the prudential standpoint, it seems Monica ought to invest. Observing this, we can reasonably suppose that a normative standpoint is, broadly, a way of seeing or thinking about situations in which an agent must decide among options. This way of seeing or thinking might involve the (conscious or otherwise) application of general principles (e.g., general moral principles, or general principles of prudence) or it might involve something like intuitions about cases, or apprehension of substantive conceptual truths like “moral fixed points” (Cuneo and Shafer-Landau 2014). For my purpose here I want to be as neutral as possible about how precisely standpoints work psychologically. What is important, for my purpose, is simply that we seem typically capable, when presented with any given situation in which multiple options are available, of viewing that situation from several different standpoints— such as the moral standpoint, the prudential standpoint, and so on. But what

148

David Killoren

makes it the case that this or that standpoint counts as the (or a) moral (or prudential, or aesthetic, etc.) standpoint? I think we can answer that question in terms of domains: A given standpoint is the moral (or e.g., prudential) standpoint if and because obligations in the moral (or e.g., prudential) domain are apprehensible from it.² In other words, I’m suggesting that a given standpoint counts as the X-standpoint because obligations of the X-domain are apprehensible from that standpoint.³ Let’s say that a zeroth-order standpoint is a standpoint of the sort that most of us are already familiar with: a standpoint from which we can consider the question of what an agent is to do in a given situation. A first-order standpoint, then, is a standpoint from which we can consider the question of which zeroth-order standpoint the agent is to adopt and act in accordance with—or, equivalently, the question of which zeroth-order domain ought to guide the agent in deciding what to do. A second-order standpoint is a standpoint from which we can consider the question of which first-order standpoint the agent ought to adopt. A third-order standpoint is a standpoint from which we can consider the question of which second-order standpoint the agent ought to adopt. And so on. Let’s say that a higher-order standpoint is any standpoint that lies above the level of zerothorder standpoints. So now we have an idea of what a higher-order standpoint is. A novel first-order standpoint is a standpoint that is distinct from any of the zerothorder standpoints under consideration in a given case. Not all higher-order standpoints need be novel in this sense. The moral standpoint, for example, seems to double as both a zeroth-order standpoint and as a first-order standpoint: The moral standpoint is a standpoint from which we can consider questions about what an agent is to do and can consider questions

² Suppose that, in a given case, one’s moral obligation is simply to do what one is prudentially obligated to do. Then you might think that in order to apprehend one’s moral obligation from the moral standpoint, one must be able to apprehend one’s prudential obligation from within that standpoint. But then the moral standpoint would (by definition) be the prudential standpoint. (Thanks to an anonymous referee for pushing me on a version of this point.) In order to avoid this result, I can say that in this sort of case, in order to apprehend one’s moral obligation from the moral standpoint, one must first already have apprehended what one’s prudential obligation is; and in order to know this, one must have previously occupied the prudential standpoint. ³ This defines what it is for a standpoint to be the X-standpoint; it does not define the notion of a standpoint itself. I take standpoints to be a familiar feature of normative experience and so I don’t think it is necessary to define the notion of a standpoint, though it is important that standpoints characteristically allow us to apprehend obligations. Other features of standpoints might also be characteristic of them; e.g., it may be that standpoints are characteristically associated with certain types of reactive attitudes (cf. McElwee 2017).

Infinitism about Cross-Domain Conflict

149

about which first-order standpoint an agent is to adopt. The moral standpoint may—or may not—be self-ratifying at the first-order level; that is, it may or may not be that from the moral standpoint it looks as if any given agent ought to rely upon the moral standpoint in order to decide what to do. For example, if something like the two-level utilitarianism described by Hare (1981) were true, then morality would not be self-ratifying at the firstorder level. Throughout this paper I will assume deontic rationalism, the view that obligations of any given domain are explained entirely by reasons from that domain. So, for example, I am assuming that moral obligations are explained entirely by moral reasons, that prudential obligations are explained entirely by prudential reasons, and so on. Further, I will assume that to be a reason of any given domain is precisely to be capable of entering into an explanation of an obligation from that domain. So, for example, to be a moral reason (as against some other sort of reason) is just to be capable of explaining moral obligations.⁴ Deontic rationalism thus conceived implies that at least some reasons are in the business of explaining obligations (so, it’s in tension with certain views, such as that of Kearns and Star 2008). But deontic rationalism does not say that all reasons are in the business of explaining obligations (so, it is consistent with the possibility of what Dancy calls ‘enticing reasons’— reasons that favor but don’t require). Deontic rationalism says that if a reason belongs to a given domain then it explains obligations in that domain.

7.2.1. From the Overriding View to Novel Higher-Order Domains Consider the following pair of views: Cross-domain reasons fundamentalism—metaphysical component: Suppose obligation OA belongs to normative domain DA and obligation OB belongs to normative domain DB (where DA and DB are different domains). If OA overrides OB, this is entirely because the reason(s) that give rise to OA outweigh the reason(s) that give rise to OB. Cross-domain reasons fundamentalism—epistemological component: Suppose obligation OA belongs to normative domain DA and obligation OB belongs to normative domain DB (where DA and DB are ⁴ This does not entail that nothing is supererogatory. The assumption that all reasons are capable of explaining moral obligations does not imply that reasons are not also capable of explaining other sorts of properties, such as supererogation.

150

David Killoren different domains). If OA overrides OB, this can be known simply by observing (i.e., coming to know) that the reason(s) that give rise to OA outweigh the reason(s) that give rise to OB.

To make the argument that I want to make in this section, I will first need to argue that the epistemological component of cross-domain reasons fundamentalism is false. So consider three possibilities: : Reasons of one domain never outweigh or are outweighed by reasons of another domain. Reasons of different domains are, in short, incomparable. C W D: There is cross-domain outweighing at the level of reasons—reasons of different domains are comparable—but cross-domain outweighing at the level of reasons cannot fully determine or explain cross-domain overriding at the level of obligations. C W D: There is cross-domain outweighing at the level of reasons, and cross-domain outweighing at the level of reasons fully determines and explains cross-domain overriding at the level of obligations.

I will argue that the epistemological component of cross-domain reasons fundamentalism is false regardless of which of the above theses is true.

7.2.1.1. Incomparability and Comparability Without Determination If Incomparability is true, the weight of Monica’s moral reason to donate can’t be compared with the weight of Monica’s prudential reason to invest. In that event there are two possibilities. The first possibility is that (I) there is no cross-domain overriding: so, e.g., Monica’s moral obligation neither overrides nor is overridden by her prudential obligation. But we have to reject this possibility because, in assuming the overriding view, we’ve assumed that there is cross-domain overriding. The second possibility— which is therefore the only one open to us here—is that (II) when a given obligation OA overrides another obligation OB (where OA and OB belong to different domains) this is explained in terms of something other than the relative weights of the reasons underlying OA and OB. For example, perhaps there is something special about morality (or about prudence, or whatever) in virtue of which moral obligations (or prudential obligations, or whatever) override other types of obligations, regardless of the relative weights underlying the obligations in question. If (II) holds, then cross-domain reasons fundamentalism—in both its metaphysical component and (therefore) its epistemological component— is false. Given this, and given that (II) is the only possibility given

Infinitism about Cross-Domain Conflict

151

Incomparability, we can conclude that if Incomparability is true then cross-domain reasons fundamentalism is false. (This will not be particularly surprising to anyone.) Now consider the case where Comparability Without Determination is true. Here the reasoning will be very similar to that in the case where Incomparability is true, so I’ll be brief. Here, as with Incomparability, we have two options: either (I) there is no cross-domain overriding—an option we have to reject given our assumption of the overriding view—or (II) crossdomain overriding is to be explained in terms of something other than the relative weights of conflicting reasons. And as we’ve seen, if (II) is true, then cross-domain reasons fundamentalism in both its components is false. And so if Comparability Without Determination is true, then cross-domain reasons fundamentalism is false.

7.2.1.2. Comparability With Determination If Comparability With Determination is true, then the metaphysical component of cross-domain reasons fundamentalism is definitely true. But I want to argue that, even in that case, the epistemological component is still false. I begin with the premise that Overriding is an antisymmetric relation If OA overrides OB, then OB doesn’t override OA.

This premise is highly plausible. When we say that Monica’s moral obligation overrides her prudential obligation (for example) we mean at least this: Monica’s moral obligation enters into some kind of conflict with her prudential obligation, and her moral obligation emerges victorious from this conflict, and her prudential obligation emerges defeated. Such a victory for morality would be impossible if it is shared equally on both sides of the conflict. In this sort of competition, if both sides win the prize, then neither side can rightfully be said to have won or lost. My next step is to make two observations: From within the prudential standpoint, it looks as if Monica’s prudential reason outweighs Monica’s moral reason; and from within the moral standpoint, it looks as if Monica’s moral reason outweighs Monica’s prudential reason. These claims, I believe, are well-supported by experiences familiar to anyone who has occupied these normative standpoints. Let me draw that phenomenological claim out a bit. My claim is that, for instance, anyone who knows what it is like to consider matters from the prudential standpoint knows that from within that standpoint it appears that—for example—Monica has most reason to do precisely what she has most prudential reason to do, which is to say that it appears that she

152

David Killoren

has most reason to invest rather than to donate. Another way to say this is to say that from the prudential standpoint, it seems utterly irrational for Monica to donate the money. If she donates the money, then that money will just be gone—it’ll do no good for her whatsoever. The fact that donating the money would help others (even to the point of saving many lives) either does not register as a reason at all from the prudential standpoint or, if it registers as a reason, it registers as a reason that is clearly outweighed by Monica’s reason to invest. A financial advisor—whose job is to view Monica’s situation from the prudential standpoint and to offer advice from within that standpoint—would surely concur. The financial advisor’s verdict would be that Monica has all the reason in the world to invest and little to no reason to donate.⁵ Further, my phenomenological claim is that something similar is true of the moral standpoint: From within the moral standpoint, it appears that Monica has most reason to do precisely what she has most moral reason to do—that is, she has most reason to donate rather than to invest. This, we can suppose, is what a competent moral advisor would say to Monica. And, further still, I want to make a general claim: that from within any given standpoint X, it appears that any given agent has most reason to do whatever she has most X-reason to do. Now, what this means is that we cannot accept the following principle: Outweighing principle As long as it appears to us that the reasons underlying OA outweigh the reasons underlying OB, then we are justified in concluding that OA overrides OB.

The reason we cannot accept the outweighing principle is that—given my phenomenological claims above—the outweighing principle would typically justify us in believing that any two conflicting obligations from different domains override one another. In the case of Monica, for example, by switching back and forth between the moral standpoint and the prudential ⁵ One might object that from the prudential standpoint, non-prudential (e.g., moral) reasons do appear to have significant weight, or would appear to have significant weight if they were given consideration—but while one is occupying the prudential standpoint one is merely bracketing or ignoring non-prudential reasons. In response, consider the fact that if Monica were to donate then she will save many others’ lives. This fact is a weighty moral reason. Yet, while Monica is occupying the prudential standpoint, she can focus her attention squarely and directly on this fact; she needn’t ignore it. Indeed, she likely needs to take account of this fact in order to thoroughly consider her situation from the prudential standpoint, because she needs to determine whether she could personally benefit in any way by saving others’ lives. So it is not the case that occupying the prudential standpoint means ignoring moral reasons. Thanks to an anonymous referee for pushing me on this point.

Infinitism about Cross-Domain Conflict

153

standpoint and applying the Outweighing Principle from within each standpoint, we’d quickly gain license to conclude that Monica’s moral obligation both outweighs and is outweighed by her prudential obligation. But we should not be able to conclude this, given the premise that overriding is an antisymmetric relation. So we should reject the outweighing principle. The upshot of this is that, even if the metaphysical component of crossdomain reasons fundamentalism is true, the epistemological component is false (because the epistemological component requires the outweighing principle—if knowledge requires justification). Summing up: Even when we assume Comparability With Determination, we should still conclude that the Outweighing Principle is false; and if the Outweighing Principle is false, then the epistemological component of cross-domain reasons fundamentalism is false; so, even if Comparability With Determination is true, the epistemological component of cross-domain reasons fundamentalism is false. At this point I’ve made the following argument (strewn between Sections 7.2.1.1 and 7.2.1.2): The argument against the epistemological component of reasons fundamentalism (A) Either Incomparability, Comparability Without Determination, or Comparability With Determination is true. (B) If either Incomparability, Comparability Without Determination, or Comparability With Determination is true, then the epistemological component of reasons fundamentalism is false. (C) Therefore, the epistemological component of reasons fundamentalism is false.

I will now go on to draw out some of the interesting implications of the conclusion of this argument.

7.2.1.3. Novel Higher-Order Domains I have argued that the epistemological component of reasons fundamentalism is false. The upshot of this is that, in order to determine that (for example) Monica’s moral obligation overrides her prudential obligation, it is not enough to observe from within the moral standpoint that Monica’s moral reason outweighs her prudential reason. Some further step is required. But what is that further step? If the needed insight is unavailable from within the moral standpoint, then it would stand to reason that we must shift to some other standpoint. But this other standpoint cannot be the prudential standpoint. That’s a non-starter. From within the prudential standpoint it appears that Monica’s

154

David Killoren

prudential reason outweighs her moral reason, and—no matter what the relation between reasons and cross-domain overriding turns out to be—this cannot be a basis for the conclusion that Monica’s moral obligation overrides her prudential obligation. So, if we can determine that Monica’s moral obligation overrides her prudential obligation, it seems that this must be determinable only by occupying some third standpoint—call it Standpoint D₁. From within Standpoint D₁, we may suppose, one gets an external view of the conflict between the moral and the prudential—and, somehow, it is from that external view that one can justifiably conclude that (in Monica’s case at least) the moral obligation overrides the prudential one but not vice versa. A natural question at this point is: What is Standpoint D₁? Many possibilities arise here: Maybe Standpoint D₁ is the rational standpoint, or is in some way an impartial standpoint, or the standpoint that results from reflective equilibrium reasoning, or (borrowing a possibility from Sidgwick 1874) the point of view of the universe, or (borrowing a possibility from Nagel 1970) the view from nowhere, or (borrowing a possibility from Foot 1978) is unscripted, or (borrowing a possibility from McLeod 2001) is the domain of just plain ought. Of course, we have to be careful here because some of these kinds of characterizations have variously been offered of the moral standpoint itself, and I have already argued that Standpoint D₁ isn’t the moral standpoint. In any event, for my purpose, I don’t require a worked-out view about what Standpoint D₁ is or how we should refer to it. What I require is only the negative characterization of Standpoint D₁: that Standpoint D₁ isn’t the moral standpoint, or the prudential standpoint, or any of the other standpoints that may be in conflict at the zeroth-order level. This means that Standpoint D₁ is, in the terms I introduced above, a novel higher-order standpoint.⁶ Now I have argued that if it is determinable that (for example) Monica’s moral obligation overrides her prudential obligation, then there must be a novel higher-order standpoint—Standpoint D₁—from which this determination can be made. This isn’t quite the same as the conclusion that there in fact is at least one novel higher-order standpoint. After all, it’s possible that we simply cannot determine that Monica’s moral obligation overrides her prudential obligation. Perhaps such things are unknowable. ⁶ Standpoint D₁ could be a standpoint that takes account of both prudential and moral reasons—call it prudence-plus-morality. Even then, Standpoint D₁ would not be the moral standpoint and wouldn’t be the prudential standpoint, for the simple reason that prudence-plus-morality takes account of more reasons than either the prudential standpoint or the moral standpoint. So prudence-plus-morality would still count as novel in my terms.

Infinitism about Cross-Domain Conflict

155

But we’ve assumed the overriding view—so we’ve assumed that crossdomain overriding occurs. And if cross-domain overriding does indeed occur, then—if we aren’t extreme skeptics—we should think that facts about cross-domain overriding are at least in principle knowable. Richard Rowland (2016) persuasively argues for a general constraint according to which any given normative fact must be knowable by some possible agent; and this general constraint is all that I require for the argument here (given that facts about overriding are normative facts). Further, my argument here is precisely that facts about cross-domain overriding are knowable only by adopting some novel higher-order standpoint. So we should now conclude that there is at least one novel higher-order standpoint, which I’ve called Standpoint D₁. Now recall the relation between standpoints and domains. I claimed above, in introducing the notion of a standpoint, that a given standpoint counts as the X-standpoint because obligations of the X-domain are apprehensible from that standpoint. Given this claim we should say that Standpoint D₁ is associated with a normative domain—and because Standpoint D₁ is a novel standpoint, the associated domain is novel also. This associated domain—which we may call D₁—must (like any normative domain) contain deontic objects and reasons. Presumably, one of the deontic objects that belongs to this domain is an obligation that points toward the overriding domain at the zeroth-order level—in the case of Monica, this would be the moral domain. So the picture that is now emerging is, as applied to the case of Monica, something like this: When we take up the prudential point of view, we see that Monica must invest; when we take up the moral point of view, we see that Monica must donate; to adjudicate this conflict, we take up Standpoint D₁ (which we may say is the rational standpoint, or the point of view of the universe, or the unsubscripted standpoint, or whatever); and once we do this—that is, once we take up Standpoint D₁—then we see that Monica must take up the moral standpoint rather than the prudential standpoint in order to decide what to do (i.e., she must act in line with her moral obligation rather than her prudential one).

7.2.2. From Novel Higher-Order Domains to Infinitism about Cross-Domain Conflict I have thus far argued that in order to know that (e.g.) a moral obligation overrides a prudential obligation, we have to be able to occupy a higherorder standpoint, Standpoint D₁, and to see, from that standpoint, that the agent has an obligation to act in line with her moral obligation rather than

156

David Killoren

her prudential obligation. Given the connection between standpoints and domains, this means that—in general—in order to know that OA overrides OB we have to know that there is an obligation belonging to a higher-order domain that requires the agent to act in line with OA rather than OB. This is an epistemic claim, not a metaphysical one. That is, it is not a claim about what it is for a moral obligation to override a prudential obligation. But this epistemic claim does support a further metaphysical claim, or so I will now argue (Section 7.2.2.1). After that, I will show how this metaphysical claim supports the argument for infinitism about crossdomain conflict (Section 7.2.2.2).

7.2.2.1. Higher-Order Domains Ground Cross-Domain Overriding I want to argue that the epistemic claim that I’ve been defending supports the following view: Higher-order domains ground cross-domain overriding Suppose obligation OA belongs to normative domain DA and obligation OB belong to normative domain DB (where DA and DB are different domains). If OA overrides OB, this is at least in part because there is a higher-order domain, D₁, which contains an obligation, O₁, that requires the agent to act in line with OA rather than OB.

My argument for this claim is a kind of inference to the best explanation. I’ve argued that in order to know that OA overrides OB we have to know that there is an obligation belonging to a higher-order domain that requires the agent to act in line with OA rather than OB. The most straightforwardly appealing explanation for this epistemic constraint, I suggest, would be that the obligation belonging to the higher-order domain is at least part of what makes it the case that OA overrides OB. Here’s an analogy to support this inference. Suppose that I’m a private in an army. SergeantA has just given me an order to (say) march twenty miles west. But a different sergeant, SergeantB, has given me an incompatible order to (say) march twenty miles east. Suppose I am of the view that each sergeant’s order creates a certain kind of obligation for me. So I am obligated to march twenty miles west and simultaneously obligated to march twenty miles east. I want to know which of these obligations overrides the other. Suppose I come to the view that I cannot know which obligation overrides the other without consulting someone who is further up the hierarchy than either of the sergeants—say, a lieutenant. If the lieutenant says that I ought to go with SergeantA’s order, then I’m in a position to know that SergeantA’s order overrides SergeantB’s order; if the lieutenant says that I ought to go with SergeantB’s order, then I’m in a position to know that SergeantB’s order overrides SergeantA’s order. Well, if I come to that epistemic view, then it

Infinitism about Cross-Domain Conflict

157

would be natural to think that the lieutenant’s say-so is also at least part of what makes it the case that one sergeant’s order overrides the other. In this example, then, an epistemic claim about how I can know that one order overrides another provides support for a metaphysical claim about what makes it the case that one order overrides the other. I do not want to claim that in every case, if I cannot know some fact F₁ without knowing some other fact F₂, I am entitled to conclude that F₂ is part of what makes F₁ the case. That principle is clearly false. For example, perhaps I cannot know that the temperature is fifty degrees without knowing that the thermometer reads fifty degrees, but this does not show that the reading on the thermometer is any part of what makes it the case that the temperature is fifty degrees. However, it is the case that in general, if I cannot know some fact F₁ without knowing some other fact F₂, this epistemic constraint calls for some sort of an explanation. And in most cases there will only be three sorts of explanations to consider: either F₂ brings F₁ about; or F₁ brings F₂ about; or some third factor brings F₁ and F₂ about. Each of these possibilities could, in principle, explain why I cannot know some fact F₁ without knowing some other fact F₂. In the present case, these three options would be: (i) The fact that [OA overrides OB] brings it about that [an obligation belonging to a higher-order domain requires the agent to act in line with OA rather than OB]. (ii) The fact that [an obligation belonging to a higher-order domain requires the agent to act in line with OA rather than OB] brings it about that [OA overrides OB]. (This is the claim that I’m defending.) (iii) Some third factor brings about both that [OA overrides OB] and that [an obligation belonging to a higher-order domain requires the agent to act in line with OA rather than OB]. Each of these three possibilities could in principle explain why we need to know that [an obligation belonging to a higher-order domain requires the agent to act in line with OA rather than OB] in order to know that [OA overrides OB]. So, in order to defend (ii) (my preferred explanation) I need to say something against (i) and (iii). (i) and (iii) have a crucial similarity. If we accept either (i) or (iii), then we are left without an explanation of the fact that OA overrides OB. So, if we accept either (i) or (iii), then we are suddenly in need of something to say about why OA overrides OB. Perhaps the relative weights of reasons could explain this. On that view, the reasons underlying OA are weightier than the reasons underlying OB, and this—rather than any facts about higher-order domains—is why OA overrides OB. But this explanation isn’t satisfactory.

David Killoren

158

Here’s why. I argued in Section 7.2.1.2 that from within the moral standpoint, moral reasons seem to outweigh prudential reasons, whereas from within the prudential standpoint, prudential reasons seem to outweigh moral reasons. This suggests that reasons can be measured on different scales (as it were) and that different scales give different results. On the moral scale, moral reasons register as heaviest; on the prudential scale, prudential reasons register as heaviest. And it is unclear to me how defenders of either (i) or (iii) can identify a basis on which one or the other of these two scales is rightly regarded as uniquely correct or as more correct than the other. We may suppose that each scale measures some quantifiable attribute of reasons; each measures something that could be called ‘weight.’ But it is unclear how defenders of either (i) or (iii) could produce a principled reason why one of these quantifiable attributes rather than the other deserves to be regarded as the true weight of reasons. Of course, at this point, in order to explain why the quantifiable attribute measured by the moral scale (or the prudential scale, or whatever scale) is the one that is correctly regarded as the true weight of reasons, proponents of (i) or (iii) might want to point to a third, higher-order normative standpoint as an arbiter between the contradictory deliverances of the diverging scales. But if we point to a third higher-order standpoint as arbiter, then the view begins to bear an uncanny resemblance to (ii). So proponents of (i) or (iii) will have difficulty explaining why OA overrides OB in terms of the relative weights of reasons, at least not without allowing their view to shade into some version of (ii). Can proponents of (i) or (iii) come up with some alternative, superior explanation, one that doesn’t appeal to the relative weights of reasons? I don’t see what that alternative explanation might be. By contrast, (ii) is precisely an explanation of why OA overrides OB. This is why I think (ii) is to be preferred over (i) or (iii). And (ii) implies the claim that higher-order domains ground cross-domain overriding.

7.2.2.2. Authority and the Beginning of the Regress Recall the argument for infinitism about cross-domain conflict. The first premise is just a definition: (1) OA overrides OB iff OA belongs to DA and OB belongs to DB and DA trumps DB.

The second premise is: (2) If DA trumps DB, this is because there is a novel higher-order domain, D₁, that explains why DA trumps DB.

The above premise follows directly from the argument I gave in Section 7.2.1. The next premise in the argument is as follows:

Infinitism about Cross-Domain Conflict

159

(3) D₁ explains why DA trumps DB only if D₁ has the requisite authority.

Intuitively, not just any domain can explain why DA trumps DB. To illustrate this, imagine a first-order analog of the esthetic domain. Suppose there is a standpoint from which we can assess the beauty (or other esthetic properties) of different zeroth-order domains. From this standpoint, suppose, it appears that the correct zeroth-order domain—that is, the zeroth-order domain that should be used to guide one’s actions—is whatever happens to be the most beautiful domain. And suppose that DA just happens to be considerably more beautiful than DB. Then, from this first-order esthetic standpoint, it would appear that one ought to act in line with the obligations of DA rather than those of DB; in other words, the domain associated with that standpoint would contain an obligation to act in line with DA rather than DB. But this would not mean that DA genuinely trumps DB. For it seems clear that we should not select our zeroth-order standpoint on esthetic grounds. The first-order esthetic domain that I am discussing here thus lacks what I’ll call the authority required in order to explain why DA trumps DB. The next step in the argument is: (4) D₁ has the requisite authority only if D₁ trumps at least some of its same-order competitors D₁*, D₁**, D₁***, . . . .

To defend this premise, I have to first explain what I mean by a same-order competitor. Imagine that from Standpoint D₁, it appears that one should act in line with DA, but from Standpoint D₁*, it appears that one should act in line with DB. Then D₁* is a same-order competitor of D₁. D₁* and D₁ are same-order competitors because they contain conflicting verdicts about domains one level below them. The thought behind (4) is that if D₁ and D₁* contain conflicting verdicts in this way, then D₁ cannot have the sort of authority mentioned in (3) unless D₁ trumps D₁*. To see the plausibility of this, let’s once again consider the example of the army private. In that example, SergeantA has just given me (the private) an order to march twenty miles west, and a different sergeant, SergeantB, has given me an incompatible order to march twenty miles east. To resolve this conflict, I go to the rank above sergeant—the rank of lieutenant. I find a lieutenant, Lieutenant₁, who tells me to go with SergeantA’s order. This could conceivably explain why SergeantA’s order overrides SergeantB’s order (i.e., it could explain why SergeantA trumps SergeantB). But now suppose that, unfortunately, a different lieutenant, Lieutenant₁*, tells me to go with SergeantB’s order. Then in this eventuality, it seems that Lieutenant₁’s verdict cannot explain why SergeantA trumps SergeantB—unless it is also the case that Lieutenant₁ trumps Lieutenant₁*. The thought behind (4) is that something similar needs to be true in the analogous case with regard to normative domains.

160

David Killoren

Here one might issue the following objection. There is no guarantee that Lieutenant₁ will have any same-rank competitors. Perhaps Lieutenant₁ is the only lieutenant. Or perhaps there are multiple lieutenants, but all of the lieutenants agree (and therefore aren’t competitors): perhaps all of them tell me to heed SergeantA rather than SergeantB. In either case, we might reasonably think that Lieutenant₁ doesn’t need to trump any same-rank competitors in order for her verdict to explain why SergeantA trumps SergeantB. Likewise, one might suggest, it is not a foregone conclusion that D₁ has any same-order competitors: perhaps D₁ is the only first-order domain; or perhaps all of the first-order domains agree with D₁. In that case we could imagine that D₁ has the requisite authority without trumping any competitors at all, and then (4) would be false. However, I think it is inevitable that D₁ will have same-order competitors. It seems that, at every level, there should be a domain for every relevant principle. (This is not the same as to say that every domain is governed by a principle. There may be unprincipled domains. Indeed, if moral particularists are correct, morality is one such domain.) At the zeroth order, the relevant principles are action-guiding principles. Given that there are infinitely many different logically possible action-guiding principles, there should be infinitely many domains at the zeroth order. At the first order, the relevant principles are principles that guide the selection of zeroth order domains, and there are infinitely many such principles. So we should think that D₁ has infinitely many same-order competitors; and the subset of these with which D₁ conflicts will likewise be infinite. So I believe there is no way for D₁ to avoid having same-order competitors. The fifth premise of the argument is as follows: (5) If D₁ trumps any of its same-order competitors, this is because there is a novel higher-order domain, D₂, that explains why D₁ trumps those same-order competitors.

This premise can be defended with a line of argument that is precisely parallel with the above defense of (2). This is where the regress begins. To explain why D₁ trumps its same-order competitors, we need D₂; in order for D₂ to do this explanatory work, D₂ needs authority; in order for D₂ to have authority, D₂ needs to trump at least some of its same-order competitors D₂*, D₂**, D₂***, . . . ; in order for D₂ to trump its same-order competitors, there needs to be a novel higher-order domain D₃ to explain this; and so on. The result: In order for OA to override OB, there needs to be an infinitely long chain of domains D₁, D₂, D3, . . . , such that D₁ contains an obligation to act in line with the obligations of DA, and D₂ contains an obligation to act in line with the obligations of D₁, and D₃ contains an obligation to act in

Infinitism about Cross-Domain Conflict

161

line with the obligations of D₂, and so on, indefinitely. And that’s infinitism about cross-domain conflict. The picture of normativity that emerges from this infinitism is something like the following. We are at least sometimes confronted by a variety of normative domains that pull us in different directions. And in at least some of these cases, one domain trumps the other—for example, perhaps in a given case morality wins out over prudence. When this happens, it happens in virtue of a higher-order domain that possesses the requisite authority. We can occupy the standpoint associated with that higher-order domain by taking a step back, as it were, and viewing both morality and prudence from the outside, that is, from the perspective of that authoritative higherorder domain. Once we do this, we can ask the question of why this higher-order domain possesses the authority that it does—that is, we can ask why this higher-order domain (rather than any of its possible alternatives) is the domain we should rely upon in order to adjudicate the conflict between morality and prudence. And in order to address that further question, we must take a step back yet again, and view from the outside that higher-order domain alongside its alternatives. The picture given by infinitism is that we will always be able to take these sorts of steps backward, no matter how many steps back we take. We will never reach a point where the authority of a given domain to adjudicate lower-order domains is brutely given.

7. 3. OBJEC T IONS AN D RE P LI ES In this section I will consider a few of the many objections that may be raised against the argument that I have just laid out. Several of these objections are directed against infinitism about cross-domain conflict, rather than against the conditional claim that if the overriding view is true, then infinitism is true. But such objections may be taken as indirectly targeting the conditional claim, given that such objections would show that the conditional claim ought to be rejected by anyone who finds the overriding view to be plausible. First objection: Infinitism requires that there is an infinite series of domains and an infinite series of obligations contained in those domains. But it is implausible to suppose that there are infinitely many of either of those sorts of things. To start with, it is not implausible to suppose that there are infinitely many obligations. Indeed, it would be surprising if there weren’t. For there seem to be infinitely many (real and hypothetical) cases in which agents have

162

David Killoren

to make up their minds about what to do and are obligated to do one thing rather than another. It is true that infinitism implies that any given agent has infinitely many obligations. In the example we’ve been working with, an agent has an obligation given by DA (this is an obligation to act in a certain way), and has an obligation given by D₁ (this is an obligation to act in line with DA rather than DB), and has an obligation given by D₂ (this is an obligation to comply with D₁ rather than D₁*, D₁**, D₁***, . . . ), and so on. Perhaps it seems strange to think that one agent can be saddled with infinitely many obligations in this way. But the idea that one agent can have infinitely many obligations is also not particularly strange. Suppose a given agent is obligated to give at least 4 percent of her income to charity. Then it would seem to follow trivially than she is obligated to give at least 3.999 percent of her income to charity, and is obligated to give at least 3.998 percent of her income to charity, and so on. In this way we could start with one obligation and use it to generate an infinite number of further obligations for one agent. So the very idea of one agent having infinitely many obligations isn’t objectionable. The other part of the present objection points out that infinitism requires an infinite series of domains as well as obligations. But I am not sure why that would be objectionable if infinitely many obligations isn’t objectionable. Second objection: Infinitism makes cross-domain overriding unknowable. In order to know that DA trumps DB, we have to know that D₁ has authority; and in order to know that D₁ has authority, we have to know that D₁ trumps its same-order competitors; and in order to know that D₁ trumps its same-order competitors, we have to know that D₂ has authority; and so on. Thus infinitism requires us to know an infinite series of these sorts of normative facts. Further, we cannot know these sorts of facts without occupying an infinite series of standpoints associated with these domains. For example, we cannot know that DA trumps DB without occupying the standpoint associated with D₁; and we cannot know that D₁ trumps its same-order competitors without occupying the standpoint associated with D₂; and so on. So, in order to know that DA trumps DB, we need to be capable of occupying an infinite series of standpoints, one after the other. It is not possible to do that (given that occupying a standpoint requires some positive amount of time; see Reisner 2015 for relevant discussion) so it is not possible to know that DA trumps DB. I think it is probably true that we cannot know that DA trumps DB without occupying the standpoint associated with D₁. And I think it is probably also true that we cannot know that D₁ trumps its same-order competitors without occupying the standpoint associated with D₂. But I don’t think this

Infinitism about Cross-Domain Conflict

163

has to be the case at every level. At some point, I suspect, an ordinary agent may be entitled to make an inductive inference. If the agent finds that, no matter how far back along the chain of domains she goes, she always finds that there is a domain one level up that continues the pattern she has thus far found to hold, she may be justified in inferring that the pattern will continue indefinitely. Of course, she may never be entitled to certainty that the pattern will continue indefinitely. But in general it is highly plausible that inductive inferences do not need to justify certainty in order to confer knowledge. Third objection: The argument requires a flawed and even absurd account of authority. On the required account, the authority of a given domain at the zeroth order is transmitted from a higher-order domain at the first order. But the domain at the first order must first receive the authority thus transmitted from a higher-order domain at the second order, and so on, indefinitely. In this account there is no highest order domain to originate the authority that gets transmitted along the infinite series of domains; and so the authority that is being transmitted from domain to domain comes from nowhere. And if it comes from nowhere, it cannot exist at all. It is a bit odd to suggest, as this objection does, that my argument requires authority to be transmitted along an infinite series of domains. This makes it sound as if authority is like a single baton being passed from runner to runner. But in my account, there isn’t a single object that is passed from domain to domain. Rather, each domain has its own authority: the authority to adjudicate conflicts at the domain below it. The authority of D₁ to adjudicate conflicts at the zeroth order is a different thing (or, more exactly, it is a different token property) from the authority of D₂ to adjudicate conflicts at the first order. The objection would be slightly improved if it were to claim that, on my account, the authority of any given domain derives from the separate authority of a higher-order domain. But once this is clarified, it is not clear to me that there is a serious objection here. Provided that we can have an infinite series of domains (a point I’ve already addressed) I don’t see what is absurd or even objectionable about a picture in which each domain in such a series receives its authority from the next one in the series. After all, “DY bestows authority on DX” just names a relation between DY and DX, and it is possible for each element in an infinite series to bear relations to the other elements; for example, each natural number in the infinite series of natural numbers is greater than the number below it. Fourth objection: According to this argument, the power of morality to override prudence, in a given instance, requires the existence of an infinite chain of domains in which D₁ favors morality over prudence, D₂ favors D₁

164

David Killoren

over D₁*, D₃ favors D₂ over D₂*, and so on. Call this the morality-favoring chain. But there must be multiple such infinite chains. For example, it must be that there is a chain such that D₁* favors prudence over morality, and D₂* favors D₁* over D₁, and so on. Call this the prudence-favoring chain. (This must be the case because, as I argued earlier, at every level, there should be a domain for every relevant principle.) If there are two of these sorts of infinite chains each pointing toward conflicting verdicts about which zeroth order domain to take up, then this seems problematic. For, if there is both a morality-favoring chain and a prudence-favoring chain, then it would follow that morality trumps and is trumped by prudence; this would be objectionable because (as we’ve seen) overriding is an antisymmetric relation. However, at no point in the argument have I laid out sufficient conditions for overriding, for authority, or for any of the other normative properties that I’ve been discussing here. So my argument does not commit me to say that the existence of a morality-favoring chain in a given case is sufficient to show that morality trumps prudence in that case and likewise does not commit me to say that the existence of a prudence-favoring chain in a given case is sufficient to show that prudence trumps morality in that case. My argument is consistent with the possibility that, in order for morality to trump prudence, there must be a morality-favoring chain and there must be some further factor present. Here is one attractive possibility along those lines. Suppose that there is a property of normative oomph. (For articulation of the hard-to-articulate notion of normative oomph, see Joyce 2006; Copp 2015: 142.) It may be that a given domain DX has authority in a given case iff (i) DX trumps at least some of its same-order competitors, and (ii) DX is, as we might say, oomphy (i.e., the normative objects contained in DX have the property of oomph). In that case, we could show that, for example, morality trumps prudence iff there is a morality-favoring chain of domains and every domain in that chain is oomphy. And this, in turn, would mean that even if there is both a morality-favoring chain and a prudence-favoring chain in a given case, it might still be the case that morality trumps prudence but not vice versa, or that prudence trumps morality but not vice versa. In this way, my argument can respect the constraint that overriding is an antisymmetric relation. Here it may be asked: What if, in a given case, there is both a moralityfavoring chain and a prudence-favoring chain, and every domain in each of those chains is oomphy? Then, in order to respect the antisymmetry constraint, we’d need to modify the account again. For example, we may say that a given domain DX has authority in a given case iff (i) DX trumps at least some of its same-order competitors, and (ii) DX is oomphy, and (iii) none of DX’s same-order competitors are oomphy. This is just one

Infinitism about Cross-Domain Conflict

165

possibility along the present lines; there are many others. But this possibility should suffice to show that I can allow the possibility of multiple infinite chains of domains in a given case without falling afoul of the antisymmetry constraint. At this point one might suggest that if the notion of oomph is brought into the picture, we can rely on this notion to explain why one domain trumps another, and we do not need to consider higher-order domains at all. But this suggestion faces a dilemma. Either oomph is defined in terms of overridingness, such that a given obligation has oomph iff the obligation overrides all obligations with which it conflicts, or it isn’t. A cost of defining oomph in terms of overridingness is that it will imply that any two same-order competitor domains cannot both be oomphy. Further, and more importantly in the present context, if oomph is defined in that way then we will not be able to rely on oomph to explain overridingness (because oomph will simply be overridingness). Alternatively, if oomph is not defined as overridingness, then it is defined in some other way or not defined at all. In either case oomph is unlikely to be sufficient on its own to explain why one domain trumps another. Suppose that the moral domain is oomphy and that the legal domain is not oomphy. Then given that oomph is not defined as overridingness, it is still an open question whether morality trumps legality. Indeed, the possibility that a non-oomphy domain could trump an oomphy one is not a mere logical possibility. For instance, if an oomphy higher-order domain favors legality over morality, then it may be plausible that legality trumps morality even on the supposition that morality is oomphy and legality isn’t. If that’s right, then it shows that oomph alone is not sufficient to explain why one domain trumps another.⁷

References Copp, David (1997). “The Ring of Gyges: Overridingness and the Unity of Reason,” Social Philosophy and Policy 14 (1): 86–106. Copp, David (2015). “Rationality and Moral Authority,” in R. Shafer-Landau (ed.), Oxford Studies in Metaethics, Volume 10. Oxford: Oxford University Press, 134–59.

⁷ For extremely valuable feedback, advice, and encouragement, I’m exceedingly grateful to Thomas Carson, Guy Fletcher, Rebecca Harrison, Richard Rowland, Russ ShaferLandau, Rob Streiffer, and many individuals who were in the audience at the Metaethics Workshop in Madison.

166

David Killoren

Copp, David (ms.). “Normative Pluralism and Skepticism about Ought Simpliciter.” Crisp, Roger (ms.). “Partiality and the Dualism of Practical Reason.” Cuneo, Terence and Shafer-Landau, Russ (2014). “Moral Fixed Points,” Philosophical Studies 171 (3): 399–443. Dorsey, Dale (2016). The Limits of Moral Authority. Oxford: Oxford University Press. Enoch, David (2019). “How Principles Ground,” in R. Shafer-Landau (ed.), Oxford Studies in Metaethics, Volume 14. Oxford: Oxford University Press, 1–22. Foot, Philippa (1978). Virtues and Vices: And Other Essays in Moral Philosophy. Oxford: Blackwell and Berkeley: University of California Press. Hare, R. M. (1981). Moral Thinking: Its Levels, Method, and Point. Oxford: Oxford University Press. Joyce, Richard (2006). The Evolution of Morality. Cambridge, MA: MIT Press. Kearns, Stephen and Star, Daniel (2008). “Reasons: Explanations or Evidence?” Ethics 119 (1): 31–56. Klein, Peter (1998). “Foundationalism and the Infinite Regress of Reasons,” Philosophy and Phenomenological Research 58 (4): 919–25. McElwee, Brian (2017). “Supererogration Across Normative Domains,” Australasian Journal of Philosophy 95 (3): 505–16. McLeod, Owen (2001). “Just Plain ‘Ought’,” The Journal of Ethics 5 (4): 269–91. McPherson, Tristram (2018). “Authoritatively Normative Concepts,” in R. ShaferLandau (ed.), Oxford Studies in Metaethics, Volume 13. Oxford: Oxford University Press, 253–77. Nagel, Thomas (1970). The Possibility of Altruism. Princeton, NJ: Princeton University Press. Parfit, Derek (2011). On What Matters: Volume 1. Oxford: Oxford University Press. Reisner, Andrew (2015). “Fittingness, Value, and Trans-World Attitudes,” Pacific Philosophical Quarterly 65 (260): 464–85. Rowland, Richard (2016). “The Significance of Significant Fundamental Moral Disagreement,” Noûs 51 (4): 802–31. Sidgwick, Henry (1874). The Methods of Ethics. London: Macmillan. Tiffany, Evan (2007). “Deflationary Normative Pluralism,” Canadian Journal of Philosophy 37 (33): 231–62. Wolf, Susan (2015). The Variety of Values. Oxford: Oxford University Press.

8 We Can Have Our Buck and Pass It, Too Zoë Johnson King 8. 1. Imagine a curry. Suppose that the curry is spicy, warm, and nourishing. And suppose that these properties make it the case that the curry has a further property: it is good. (Perhaps it is good qua curry, or perhaps it is good absolutely; this does not matter for present purposes.) Now imagine that you are deliberating about whether to eat this curry, and you ask an experienced friend for advice. Your experienced friend might mention any of the foregoing considerations. She might say, “You should try it, it’s really spicy!” or, “You should try it, it’s really warm and nourishing!” or, “You should try it, it’s really good!” Some people think the last of these considerations is not like the other ones. They hold that the fact that the curry is spicy, the fact that it is warm, and the fact that it is nourishing can all be reasons for you to eat it. But they hold that the fact that the curry is good cannot be a reason for you to eat it. This is because they hold that facts about the goodness of objects or states of affairs are never reasons for anyone to do anything.¹ On this view, facts about goodness obtain in virtue of further facts that are reasons—the facts about the features of objects or states that make them good—but are not ¹ When I speak of “reasons” in this paper, I mean objective normative reasons throughout. So, the claim that P is a reason to ϕ should be understood to mean that P counts in favor of ϕ-ing—regardless of whether anybody is aware of this, and regardless of whether anybody is motivated to ϕ on the grounds that P. (Nonetheless, I think there are important relationships between the idea of an objective normative reason and those of a possessed reason and a motivating reason, which I discuss in Section 8.4.) For brevity I sometimes talk of properties as reasons, but I am happy to make the orthodox assumption that objective normative reasons are facts; all my talk of a property as a reason may be understood as elliptical for the claim that the fact that the property is instantiated is a reason.

168

Zoë Johnson King

themselves reasons. This view is called “buck-passing about goodness.” It gets its name because it “passes the normative buck” from goodness down to the features that goodness consists in.² One can be a buck-passer about other properties, too. In general, to “pass the buck” with respect to a property F is to make two claims about F: first, that something’s being F is not itself a reason for action, and second, that F is instead a status that something has in virtue of our (non-F ) reasons for action. This paper concerns buck-passing about the moral rightness of acts.³ A buck-passer about rightness holds that the fact that an act is (morally) right is never a reason to perform it—nor to do anything else, such as encouraging someone to perform it, resenting someone for failing to perform it, and so on⁴—and that our reasons are instead the act’s right-making features. Buck-passers about rightness typically remain neutral on the question of how many and which features can make acts morally right. Their claim is the metaethical claim that, whatever the right-making features are, it is these features that are our reasons to perform the relevant acts. The moral rightness of the acts is no such thing. I hold that the fact that an act is morally right is a reason to perform it. I take this claim to have considerable intuitive plausibility. Intuitively, the fact that an act is morally right is a consideration that counts in favor of performing it; after all, the fact that an act is morally right surely has some bearing on whether to perform it, and presumably it does not count against performing it! So I think that buck-passers are wrong about this. Nonetheless, I agree with buck-passers that facts about acts’ right-making features are (also) reasons to perform the acts. On the view that I will defend here, we can have our buck and pass it, too. The question of whether an act’s rightness is a reason to perform it is not an idle theoretical question. On the contrary, the claim that rightness cannot be a reason is used as a premise in some important arguments. Most notably, many philosophers hold that there are positive evaluative statuses that an ² Buck-passing about goodness has been the subject of much discussion. For defenses, see e.g., Scanlon (1998), Parfit (2001), Olson (2004), Suikkanen (2004), Stratton-Lake and Hooker (2006), Skorupski (2010); for criticisms, see e.g., Rabinowicz and RønnowRasmussen (2004), Crisp (2005), Väyrynen (2006), Liao (2009), Gregory (2014). ³ For defenses, see especially Dancy (2000), Stratton-Lake (2003); cf. Darwall (2010), Bedke (2011). ⁴ Here I will only discuss an act’s rightness being a reason to perform it, eliding the additional “or to encourage others to perform it, or rebuke them for failing to perform it, or . . . .” This is just to save words and improve readability. There are some (e.g., Darwall 2010) who think that an act’s rightness can be a reason to do certain of these things, but not others. My argument applies to whatever a buck-passer thinks rightness cannot be a reason for.

We Can Have Our Buck and Pass It, Too

169

agent and/or her action can attain only if she acts “for the right reasons.” If one holds such a view, and also assumes that rightness is not a reason—so, a fortiori, it cannot be among the right reasons—it follows that when someone does the right thing because it’s the right thing to do, she and/or her action cannot attain the relevant positive status. Thus buck-passing is used to denigrate motivation by rightness de dicto. For example, Julia Markovits (2010: 207) offers an argument of this form for the claim that people who are motivated by rightness de dicto cannot perform acts with full moral worth, and David Shoemaker (2007: 88) offers an argument of this form for the claim that such people are not full-fledged members of the moral community. If an act’s rightness is in fact a perfectly good reason to perform it, then these arguments are all unsound. That is my primary motivation for writing this paper; if the view that I defend here is correct, it shows that a family of criticisms of people who do the right thing because it’s the right thing to do are all based on unsound arguments.

8. 2. When buck-passers argue that facts about goodness or rightness are not reasons, they typically use a certain type of argument. The argument begins by noting that the lower-order features that make it the case that goodness or rightness is instantiated already seem like reasons; the good-making features of an object or state seem to be perfectly respectable reasons to admire it, acquire it, promote it, protect it, realize it, or perform whatever act or adopt whatever attitude is in question,⁵ and similarly the right-making features of an act seem like perfectly respectable reasons to perform it. Buck-passers then observe that it seems redundant to take facts about goodness or rightness to be additional reasons to do these things, once the facts about the good- or right-making features have already been taken into account. Intuitively, those lower-order facts were already reason enough. The further, higher-order fact that the object or state is good, or the act is right, doesn’t seem to add anything. Since it seems redundant to take these higher-order facts to be reasons, buck-passers conclude that we should not do so. Call this the redundancy argument.

⁵ Some reasons are reasons to adopt attitudes, instead of or as well as being reasons to perform acts. But, for simplicity, I’ll just talk about acts. I intend for everything I say to apply mutatis mutandis to reasons to adopt attitudes.

170

Zoë Johnson King

This style of argument began with buck-passing about goodness. Here is T. M. Scanlon (1998: 97): [T]he natural properties that make a thing good or valuable . . . provide a complete explanation of the reasons we have for reacting in these ways to things that are good or valuable. It is not clear what further work could be done by special reasonproviding properties of goodness and value.

To see Scanlon’s point, imagine the curry again. The idea is that its spiciness, warmth, and nourishing-ness are surely reason enough to eat it, without its goodness being a further reason to eat it. To Scanlon, it seems that there is simply no point in saying that the curry’s goodness is also a reason; this property seems to make no difference to an agent’s normative situation, once the lower-order properties that explain its instantiation have already been taken into account. This is the idea that Scanlon expresses by saying that “it is not clear what further work could be done” by the reasonproviding property of goodness. Scanlon’s critics have pointed out that parallel remarks apply to his own view that an act’s rightness is a reason to perform it, and wrongness a reason not to perform it. Here, for instance, is Philip Stratton-Lake (2003: 75–6): Once Scanlon has identified wrongness with his contractualist principle he must abandon his intuition that moral wrongness is reason-giving. For unless he does, he will be committed to the implausible view that the fact that some act is permitted by a principle that others could reasonably reject provides us with a reason not to do it over and above the reasons others have to reject this principle . . . [So] Scanlon should abandon his view that wrongness is a reason-providing property.

This problem is not unique to Scanlon’s contractualism, as we can see by considering Stratton-Lake’s (2002: 15) positive view: I can see no reason why [we] cannot understand rightness as well as goodness in terms of reasons. [We]⁶ could (and in my view should) embrace not only a buck-passing account of goodness, but also a buck-passing account of rightness. According to such an account, the fact that φ-ing is right is the same as the fact that φ-ing has properties that give us conclusive reason to do it. Similarly, the fact that φ-ing is wrong is the same as the fact that it has properties that give us conclusive reason not to do it.

Stratton-Lake’s thought is as follows. The fact that an act is right is never a brute fact, insusceptible of further explanation. It is rather a fact that obtains in virtue of further features of the act. Since they make the act right, these further features must give us conclusive reason to perform it. But then there ⁶ Scanlon writes “intuitionists” where I am writing “we.” But Scanlon is an intuitionist. The things that he recommends to intuitionists are the things that he thinks are correct, and thus things that we all should do.

We Can Have Our Buck and Pass It, Too

171

is simply no point in saying that the act’s rightness is also a reason to perform it, on top of the reasons provided by the right-making features—by stipulation, those reasons were already conclusive! So the act’s rightness seems to make no difference to an agent’s normative situation; to paraphrase Scanlon, it is not clear what further work could be done by taking rightness to be a reason, since the right-making features are already conclusive reasons. Stratton-Lake thus suggests that we identify the fact that an act is right with the fact that it has (other) properties that give us conclusive reason to do it, rather than seeing it as a further fact that may be a reason in its own right. I have also heard it said that it is not only redundant but positively inappropriate to regard facts about goodness and rightness as reasons, in addition to facts about good- or right-making features; it has been suggested to me in conversation that this is an illegitimate form of double-counting. This thought is surely closely related to the redundancy argument. For present purposes I will construe it as a species of the redundancy argument. To repeat: The redundancy argument works by taking a certain feature, observing that the facts that explain its instantiation seem like reasons, and then suggesting that it seems redundant to take the fact that the feature is instantiated to be a further reason to do the same things. Buck-passers then suggest that, since it seems redundant to take the fact about the higher-order feature to be a reason, we should not do so. This argument dramatically overgeneralizes. We can see this initially by considering the features of acts that buck-passers champion: the rightmaking features. Whatever these features are, it is not plausible that facts about their instantiation are brute facts, insusceptible of further explanation. For example, suppose that an act is morally right because it is fair. The act’s being fair is not then a brute fact, with no further explanation. On the contrary, whenever an act is fair, there is some explanation of why it is fair. The explanation has something to do with the act’s distributing social benefits and burdens on reasonable, non-arbitrary grounds. Here, then, is a fact about the act that makes it fair (a “fair-making” fact, if you like): it distributes social benefits and burdens on reasonable, non-arbitrary grounds. And this fact is not brute, either. It has its own explanation, which spells out what the grounds are. Perhaps the act is meritocratic, or perhaps it makes reparations for past injustice, or perhaps it distributes resources based on need. These are then further facts about further features of the act that make it the case that it distributes social benefits and burdens on reasonable, non-arbitrary grounds. And those facts are not brute, either. So on we might go, spelling out a metaphysical hierarchy of features of acts that ranges from the less to the more fundamental. Parallel remarks hold for any other plausible candidate for being a right-making feature.

172

Zoë Johnson King

Now, consider the fact that an act distributes social benefits and burdens on reasonable, non-arbitrary grounds. Is this a reason to perform it? Intuitively, the answer is “yes.” This sure seems like a consideration that counts in favor of performing the act. But, if this fair-making fact is already a reason to perform the act, then it would be redundant to take the fact that the act is fair to be a further reason to perform it. The fact that it distributes social benefits and burdens on reasonable, non-arbitrary grounds is already reason enough. So the fact that the act is fair is out of the running, according to the redundancy argument; it is the lower-order fact that is the real reason. But wait! Consider the fact that the act is meritocratic, that it makes reparations for past injustice, or that it distributes resources based on need. These facts also seem, intuitively, to be perfectly respectable reasons to perform the act—considerations that count in favor of performing it. But it would be redundant to take the higher-order fact that the act distributes social benefits and burdens on reasonable, non-arbitrary grounds to be a further reason to perform it, once we have conceded that these facts about the grounds themselves are already reasons. So the fair-making fact is out, too, according to the redundancy argument. We spoke too soon in the previous paragraph; it is neither the right-making fact nor the fair-making fact, but actually these even-lower-order facts that are the real reasons. Shall we go on? This points toward a problem for buck-passers about rightness. Buckpassers hope to use the redundancy argument to show both that an act’s rightness is not a reason to perform it and that the right-making features are reasons to perform it. But it will be difficult for them to accomplish both of these aims, because facts about right-making features (such as fairness) can be made to seem redundant equally as easily as facts about rightness. The redundancy argument disqualifies these right-making features from being reasons just as much and in the same way as it disqualifies rightness itself. In other words, buck-passers’ strategy for denying that rightness is a reason undermines their positive claim that the right-making features are reasons. This is already quite bad for the buck-passer. But things get worse, because the redundancy argument overgeneralizes further. It is not restricted to cases involving moral properties (like goodness and rightness). Here is a non-moral example: 5--: In the UK in the early 2000s, there was a public health campaign to get people to eat at least five portions of fruit or vegetables each day. As a result, supermarkets now put stickers on their prepared food that say “1 of your 5-a-day!”, “2 of your 5-a-day!”, etc. Supermarkets produce stickers reporting the number of portions of fruit or vegetables in their food; if a salad contains, say, three portions of vegetables, then it is labeled with a single “3 of your 5-a-day!” sticker

We Can Have Our Buck and Pass It, Too

173

rather than three “1 of your 5-a-day!” stickers. But supermarkets do not produce stickers naming the particular fruits or vegetables in their food. So, a snack pack containing one portion of apple will have a “1 of your 5-a-day!” sticker, rather than a “contains apple!” sticker. The latter are not manufactured.

If the redundancy argument is correct, then British supermarkets’ behavior is quite mysterious. They label a salad containing three portions of vegetables with a single “3 of your 5-a-day” sticker, ostensibly alerting customers to a reason to eat it. But the fact that the salad contains three of your 5-a-day clearly consists in the fact that it contains one of your 5-a-day, and then one more one, and then one more one, since that is what it is to contain three of something. And this lower-order fact is surely a perfectly good reason to eat the salad. So why produce stickers mentioning any number of portions greater than 1? Isn’t this redundant, if we already have multiple one-portion stickers? Moreover, the salad’s containing one of your 5-a-day, and then one more one, and then one more one may itself consist (for example) in its containing one portion each of lettuce, tomato, and cucumber. And its containing these particular vegetables surely counts in favor of eating it. Similarly, the snack pack’s containing apple surely counts in favor of eating it. So why count facts about portion numbers as reasons at all? Or, for that matter, why count facts about the presence of fruits and vegetables as reasons, rather than facts about their nutritional properties? Wouldn’t it be redundant to note that a salad contains tomato if we had already observed that it contains something rich in lycopene? And what should we say about the fact that the salad is rich in nutrients, or the fact that it is healthy? And so on, and so on. Examples like this show that versions of the redundancy argument can be constructed whenever one fact that seems to count in favor of performing some act is wholly or partially constituted by another fact that also seems to count in favor of performing the same act. The redundancy argument, if correct, applies in all these cases too. But some such facts very plausibly are reasons. Indeed, some are such that it is more plausible that they are reasons than that the redundancy argument is correct. So there must be something wrong with the redundancy argument. Redundancy is rampant.

8. 3. Here is what I think is wrong with the redundancy argument. The argument identifies a metaphysical hierarchy wherein one fact that seems to count in favor of performing some act is wholly or partially constituted by further

174

Zoë Johnson King

facts, at least some of which also seem to count in favor of performing the same act. The argument then assumes that, for each such hierarchy, at most one of the facts can be a “real” reason. The metaphor of “buck-passing” unhelpfully encourages this way of thinking. This metaphor conjures up an image of the reason—the “buck”—originally being held by the top fact in a metaphysical hierarchy, and then being passed down to progressively more fundamental facts, until at some point the music stops and one lucky fact is left holding the buck. This suggests that if we examine each hierarchy carefully enough, we will eventually identify the special fact that holds the buck. On this view, identifying genuine reasons is like spotting Waldo in a crowd. But this is silly. We need not locate the normative buck on any particular fact in a metaphysical hierarchy. Rather, we can and should say that the facts in these hierarchies can all be reasons. Distinguish two views: “ ” : In a metaphysical hierarchy wherein some facts that seem to count in favor of performing an act are metaphysically constituted by others that seem to count in favor of performing the same act, there is always one fact that is where the buck stops—a special fact that bears all the normative weight. “  ” : In a metaphysical hierarchy wherein some facts that seem to count in favor of performing an act are metaphysically constituted by others that seem to count in favor of performing the same act, it can be that all of the facts that seem to count in favor of performing the act really do count in favor of performing it. The buck doesn’t stop anywhere. The normative weight is shared by all the facts in the hierarchy rather than resting on some particular fact.

Buck-passers about rightness accept the “special fact” view and hold that facts about right-making features are among the special facts. But, given that the redundancy argument dramatically overgeneralizes, the “share the weight” view is the more attractive option. Faced with metaphysical hierarchies of facts that each seem to count in favor of performing a certain act, we should abandon the project of examining them to hunt for a buck nestling on a special fact. Instead, we should embrace the possibility that most or even all of the facts that seem to count in favor of performing the act really do count in favor of performing it, and thus are genuine reasons to perform it. To clarify, the share the weight view is not the view that each fact in a metaphysical hierarchy possesses 1/n of the normative weight, where n is the number of facts in the hierarchy. On the contrary, it can be that two or more

We Can Have Our Buck and Pass It, Too

175

facts both possess all the weight. This can happen when one fact is wholly constituted by another. For example, the fact that a salad contains three of your 5-a-day wholly consists in the fact that it contains one of your 5-a-day, and then one more one, and then one more one. And, if it turns out that there is just one right-making feature, then the fact that an act is right wholly consists in a single further fact—the fact that it possesses this feature. On the share the weight view, if one fact is wholly constituted by another, then they may possess the same normative weight (and thus may be equally good reasons to perform the relevant act). If the weight that they share is all of the normative weight in their metaphysical hierarchy, then they both possess all the weight. Nonetheless, it does not follow from the share the weight view that every fact in a metaphysical hierarchy always bears all the weight. If a single fact that bears all the weight is jointly constituted by three further facts (specified at a certain level of generality), then they cannot each bear all the weight. Rather, in cases of partial constitution, the lower-order facts might each bear 1/n of the total weight of the higher-order fact that they jointly constitute, where n is the number of facts doing the constituting. This can happen if they make equal constitutive contributions to the higher-order fact’s obtaining. For example, the weight of the fact that a salad contains three of your 5-a-day is presumably borne equally by the facts that it contains lettuce, that it contains tomato, and that it contains cucumber. But it may also happen that some lower-order facts bear more of the total normative weight than others, in virtue of their greater constitutive contribution to the higher-order fact’s obtaining. For example, suppose that the standards for curry evaluation entail that the goodness of curries depends primarily on their spiciness, with their warmth and nourishing-ness making small additional contributions. In this case, the normative weight of the fact that the curry from Section 8.1 is good does not divide equally between the facts that it is spicy, that it is warm, and that it is nourishing. Rather, the fact that the curry is spicy bears more weight than the others—and thus counts more strongly in favor of eating it. Typically, then, as we move down a metaphysical hierarchy, the amount of normative weight borne by each fact is conserved (in cases of full constitution) or lessened (in cases of partial constitution). Nonetheless, there are also cases in which a lower-order fact bears more total normative weight than one of the higher-order facts that consists in it. This is because a single fact can appear in multiple metaphysical hierarchies, each of which bears some normative weight favoring a certain act. For example, suppose that Justin is deciding whether to buy flowers for his wife Emmy. Buying the flowers would make Emmy happy. This makes it the case that buying

176

Zoë Johnson King

the flowers would make someone happy. And that higher-order fact bears some normative weight: in general, the fact that φ-ing would make someone happy is a reason to φ. But this higher-order fact bears less total weight than the lower-order fact that buying the flowers would make Emmy happy. This is because there are other higher-order facts that also consist in the fact that buying the flowers would make Emmy happy, and that bear some weight; notably, there is the fact that buying the flowers would amount to doing a nice thing for one’s partner. The quantity of normative weight favoring doing a nice thing for one’s partner is greater than the amount that simply favors making someone (slightly) happy. And the fact that buying the flowers would make Emmy happy shares in this greater amount of normative weight, occurring slightly lower down in the relevant metaphysical hierarchy. So, the fact that buying the flowers would make Emmy happy bears more total normative weight than the fact that buying the flowers would make someone happy, since the lower-order fact occurs in multiple weightbearing hierarchies.⁷ Here is one more point of clarification. The share the weight view does not entail that every fact in a metaphysical hierarchy must be a reason if any fact in it is a reason. Normative weight can be shared by some facts in a hierarchy, but not others. For example, in addition to making it the case (we are supposing) that an act is right, the fact that an act is fair makes true all disjunctions with the fact that the act is fair as one disjunct: the fact that the act is fair or fluffy, that the act is fair or I am wearing a hat, and so on. But, intuitively, these do not seem like reasons to perform the act. At the other end of the scale, it might be that all metaphysical hierarchies bottom out in facts about the speed and location of fundamental physical particles. But, again, these facts intuitively do not seem like reasons to do anything. Perhaps this is because they aren’t reasons to do anything. That is possible if normative weight is shared by some of the facts in a metaphysical hierarchy, but it “tops off” and “bottoms out” at certain points. This is consistent with the share the weight view, but not entailed by it; questions about precisely which facts share which quantities of weight cannot be settled by the view itself and must be discussed separately. All of this may seem strange, until we remember that parallel phenomena occur throughout metaphysics. For example, no one is surprised to learn that a statue weighing 200lb and a lump of clay weighing 200lb, lain together on a scale, jointly weigh only 200lb. Once we recognize that the statue and clay share their physical weight, in virtue of the relationship of metaphysical constitution between them, no mystery remains. Similarly, ⁷ This example is named after Justin Snedegar, who suggested it to me. Thanks Justin!

We Can Have Our Buck and Pass It, Too

177

nobody is surprised to learn that a whole and the sum of its parts occupy the same physical space. But a single part does not occupy exactly the same space as the whole, just as the fact that a salad contains tomato does not bear exactly the same amount of normative weight as the fact that it contains three of your 5-a-day (modulo the possible contribution of other weightbearing hierarchies—e.g., if someone loves eating tomato and so eating it constitutes doing something she loves as well as getting one of her 5-a-day). Similarly, there is nothing untoward going on when the fact that a red triangle is presented and the fact that a scarlet triangle is presented are both causes of a trained pigeon’s starting to peck (on this see Yablo 1992: 257). Once we see that determinates do not compete with their determinables for causal relevance (Yablo 1992: 257–9; cf. Swanson 2010), we can relax. My view is that, just as facts in a metaphysical hierarchy do not compete for causal relevance, nor do they compete for normative relevance. They can all be reasons.

8. 4. The foregoing discussion of normative weight “topping off” and “bottoming out” with certain facts in a metaphysical hierarchy invites a host of further questions. How far up, and how far down, does normative weight go? Are there any gaps? And how do we tell which facts share in which quantities of weight? I doubt that there are easy answers to these questions. Indeed, I think they are properly understood not as metaethical questions, but as hard questions of first-order ethical theory. These are questions about which facts count in favor of performing which acts, and which of these facts are wholly or partially constituted by which others (such that their normative weight is shared). The task of answering such questions just is the task of doing first-order ethical theory: it is the task of figuring out what things matter morally, what the things that matter morally consist in, and which of these constituents themselves also matter morally. So, I doubt that we can say much to answer these questions while remaining first-order neutral, as I intend to do here. Nonetheless, there are tests we can use to provide defeasible support for the claim that a certain fact is a reason, and thus a fact that bears some normative weight. To repeat the popular refrain: a reason to φ is a consideration that counts in favor of φ-ing. So we can provide defeasible support for the claim that the fact that P is a reason to φ by noting that P seems, intuitively, to count in favor of φ-ing. Exercises that allow us to test our

178

Zoë Johnson King

intuitions as to whether P counts in favor of φ-ing, for some φ, thus help to determine whether P is a fact that bears some normative weight. Here is one such test. Imagine that you initially know nothing about φ-ing. At this point, I assume, you have no reason to φ and no reason not to φ. There may be objective normative reasons for or against your φ-ing— considerations that count for or against it. But reasons that there are aren’t always reasons that you have. The reasons you have to φ are the facts that count in favor of φ-ing to which you have epistemic access. Since you know nothing about φ-ing, there are none of those. Now imagine that your favorite omniscient, omnibenevolent, and trustworthy interlocutor descends from the heavens and informs you that P. Ask yourself: Intuitively, now that you know that P, do you have some reason to φ? If the answer is “yes,” this suggests that P is a reason to φ. We may call this the isolation test, since it asks us to imagine learning a single fact in isolation and seeing whether this seems to change our normative situation. The isolation test is far-fetched; omniscient interlocutors rarely descend from the heavens, so it is unrealistic to imagine that they do. But we can imagine realistic scenarios that are close enough. We can imagine testimony from a real interlocutor, with a good track record, informing us that P. Or we can imagine cases of partial forgetting, in which we forget almost everything about φ-ing, but do recall that P. (Here the “testifier” is our own past self.) So long as it is possible to imagine that someone recalls or is told nothing about φ-ing besides the fact that P, we can ask whether this learning or recollection nonetheless seems to make a difference to her normative situation. These are versions of the isolation test. The isolation test trades on a relationship between normative reasons and possessed reasons (i.e., between the reasons that there are and the reasons that someone has). Another test is based on a relationship between normative reasons and motivating reasons. When we learn of someone’s motivating reason, we can evaluate it in various ways: we can ask whether it was a good reason, and whether the action was reasonable or rational. Here we are looking for a match between motivating and normative reasons. There is a characteristic sense of mismatch accompanying the intuition that someone’s motivating reason was not a genuine normative reason for them to perform the act. For example, we may experience this sense of mismatch if we encounter someone who goes around throwing pencils at people if she dislikes the way they smell, or someone who rushes home to make a cup of tea whenever she sees a rabbit, doing so “because there was a rabbit.” These agents’ motivating reasons are not normative reasons; disliking someone’s smell is not a reason to besiege them with pencils, nor is the presence of a rabbit a reason to make tea. So, we can test our intuitions as to whether the fact that P is a reason to φ by imagining someone who is

We Can Have Our Buck and Pass It, Too

179

motivated to φ by the fact that P and seeing whether this elicits the characteristic sense of mismatch. Call this the motivation test. The isolation and motivation tests can help us to figure out the contours of quantities of normative weight shared by facts in a metaphysical hierarchy. Facts that fail these tests do not bear normative weight—as is the case, for instance, for facts about fundamental physical particles and the disjunctive fact that an act is fair or I am wearing a hat. These would be odd motivating reasons, and do not seem to count in favor of performing the act when learned in isolation. But if a fact passes the isolation and motivation tests, this is (defeasible) evidence for its being a genuine normative reason. And sometimes multiple facts in a metaphysical hierarchy all pass the tests. For example, suppose that someone learns that a certain act is the fair thing to do under her circumstances. She has intuitively learned something that changes her normative situation with respect to this act, and that would be a perfectly fine motivating reason. Moreover, the very same thing holds of her learning that the act distributes social benefits and burdens on reasonable, non-arbitrary grounds; this also changes her normative situation, and would be a perfectly fine motivating reason. These observations suggest that the two facts are both reasons to perform the act, notwithstanding the metaphysical relationship between them. I hold that this is because normative weight is shared by the facts in a metaphysical hierarchy, including these two facts. I submit that the fact that an act is morally right passes the isolation and motivation tests. Here are some real-life examples: P F: Efua is a vegan. The arguments for veganism are complex, requiring hard-to-find empirical information and subtle moral reasoning. But Efua has obtained this information, done this reasoning, and concluded that veganism is morally required, on multiple long dark nights of the soul. Efua has no reason to distrust her past reasoning or her past evidence-gathering, especially since she has gone over the issue many times and reached the same conclusion every time. But Efua cannot always recall all of the subtleties of the moral arguments that she has repeatedly gone over, nor all the empirical facts on whose basis she reached this verdict. Nonetheless, Efua maintains a vegan diet, and she does so because it is morally right. T: Fliers containing racist messages are posted all over Evelyn’s college campus. Evelyn, a first-time teacher, asks a friend with substantial pedagogical expertise and good character whether she should deviate from the syllabus to mention the fliers in class, and what she should say if so. Her friend tells Evelyn that she should ignore the syllabus and use this opportunity to encourage students to discuss strategies that they and their instructors can implement to make the campus a more welcoming place for students of color. Evelyn can tell

180

Zoë Johnson King that her trustworthy friend is well aware of the complex combination of facts that collectively make this the right thing to do. But she doesn’t have time to ask her friend to explain it all. Nonetheless, she performs the intervention that her friend suggests, and does so because it’s the right thing to do.⁸

Evelyn learns and Efua remembers that a certain course of action is morally right, and they each choose to take the relevant course of action on this basis. This is all perfectly fine. Indeed, Efua and Evelyn seem intuitively to be doing a much better job of responding to reasons than the smell-pencil-thrower and the rabbit-tea-drinker—they do not seem unreasonable or irrational, and their actions make perfect sense. The fact that a course of action is morally right therefore passes the motivation test in both cases. Moreover, when Evelyn’s friend tells her that a certain intervention is morally right, this changes Evelyn’s normative situation; she goes from having no reason to perform it to having very strong reason to perform it. Similarly, if Efua forgot all her reasoning concerning veganism, and then suddenly remembered that she had repeatedly concluded that it is morally right, this would change her normative situation. So the fact that an act is right passes the isolation test, too. One clear and simple explanation of these data is that the fact that an act is morally right is a reason to perform it.⁹ Whatever the right-making features are, I assume that they will also pass the isolation and motivation tests. In that case, I think we should say that the fact that an act possesses such a feature and the fact that it is right are both reasons to perform it. We can—and should—have our buck and pass it, too.

8. 5. I have argued that when multiple facts in a metaphysical hierarchy each seem, intuitively, to count in favor of performing a certain act, this can be because they do all count in favor of performing it. They can all be reasons, with shared normative weight. And these metaphysical hierarchies of facts

⁸ This example is structurally similar to the famous “surprise party” example offered by Schroeder (2009), which is used to argue that the fact that there is a reason to do something can itself be a reason to do it. The share-the-weight-view supports the position that Schroeder defends in this paper. ⁹ I grant that this is not the only possible explanation. I will discuss some alternatives in Section 8.6.

We Can Have Our Buck and Pass It, Too

181

with shared normative weight can include the fact that an act is right and the fact that it possesses a certain right-making feature. What, then, of the redundancy argument? On my view, there is a kernel of truth to the redundancy argument. Redundancy intuitions do not tell us which facts can be reasons and which cannot. But they may alert us to metaphysical relationships between reasons. So, these intuitions are not completely useless or misleading. Buck-passers simply drew the wrong conclusion from them. The correct conclusion is that, when two or more reasons share some weight, we should be careful when aggregating them to determine the total amount of normative weight favoring each act, and we should be careful when describing the reasons for performing these acts to another person. In deliberation about what to do, and in evaluation of our own or others’ behavior, we are often interested in determining the total amount of normative weight favoring the performance of an act. This is because we want to know whether there is sufficient reason to perform it, or more reason to perform it than an alternative. When some reasons to perform the act occur in a metaphysical hierarchy with shared weight, then, we must tread carefully. It is this shared weight that we risk double-, triple-, or n-tuple-counting if we do not pay attention to the ways in which it is shared. But that is not because only some facts in the hierarchy are special facts that are the “real” reasons, with the rest being spurious pseudo-reasons. They can all be reasons, and it can nonetheless be the case that we must be careful in determining their aggregate weight, given the ways in which normative weight is shared among them. So, we must do the hard work of figuring out which of the cases described in Section 8.3 obtains: whether normative weight is conserved by full constitution or divided by partial constitution at each level in a hierarchy, how these divisions work, which facts occur in multiple weight-bearing hierarchies, and so on. Since reasons can share their weight, we can’t aggregate them just by adding up their weights. The redundancy argument gets this right. We must also be careful when describing reasons to one another. When we list reasons for doing something, we typically convey that they are metaphysically independent, such that their total weight could be determined by simple addition. If some reasons share some weight, this can be misleading. For example, it is misleading to say, “this salad contains three of your 5-a-day, and it contains lettuce, cucumber, and tomato!”—as if the salad contained six vegetables in total. This would misrepresent the total amount of normative weight favoring eating the salad. But the phenomena here are pragmatic, rather than metaphysical. Reasons that it is infelicitous to mention are still reasons. We can see this by noting that listing multiple facts in a single hierarchy as reasons can be helpful, rather than misleading, if

182

Zoë Johnson King

we use language that clarifies the metaphysical relationships between them. For example, it is fine to say, “this salad contains three of your five-a-day— namely, lettuce, cucumber and tomato!” or, “this salad contains lettuce, cucumber and tomato—so that’s three of your five-a-day!” By using words and phrases like “namely,” “viz.,” “that is to say,” “which makes,” and “so that’s,” we avoid misleading our audience. Far from being infelicitous, this can be a particularly helpful contribution to the conversation, conveying more to our audience about the structure of normative reality than we would by mentioning one reason alone. This suggests that there is another sense in which buck-passers are on to something. They are correct that it can seem redundant to mention the fact that an act is right once the right-making features have all already been taken into account. That is because all of the normative weight of the act’s rightness has thereby already been considered. But this kind of redundancy cuts both ways. Either lower-order or higher-order facts in a metaphysical hierarchy can be made redundant by taking account of other facts in the same hierarchy that share all their weight. For example, it can be redundant to mention the fact that a salad contains tomato if the fact that it contains a vegetable has already been taken into account, and equally redundant to add that the salad contains a vegetable once the presence of tomato has been noted. The conversational constraint that we are under here is a constraint enjoining us to convey the total amount of normative weight favoring the performance of each act, no more, and no less. This might be accomplished by saying that an act is right, or by mentioning its right-making feature/s, or by describing the lower-order features that make it the case that the right-making features are instantiated, or by mentioning a combination of right-making features and lower-order features that together bear all of the weight, and so on. Whichever facts we choose to mention, the others will thereby be made redundant. But this is not because they stop being reasons. Rather, they are reasons whose normative weight has already been taken into account. Features of our conversational context can make it more appropriate to mention certain reasons than others. For example, certain facts in a weightbearing hierarchy can be salient because they were just discussed, or because one party to the conversation is especially interested in them. Or certain reasons can be useful to mention because they are the only facts in a metaphysical hierarchy whose normative significance the audience appreciates, meaning that the audience will underestimate the amount of normative weight favoring an act if other facts are mentioned instead. Or some reasons can be more helpful to mention than others because they hold of fewer of the agent’s available acts, thereby facilitating comparison of the total amounts of normative weight favoring each of them; it is unhelpful to list reasons that are so coarse-grained that they hold of all available acts, thus

We Can Have Our Buck and Pass It, Too

183

failing to distinguish between them. In some contexts, one reason is particularly appropriate to mention because it is the only fact in the relevant hierarchy to which the agent under discussion has epistemic access, and thus the only one that might motivate her to act; in such contexts, mentioning other facts as her reasons can misleadingly suggest that her motivating reason is not a genuine normative reason, when in fact it is a different reason in the same hierarchy. And the list goes on. The features of conversational context that can favor mentioning some reasons over others are as many and varied as features of conversational context usually are. Not all of these factors favor detail over summary. On the contrary, some factors can make it more appropriate to be succinct, focusing on higherorder facts. For example, if a hungry person must choose between a nutritious but boring salad and a delicious yet ludicrously unhealthy chip butty for lunch, then we can adequately convey the difficulty of her decision by saying, “Well, a salad would be three of her 5-a-day, but a chip butty sure is tasty!” Similarly, if a reluctant hero must choose between doing what’s right and doing what’s easy, then we can adequately convey the difficulty of her decision in precisely these terms. This level of detail is suitable for many everyday conversational contexts. So the fact that an act is morally right will often be one of the most conversationally appropriate reasons to mention in a context. And, through our mentioning it, the right-making features will be made redundant. Again, this all might be surprising until we remember that similar phenomena are already well-recognized. It is well-recognized that the normative weight of a set of reasons cannot always be calculated by simple addition of their individual weights, and that context can make it inappropriate to consider certain reasons. For example, there are combinatorial effects between reasons, such that two facts that would each count in favor of an act if they obtained individually do not always do so when they obtain together (on this, see e.g., Horty 2012, Nair 2016). And there are what Joseph Raz (1990) calls “exclusionary” reasons, that is, facts whose obtaining makes the case that we ought not to take certain other reasons into account. And there are what Daniel Fogal (2017) calls “normative clusters,” where a “cluster” can include items of different metaphysical types (e.g., an event and the fact of the event’s occurrence) or facts that count in favor of performing an act taken collectively but not severally (e.g., the facts that there is dancing at a party and that Suwrial enjoys dancing), any of which we might felicitously mention by itself and call a reason. Parallel phenomena to those I have described for facts in a metaphysical hierarchy obtain for facts in a normative cluster. In short, the idea that normative weight is non-additive is not new. But the argument of this paper suggests another, hitherto underexplored, way in which normative weight is non-additive.

184

Zoë Johnson King 8. 6.

Some alternatives to the share the weight view remain on the table. One is the view that the rightness of an act is a reason to perform it, but a derivative one. Derek Parfit held a view like this. Discussing Scanlon’s buckpassing, Parfit (2011: 39) says the following: This view needs, I think, one small revision. If some medicine or book is the best, these facts could be truly claimed to give us reasons to take this medicine, or to read this book. But these would not be further, independent reasons. These reasons would be derivative, since their normative force would derive entirely from the facts that made this medicine or book the best. That is why it would be odd to claim that we had three reasons to take some medicine: reasons that are given by the facts that this medicine is the safest, the most effective, and the best. Since such derivative reasons have no independent normative force, it would be misleading to mention them in such a claim.

I accept the claim that Parfit calls “odd.” But I agree with some things Parfit says here. I agree that it is often misleading to say that there are three reasons to take the medicine, and I agree that this is because doing so elides the fact that the reasons share their normative weight, which may confuse a hapless hearer into thinking that there is more normative weight favoring taking the medicine than there really is. But, unlike Parfit, I do not think that this means that the normative force of the fact that the medicine is best “derive [s] entirely from the facts that made this medicine . . . the best.” This is just the special fact view. And I doubt that the special fact view can be coherently maintained. We can see the problem for Parfit by attempting to identify the special facts in his hierarchy from which normative force supposedly “derives” (whatever this means). Parfit tells us that the fact that a medicine is the best is not one such fact. He seems to think that the special facts are the facts that the medicine is the safest and that it is the most effective. But what about some slightly lower-order facts in which these facts consist, such as the fact that the medicine is the least likely to lead to painful side-effects (which partially constitutes its being the safest), and the fact that it contains a higher dose of the active ingredient than all available alternatives (which partially constitutes its being the most effective)? Aren’t these perfectly good reasons to take it? Or, what about the slightly higher-order conjunctive fact that the medicine is the safest and the most effective? If you want to embark on the project of determining which of these are the special facts from which normative force “derives,” then good luck to you. I do not think that this can be done.

We Can Have Our Buck and Pass It, Too

185

Another possible view is that higher-order facts are not themselves reasons, but each one is evidence that a certain lower-order fact obtains, which is the “real” reason. On this view, the stickers in British supermarkets, Efua’s memory, and Evelyn’s friend’s testimony do not directly inform the agents of a fact that is a reason, but merely provide them with evidence that some fact that is a reason obtains—perhaps some fact about the vegetables in the salad, or their nutritional properties, and perhaps some fact about the right-making features of Efua and Evelyn’s courses of action, or the lowerorder features that constitute these right-making features. As I hope is now clear, this is yet another version of the special fact view. To repeat: I think that the project of hunting for the special fact in each metaphysical hierarchy that is the “real” reason, such that all other facts are at best evidence that the special fact obtains, is hopeless. If you want to think about whether the fact that veganism avoids promoting unnecessary suffering or the fact that it avoids promoting something cruel is Efua’s “real” reason to be vegan, or whether the fact that a salad is healthy or the fact that it contains tomato or the fact that it contains something rich in lycopene is my “real” reason to eat it, then go ahead. But I think that this would be a wild goose chase. A buck-passer might try saying that it is always the most metaphysically fundamental weight-bearing facts in a hierarchy that are the “real” reasons. On this view, each buck passes down to the most fundamental facts that seem intuitively to count in favor of something, which are the special facts. This is a coherent view. But it is an unappealing view. If we push the buck all the way down to the most fundamental weight-bearing facts, then we will push it to facts that are too complicated to be the objects of motivation for ordinary agents. For example, lots of ordinary people are motivated to act justly. But there are surely very few people who are motivated by (a) each person’s having the maximum degree of liberty compatible with equal liberty being granted to all and (b) social and economic inequalities’ being distributed such that they (i) are open to all under fair equality of opportunity and (ii) always benefit the least well-off, with (b) taking lexical priority over (a). Yet the most famous and influential theory of justice (Rawls 1971) says that this is what justice consists in, which entails that the fact that an act would bring about conditions (a) and (b) falls below the fact that it is just in any metaphysical hierarchy. And the fact that an act would bring about conditions (a) and (b) certainly seems to count in favor of performing it. So, if only the most fundamental weight-bearing facts are reasons and anything like Rawls’ view is correct, then someone who wants to perform an act because it is just does not count as performing it for a genuine normative reason. For her motivating reason to be a genuine normative reason, she must be motivated by a fact at least as complicated

186

Zoë Johnson King

as the fact that it brings about conditions (a) and (b). But no ordinary people are motivated by this. Parallel phenomena hold for all other morally significant features—their full analyses would specify conditions that intuitively bear normative weight, but that motivate nobody. On this version of the special fact view, then, we very rarely act for genuine normative reasons. This view gets even less appealing if we recall the connections between normative and motivating reasons that underpin the criticisms of motivation by rightness de dicto mentioned in Section 8.1. Critics of motivation by rightness de dicto hold that someone performs an action with full moral worth (Markovits 2010) and is a full-fledged member of the moral community (Shoemaker 2007) only if she is motivated by facts that are genuine moral reasons. But this entails that, if only the most fundamental weightbearing facts in a hierarchy are genuine moral reasons, then moral worth is extremely difficult to come by, and almost no one is a member of the moral community. So buck-passers’ argument overgeneralizes again—this time with unexpectedly harsh evaluations of all of us. The buck-passer might try to avoid such harsh verdicts by insisting that we are, in fact, motivated by the facts that are the real reasons, unbeknownst to us. But this would be a strange view of motivation. Ordinary agents who seem to act for reasons are frequently completely unaware of the most fundamental weight-bearing facts in the relevant metaphysical hierarchies. We are like Efua or Evelyn, having forgotten or never having known these facts. The buck-passer would therefore be committed to the view that it is possible for someone to be motivated by a fact to which she has no epistemic access. This is bizarre. It is very implausible that someone can be motivated by a fact of which she is completely unaware—not because it is buried deep within her subconscious, but because she has no epistemic access to it. Facts that are completely outside of our heads in this way cannot motivate us. So this strategy seems unpromising. In short, buck-passers’ hunt for the “real,” non-derivative reason in each metaphysical hierarchy leaves us in a pickle. We must draw implausibly sharp lines between facts at very slightly different levels of metaphysical grain, all of which pass the isolation and motivation tests, out of the (misguided) conviction that at most one of them can be a reason. To avoid a sense of arbitrariness in this task, we might try insisting that only the most fundamental weight-bearing facts in each hierarchy are the genuine reasons. But this has the equally implausible implication that very few ordinary agents ever act on genuine reasons—thus denying everyone the special evaluative statuses that buck-passers about rightness wanted to deny to people who do things because they’re morally right. Fortunately, though, we needn’t say any of this. By embracing the view that I have defended in this paper, we can understand how multiple facts in

We Can Have Our Buck and Pass It, Too

187

a metaphysical hierarchy can all be reasons. Rather than passing the buck, we can share it.¹⁰

References Bedke, Matthew (2011). “Passing the Deontic Buck.” In R. Shafer-Landau (ed.), Oxford Studies in Metaethics, Volume 6. Oxford: Oxford University Press, 128–51. Crisp, Roger (2005). “Value, Reasons and the Structure of Justification: How to Avoid Passing the Buck.” Analysis 65 (285): 80–5. Dancy, Jonathan (2000). “Should We Pass the Buck?” In A. O’Hear (ed.), Philosophy: The Good, the True, and the Beautiful. Cambridge: Cambridge University Press, 159–74. Darwall, Stephen (2010). “But It Would Be Wrong.” Social Philosophy and Policy 27 (2): 135–57. Fogal, Daniel (2017). “Reasons, Reason, and Context.” In E. Lord and B. Maguire (eds.), Weighing Reasons. Oxford: Oxford University Press, 74–103. Gregory, Alex (2014). “A Very Good Reason to Reject the Buck-Passing Account.” Australasian Journal of Philosophy 92 (2): 287–303. Horty, John F. (2012). Reasons as Defaults. Oxford: Oxford University Press. Liao, S. M. (2009). “The Buck-Passing Account of Value: Lessons from Crisp.” Philosophical Studies 151 (3): 421–32. Markovits, Julia (2010). “Acting for the Right Reasons.” Philosophical Review 119 (2): 201–42. Nair, Shyam (2016). “How do Reasons Accrue?” In E. Lord and B. Maguire (eds.), Weighing Reasons. Oxford: Oxford University Press, 56–73. Olson, Jonas (2004). “Buck-Passing and the Wrong Kind of Reasons.” Philosophical Quarterly 54 (215): 295–300. Parfit, Derek (2001). “Rationality and Reasons.” In D. Egonsson, J. Josefsson, B. Petersson and T. Rønnow-Rasmussen (eds.), Exploring Practical Philosophy: From Action to Values. Aldershot: Ashgate, 17–39. Parfit, Derek (2011). On What Matters. Oxford: Oxford University Press. Rabinowicz, Wlodek and Rønnow-Rasmussen, T. (2004). “The Strike of the Demon: On Fitting Pro-Attitudes and Value.” Ethics 114 (3): 391–423. Rawls, John (1971). A Theory of Justice. Cambridge, MA: Harvard University Press. Raz, Joseph (1990). Practical Reason and Norms. Princeton, NJ: Princeton University Press. Scanlon, T. M. (1998). What We Owe To Each Other. Cambridge, MA: Harvard University Press. ¹⁰ I am grateful to Matt Bedke, Sarah Buss, Bruno Guindon, Maria Lasonen-Aarnio, Rob Long, Grant Rozeboom, Mark Schroeder, Kenneth Shields, Brian Weatherson, two anonymous referees for this volume, and to audiences at the Vancouver Summer Philosophy Conference 2017, the Wisconsin Metaethics Workshop 2017, and the Central APA 2018, for helpful feedback on earlier versions of this paper.

188

Zoë Johnson King

Schroeder, Mark (2009). “Buck-Passers’ Negative Thesis.” Philosophical Explorations 12: 341–7. Shoemaker, David (2007). “Moral Address, Moral Responsibility, and the Boundaries of the Moral Community.” Ethics 118 (1): 70–108. Skorupski, John (2010). The Domain of Reasons. Oxford: Oxford University Press. Stratton-Lake, Philip (2002). “Introduction.” In P. Stratton-Lake (ed.), Ethical Intuitionism: Re-evaluations. Oxford: Clarendon Press, 1–28. Stratton-Lake, Philip (2003). “Scanlon’s Contractualism and the Redundancy Objection.” Analysis 63 (1): 70–6. Stratton-Lake, Philip and Hooker, Brad (2006). “Scanlon Versus Moore on Goodness.” In T. Horgan and M. Timmons (eds.), Metaethics after Moore. Oxford: Clarendon Press, 149–68. Suikkanen, Jussi (2004). “Reasons and Value—in Defence of the Buck-Passing Account.” Ethical Theory and Moral Practice 7 (5): 513–35. Swanson, Eric (2010). “Lessons from the Context Sensitivity of Causal Talk.” Journal of Philosophy 107 (5): 221–42. Väyrynen, Pekka (2006). “Resisting the Buck-Passing Account of Value.” In R. Shafer-Landau (ed.), Oxford Studies in Metaethics, Volume 1. Oxford: Oxford University Press, 296–324. Yablo, Stephen (1992). “Mental Causation.” The Philosophical Review 101 (2): 245–80.

9 Reason without Reasons For Michael G. Titelbaum

In 1772, Joseph Priestley was invited by Lord Shelburne to become his librarian. Conflicted about whether to leave behind his happy life in Leeds, Priestley wrote to Benjamin Franklin for advice. Franklin replied as follows:¹ In the Affair of so much Importance to you, wherein you ask my Advice, I cannot for want of sufficient Premises, advise you what to determine, but if you please I will tell you how. When these difficult Cases occur, they are difficult chiefly because while we have them under Consideration all the Reasons pro and con are not present to the Mind at the same time; but sometimes one Set present themselves, and at other times another, the first being out of Sight. Hence the various Purposes or Inclinations that alternately prevail, and the Uncertainty that perplexes us. To get over this, my Way is, to divide half a Sheet of Paper by a Line into two Columns, writing over the one Pro, and over the other Con. Then during three or four Days Consideration I put down under the different Heads short Hints of the different Motives that at different Times occur to me for or against the Measure. When I have thus got them all together in one View, I endeavour to estimate their respective Weights; and where I find two, one on each side, that seem equal, I strike them both out: If I find a Reason pro equal to some two Reasons con, I strike out the three. If I judge some two Reasons con equal to some three Reasons pro, I strike out the five; and thus proceeding I find at length where the Ballance lies; and if after a Day or two of farther Consideration nothing new that is of Importance occurs on either side, I come to a Determination accordingly. And tho’ the Weight of Reasons cannot be taken with the Precision of Algebraic Quantities, yet when each is thus considered separately and comparatively, and the whole lies before me, I think I can judge better, and am less likely to take a rash Step; and in fact I have found great Advantage from this kind of Equation, in what may be called Moral or Prudential Algebra. (Franklin 1772/1975, emphases in original)

¹ John Horty (2012: 3) also quotes from Franklin’s letter to characterize much the same view I will be addressing here. Thanks to Roy Sorensen for originally bringing Franklin’s letter to my attention, and to Mark Schroeder for pointing out Horty’s use of it.

190

Michael G. Titelbaum

The assessment of normative situations by tallying up reasons pro and con has attracted increasing philosophical interest over the past few decades. Philosophers have considered such questions as: What makes a particular consideration a reason for or a reason against a particular decision? What determines the weights of such reasons? How do the weights of reasons interact? and so forth. I want to ask a slightly different question, one that I do not believe has been adequately addressed—I want to ask how generally Franklin’s “Algebra” can be applied. Franklin is confident that whatever the details of Priestley’s affairs, Priestley’s dilemma can be resolved by weighing the pros and cons. Can every normative issue that is resolvable be resolved in this fashion? Whenever we set out to make up our minds, will it always be the case that if a determination can properly be reached, it can be reached by tallying up reasons for and against? Or are there cases in which, while the relevant considerations definitely combine to point towards some conclusion, they don’t do so by offering reasons for that conclusion? In Section 9.1 of this chapter, I will specify more precisely how I understand the claim that every normative case is amenable to something like Franklin’s approach. In Section 9.2, I will offer a series of counterexamples to this claim. In Section 9.3, I will consider the philosophical significance of the fact that one may have all-things-considered reason to draw a conclusion without having any individual reasons for doing so.

9. 1.

9.1.1. I will label a normative situation that can be correctly analyzed by something like Franklin’s method a “Franklin case.” This section specifies more precisely what I mean by a Franklin case. To help us keep a fixed example before our minds, here’s a passage from Shyam Nair: Suppose for example that there is a movie theater and a restaurant across town. And suppose that in order to get to that side of town I must cross a bridge that has a $25 toll. The toll is a reason not to cross the bridge. The movie is a reason to cross the bridge and the restaurant is also a reason to cross the bridge. It may be that if there were just the movie to see, it wouldn’t be worth it to pay the toll and if there were just the restaurant, it wouldn’t be worth it to pay the toll. But given that there is both the movie and the restaurant, it is worth it to pay the toll. In this case, the reason to cross the bridge provided by the movie and the reason to cross the bridge provided by the restaurant are individually worse than the reason

Reason without Reasons For

191

to not cross the bridge provided by the toll. But together these reasons are better than the reason to not cross the bridge so that you ought to cross the bridge. More generally, cases like this show that we can have two reasons to do an act that are individually worse than a reason to not do that act, but the reasons together are better than the reason to not do that act. (Nair 2016: 56–7)

This is a canonical example of what I will call a Franklin case. The agent in the example has all-things-considered reason to cross the bridge. This is because the agent’s reasons for crossing (having to do with the movie and the restaurant) together outweigh the agent’s reason against (having to do with the toll). The pros outweigh the cons, so the agent ought to cross. In Nair’s example, a number of considerations bear on the agent’s decision: that it costs $25 to cross the bridge, that the movie theater is across town, that the restaurant is across town. Going forward, I will assume that such normatively relevant considerations are always facts. Here I follow the trend in the metaethical literature of assuming that reasons are facts. Yet I do so as a convenience; I don’t think anything important in my argument hangs on this assumption. I will label an example a Franklin case when it satisfies three conditions. The first is: (1) If the agent in the case ought to do something (perform an action, form a belief, etc.), then some set of facts together provides that agent with all-things-considered reason to do so.

To use a bit of contemporary jargon, we might say that Franklin cases instantiate a mild, highly general form of “rationalism.” Why mild? Because rationalism is usually a thesis about reasons in particular; moral rationalism, for instance, is the thesis that moral oughts always entail the existence of reasons. Condition 1 merely says that when an agent ought to do something, there is some set of facts that together normatively support the agent’s doing so. It says nothing about the nature of the facts in the set, or what about them allows them to provide such support. What kinds of thing might be required by an ought? I am interested in cases of both practical and theoretical support, and it would be convenient to be able to discuss both kinds of case with a unified vocabulary. So (perhaps somewhat artificially) I will talk about a set of facts’ together requiring an agent to “conclude C,” where C is some proposition. In theoretical cases, to conclude C will simply be to believe C. In practical reason cases, to conclude C might be to form an intention with the content C. Or it might be to act in the manner described by C. (For example, in Nair’s case C is the proposition that I cross the bridge.) I realize it’s controversial what sorts of entities are supported by reasons; hopefully all of the available approaches can be shoehorned into this way of talking. And again,

192

Michael G. Titelbaum

I don’t think anything important in my argument depends on the ontology of the conclusions of reasoning. The next necessary condition for a Franklin case is: (2) If in the case a particular set of facts provides at least some allthings-considered reason to conclude C, then it does so because of local relations among individual facts within the set and C. These local relations make it the case that the set provides all-things-considered reason to conclude C, and the local relations explain why the set provides all-things-considered reason.

It’s important to see that this condition goes beyond what’s required by Condition 1. Many philosophers would endorse the kind of mild rationalism that takes all normative cases to satisfy Condition 1; they would grant that oughts are always accompanied by supporting sets of facts. But Condition 2 requires that support to be structured in a very particular way: it has to flow bottom-up from local relations among those facts.² Contrast this structure with Jonathan Way’s account of reasons. Way maintains that For the fact that p to be a reason for S to ϕ is for there to be a good pattern of reasoning from the belief that p, perhaps together with other correct attitudes which S has, to ϕ-ing. (Way 2017: 254)

Notice that on this view, for p to be a reason to ϕ is for it to factor in a good pattern of reasoning in a particular way. More generally, the facts in a set count as reasons for a conclusion C just in case—and because— those facts can act as premises in a piece of good reasoning to C. Way is explicit that his view “takes the correctness of attitudes and the goodness of reasoning to be prior to reasons” (Way 2017: 254).³ Thus he has a top-down approach: The facts in the set are reasons for C because the set has some global property that can be independently understood. In a Franklin case, the determination works the other way; the set provides all-things-considered support for C because it contains reasons for C. This is clearly the way we’re meant to read Nair’s case; the movie and the

² Compare Dancy (2004: 16): “To talk of what there is overall reason to do (and note that ‘reason’ in this phrase is not a count noun) is to talk about where the contributory reasons come down—on this side or that.” ³ Kieran Setiya (2014: 228) also affirms that “A reason to ϕ counts in favour of doing so by contributing a premise for sound reasoning to the desire or motivation to ϕ.” But Setiya remains explicitly non-committal about whether the reason counts as a reason because it is a premise in a piece of sound reasoning, or the reasoning is sound because its premises count as reasons. (See Setiya 2007: 13.)

Reason without Reasons For

193

restaurant provide reasons to cross the bridge that make it worth paying the toll.⁴ The final condition for being a Franklin case is: (3) In the case, the local relations among the facts and C that make the set provide all-things-considered reason to conclude C all belong to one of the following types: • Qualitative, comparative, or quantitative reason for or against relations; • Being a condition on a fact in the set; • Being a modifier on a fact in the set.

These relation-types require some explaining. First, the reason for and against relations: Here I’ve borrowed Carnap’s famous (1950) tripartite division of concepts. The qualitative relations in question are being a reason for C and being a reason against. (These are sometimes called pro tanto reason relations.) For example, that the restaurant is on the other side of the bridge is a reason for crossing the bridge. Comparative relations include one fact’s being a stronger reason for or against C than another. For example, Nair suggests that the amount of the toll is a stronger reason against crossing than the restaurant’s location is a reason for. Finally, quantitative relations say that a particular fact supports C to a particular degree; the fact has a specific weight as a reason for or against C. These weights may be combined, as when Nair suggests that the movie and restaurant locations together outweigh the reason against crossing provided by the toll. What does it take for a fact to be a reason for or against some conclusion, for one fact to be a stronger reason than another, or for a fact to provide a reason of a particular weight? I am not going to provide any such analysis here (though I will mention one possibility later). I will assume only that these relations can be functionally characterized,⁵ and that they obey particular platitudes. For example: • Whenever a set of facts provides all-things-considered reason to conclude C, at least one of the facts in the set is a reason for C.

⁴ On either a top-down or a bottom-up view, the facts in a set together provide reason to draw the conclusion; they make it the case that the agent ought to conclude C. That’s what Condition 1 is about. Condition 2, and the top-down/bottom-up distinction, is about what makes it the case that the set has that global property—why it is that those facts make it the case that the agent ought to conclude C. In a Franklin case, the set provides reason to conclude C because the facts in that set stand in particular local relations to each other and to C. ⁵ Not necessarily defined or reduced—just characterized!

194

Michael G. Titelbaum

• If a set of facts contains at least one reason for C and no reasons against C, then the set provides at least some all-things-considered reason to conclude C.⁶ • Combining two reasons for C produces a weight in favor of C no less than that of either reason taken alone.⁷ The “condition” and “modifier” terminologies come from Bader (2016); other authors employ similar notions under different names.⁸ A condition is a fact that, while perhaps not itself a reason for or against C, affects whether some fact counts as a reason for or against. For instance, the fact that the agent enjoys movies is a condition making the location of the movie a reason to cross the bridge. A modifier affects the weight of a reason for or against C. For instance, if the agent has ten billion dollars in the bank, this may weaken the $25 toll as a reason against crossing the bridge.⁹ I will continue to describe reasons for and against C, conditions on those reasons, and modifiers on them, as being among the facts that “bear on” C. It might seem more natural to describe all these kinds of facts as “reasons” relevant to C. But the word “reason” is used in two different ways in the literature. Some authors define “reason” so that a reason must always be for or against, as when Setiya (2007: 9) writes, “ ‘Reasons,’ here, are considerations that count in favour.” Other authors use “reason” more expansively, so that conditions and modifiers also count as reasons.¹⁰ To avoid having to ⁶ Compare Berker (2007: 125): “According to one common conception, a reason for action is a consideration that would decisively count in favor of a given action were no other reasons present.” ⁷ Again, Berker (2007: 130): “The individual contribution made by a reason of positive valence always positively affects the total reason in favor of the action in question.” ⁸ For more on conditions and modifiers, see Lord and Maguire (2016). Dancy (2004) draws similar distinctions using different terminology—our “conditions” would include his “enablers” and “disablers,” while our “modifiers” would include his “intensifiers” and “attenuators.” ⁹ Franklin’s description of his method doesn’t seem to allow for conditions and modifiers, so the “Franklin case” moniker is a bit of an anachronism. I include cases with conditions and modifiers in the class under consideration both to bring our discussion in line with the contemporary reasons literature, and to make it at least somewhat plausible that the class might be fully general. If Franklin cases were not allowed to include conditions and modifiers, their claim to full generality would fall quickly to Shelly Kagan’s (1988) arguments. While Kagan’s piece is a clear precursor to and inspiration for my efforts here, contemporary reasons theorists have blunted the force of Kagan’s arguments by working conditions and modifiers into their view. (Thanks to Andrew Kernohan for bringing Kagan’s article to my attention.) ¹⁰ One example is Scanlon (1998: 51): “Reasons can be related to one another in more complex ways. I may, for example, judge one consideration, C, to be a reason for taking another consideration, D, not to be relevant to my decision whether or not to pursue a certain line of action.”

Reason without Reasons For

195

settle this issue, I will describe all the facts in a set that normatively supports concluding C as facts that “bear on” C.¹¹ Condition 3 says that in a Franklin case, all of the facts bearing on C are reasons for or against, conditions, or modifiers. Together, Conditions 1 through 3 entail that in any Franklin case in which an agent ought to conclude C, there exists at least one fact that is a reason for C (perhaps in the context of other facts acting as conditions or modifiers). To deny that all cases involving normative support are Franklin cases, I will provide examples in which an agent ought to conclude C even though no fact is a reason for C.¹²

9.1.2. Before I do so, however, a few more clarifications. Consider the following six facts: A movie is playing across town. A restaurant is across town. The bridge to cross town has a $25 toll. Highways 101 and 280 would each get me across the bridge. 101 has more interesting sights than 280. 280 has less traffic than 101, and is faster.

Suppose we say that this collection of facts together supports concluding C, that I take Highway 280. I want to count this example as a Franklin case. However, its bottom-up structure is a bit more complicated than a straightforward reading of Conditions 1 through 3 would allow. The first three facts in the collection combine to support the intermediate conclusion that I cross the bridge. The mandate to cross the bridge then combines with the final three facts to recommend Highway 280. While it’s debatable whether any of these six facts counts by itself as a reason for taking 280, the example nevertheless seems to me firmly in the bottom-up spirit of a Franklin case. We have six facts bearing on C. The first In Chapter 10 in this volume, Chris Howard employs both characterizations of reasons: He first writes that “Normative reasons count in favor of what they’re reasons for.” (p. 217) A bit later he writes, “Reasons for an attitude are facts that are relevant to determining whether you ought to have that attitude, all-things-considered.” (See pp. 218.) ¹¹ We could also describe these facts as “relevant to” C. Yet I will try to avoid “relevance” talk because it has a technical usage, which I will discuss later. ¹² Kearns and Star (2008: 51–2) write: “The following is a very plausible principle to many people: If one ought to F, then there is a reason for one to F. It seems impossible for it to be the case that one ought to perform an action if there is no reason at all to do so” (emphasis in original). I intend to demonstrate the possibility of what seems to them impossible.

196

Michael G. Titelbaum

three of them offer reasons for and against an intermediate conclusion, ultimately combining to support that intermediate conclusion. When the intermediate conclusion is added to the original six, the resulting set provides all-things-considered reason to conclude C in the traditional forand-against way. Thus there’s an explanation of why the set of six facts together provides reason to conclude C, and that explanation works with the local relation-types delineated by Condition 3. It’s just that some of those local relations are used to establish the intermediate conclusion, rather than contributing directly towards the ultimate conclusion C. If we wanted, we could modify our three conditions to explicitly allow for this kind of recursive structure. I hope the general idea is clear enough that I don’t have to work through the details. I do, however, want to emphasize one point about intermediate conclusions: For an example involving an intermediate conclusion to count as a Franklin case, it must be possible to establish that intermediate conclusion in Franklinian fashion. Franklin reasoning always begins by identifying some fact (perhaps in the context of others) as a reason for some conclusion, even if that conclusion isn’t the ultimate conclusion sought after. Which brings me to the question of how normative support relates to reasoning. I believe that if a set of facts all-things-considered supports drawing a conclusion, and does so by virtue of local relations among those facts, then it ought to be possible for an agent to reason from those facts to that conclusion by recognizing those relations (and perhaps writing them down to keep track, as Franklin suggests). This need not be the only way to reason from those facts to that conclusion, and it need not even be the best way in some broad sense. We’re all familiar with the Franklinian image of making a decision by “weighing up the pros and cons,” but it’s unclear how often we actually do that in our heads. (In fact, I started thinking about the topic of this chapter when I noticed that it almost never seems to me like I’m making a decision that way.) Psychologists study reasoning heuristics that skip over important normative details, but confer advantages in speed or reduced cognitive load. They also repeatedly warn against reading the structure of reasoning off of what seems to be happening while one’s doing it.¹³ Among other things, we sometimes reason so quickly and automatically that there’s nothing the process “seems like” at all. So I am not interested in classifying cases according to how the agent’s reasoning in the case actually goes, or seems to go (even if we confine our attention to cases in which the agent is reasoning well). Instead, I am

¹³ See, for example, Nisbett and Wilson (1977).

Reason without Reasons For

197

interested in how the reasoning in a case could go:¹⁴ In every Franklin case in which an agent considers a set of facts, and (correctly) determines that they together provide all-things-considered reason to draw some conclusion, it should be possible for that agent to slow down, work explicitly through a series of correct reasoning steps, and tell a bottom-up story about how that all-things-considered feature of the set arises from individual reasons for and against the conclusion within the set. As Jay Wallace (2012: 19) puts it: “We resolve the question of what we are to think or to do by . . . reflecting on the considerations that count in favor of holding certain beliefs or acting in certain ways.” Finally, while certain kinds of holists about normativity will immediately reject the suggestion that all (or even any) cases of normative support are Franklin cases, I am not a holist. I believe there is a bottom-up, atomistic, compositional process by which facts combine to support drawing conclusions.¹⁵ That is, I believe that every case of normative support satisfies Conditions 1 and 2 of the definition of a Franklin case. My issue is with Condition 3. I think that the ideology of reasons, conditions, and modifiers misdescribes the functional structure of normative support in many cases. That’s what the counterexamples are meant to illustrate.

9. 2.

9.2.1. My first example is inspired by one from Richard C. Jeffrey’s (1983) Logic of Decision. You have been invited to a dinner party, and tasked with bringing the wine. You must choose among a red, a white, and a rosé. Before you arrive, the hosts will choose between two entrées, steak or a delicate fish. The pairing of your wine with their entrée yields the following possible utility payoffs: utilities red white rosé

steak 5 10 1

delicate fish 10 5 1

¹⁴ Compare Silverstein (2016: 2, n. 5). ¹⁵ I also tend to think that this compositional process is systematic, and might be expressible in general principles. So I tend to reject particularism. But I will not address that issue in this chapter.

198

Michael G. Titelbaum

You are equally confident that the hosts will serve steak or fish. Which wine should you bring? When I reason about this case, I conclude that you should bring the rosé, on decision-theoretic grounds. Nowhere in applying the decision theory do I search for reasons for or against bringing rosé among the stated facts of the case. While decision theory offers a machinery for composing sets of facts into normative conclusions, reasons for are no part of that machinery.¹⁶ As I noted above, one need not accept my reasoning, nor my reports about it, at face value. First, whenever a conclusion is drawn, multiple ways of reasoning to it are available; just because one way of reasoning to rosé contains no reasons for or against doesn’t mean there aren’t other ways of reasoning that do. Or second, just because I don’t explicitly think about reasons for or against while doing decision theory, that doesn’t mean they aren’t in there somewhere. All right, then: Of the facts provided in the story above, which could be considered a reason for bringing rosé? Is it that rosé yields an expected utility of 1 when steak is served? But red yields an even higher utility in that case! Similarly for the utility fact about rosé and fish. What about the fact that you divide equal confidence between the two possible entrées? Well, that doesn’t seem to be a reason to bring one wine rather than another . . . . So it’s not obvious which fact could be a reason for rosé, and this example doesn’t seem to be a Franklin case. But there are maneuvers available that might bring this example back into the fold. First, one could simply insist that there must be a reason for rosé in there somewhere. Call this the FootStomping accommodation strategy. Perhaps the equal confidence split is in fact a reason to bring rosé. If that were the only fact in the story it would not act as a reason for rosé, but one could try to spin a yarn about how the other facts act as conditions and modifiers to make it so. (Maybe there’s something in there about radical entrée uncertainty making it best to bring the safest wine.) A second accommodation strategy would be Reasons Contrastivism. (See especially Snedegar 2015.) On a contrastive view, a fact is a reason for a particular conclusion only relative to specified other options. The contrastivist might say that the fact that rosé yields a higher utility than white when steak is served is a reason for bringing rosé rather than white in case of steak. Similarly, the fact that rosé yields a higher utility than red when fish is served

¹⁶ Dietrich and List (2013: 105) note that “reasons are largely absent from modern rational choice theory.” They then propose a reason-based theory of rational choice, but their theory largely uses reasons for and against to determine an agent’s utility distribution over possible worlds. When it comes time to combine utilities and probabilities, Dietrich and List do so in the traditional fashion, which does not count either the utility facts or the credence facts as reasons for performing a particular act. (Thanks to Reuben Stern for bringing the Dietrich and List proposal to my attention.)

Reason without Reasons For

199

is a reason for bringing rosé rather than red. Now the constrastivist still has some work to do: We need to combine these reasons that obtain only on particular entrée selections into an overall decision in light of entrée ignorance. But at least the contrastivist has identified some of the facts in our story as reasons (in some sense) to bring rosé. Finally, there’s the Summary Fact accommodation strategy. The idea here is to identify some other fact in the vicinity that summarizes the relevant property of the facts presented in the story, then treat that summary fact as a reason for the conclusion (even if the other facts aren’t). For instance, one could suggest that the fact that bringing rosé maximizes expected utility is your reason for bringing rosé.¹⁷ The Summary Fact strategy will often be tempting in the examples to come. It suffers from one universal problem, and one problem particular to this case. The universal problem is that we will always need to know how the agent is supposed to reason to the summary fact. A summary fact is meant to operate as an intermediate conclusion, of the type I suggested modifying the Franklin case definition to incorporate. But it was crucial to that modification that the intermediate conclusion be supported by some subset of the original facts in the case, in the traditional Franklinan fashion. In the present example, you should be able to take the facts given in the case, reason to the conclusion that rosé maximizes expected utility, and then treat that fact as your reason to bring the rosé. But this pushes the problem back to the reasoning yielding the conclusion that rosé maximizes expected utility. If this truly is a Franklin case, then the utility calculations that get you the intermediate conclusion should operate with reasons for and against. And again, I don’t see which of the facts outlined in the original story is a reason for the conclusion that rosé maximizes expected utility. A version of this problem will come up every time we consider the Summary Fact strategy below. However, the Summary Fact approach also has a particular problem in this example, having to do with how we understand utility in the case. Sometimes utility is understood as a natural entity or property, which our actions may produce in greater or lesser quantities. (Canonical examples include pleasure, happiness, etc.) If we understand utility this way, then even once an agent has determined what maximizes utility (or expected utility), there can always be a further question whether utility is something she has all-things-considered reason to maximize. I take it that on the proposed interpretation of the wine case, the agent calculates that rosé maximizes expected utility, then some other facts in the vicinity make it the case that she has reason to choose the act which maximizes expected utility. These further facts are then conditions which

¹⁷ Thanks to Eric Sampson, among others, for this suggestion.

200

Michael G. Titelbaum

convert the fact that bringing rosé would maximize expected utility into a reason for bringing rosé. Yet that’s not how I meant utility to be understood in setting up the case. I meant utility to be a direct measure of value—genuinely normative value— not a measure of some natural product that might or might not be valuable.¹⁸ You are not meant in the example to first determine what would maximize expected utility, then ask the further question what you have all-things-considered reason to do. The expected utility calculation is how you determine whether the facts in the story give you all-thingsconsidered reason to bring rosé. On this reading of the utilities—a reading on which doing decision theory just is a form of practical reasoning—it’s much more difficult to get the proposed Summary Fact response going.¹⁹

9.2.2. While I doubt the Summary Fact strategy can be made to work for the decision-theoretic example, the two other strategies might be able to recast that example as a Franklin case. So let’s move on to a second example, in which neither of those strategies fares as well. In the days leading up to the 2016 American presidential election, the RealClearPolitics website offered daily averages of the most recent polls. At 9:30 a.m. on the morning of November 7, they reported an average lead for Hillary Clinton of 2.5 percent. They also reported the individual poll results that factored into their average; for instance, a CBS News/New York Times poll showed Clinton leading by 4 percent. While some of the individual polls were closer to 2.5 percent than that, and some were farther, none of the polls reported an exactly 2.5 percent Clinton lead. The folks at RealClearPolitics began with a set of facts about the results of individual polls, and reasoned their way to a conclusion of what to report. So which of those facts was a reason for the conclusion that they should report an average of exactly 2.5 percent? The CBS/NYT result clearly bears on that conclusion, but is a 4 percent poll a reason to report a 2.5 percent average? Was any of the individual poll results a reason to report that average? ¹⁸ This is how many contemporary utilitarians in ethics interpret the utilities appearing in their theories. ¹⁹ An anonymous referee for this volume suggests that on the Summary Fact strategy utility could be read as a genuinely normative property, just a different kind of normative property from reasons. We could then move from facts about one kind of normative property (utility) to another (reasons). But if we do that, then we’re no longer offering a Franklinian explanation of the given facts’ providing all-things-considered reason to bring rosé. Our explanation for that support now bottoms out in local normative relations having to do with utility rather than reasons, and so fails to satisfy Condition 3 for a Franklin case.

Reason without Reasons For

201

Here our accommodation strategies from above look hopeless: We can’t pick one of the poll results and argue that that one (in the context of all the others) is the reason to report an average of 2.5 percent. Which one would we pick? We could try going contrastivist again. Perhaps a 4 percent poll result is a reason to report an average of 2.5 percent rather than, say, an average of 50 percent. But if that’s right,²⁰ then a 4 percent result is also a reason to report an average of 3.9 percent rather than 2.5 percent. And now all the interesting, important compositional work is going to be done by whatever story combines these reasons relative to pairs of possible averages into a selection amongst the full set of averages available.²¹ Perhaps we have found some (contrastive) reasons for reporting 2.5 percent in the mix, but they look awfully trivial and inconsequential. Is a summary fact available? Perhaps RealClearPolitics’ reason for reporting an average Clinton lead of 2.5 percent is that, given all the individual poll results, the mathematically-calculated mean is 2.5 percent. This mathematical fact supplies a reason for reporting 2.5 percent (in light of various other reasons against misleading the American people with respect to basic mathematics). But now we have the same universal problem as the first time we considered the Summary Fact strategy: Shift focus to the intermediate conclusion. Before the RCP folks reasoned to the intention or action of reporting an average of 2.5 percent, they had to reason from the individual polls to a belief that the average at that time was 2.5 percent. That is, they had to do a bit of theoretical reasoning in order to do their practical reasoning. The conclusion of that theoretical reasoning was the intermediate conclusion that the current polling average is 2.5 percent, and the individual poll results were the relevant facts bearing on that conclusion. So which polling fact was the reason for that conclusion? Here it won’t do to invoke the summary fact that the current polling average is 2.5 percent as a reason for that conclusion—we cannot reason from a fact to itself ! Yet a new accommodation strategy is available in this case.²² Suppose that prior to receiving any of the polling data, the RCP folks had some opinions about what the average polling gap between Clinton and Trump might be. Perhaps they even went so far as to assign degrees of confidence across all the available numerical values. When the 4 percent CBS News result comes in, it might increase their confidence that the ultimate polling average will be

²⁰ And I’m not entirely sure that it is. Depending on the other polling results in the collection, why couldn’t a 4 percent result be a stronger reason to report an average of 50 percent than 2.5 percent? ²¹ In Snedegar’s (2015) terminology, the reasons relative to normative questions Q containing only two alternatives have to be combined into an ought relative to the question Q containing all the averages RealClearPolitics might report. ²² Thanks to Sophie Horowitz for this suggestion.

202

Michael G. Titelbaum

2.5 percent. That doesn’t mean that their confidence in a 2.5 percent average given a single 4 percent result need be very high. But it will have increased, while other possible averages (such as a 2.5 percent Clinton deficit) will have decreased. Focusing on the deviation the CBS News result causes from the pollsters’ default confidence in an average of 2.5 percent, the 4 percent result counts as a reason in favor of reporting 2.5 percent. This Default-Comparison strategy is also appealing because it meshes with theories of reasons available in the literature. To this point I have relied on intuitive judgments about whether a particular fact counts as a reason for a particular conclusion. For theorists who take reasons to be basic, there may not be too much more to say. But some philosophers have offered detailed theories about when a fact counts as a reason for a conclusion. Many of these theories (such as those of Finlay 2006, Schroeder 2007, and Kearns and Star 2009) involve confidence- or probability-raising of some sort. They differ on precisely what proposition needs its probability raised, and raised relative to what. But the general theme that a reason raises the probability of some proposition relative to a default value is there. So it’s a strategy worth considering as we go forward.

9.2.3. Let’s test that strategy on another case. Suppose that initially I know nothing about Tiresias. Then I learn the following facts: All men are mortal, and all women are immortal. Only women serve Hera. Tiresias serves Hera.

From these three facts I reason to the conclusion that Tiresias is immortal. Which of the facts I learned is a reason for that conclusion? Take a minute and consider each of those facts one at a time. Does any of them look like a reason to conclude that Tiresias is immortal? This certainly doesn’t seem to be a Franklin case. The Foot-Stomper will grant that of course none of the facts above is a reason on its own for the conclusion that Tiresias is immortal—the reason for that conclusion only becomes a reason for in light of the other two. But which one is that one? Can the Foot-Stomper provide any non-arbitrary grounds for designating one as the reason and the others as conditions or modifiers? In case he thinks he can, consider these three facts: 1. (A & –B & C) ∨ (A & –B & –C) ∨ (–A & B & C) ∨ (–A & B & –C) 2. (A & B & –C) ∨ (A & –B & –C) ∨ (–A & B & C) ∨ (–A & –B & C) 3. (A & B & C) ∨ (A & –B & –C) ∨ (–A & B & –C) ∨ (–A & –B & C)

Reason without Reasons For

203

An agent wondering whether proposition A is true ought, in light of these three facts, conclude that it is. Now which of these facts is the reason for A, made such by the presence of the other two? The lesson is the same for both the Tiresias example and this more abstract one: When multiple premises jointly entail a conclusion, none of them needs be a reason for that conclusion. I think it’s with these logical cases that the accommodation strategies we’ve tried so far begin to look truly desperate. There’s no non-arbitrary way to argue for one of the premises as a reason for and the others as conditions or modifiers. Contrastivism is also useless here: In the Tiresias deduction, the clear contrast is between the conclusion that Tiresias is immortal and the conclusion that Tiresias is mortal; in the more abstract deduction it’s between A and –A. In neither case does one of the premises look like a reason for the former conclusion over the latter. What about the Summary Fact strategy? In each case we could take the fact that the conclusion is logically entailed by the premises as a summary fact providing reason for the ultimate conclusion. But then we have the same problem the Summary Fact strategy always generates: We need to know how the compositional reasoning from the premises to the summary fact goes, and which of the premises is a reason for that fact. In deductive cases all the really interesting reasoning happens when the agent determines that the conclusion is entailed; the Summary Fact response needs something to say about that reasoning. Which brings us back to the Default-Comparison strategy. We can’t apply that strategy to these examples yet, because we don’t know what the default confidences/probabilities are supposed to be going into our deductions. So let’s make some reasonable ones up. In the more abstract deduction case, each of the conjunctive disjuncts appearing in the premises is what’s known as a “state-description.” (In a language with only the three atomic sentences A, B, and C, these state-descriptions give the fullest description available of the state of a possible world.) With three atomic sentences there are eight possible state-descriptions. So let’s suppose that before any of the premises is given, each of the eight state-descriptions (each possible arrangement of the world in terms of those atomic sentences) is equally probable. In that case, the default probability of A is 1/2. With a bit of calculation, you should be able to determine that none of the premises in the argument raises the probability of A. That is, if you start with no information, and take any one of those premises by itself, the probability of A in light of that premise is still 1/2. (In fact, that remains true of any pair of premises within the set; it’s only once all three premises are added that the probability of A increases above 1/2.) So even if we think of a reason for a conclusion as a fact that raises the probability of that conclusion above some default value, none of the premises in the abstract deductive argument is a reason for that argument’s conclusion.

Michael G. Titelbaum

204

9.2.4. This last effect may be easier to grasp in an example of inductive theoretical reasoning.²³ Consider the twenty-four shapes in Figure 9.1. Each is either large or small, black or white, a circle or a square. Suppose I tell you that I have selected one of them at random, with each shape having an equal chance of being selected. I then provide you with the following facts: My shape is large. My shape is a square. My shape is white.

I ask you to consider whether my shape is to the left or the right of the dividing line. (Feel free to think about it for a moment, and do some counting.) I submit that taken together, those three facts give you some all-thingsconsidered reason to conclude that my shape is to the left of the line. The reason is not conclusive, but it is there. Now if this example is a Franklin case, at least one of the three facts must be a reason for the left-of-line conclusion, either considered in isolation or in light of the other two. So which one(s) is it?²⁴

Figure 9.1

²³ Inductive examples may also be preferable to deductive because logical entailments are sometimes viewed as special limit cases of more commonplace reasoning practices. To the extent I understand the legal aphorism, developing a theory of reasoning by attending primarily to deductive cases may be a situation in which “hard cases make bad law.” ²⁴ I find that whenever I present a statistical example, some readers have a tendency to fight against the example (perhaps suggesting that there isn’t really any all-thingsconsidered support in this case). Rest assured that if I wanted, I could tweak the numbers

Reason without Reasons For

205

Let’s work through our accommodation strategies in turn. First, there doesn’t seem to be any way to single out one of the facts as the genuine reason for the conclusion in the context of the other two. The three facts are perfectly symmetrically situated relative to the conclusion that my shape is on the left-hand side. Second, a contrastive approach will not help, as we are already considering a clear contrast of conclusions between left and right. Third, the relevant summary fact in the vicinity (that the set raises the probability of the conclusion) can be dismissed for reasons we’ve already covered. Fourth and finally, what about comparison to defaults? Given the background information about how I selected my shape, each shape should clearly start with a probability of 1/24. Since there are twelve shapes to the left of the line, the left conclusion has a default confidence of 1/2. Now take any of the three facts above and add it to your evidence. Your confidence in left should remain at 1/2. (For example, twelve of the shapes are squares, and six squares lie left of the line.) None of the three facts increases the probability of left above its initial value. So on a probability-raising theory of reasons, none of those facts is a reason for the conclusion. In fact, to make matters worse, no conjunction of two of the facts raises the probability of left above 1/2; put another way, assuming one of the facts doesn’t make another of them a probability-raiser. (For example, the number of large squares is the same on each side of the line.) The probability that my shape is left of the line goes up to 2/3 only once all three facts are incorporated. To sum up: In this example there are no grounds to claim that any of the facts provided is a reason for the conclusion that my shape lies to the left of the line. Yet together those facts provide all-things-considered reason to draw that conclusion. So the example, while clearly an instance of normative support, is not a Franklin case.

9.2.5. I would like to pause for a moment to relate the shapes example to some points I made earlier, and to some notions common in discussions of inductive reasoning. In Section 9.1.1, I listed some platitudes that come along with a functional picture of reasons for and against. The first platitude was that whenever a set of facts provides all-things-considered reason to conclude C, at least one of the here so that the ratio between large white squares on the left and right sides is as high as you like. I could also arbitrarily increase the number of facts provided as evidence, while keeping intact all the relevant symmetries. I presented a case with small numbers simply to aid in visual counting.

206

Michael G. Titelbaum

facts in the set is a reason for C. I have just argued that the shapes example violates that platitude.²⁵ Since the shapes case involves inductive reasoning, and in particular probabilistic inductive reasoning, we can recast that platitude in probabilistic terms. Bayesian analyses of induction typically start with the assertion that evidence E inductively confirms hypothesis H just in case the probability of H given E is greater than the default probability of H. This condition is also known as positive relevance, or positive correlation; Bayesians also take it to be necessary and sufficient for E’s constituting evidence for H. If that’s right, then in the shapes example a set of facts provides evidence for a conclusion even though no fact within the set is evidence for that conclusion. Put another way, an entire set may be relevant to a conclusion even if no fact in that set is relevant.²⁶ This is no strike against Bayesian confirmation theory; it simply shows that the Bayesian approach to evidence doesn’t satisfy a platitude central to the structure of reasons for. What about the other reasons-for platitudes? Well, consider this example: I randomly select a positive integer between 1 and 10, inclusive, such that each integer has an equal chance of being selected. You are interested in the conclusion C that my integer is prime. Consider these facts: My integer is odd. My integer is less than 3.

Each of these facts, considered in isolation, raises the probability of C, and so on a probability-raising account supplies a reason for C. Yet combined they provide conclusive reason for –C. (Recall that 1 is not prime.) Thus this example violates the platitude already mentioned. It also violates the other two platitudes I mentioned earlier: • If a set of facts contains at least one reason for C and no reasons against C, then the set provides at least some all-things-considered reason to conclude C. • Combining two reasons for C produces a weight in favor of C no less than that of either reason taken alone.

²⁵ Some reasons theorists are suspicious of comparative or especially quantitative reason relations. So suppose we pared down the definition of a Franklin case to include only qualitative relations. The new definition would still invoke the platitude I’ve just discussed, and so still would not apply to the shapes example (and many of the others I have presented). ²⁶ This is why I opted earlier for “bears on” rather than “relevance” talk—see footnote 11.

Reason without Reasons For

207

Examples like this violate everything an account of normative support based on reasons for would lead us to expect.²⁷ This is what I mean when I say that, while I’m comfortable with the idea that normative support has a bottom-up, compositional structure, I don’t think that structure always has the functional shape suggested by an account based on reasons for.

9.2.6. There is one last, Hail Mary strategy that has the potential to cast all of these examples as Franklin cases. In each of the examples I have offered, one could maintain that the conjunction of the facts bearing on the conclusion counts as a reason for that conclusion, and the individual facts count as reasons for that conjunction.²⁸ This Conjunctive strategy has the advantage that, at least in cases of theoretical reasoning, it meshes well with a probabilityraising account of reasons for. A standard result in probability theory demonstrates that as long as a conjunction and each of its conjuncts has non-extreme probability, each of the conjuncts individually raises the probability of the conjunction.²⁹ So if we treat the conjunction of the original ²⁷ Sorensen (2016: 8–9) offers an example in which “reasons that separately raise hope can jointly dash that hope.” His moral is “that what is individually confirming can be collectively disconfiming.” Meanwhile, Salmon (1975) finds in Carnap’s (1950) work on probability various examples with the following structures: • Two facts each raise the probability of a conclusion, but their conjunction lowers that probability. • Two facts each raise the probability of a conclusion, but their disjunction lowers that probability. • A fact raises the probability of each of two conclusions, but lowers the probability of their conjunction. • A fact raises the probability of each of two conclusions, but lowers the probability of their disjunction. In personal communication, Branden Fitelson notes that if two facts each raise the probability of a conclusion, then either their conjunction or their disjunction must also raise its probability. And if a fact raises the probability of each of two conclusions, then it must also raise the probability of either their conjunction or their disjunction. While these are fascinating theorems, they underscore for me the ways in which the functional structure of probability-raising is very different from that of additive reason weights. ²⁸ cf. Kagan (1988: 21, n. 9). ²⁹ The “non-extreme” qualifier is required to make that standard result work. It means that a particular class of cases—cases in which the facts bearing on a conclusion and the conclusion itself are all mathematical theorems—cannot be covered by the strategy under consideration. But cases in which reasons for a conclusion and the conclusion itself all have probability 1 were never going to be amenable to a probability-raising account of reasons for, due to the general “logical omniscience” problem in probability theory. So this isn’t a special problem for the Conjunctive strategy I’m considering at this point in the text.

208

Michael G. Titelbaum

facts as an intermediate conclusion, we don’t have the problem (suffered by the Summary Fact strategy) of explaining how the original facts count as reasons for the intermediate conclusion. Moreover, it’s clear at least in the deductive and inductive reasoning cases I offered above that this intermediate conclusion (the conjunction of the pieces of evidence described) does in fact raise the probability of the ultimate conclusion C.³⁰ Strictly speaking, this Conjunctive strategy squares every example I’ve offered with the three Franklinian conditions (modified so as to accommodate intermediate conclusions). But it doesn’t respect the motivations that drove those conditions in the first place.³¹ By itself, Condition 1 required only that whenever an agent ought to do something, some set of facts makes that so. This condition was weak enough to be satisfied by a top-down position like Jonathan Way’s. We added Conditions 2 and 3 because the Franklinian wanted explanations of fact-sets’ normative properties. She thought global normative features of sets could be explained by local reasons for and reasons against interactions among individual facts in those sets. The Conjunctive strategy does indeed explain why some sets of facts provide all-things-considered reason to draw particular conclusions while other sets of facts don’t. The explanation is that a set of facts provides allthings-considered reason to conclude C because the conjunction of facts in that set is a reason for C. But given the Franklinian drive for bottom-up explanations of significant normative features, the explanatory progress made here is almost negligible. Immediately we want to ask what explains why some fact-conjunctions constitute a reason for some conclusions C, while other fact-conjunctions don’t.³² The Conjunctive strategy has merely shifted the burden from explaining the normative features of sets to explaining the normative features of their associated conjunctions. It’s possible for the Conjunctive strategist to dig in her heels here. Most reasons theorists are willing to grant that some reasons relations are just brute and inexplicable. (I’d be happy to concede that as well.) The Conjunctive strategist might insist that the normative explanations in this vicinity always end at the conjunction—we explain the normative feature ³⁰ In the abstract deductive case the conjunction is equivalent to A & –B & –C, which raises the probability of A from 1/2 to 1. In the inductive case the conjunction says that my shape is large, square, and white, which raises the probability of left from 1/2 to 2/3. ³¹ Bader (2016, §5) provides a very different set of arguments from the one I’m about to offer against rolling up all of the facts bearing on a conclusion into a conjunction that’s reason for that conclusion. ³² This question isn’t epistemological: I’m not asking how agents detect that a conjunction is or isn’t a reason for a conclusion. (Perhaps rational intuition directly reveals to the agent which conjunctions do the job.) I’m asking what explains why that relation holds in some cases but not others.

Reason without Reasons For

209

of a set in terms of its associated conjunction’s providing a reason for, then no explanation of the latter relation is ever available.³³ But we know this is wrong. There are plenty of cases in which deeper explanations of a fact-set’s normative standing are readily available, and available on Franklinian terms. In Nair’s bridge example (Section 9.1.1), a set of facts bears on an ultimate conclusion, the set contains reasons for and reasons against that conclusion, the reasons for together outweigh the reasons against, and this explains why the set as a whole provides all-things-considered reason to draw the conclusion. This excellent explanation of the set’s global normative properties engages the complex interplay of reasons, conditions, and modifiers within the set; it doesn’t just say that the facts in the set count as reasons for their conjunction and that conjunction is a reason to cross the bridge. So the Conjunctive strategist shouldn’t insist that pointing to the conjunction associated with a set always offers the deepest, most fine-grained explanation of that set’s normative features available. Perhaps, then, we should pursue a hybrid Conjunctive strategy? In cases like Nair’s, the factset’s all-things-considered support for concluding C is explained by local reasons for and reasons against relations among individual facts in the set and C. But in cases where that won’t work (like some of my examples above), the example is assimilated to the Franklinian framework by saying that the individual facts are reasons for their conjunction, that conjunction is a reason for C, and that’s all the explanation available for the set’s normative relation to C. One way or another, every case is a Franklin case, though in some of those cases our normative explanations turn out not to run very deep. But now the strategy has become blinkered and ad hoc. Once we admit that in at least some cases, a set’s all-things-considered support for concluding C can be explained directly by local relations among the facts in that set and C (and without working through the associated conjunction), we should be honest about the variety of ways in which local relations can do the job. ³³ An anonymous referee for this volume points out that generalists about reasons have responded to particularist attacks by saying that genuine reasons are always large conjunctive facts specifying copious details about the circumstances in which an agent ought to draw a conclusion. (The real reason to fulfill a promise is that the agent made that promise, and she never withdrew it, and she’s capable of fulfilling it, and no pressing moral concerns stand in the way of her doing so, and . . . , etc.) The question I would ask such a generalist is what explains why that conjunction is a reason to fulfill the promise while other conjunctions are not. Perhaps the explanation has to do with how that conjunction would enter into good patterns of reasoning, in which case the generalist has a top-down approach à la Way and Setiya, not a Franklinian Conjunctive strategy. But if the generalist maintains that such explanations are never available, then I have the same problem with her position as the one I’m about to describe for the Conjunctive strategy.

210

Michael G. Titelbaum

If we take my examples on their own terms, we can find in each case a perfectly good local explanation for why the collected facts provide reason to draw the conclusion. In the wine example, that explanation is decisiontheoretic; in the polling example, it’s mathematical; in the deduction examples, it’s logical; and in the inductive examples, it’s probabilistic.³⁴ In each of these cases, we know full well what the true explanation is of the factset’s crucial normative feature. It’s just that that explanation doesn’t operate with reasons for and against. Sometimes reasons relations are inexplicable. But in the examples I’ve offered, there are explanations for why the set as a whole supports the ultimate conclusion. Some of those explanations have the Franklin structure, and some of them do not.³⁵ Rather than obscure the true explanations with an artificial conjunctive substitute, we should admit that some examples are Franklin cases and some just aren’t.³⁶

³⁴ A few people have suggested to me a symmetrical reading of the Tiresias example (Section 9.2.3) on which the second and third facts are conditions making the first fact a reason for the conclusion, and also the first and third facts are conditions making the second fact a reason for, and also the first and second facts are conditions for the third. This reasons-all-around proposal suffers from the same problem as the Conjunctive strategy. I will want to know what explains why some sets of facts have this reasons-allaround structure, while others don’t. Though that might be inexplicable in some cases, it’s surely explicable in others. I would count the Tiresias example as one of the latter, with the relevant explanation provided by logical relations. ³⁵ I suppose it could be suggested that the non-Franklinian explanations are not normative explanations. For example, in the polling case, one could maintain that the mathematical facts are all descriptive, the first normative fact comes onto the scene once the mathematics is completed, and that normative fact (about the set as a whole) has no deeper normative explanation. To make this work in the Tiresias example, one would need an account of deductive logic free of normative notions. (Whether that’s possible gets us into a debate about the intrinsic normativity of logic, in which folks from Aristotle to Beall and Restall (2006) would stand on one side and folks like Harman (1986) would stand on the other.) But past that point, I think the suggestion becomes hopeless. I have already explained decision theory as an essentially normative enterprise. And evidential probabilities—whether subjective or objective—cannot be reduced to purely descriptive terms. (The physical world doesn’t contain enough objective chances to cover so vast a job.) ³⁶ I would have no quarrel with someone who wanted to maintain that in, say, the wine example (Section 9.2.1), non-Franklinian decision-theoretic relations explain why the set of facts supports rosé, but also the facts in that set provide reason for their conjunction and that conjunction is a reason for rosé. On this view, a reason for (even if only a conjunctive one) is always present whenever a conclusion is supported. I would simply insist that the decision-theoretic relations—not the reasons for/against relations— are what explain why the set provides overall support, and indeed why the conjunction is a reason for. The example still wouldn’t satisfy Condition 3, because the local relations explaining the global property wouldn’t be relations of the type enumerated there. So the example would still not be a Franklin case.

Reason without Reasons For

211

9. 3. We are philosophers. There is always another move available. Perhaps we could go epistemicist: in each example above, one of the facts is the true reason for the conclusion; it’s just that there’s no way of knowing which one it is. But it’s a bit difficult to make that work. In at least two of the examples (abstract deduction and shapes), the facts are all situated perfectly symmetrically with respect to the conclusion, so it’s difficult from a metaphysical point of view to figure out how there could even be a difference that would make it the case that one is a reason while the others aren’t. So maybe instead we go supervaluationist . . . . At this point we’re trying to save a generalization that doesn’t want saving. As I said a moment ago, in each of my examples we know what the correct explanation is of the relevant normative support; that explanation just doesn’t have the structure Franklin envisioned. So let’s accept this, and move on to what follows. What are the consequences of the fact that not every instance of all-things-considered reason can be explained by a reason for? Well, one question will be whether an alternative, fully general compositional account of normative support is available. I think this question remains wide open. In the meantime, I don’t want to deny that at least some examples of normative support are genuine Franklin cases; in certain situations (like Nair’s) it does seem appropriate to make a decision by tallying up reasons pro and con. And in other situations, the reasons for approach may provide a useful heuristic even if it doesn’t track the true normative structure of the case. It would certainly be helpful to have general guidelines for when the Franklin approach can and can’t be fruitfully applied. While we await such guidelines, we should stop treating the Franklin approach as a default, walking into every normative situation blithely assuming it can be analyzed with reasons for and against.³⁷ As Michael Conrad used to say,³⁸ let’s be careful out there. I can’t tell you how many times I’ve read a philosophical paper that otherwise doesn’t even mention reasons, but at a crucial juncture the author defends a normative claim by suddenly invoking the weight of reasons on each side.³⁹ This is particularly egregious when the topic is probabilistic relations; we’ve seen that probabilistic confirmation often doesn’t pattern with reasons for and

³⁷ Paula Gottlieb points out to me that ancient philosophers didn’t have this default— they didn’t discuss practical reasoning in terms of anything like reasons for and against. ³⁸ In his role as Sgt. Phil Esterhaus on Hill Street Blues. ³⁹ Once you’re attuned to the phenomenon, I bet you’ll start noticing it too!

212

Michael G. Titelbaum

against. But it also can cause problems with one’s philosophical methodology.⁴⁰ Often, a philosopher will defend a thesis about a particular area with an argument structured as follows: There are a variety of popular positions in this area; the thesis is plausible on this position, the thesis is plausible on this other position, and so on; so in general my thesis is well-supported. To the extent I can make sense of this argumentative strategy, the idea seems to be that the plausibility of the thesis on each position is a reason in favor of that thesis; those reasons stack up; so the thesis is well-supported in general. Yet this assumes exactly the sort of additive normative structure that I’ve been arguing doesn’t universally apply.⁴¹ Are the enumerated positions mutually exclusive? Could the plausibility of the thesis on one position undermine its plausibility on another? Is the plausibility in question meant to be probabilistically construed?⁴² David Enoch (2011: 14ff.) likes to assess philosophical theories by tallying up their “plausibility points”; the metaphor is apt only if the relevant form of plausibility can be added and subtracted like scores in a game.⁴³ But the lesson of this chapter shouldn’t just affect how we defend philosophical theses; it should also influence which theses we defend. Over the past few decades, a particular “reasons-first” metanormative account has gained in popularity. Chris Howard explains: The recently influential “reasons-first” approach to normativity says that reasons are the fundamental elements of the normative domain, and that all other normative facts, properties, and relations can be analyzed or accounted for in terms of the reason relation. (Howard 2019: 216)

In a similar vein, Lord and Maguire write: It is . . . increasingly common to think that the weighted [normative] notions will explain the non-weighted, and that within the weighted notions, reasons will explain the rest. (Lord and Maguire 2016: 4)

And Selim Berker discusses ⁴⁰ Thanks here to Luca Ferrero, who after hearing me present this chapter asked, “That’s all fine, but now how am I supposed to do philosophy?” ⁴¹ Kagan (1988) locates a similar additive assumption behind the argumentative structure in which we construct two cases that are identical except for one factor, attribute any moral difference between the cases to that factor, then assume that the factor has similar moral effects whenever it appears. ⁴² For what it’s worth, the only formal theory of plausibility I know (a component of Demster–Shafer theory) is intimately related to probability theory and bears nothing like the functional structure of reasons for. ⁴³ Enoch takes his inspiration from Lewis and Lewis (1970), who prefer to speak of plausibility “prices.”

Reason without Reasons For

213

a widely held framework according to which reasons for action are the fundamental normative units whose interactions determine, through a metaphorical balancing of the weight of reason, all other normative properties of actions. (Berker 2007: 112)

These are claims about the fundamental structure of normativity, and they have consequences. Suppose that interactions among reasons for and against explain (or determine) all other normative properties. Then in any case in which a normative property is present, reasons for or against must be as well, and the former must be explained by the latter. This implies that all examples of normative support must be Franklin cases. So my argument in this chapter rules out certain kind of reasons fundamentalism.⁴⁴ Finally, I don’t think it’s a coincidence that the “reasons-first” approach has grown in tandem with nonnaturalist metanormative realism, and is often defended by the same people. I wouldn’t argue that either position entails the other,⁴⁵ but there may be explanations for the correlation. On a rival metaethics like, say, constructivism, normative properties are grounded in features of the agent or her agency. Those features give rise to a limited collection of principles or standards, which in turn generate all else that is normative (including reasons). Reasons-first, on which facts out in the universe stand in reasons relations without those relations’ deriving from anything else, is much more at home in a realist metaethics. Going in the other direction, explanations of normative support that do not boil down to reasons relations may undermine realism. In my wine and shapes examples, the relevant norms are most naturally explained by decision theory and Bayesian confirmation theory. Over the course of the twentieth century, proponents of these two theories confronted a number of challenges, and tended to overcome them by moving the theories in a more subjectivist, non-realist direction.⁴⁶ By couching their normative explanations in reasons-first terms, recent metanormative realists have avoided those challenges. In a sense, reading all cases as Franklin cases has given them cover not

⁴⁴ In a series of articles, Daniel Fogal (e.g., 2016) has attacked the reasons-first program and defended a top-down account of reasons. Many of his arguments start from the way we use “reasons” talk in ordinary language. While I agree with many of Fogal’s conclusions, I hope it’s clear how those arguments differ from mine. I don’t much care about the use of the word “reasons”; I’m more interested in the assertion that in every example of normative support, that support arises from facts (call them what you will) that are related to the conclusion via functional relations with a particular structure. ⁴⁵ For instance, some realists now adopt a fittingness-first position rather than a reasons-first one. ⁴⁶ In general terms, probability theorists were challenged to explain what probabilities are, what sets their values, and why they are constrained by particular axioms. More specifically, Titelbaum (2010) describes a particular challenge to theories of evidential support for which subjectivism seems the most plausible answer.

214

Michael G. Titelbaum

to address particular concerns. Again, I don’t think there’s anything like a knockdown argument here. But I do think that if we attend to the functional structure of normative cases for long enough, and adjudicate our compositional story on the basis of what we find in the phenomena, the results may be less comfortable for some metanormative approaches than others.⁴⁷

References Bader, R. (2016). “Conditions, Modifiers, and Holism.” In E. Lord and B. Maguire (eds.), Weighing Reasons. Oxford: Oxford University Press, pp. 27–55. Beall, J. C. and G. Restall (2006). Logical Pluralism. Oxford: Oxford University Press. Berker, S. (2007). “Particular Reasons.” Ethics 118 (1): 109–39. Carnap, R. (1950). Logical Foundations of Probability. Chicago, IL: University of Chicago Press. Dancy, J. (2004). Ethics without Principles. Oxford: Oxford University Press. Dietrich, F. and C. List (2013). “A Reason-Based Theory of Rational Choice.” Noûs 47 (1): 104–34. Enoch, D. (2011). Taking Morality Seriously: A Defense of Robust Realism. Oxford: Oxford University Press. Finlay, S. (2006). “The Reasons that Matter.” Australasian Journal of Philosophy 84 (1): 1–20. Fogal, D. (2016). “Reasons, Reason, and Context.” In E. Lord and B. Maguire (eds.), Weighing Reasons. Oxford: Oxford University Press, pp. 74–103. Franklin, B. (1772/1975). “Letter to Joseph Priestley.” In W. B. Willcox (ed.), The Papers of Benjamin Franklin, Volume 19, New Haven, CT and London: Yale University Press, pp. 299–300. Harman, G. (1986). Change in View. Boston, MA: MIT Press. Horty, J. F. (2012). Reasons as Defaults. Oxford: Oxford University Press. Howard, C. (2019). “The Fundamentality of Fit.” In R. Shafer-Landau (ed.), Oxford Studies in Metaethics, Volume 14. Oxford: Oxford University Press, pp. 216–36. Jeffrey, R. C. (1983). The Logic of Decision (2nd edn). Chicago, IL: University of Chicago Press. ⁴⁷ Thanks to Michael Barkasi, Jeff Behrends, Elizabeth Bell, Josh DiPaolo, Luca Ferrero, Branden Fitelson, Paula Gottlieb, Sophie Horowitz, Zoe Johnson King, Andrew Kernohan, Zak Kopeikin, Ram Neta, Eric Sampson, Mark Schroeder, Russ ShaferLandau, Justin Snedegar, Roy Sorensen, Reuben Stern, Sergio Tenenbaum, Jonathan Weisberg, two anonymous referees for this volume, audiences at the 2016 Iowa Philosophical Society conference, the 2017 St. Louis Annual Conference on Reasons and Rationality, the 2017 Bergen Conference on the Normativity of Logic, and the 2017 Madison Metaethics Workshop, and the folks I button-holed between sessions at the 2016 Chapel Hill Metaethics Workshop. My work on this chapter was supported by a Romnes Faculty Fellowship from the University of Wisconsin-Madison.

Reason without Reasons For

215

Kagan, S. (1988). “The Additive Fallacy.” Ethics 99 (1): 5–31. Kearns, S. and D. Star (2008). “Reasons: Explanations or Evidence?” Ethics 119 (1): 31–56. Kearns, S. and D. Star (2009). “Reasons as Evidence.” In R. Shafer-Landau (ed.), Oxford Studies in Metaethics, Volume 4. Oxford: Oxford University Press, pp. 215–42. Lewis, D. and S. Lewis (1970). “Holes.” Australasian Journal of Philosophy 48 (2), 206–12. Lord, E. and B. Maguire (2016). “An Opinionated Guide to the Weight of Reasons.” In E. Lord and B. Maguire (eds.), Weighing Reasons. Oxford: Oxford University Press, pp. 3–26. Nair, S. (2016). “How do Reasons Accrue?” In E. Lord and B. Maguire (eds.), Weighing Reasons. Oxford: Oxford University Press, pp. 56–73. Nisbett, R. E. and T. D. Wilson (1977). “Telling More Than We Can Know: Verbal Reports on Mental Processes.” Psychological Review 84 (3): 231–59. Salmon, W. C. (1975). “Confirmation and Relevance.” In G. Maxwell and R. M. Anderson Jr. (eds.), Minnesota Studies in the Philosophy of Science, Volume VI. Minneapolis: University of Minnesota Press, pp. 3–36. Scanlon, T. (1998). What We Owe to Each Other. Cambridge, MA: Harvard University Press. Schroeder, M. (2007). Slaves of the Passions. Oxford: Oxford University Press. Setiya, K. (2007). Reasons without Rationalism. Princeton, NJ and Oxford: Princeton University Press. Setiya, K. (2014). “What Is a Reason to Act?” Philosophical Studies 167 (2): 221–35. Silverstein, M. (2016). “Reducing Reasons.” Journal of Ethics & Social Philosophy 10 (1): 1–22. Snedegar, J. (2015). “Contrastivism about Reasons and Ought.” Philosophy Compass 10 (6): 379–88. Sorensen, R. (2016). A Cabinet of Philosophical Curiosities: A Collection of Puzzles, Oddities, Riddles and Dilemmas. Oxford: Oxford University Press. Titelbaum, M. G. (2010). “Not Enough There There: Evidence, Reasons, and Language Independence.” Philosophical Perspectives 24 (1): 477–528. Wallace, R. J. (2012). “Constructivism about Normativity: Some Pitfalls.” In J. Lenman and Y. Shemmer (eds.), Constructivism in Practical Philosophy. Oxford: Oxford University Press, pp. 18–39. Way, J. (2017). “Reasons as Premises of Good Reasoning.” Pacific Philosophical Quarterly 98 (2): 251–70.

10 The Fundamentality of Fit Christopher Howard 10 .1 . INTR O DU CT ION The recently influential “reasons-first” approach to normativity says that reasons are the fundamental elements of the normative domain, and that all other normative facts, properties, and relations can be analyzed or accounted for in terms of the reason relation (Scanlon 1998, Schroeder 2007, Parfit 2011). A central alternative—the “value-first” approach—says that value or goodness occupies the fundamental position in the ontology of normativity, and that the rest of the normative can be accounted for in terms of the property of being good (Moore 1903). In the first part of this paper (Sections 10.2–10.4), I argue that both of these approaches should be rejected. An ontology of normativity should accommodate our commonsense normative intuitions. In particular, it should accommodate the existence of all of the normative reasons that, intuitively, there are. But both the reasons- and value-first approaches fail in this regard: neither can be reconciled with the existence of a certain kind of reason. In the second part of the paper (Section 10.5), I advance an alternative ontology of normativity—one that takes fittingness as fundamental. The normative relation of fittingness is the relation in which a response stands to an object when the object merits—or is worthy of—that response. Call the approach that says that fittingness is the fundamental feature of normativity, and that the rest of the normative can be accounted for in terms of fittingness, the “fittingness-first” approach to normativity. The fittingness-first approach has a rich history. Versions of the ontology were originally suggested by Franz Brentano (1889/2009) and A. C. Ewing (1948), and have recently been proposed by more contemporary authors (Chappell 2012, McHugh and Way 2016). I will argue, however, that existing versions of the fittingness-first approach to normativity suffer the same problem as the reasons- and value-first approaches: none can accommodate all the normative reasons there are.

The Fundamentality of Fit

217

Because my version of the fittingness-first approach similarly accounts for all of normativity in terms of just a single normative relation, it’s no less parsimonious than the reasons-first approach, the value-first approach, or existing versions of the fittingness-first approach. And since, as I’ll argue, my fittingness-first view can plausibly accommodate and explain the existence of all the normative reasons there are, it provides a substantively superior and so overall more attractive ontology of normativity. Thus, as compared to the reasons- and value-first approaches, as well as existing versions of the fittingness-first approach, my version of the fittingness-first approach ought to be preferred. This is the central thesis of the paper. Two preliminary remarks are in order. First, the “analyses” or “accounts” of various normative items that I’ll be interested in throughout the paper are intended as metaphysical accounts of the natures or essences of those items, as opposed to linguistic or conceptual analyses.¹ Second, I should be clear that my fittingness-first view claims only that fittingness is normatively fundamental—that is, fundamental relative to the rest of the normative domain. This leaves open the possibility that fittingness isn’t fundamental simpliciter—that the relation can be fully accounted for in naturalistic or descriptive terms. I will remain agnostic about this possibility here.

1 0 . 2 . T W O K I N D S O F RE A S O N S Normative reasons count in favor of what they’re reasons for.² So when it comes to attitudes like belief and admiration, something is a reason when it counts in favor of the attitude. But it’s familiar that a fact can count in favor of an attitude in two very different ways.³

¹ For discussion of the notion of a metaphysical account of a property/relation, see, e.g., Fine (1994), Jackson (1997), Rosen (2015b), Schroeder (2007: 61–83), and Wedgwood (2007: 135–52). ² For the rest of this paper, unless I indicate otherwise, ‘reason’ means ‘normative reason.’ ³ This distinction is commonly analyzed in terms of ‘object-given’ versus ‘state-given’ reasons, and plausibly tracks Raz’s (2009) distinction between ‘adaptive’ and ‘practical’ reasons for attitudes. I avoid the latter terminology, since I want to steer clear of Raz’s idiosyncratic theoretical commitments concerning the ‘adaptive’/‘practical’ distinction. And I reject the former analysis because, as is now well-recognized, it’s extensionally inadequate to the phenomenon (see footnote 18 for further discussion). A plausible hypothesis is that the relevant distinction is coextensive with the familiar distinction between reasons of the ‘right’ and ‘wrong’ kind. But this requires defense, since the question of how to distinguish right- from wrong-kind reasons is a matter of debate. Later, I’ll offer some support for this hypothesis.

218

Christopher Howard

One way in which a fact can count in favor of an attitude is if it explains why the attitude would be fitting to its object. For example, the fact that Sharon spends a great deal of her time doing charity work is a fact that counts in favor of admiring Sharon, since it’s a fact that explains why she’s admirable, and so worthy of—or fit for—admiration. Similarly, the fact that it’s raining counts in favor of believing that it’s wet out, since it’s a fact that explains why this proposition is credible, and so belief-worthy, or fitting to believe.⁴ A second way in which a fact can count in favor of an attitude is if it explains why the attitude would be somehow valuable or good. For example, the fact that a deplorable dictator will order your execution unless you admire him is a fact that counts in favor of your admiring the dictator, since it’s a fact that explains why your admiring him would be good. Likewise, the fact that Sam will survive her illness only if she believes that she will is a fact that counts in favor of Sam’s believing that she’ll survive, since it’s a fact that explains why it would be good if Sam had this belief. So, intuitively, if a fact explains why some attitude would be fitting, that fact counts in favor of that attitude. And, intuitively, if a fact explains why some attitude would be good to have, that fact counts in favor of that attitude. So if reasons for attitudes are facts that count in favor of those attitudes, then reasons for attitudes seem to come in two quite different kinds: those that explain why an attitude would be fitting and those that explain why an attitude would be good. Call reasons of the former kind fitrelated reasons. Call reasons of the latter kind value-related reasons.⁵ A normative view that says that both fit- and value-related reasons are genuine reasons is highly plausible. Reasons for an attitude are facts that are relevant to determining whether you ought to have that attitude, all-thingsconsidered. And both fit- and value-related reasons seem entirely relevant to determining what attitudes you ought, all-things-considered, to have. It’s an attractive normative hypothesis that we ought normally to have fitting attitudes: that, typically, we ought to believe only what’s credible, admire only what’s admirable, and intend to do only what’s worth doing. But in circumstances in which your having an unfitting attitude is necessary to bring about a great good (or to prevent a great evil) it seems that, ⁴ The issue of what properties comprise credibility is, in principle, just as much a matter of substantive normative dispute as the issue of what properties comprise, e.g., admirability. There is wide agreement that credibility—what merits belief—is truth. But disagreement is possible (if not actual), and so I’ll remain neutral with respect to this issue as far as possible. ⁵ The distinction between ‘fit-related’ and ‘value-related’ reasons generalizes to apply to a wide range of attitudes, including, e.g., desire, resentment, intention, trust, blame, and fear.

The Fundamentality of Fit

219

all-things-considered, you ought to have that attitude, despite its being unfitting. Any contrary normative view would, it seems, stray too far from common sense.⁶ But neither the reasons- nor the value-first approach to normativity can accommodate a normative view on which both fit- and value-related reasons are genuine reasons. The following two sections explain why.

10 . 3. T H E R E A S O N S -F I R S T A P P R O AC H A N D V AL U E - R E L A T E D R E AS O N S Reasons-firsters can’t accommodate a view on which both fit- and valuerelated reasons are genuine reasons, because if value-related reasons are genuine reasons, their account of value is false. Consider the reasons-firster’s account of value: ---- (VR): For p to be a respect in which x is good is for p to be a reason to desire x.⁷

Since VR tells us what it is for p to be a respect in which x is good, it entails that p is a respect in which x is good just in case p is a reason to desire x. And this entails that p can’t be a reason to desire x unless p is a respect in which x is good. But not all facts that make desiring x good—that is, value-related reasons to desire x—are also respects in which x is good. For example, the fact that a demon will kill you unless you desire a cup of mud is a fact that makes your desiring a cup of mud good, but it’s not also a respect in which a cup of mud is good.⁸ And we needn’t resort to such fantastical cases. For instance, the fact that wanting to quit smoking will help you to quit is a fact ⁶ This isn’t to say that a normative view on which both fit- and value-related reasons are genuine reasons is uncontroversial. While the view that fit-related reasons are genuine reasons is among the most widely shared in contemporary normative theory, the issue of whether value-related reasons are genuine reasons for attitudes is contentious. In the following section, I discuss the prospects for skepticism about value-related reasons in some detail. ⁷ VR is in the first instance an account of both final and instrumental value (though, according to some, there are prospects for generalizing the account to other kinds of value, e.g., to goodness-for and the varieties of attributive goodness (see Schroeder 2010: 25–55, Skorupski 2010). Reasons-firsters will vary on the exact formulation. For example, some might want to account for value in terms of reasons for a range of pro-responses, including, but not limited to, that of desire; e.g., admiration, choice, wish, taking pleasure in, being glad that, and so on. And some reasons-firsters might want to specify whether a fact gets to be a respect in which x is good only if it’s a reason for everyone to desire x, or whether it suffices for a fact’s being a respect in which x is good that it be a reason for only some people to desire x. But for our purposes, these details won’t matter. My argument can be extended to all variants of VR. ⁸ This example famously comes from Crisp (2000).

220

Christopher Howard

that makes this desire good to have, but it’s not also a respect in which your quitting smoking is good. So, if value-related reasons are genuine reasons, VR is false. The observation that VR is incompatible with the existence of valuerelated reasons is by no means new. Famously, this is known as the “Wrong Kind of Reason” problem (WKR problem), because if value-related reasons are genuine reasons for attitudes, then these reasons are of the ‘wrong kind’ to figure in VR.⁹ Reasons-firsters have tried to answer the WKR problem in either of two ways. Some opt to reject the counterexamples that constitute the problem— they simply deny that value-related reasons are genuine reasons for attitudes. Others concede the relevant counterexamples, try to draw a distinction between reasons that are of the ‘right’ and ‘wrong’ kind to figure in VR, and revise the account such that it makes reference only to reasons of the ‘right’ kind. But as I’ll now explain, both of these responses are highly difficult to defend, especially for reasons-firsters. Consider first those who try to answer the WKR problem by rejecting the counterexamples that constitute it—that is, by denying that value-related reasons are genuine reasons for attitudes. Call such theorists WKR skeptics. According to WKR skeptics, value-related reasons aren’t reasons, but incentives, where an incentive for an attitude A is a reason to want to have A and to try to bring it about that you have A, but not a reason to have A.¹⁰ Thus, according to the WKR skeptic, the fact that a demon will kill you unless you desire a cup of mud doesn’t strictly speaking give you a reason to desire a cup of mud. Rather, this fact merely gives you a reason to want to desire a cup of mud and to bring this about if you can (Skorupski 2010, Way 2012). Now perhaps we shouldn’t go so far as to call WKR skepticism a “crazy normative position” (D’Arms and Jacobson 2000b: n. 61). But the suggestion that the threat of your own death doesn’t even give you some reason to desire the mud does seem to me deeply implausible. And the WKR skeptic’s attempt to save the phenomenon by claiming that, although there is no reason for you to desire the mud, there are reasons for you to want and to try to desire the mud just doesn’t suffice. For one, this view entails that, in the case of the demon, if you refrain from desiring a cup of mud while simultaneously wanting and trying to desire a cup of mud, then your mental life will be exactly as it should be, all-things-considered. However, as a number of authors recognize, this seems intuitively unacceptable (D’Arms and Jacobson 2000b: n. 61, Rabinowicz and Rønnow-Rasmussen 2004, ⁹ See D’Arms and Jacobson (2000a) and Rabinowicz and Rønnow-Rasmussen (2004). ¹⁰ I borrow this characterization of “incentives” from Way (2012: 492).

The Fundamentality of Fit

221

Rosen 2015a). For another, as Sven Danielsson and Jonas Olson (2007: 514) rightly observe, once we specify that whether the demon delivers on his threat depends on whether you actually come to desire the mud, it becomes even harder to see why the demon’s threat merely gives you reason to want and to try to desire a cup of mud, but no reason to desire a cup of mud. But even setting these worries aside, the real trouble is, as Danielsson and Olson (2007: 513) put it, “that it’s difficult to find an independent rationale for why we should distinguish in the relevant way between reasons for an attitude and [reasons to want] that attitude”. To be sure, facts that make it good to have an attitude—that is, value-related reasons for the attitude—are reasons to want that attitude: such facts make the attitude fitting to want, and are thus fit-related reasons to want it.¹¹ But why shouldn’t facts that are fitrelated reasons to want some attitude also be reasons for that attitude? Consider the normative thesis that p is a fit-related reason to want to do some act only if p is a reason to do that act. This thesis is attractive and widely accepted.¹² Why, then, shouldn’t we think that a parallel thesis holds in the case of reasons for attitudes? If fit-related reasons to want to do some act are reasons to do that act, then why shouldn’t fit-related reasons to want some attitude be reasons for that attitude? And indeed, many theorists take it as a datum that fit-related reasons to want an attitude count as, or “add up to,” reasons for that attitude (Danielsson and Olson (2007: 513), Markovits 2011, Rabinowicz and Rønnow-Rasmussen 2004).¹³ So WKR skepticism isn’t justified on its face. And so in order to provide reasons-firsters with an adequate answer to the WKR problem, WKR skepticism needs independent support. But providing WKR skepticism with the independent support it needs is no easy task, especially for reasons-firsters. Perhaps the most prominent line of argument in favor of WKR skepticism claims that facts like that of the demon’s threat—that is, value-related reasons—aren’t genuine reasons for attitudes, on the grounds that such facts fail a ‘response-condition’ on reasons, according to which a fact p can be a reason to A only if it’s possible to A for the reason that p (Kelly 2002, Shah 2006, Parfit 2011, Gibbons

¹¹ This proposed connection between facts that make an attitude good to have and facts that make that attitude fitting to want is explicitly motivated and defended in Section 10.5.2. ¹² What’s more controversial is the thesis that something is a reason to do some act only if it’s a fit-related reason to want to do that act (though see Portmore 2011 for a defense). ¹³ While the thesis that fit-related reasons to want some attitude are reasons for that attitude is plausible, those who accept it (and its parallel in the case of reasons to act) ultimately owe some explanation of why these theses are true—they shouldn’t be taken as brute. So far, this explanatory task has been neglected in the literature. But later (in Section 10.5.2), I satisfy it.

222

Christopher Howard

2013). But this line of argument is unavailable to reasons-firsters, since these theorists are committed to there being reasons that violate a responsecondition, insofar as they accept VR. To see this, suppose that a deplorable dictator threatens to start a third world war unless you admire him without also wanting to admire him.¹⁴ Since admiring the dictator would be good insofar as it would prevent another world war, VR entails that the fact that it would prevent another world war is a reason to want to admire the dictator. However, if you form a desire to admire the dictator, then the reason that there is for you to desire to admire him will disappear, since, by stipulation of the case, if you desire to admire the dictator, he’ll start another world war anyway. So VR predicts that there’s a reason to want to admire the dictator that can’t survive being responded to.¹⁵ And so defenders of VR are committed to there being reasons that violate a response-condition.¹⁶ So the WKR skeptic’s appeal to a response-condition in order to explain why valuerelated reasons aren’t genuine reasons doesn’t seem promising, especially for reasons-firsters.¹⁷ So what is perhaps the most prominent line of argument for WKR skepticism doesn’t suffice to provide the view with the independent support that it needs in order to offer reasons-firsters an adequate answer to the WKR problem. And so WKR skepticism is, I think, untenable as a response to this problem.

¹⁴ Reisner (2009: 271) deploys cases with a similar structure in the course of raising a different, albeit related, objection to WKR skepticism. He calls these cases of “blocked ascent.” ¹⁵ This case is parallel in structure to Schroeder’s (2007) well-known surprise party case, which Schroeder takes to be a counterexample to a response-condition on reasons (the case also appears in Markovits 2011). In Schroeder’s case, we’re asked to consider Nate, who loves successful surprise parties thrown in his honor, but hates unsuccessful surprise parties. If there is an unsuspected surprise party waiting for Nate next door, then that fact is plausibly a reason for Nate to go next door. But, as Schroeder (2007: 165) observes, it’s simply impossible for Nate to go next door for this reason: “for as soon as you tell him about it, it will go away.” ¹⁶ Way (2012: 512f.) makes a similar point in the context of providing an argument for WKR skepticism that doesn’t rely on the truth of a response-condition. For reasons of space, I won’t discuss Way’s argument for WKR skepticism here, but see Howard 2016a for my response. Also, it’s worth noting that, as McHugh and Way (2016) acknowledge, Way (2012) provides only an argument that value-related reasons aren’t genuine reasons; he doesn’t attempt to give an explanation of why such reasons aren’t reasons. And in order to make an adequate case for WKR skepticism, particularly as an answer to the WKR problem, it seems WKR skeptics will need to succeed in both of these tasks. Establishing and explaining the truth of a response-condition on reasons would plausibly allow WKR skeptics to do just this. However, as I’ve just argued, the prospects for this strategy seem dim for reasons-firsters. ¹⁷ For a very different sort of argument against a response-condition, see Howard (2016b).

The Fundamentality of Fit

223

Consider next those who respond to the WKR problem by trying to draw a distinction between reasons that are of the ‘right’ and ‘wrong’ kind to figure in VR. If such a distinction can be drawn, reasons-firsters can revise VR. They can say that: V----RKR (VRKR): For p to be a respect in which x is good is for p to be a right-kind reason to desire x.

Since this second strategy doesn’t require reasons-firsters to accept and defend the highly committing hypothesis that value-related reasons aren’t genuine reasons for attitudes, it’s in this respect more attractive than WKR skepticism. Nonetheless, this second strategy is also highly difficult to defend, particularly for reasons-firsters. Perhaps the most obvious way to draw the right-/wrong-kind reason distinction is in terms of goodness: Right-kind reasons to want x are reasons to want x that are also facts about x that make x good. Wrongkind reasons to want x are reasons to want x that are not also facts about x that make x good. But of course, reasons-firsters can’t draw the distinction in this way, on pain of circularity. A second, and I think tellingly natural, way of the drawing the right-/wrong-kind reason distinction is in terms of fittingness. Right-kind reasons to want x are reasons to want x that are also facts about x that make x fitting to want. Wrong-kind reasons to want x are reasons to want x that aren’t also facts about x that make x fitting to want. This way of drawing the distinction is attractive because, intuitively, facts about x that make x good are also facts that make x worthy of being desired—that is, fitting to want. So if rightkind reasons to want x are facts about x that make x good, then, intuitively, they’re also facts that make x fitting to want. And because wrong-kind reasons to want x aren’t also facts about x that make x good, they’re not also facts that make x fitting to want. So, I think it’s highly natural to think that the right-/wrongkind reason distinction can be drawn in terms of fittingness. But reasons-firsters can’t draw the right-/wrong-kind reason distinction in terms of fittingness, or in terms of any normative property other than that of being a reason. This is because, according to the reasons-first approach, all normative properties ultimately need to be accounted for in terms of reasons. So reasons-firsters can draw the right-/wrong-kind reason distinction in terms of some normative property other than that of being a reason only if the relevant property can itself be accounted for in terms of reasons. But accounting for any normative property used to distinguish right- from wrong-kind reasons in terms of reasons would itself require a solution to the WKR problem. So reasons-firsters have very few resources with which to distinguish right- from wrong-kind reasons: they must draw this distinction in terms of reasons and non-normative properties alone—an exceedingly tall

224

Christopher Howard

order. Although several authors have tried to offer an account of the distinction that meets this constraint, no such account has gained wide acceptance. Indeed, all such accounts of the distinction seem either subject to counterexample or to turn out to tacitly presuppose the property of being good, thus making VR viciously circular.¹⁸ I think we should conclude that neither of the reasons-firster’s strategies for responding to the WKR problem can plausibly succeed. The upshot: Reasons-firsters can neither plausibly deny nor reconcile their approach to normativity with a normative view that says that value-related reasons are genuine reasons for attitudes.

10 . 4. T H E V A L U E -F I R S T A P P R O AC H AN D F IT -R EL AT ED R EA S O N S Unlike reasons-firsters, value-firsters can accommodate value-related reasons. But they can’t accommodate fit-related reasons, since if fit-related ¹⁸ Three main accounts of the right-/wrong-kind reason distinction in terms of reasons and non-normative properties alone have been advanced in the literature. The most famous of these is the ‘object-given/state-given’ theory, put forward by Stratton-Lake (2005), Piller (2006), and Parfit (2011). However, as I noted earlier (see footnote 3), this account is extensionally inadequate to the phenomenon. According to the objectgiven/state-given theory, right-kind reasons are equivalent to object-given reasons and wrong-kind reasons are equivalent to state-given reasons. But, as is now familiar, there are object-given wrong-kind reasons (Rabinowicz and Rønnow-Rasmussen 2004) as well as state-given right-kind reasons (Schroeder 2012). So the object-given/state-given theory is extensionally inadequate. A second account is that of Schroeder (2010), according to which right-kind reasons with respect to A-ing are those reasons shared by necessarily everyone engaged in A-ing, just because they’re so engaged. Schroeder’s account requires Schroeder to establish two claims: (1) that there’s a set of reasons shared by necessarily everyone engaged in A-ing, just because they’re so engaged and (2) that this shared set of reasons is coextensive with the set of right-kind reasons for A-ing. Schroeder offers two separate strategies for establishing these claims. But Sharadin (2013) convincingly argues that both of these strategies fail. And McHugh and Way (2016) raise further worries for Schroeder’s account. And Schroeder himself has abandoned it (2012). A third account comes from D’Arms and Jacobson (2000a). According to it, right- but not wrong-kind reasons to A are reasons to A which bear on the truth of A’s content (see also Rosen 2015a). This account faces two related problems: (1) it requires us to take on substantial and contentious commitments concerning the contents of various types of attitude in order to ensure the account’s extensional correctness (McHugh and Way 2016, Schroeder 2010); (2) these commitments would seem to render a reasons-based account of value circular, on the plausible assumption that pro-responses like desire have normative contents—for example, if desiring x involves seeing x as somehow good (Ross 1939: 278–9). These are the most well-known, but not the only, accounts advanced in the literature. Certain others have yet to be adequately investigated—see esp. Rowland (2013, 2017)—but unfortunately I don’t have the space to do this here.

The Fundamentality of Fit

225

reasons are genuine reasons, their account of reasons is false. On the value-firster’s account of reasons: R----V (R): For p to be a reason to φ (where φ ranges over actions and attitudes) is for p to explain why φ-ing would be good in some respect and to some degree.¹⁹

Since RV tells us what it is for p to be a reason to φ, it entails that p is a reason to φ just in case p explains why φ-ing would be somehow good. But not all facts that make an attitude fitting also explain why it would be good to have that attitude. For example, the fact that there are eight specks of dust on my desk is a fact that makes it fitting to believe that the number of specks of dust on my desk is even, but this fact doesn’t also explain why it would be good if I had this belief. For there is plausibly no value at all in my having even a true and warranted belief about the number of specks of dust on my desk: Worlds in which I have such a belief are no better than those in which I don’t.²⁰ And there are still further examples, using other attitudes. For instance, the fact that your colleague just received a promotion is a fact that makes it fitting to envy her, but it’s not also a fact that explains why your envying her would be good. And the fact that your friend insists on talking to you (incessantly) throughout the movie is a fact that makes it fitting to be annoyed with him, but this fact doesn’t also explain why it would be good if you were annoyed with your friend. So if fit-related reasons are genuine reasons, RV is false. Like the observation that VR is incompatible with the existence of valuerelated reasons, the observation that RV is incompatible with the existence of fit-related reasons isn’t new. Though the problem hasn’t received nearly as much attention as the WKR problem, it gets some discussion in recent work by John Brunero (2018) and Jonathan Way (2013). These authors don’t give the problem a name, but for parity’s sake, we can call it the Right Kind of Reason problem (RKR problem) since, as I’ve already suggested, the reasons that cause trouble for RV—fit-related reasons—are very plausibly the right kind of reasons to figure in VR.

¹⁹ The phrase “in some respect and to some degree” is borrowed from Raz (1999: 23), who suggests a similar view. See also Finlay (2006: 7–8, and esp. 2012) for an extended defense. ²⁰ Even most ‘epistemic teleologists’ are willing to admit as much. See, e.g., Goldman (1999) and Alston (2005). For an exception, see Lynch (2004), who insists that a true belief about the number of dust-specks on my desk would be good, insofar as it’s true. I find this implausible. But nothing crucial hinges on this—I go on to provide further counterexamples to RV.

226

Christopher Howard

Now, as far as I know, no value-firster has offered a systematic response to the RKR problem in print.²¹ But there are two clear options. First, valuefirsters might be “RKR skeptics,” denying that fit-related reasons are genuine reasons for attitudes. By contrast with WKR skepticism, however, RKR skepticism has no following to speak of. The view that fit-related reasons are genuine reasons is among the most widely shared in contemporary normative theory. And for good reason. It’s highly plausible that the facts that make something admirable, and so fitting to admire, are reasons to admire it; that the facts that make something enviable, and so fitting to envy, are reasons to envy it; that the facts that make something credible, and so fitting to believe, are reasons to believe it; and that the facts that make something fearsome, and so fitting to fear, are reasons to fear it. So I propose to set RKR skepticism aside. If I’m wrong to reject RKR skepticism—if, contrary to what I expect, a view which denies that fit-related reasons are genuine reasons can be made plausible—then the discussion to follow would have to be modified accordingly. Second, value-firsters might adopt a theory of value according to which facts that make an attitude fitting are also facts that explain why having the attitude would be good in some respect and to some degree. How might this go? Some philosophers find plausible a theory of value on which it’s good to have fitting attitudes. And there does seem to be something to this idea. It may seem good to admire admirable persons and objects, to intend to do things that are worth doing, to believe credible propositions, and to desire desirable outcomes. Of course, such attitudes needn’t be of instrumental value. But such attitudes might be thought to be good for their own sakes— or of final value—insofar as they fit their objects.²² So suppose fitting attitudes have final value. Then such attitudes are, presumably, proportionally valuable—that is, valuable in proportion to the

²¹ Finlay (2012) comes the closest. Finlay gestures toward an interesting way to account for fit-related reasons in terms of goodness. He develops an “end-relational” semantics for “good,” according to which something is good just in case it’s good for some end (where the end in question is stated explicitly or otherwise made relevant by the context). “Good for an end” is ultimately understood naturalistically—specifically, in terms of what raises the probability of the relevant end. In the case of fit-related reasons for belief, your fit-related reasons for believing that p are considerations that explain why your believing that p would be ‘epistemically good,’ where epistemic goodness, on Finlay’s semantics, is explained in terms of the promotion of an end—namely, the end of believing that p just in case p is true. Finlay suggests that similar stories can plausibly be told to account for fit-related reasons for other attitudes in terms of goodness. However, he only gestures at this possibility and, although interesting, this way of accounting for fit-related reasons in terms of goodness is highly committal. For one, it requires adopting Finlay’s end-relational semantics for “good.” ²² See Hurka (2001), esp. ch. 1, and Sylvan (2012).

The Fundamentality of Fit

227

degree to which they’re fitting.²³ Thus, given a theory of value on which fitting attitudes are of final value, facts that make an attitude to some degree fitting—that is, fit-related reasons for the attitude—plausibly count as, or “add up to,” facts that make having that attitude to some degree finally valuable. Consequently, a theory of value according to which fitting attitudes are of final value plausibly yields the following result: F-V L (FVL): Necessarily, a fact p makes φ-ing fitting (to some degree) only if p makes φ-ing finally valuable (to some degree).

FVL renders RV extensionally consistent with the existence of fit-related reasons. But the proposal faces three serious worries. I will raise the first two, consider a response to both, and then raise the third, to which there is no good response. First, FVL fares poorly when it comes to cases. Recall the initial counterexamples to RV. It’s implausible that the fact that your colleague just received a promotion makes it good to envy her; or that the fact that your friend insists on talking to you throughout the movie makes it good to be annoyed with him; or that the fact that there are eight specks of dust on my desk makes it good to believe that the number of dust-specks on my desk is even. Second, the conjunction of RV and FVL has absurd implications. FVL entails that if a fact makes some attitude fitting to some degree, that fact makes that attitude good to some degree. And RV entails that if a fact makes doing some act good to some degree, that fact is a reason to do that act. So together, RV and FVL entail that whenever your doing some act would help to bring it about that you have some attitude that’s fitting, that fact is a reason for you to do that act. But this is absurd. Suppose you believe now that, tomorrow, you’ll have a bad day. Is the fact that it will make credible the proposition that you’ll have a bad day tomorrow a reason to stab yourself in the leg today? Or suppose you fear flying. Is the fact that it will make your next flight genuinely fearsome a reason to tamper with the plane’s engine? Together, RV and FVL imply that the answer to both of these questions is “Yes.” Defenders of FVL might respond to each of the above worries in the following way. Mark Schroeder has argued that our intuitions to the effect that there is no reason for a response are unreliable.²⁴ When the reasons for a response are very weak, Schroeder claims, we’re easily misled into thinking there is no reason for the response. And Schroeder’s point easily extends to ²³ For an extended defense of this proportionality thesis, see Hurka (2001), esp. ch. 2. ²⁴ See especially Schroeder (2007: 92–7).

228

Christopher Howard

intuitions to the effect that there is nothing good about something. Perhaps when something is just a little good, but very bad overall, we’re easily misled into thinking it’s in no way good. Suppose all of this is right. Then fans of FVL can respond to each of the above worries in the following way. With regard to the first, they can accept that envying your colleague and being annoyed with your friend are bad overall, but maintain that it’s also good to envy your colleague and to be annoyed with your friend—it’s just that the value of having these attitudes is heavily outweighed. With regard to the second, they can accept that the reasons to stab yourself in the leg and to tamper with the plane’s engine are very weak, since there are so many reasons not to do these things, but maintain that there’s still some reason to stab yourself in the leg and to tamper with the plane—it’s just that the relevant reasons are, again, heavily outweighed. These are fair responses, and I have no quick answer to them. However, I will say this. FVL seems to me a highly substantial and unattractive commitment, especially when paired with RV. And if we can avoid making such commitments and still accommodate a view on which both fit- and value-related reasons are genuine reasons, then I think that we should. And in just a moment, I will explain how we can do exactly this. But before I do, let me raise a third worry about the strategy currently under consideration— one which value-firsters can’t so easily avoid. As a metaphysical account of the nature of reasons, RV isn’t merely an extensional thesis—it’s also an explanatory one. RV entails that whenever a fact is a reason for some response, that’s because that fact explains why it would be good to have that response. And while this seems spot on in the case of value-related reasons, this explanatory commitment is intuitively very implausible when it comes to fit-related reasons. The fact that Sharon spends a great deal of time doing charity work isn’t a reason for you to admire her because it explains why your admiring her would be good. (And this even if, as per FVL, we think that as a matter of extensional fact, this fact does make your admiring Sharon good.) Indeed, the explanation for why the fact that Sharon spends much of her time doing charity work is a reason for you to admire Sharon plausibly has nothing to do with you and your attitudes. Instead, it has everything to do with Sharon: this fact is a reason to admire Sharon simply because it explains why she merits admiration, or is worthy of admiration, or is fit to admire. So even if we grant that value-firsters who accept FVL can plausibly render RV extensionally consistent with the existence of fit-related reasons, their approach to normativity remains explanatorily inadequate. And there is, I think, no good response to this worry. It’s an essential commitment of the value-first approach that all normative facts ultimately need to be explained by appeal to—or grounded in—facts about value. But, as I’ve

The Fundamentality of Fit

229

just now suggested, not all reasons-facts—that is, facts about which facts are reasons—can be felicitously explained in this way. So at best, FVL provides value-firsters with only a partial response to the RKR problem: While the principle perhaps allows value-firsters to say that fit-related reasons are genuine reasons, it fails to secure them an adequate account of why such reasons are reasons. So although this second possible response to the RKR problem is much more plausible than RKR skepticism, it fails to provide value-firsters with a fully adequate answer to the problem. And so I think neither of the ways in which value-firsters might respond to the RKR problem can plausibly succeed. The upshot: Value-firsters can neither plausibly deny nor reconcile their approach to normativity with a normative view that says that fit-related reasons are genuine reasons for attitudes.

10 .5 . T HE FITT ING N ESS- FIR ST A PPR O A CH In the previous sections I have argued that neither the reasons- nor the value-first approach to normativity can accommodate a view on which both fit- and value-related reasons are genuine reasons. Now I will explain how the fittingness-first approach can.

10.5.1. The Fittingness-First Account of Value Consider first the fittingness-first account of value. According to it: V----F (VF): For p to be a respect in which x is good is for p to explain why it would be fitting (to some degree) to desire x.²⁵

The first thing to note about VF is that it’s not new. Accounts of value in its spirit were famously proposed by Brentano (1889/2009) and Ewing (1948), and also suggested by, for example, Broad (1930), Brandt (1946), Wiggins (1987), and McDowell (1998).²⁶ ²⁵ Like VR, VF is in the first instance an account of both final and instrumental value (and, as is the case with VR, there are prospects for generalizing the account to goodnessfor and the varieties of attributive goodness: see, e.g., McHugh and Way 2016). The ‘to some degree’ qualifier is necessary to accommodate the fact that, in many cases, a respect in which something is good will make wanting that thing fitting to some degree, though not fitting overall. ²⁶ Fittingness-based accounts of value are also defended by more contemporary authors, including Danielsson and Olson (2007), Chappell (2012), and McHugh and Way (2016). As I indicated earlier, Chappell and McHugh and Way defend

230

Christopher Howard

The second thing to note about VF is that, unlike VR, it doesn’t face the WKR problem. The fact that a demon will kill you unless you desire a cup of mud is a reason for you to desire a cup of mud—this fact explains why your desiring a cup of mud would be good. But the fact of the demon’s threat doesn’t also explain why your desiring a cup of mud would be fitting—this fact doesn’t make a cup of mud fit to desire. This judgment is prevalent in the literature and certainly would have been shared by VF’s historical defenders.²⁷ Moreover, it seems intuitively quite clear. Remember: The fittingness relation is the relation in which a response stands to an object when the object merits—or is worthy of—that response. And while there’s certainly something to be said for desiring a cup of mud in circumstances in which you’ll be killed if you don’t, it would seem infelicitous to say that the mud merits being desired, or that it’s worthy of being desired. Rather, it seems, the fitting response to have toward such an object would be closer to indifference, or perhaps disgust.²⁸ Still, it might be suggested, perhaps there is a ‘wrong kind of fittingness,’ or a notion of ‘pragmatic fittingness,’ such that, in this sense, it would be fitting to desire the mud. If so, then VF would face a problem parallel to the WKR problem: The demon’s threat would make it (pragmatically) fitting to desire a cup of mud, although the demon’s threat wouldn’t also make a cup of mud valuable or good. Counterexample. Call this the “Wrong Kind of Fittingness” problem (WKF problem). The “WKF problem” is not, I think, a problem. The notion of ‘pragmatic fittingness’ invoked in its statement seems to me spurious.²⁹ To say that it would be pragmatically fitting to want x is, I think, just to say that it would be instrumentally good to want x—or, on my account, that it would be fitting to want to want x. But, as should by now be clear, facts that make it fittingness-first views similar in spirit to that proposed in this paper (though Chappell is specifically concerned to defend the conceptual thesis that ‘fittingness’ is the sole primitive normative concept, whereas McHugh and Way and myself are concerned to defend the metaphysical thesis that fittingness is the fundamental normative relation). There are a number of interesting and consequential differences among contemporary fittingnessfirsters, but perhaps the most crucial lies in our importantly dissimilar accounts of reasons. More on this in Section 10.5.2. ²⁷ Here’s a non-exhaustive list of authors who endorse this judgment, either explicitly or tacitly: Brentano (1889/ 2009); Broad (1930); Brandt (1946); Ewing (1948); Chisholm (1986); Wiggins (1987); McDowell (1998); D’Arms and Jacobson (2000b); Rabinowicz and Rønnow-Rasmussen (2004); Danielsson and Olson (2007); Thomson (2008); Schroeder (2010); Zimmerman (2011); Chappell (2012); Svavarsdottir (2014); Sharadin (2015); McHugh and Way (2016). ²⁸ For a more detailed discussion, see Howard (2018). ²⁹ What could it mean to say that something is ‘pragmatically worthy of ’ a certain response?

The Fundamentality of Fit

231

instrumentally good to want x needn’t also make it fitting to want x, any more than facts that make it instrumentally good to admire a deplorable dictator need also make it fitting to admire the dictator. So VF doesn’t face the WKR problem (or the WKF problem). But by itself, VF doesn’t provide the resources necessary to accommodate and explain the existence of both fit- and value-related reasons. This is because, by itself, VF implies nothing about reasons. What we need, then, is a fittingness-based account of reasons.

10.5.2. The Fittingness-First Account of Reasons Three fittingness-based accounts of reasons are already suggested in the literature: (1) For p to be a reason to φ is for p to be evidence that it is fitting to φ (Thomson 2008). (2) For p to be a reason to φ is for p to explain why it is fitting to φ (Chappell 2012). (3) For p to be a reason to φ is for p to be a premise of a fittingnesspreserving pattern of reasoning from fitting responses to φ-ing (McHugh and Way 2016). The main problem with each of (1)–(3) is the same: all three accounts are incompatible with the existence of value-related reasons for attitudes. (1) and (2) entail that something is a reason for a response only if it explains why (or is evidence that) it would be fitting to have that response. But, as we’ve already seen, not all value-related reasons for attitudes also explain why (or are evidence that) it would be fitting to have those attitudes. (3) entails that something is a reason for a response only if it’s a premise of a fittingnesspreserving pattern of reasoning from fitting responses to that response. But not all value-related reasons will satisfy this condition. For example, reasoning from the belief that the demon will kill you unless you desire a cup of mud to desiring a cup of mud doesn’t preserve fittingness, since desiring a cup of mud isn’t fitting. So, according to (3), the fact that the demon will kill you unless you desire a cup of mud isn’t a reason for you to desire a cup of mud. So none of (1)–(3) are compatible with the existence of valuerelated reasons for attitudes. So how might an adequate fittingness-based account of reasons go? Recall that I began this paper with the observation that reasons for attitudes are facts that count in favor of those attitudes, but a fact can count in favor of an attitude in two different ways: by making the attitude fitting or by making the attitude good to have. This observation suggests that reasons for

232

Christopher Howard

attitudes come in two different kinds: fit-and value-related. The accounts of reasons so far considered in this paper have been able to account for one or the other of these kinds of reasons, but none has been able to account for both. The value-firster’s account of reasons easily accounts for value-related reasons, but founders when it comes to fit-related reasons. And each of the above fittingness-based accounts of reasons accounts for the existence of fitrelated reasons, but fails to account for value-related reasons. So what we need is an account of reasons that accounts for both fit- and value-related reasons, an account which predicts that—and explains why—a fact gets to be a reason either when it makes an attitude fitting or when it makes an attitude good. The fittingness-firster has the resources necessary to provide such an account. So consider: R----F (RF): For p to be a reason to φ is for p to explain either why it is fitting to φ or why it is fitting to want to φ.

By itself, RF entails that whenever a fact makes some attitude fitting, that fact is a reason for that attitude. So the fact that fit-related reasons are genuine reasons for attitudes follows easily. But paired with VF, RF also entails that value-related reasons are genuine reasons. This is easy to see. VF entails that whenever a fact makes some attitude good to have, that fact makes that attitude fitting to want. And RF entails that whenever a fact makes some attitude fitting to want, that fact is a reason for that attitude. So VF and RF jointly entail that value-related reasons are genuine reasons for attitudes. And so together with VF, RF predicts that—and explains why— both fit- and value-related reasons are genuine reasons. Besides this, RF has two further, notable attractions. First, recall that in Section 10.3, I highlighted the attractiveness of the thesis that fit-related reasons to want an attitude count as, or “add up to,” reasons for that attitude—that is, that p is a fit-related reason to want some attitude only if p is a reason for that attitude. I motivated this thesis by appealing to the plausibility of its parallel in the case of reasons to act—that is, that p is a fitrelated reason to want to do some act only if p is a reason to do that act. And I noted that, if true, these theses shouldn’t be taken as brute truths—that, if true, we ultimately need to be able to provide some explanation of why they’re true (see footnote 13). But now, notice that RF allows us to do this: Facts that make it fitting to want some response (whether it be an act or attitude) are reasons for that response, because it lies in the nature of reasons that if a fact makes some response fitting to want, then that fact is a reason for that response. Next, recall that in Section 10.4, I charged the value-first approach with explanatory inadequacy. I argued that while the value-firster’s commitment to grounding all reasons-facts in facts about value looks plausible when it

The Fundamentality of Fit

233

comes to value-related reasons, this commitment yields intuitively implausible results when it comes to fit-related reasons. Thus, although it’s plausible that the explanation for why the fact that the dictator will kill you unless you admire him is a reason for you to admire the dictator has to do with the value of your admiring him, it’s not at all plausible that the explanation for why the fact that Sharon spends much of her time doing charity work is a reason for you to admire Sharon has to do with the value of your admiring her. But now notice RF doesn’t face this difficulty and, in fact, that it gets the best of both worlds. For the fittingness-firster who accepts RF can say that what makes the fact that the dictator will kill you unless you admire him a reason for you to admire the dictator is that this fact explains why it would be fitting to want—that is, good—to admire him. And she can say that what makes the fact that Sharon spends much of her time doing charity work a reason for you to admire Sharon is that this fact explains why she merits admiration, or is worthy of admiration, or is fitting to admire.³⁰

10 .6 . CONC L U S ION In the first part of this paper, I argued that neither the reasons- nor the valuefirst approach to normativity can accommodate a normative view on which both fit- and value-related reasons are genuine reasons. In the second part of the paper, I explained how my version of the fittingness-first approach can accommodate such a view, and argued that alternative, existing fittingness-first views cannot. Because my version of the fittingness-first approach similarly accounts for all other normative properties and relations in terms of just a single normative relation, it’s no less parsimonious than the reasons-first approach, the value-first approach, or existing versions of the fittingness-first approach. And since, unlike any of these alternatives, my version of the fittingness-first approach can plausibly accommodate and explain the existence of all of the reasons there are, it provides a substantively superior and so overall more attractive ontology of normativity. So, as compared to the reasons- and value-first approaches, as well as existing versions of the fittingness-first approach, my version of the fittingness-first approach ought to be preferred. ³⁰ RF has further attractions as well, which, although shared by the other accounts of reasons considered in this paper, are worth noting here. For example, RF offers a unified account of reasons, in the sense that it applies to reasons for responses of all types, i.e., to reasons to act, believe, desire, and so forth. And RF easily accounts for the existence of outweighed reasons, since a response can be fitting to some degree, without being fitting overall, or most fitting.

234

Christopher Howard

Of course, much more would need to be done for a full defense of the fittingness-first view I have advanced in this paper. I have argued only that the view should be preferred to prominent alternatives, not that it should be accepted full stop. A more thorough defense is a task I leave for future work. For now, I hope only to have offered a strong initial case for my fittingnessfirst view, and thus to have made the picture of normativity that it provides seem worthy of further investigation.³¹

References Alston, W. 2005. Beyond Justification: Dimensions of Epistemic Evaluation. Ithaca, NY: Cornell University Press. Brandt, R. 1946. “Moral Valuation.” Ethics 56 (1): 106–21. Brentano, F. 1889/2009. The Origin of Our Knowledge of Right and Wrong. Oxford: Routledge. Broad, C. D. 1930. Five Types of Ethical Theory. London: Routledge. Brunero, J. 2018. “Reasons, Evidence, and Explanations.” In D. Star (ed.), The Oxford Handbook of Reasons and Normativity. Oxford: Oxford University Press, 321–41. Chappell, R. 2012. “Fittingness: The Sole Normative Primitive.” Philosophical Quarterly 62 (249): 684–704. Chisholm, R. 1986. Brentano and Intrinsic Value. Cambridge: Cambridge University Press. Crisp, R. 2000. “Review of John Kupperman, Value . . . and What Follows.” Philosophy 75 (3): 452–62. D’Arms, J. and D. Jacobson. 2000a. “The Moralistic Fallacy: On the ‘Appropriateness’ of Emotions.” Philosophy and Phenomenological Research 61 (1): 65–90. D’Arms, J. and D. Jacobson. 2000b. “Sentiment and Value.” Ethics 110 (4): 722–48. Danielsson, S. and J. Olson. 2007. “Brentano and the Buck-Passers.” Mind 116 (463): 511–22. Ewing, A. C. 1948. The Definition of Good. London: Routledge & Kegan Paul.

³¹ For very helpful written comments on earlier versions of this paper, I would like to thank Jacob Barrett, Selim Berker, Garrett Cullity, Guy Fletcher, Kristen Howard, Victor Kumar, Barry Maguire, Connie Rosati, Richard Rowland, Santiago Sanchez, Mark Timmons, Jonathan Way, Daniel Wodak, Alex Worsnip, and two anonymous referees. I would also like to thank audiences at the British Society for Ethical Theory Conference, the Edinburgh Foundations of Normativity Workshop, Harvard University, the Madison Metaethics Workshop, the Nebraska Practical Reason and Metaethics Conference, the University of Southampton, the University of Arizona, the University of St. Andrews, and Stanford University.

The Fundamentality of Fit

235

Fine, K. 1994. “Essence and Modality.” Philosophical Perspectives 8: 1–16. Finlay, S. 2006. “The Reasons that Matter.” Australasian Journal of Philosophy 84 (1): 1–20. Finlay, S. 2012. “Explaining Reasons.” Deutsche Jahrbuch für Philosophie 4: 112–26. Gibbons, J. 2013. The Norm of Belief. Oxford: Oxford University Press. Goldman, A. 1999. Knowledge in a Social World. Oxford: Oxford University Press. Howard, C. 2016a. “In Defense of the Wrong Kind of Reason.” Thought: A Journal of Philosophy 5 (1): 53–62. Howard, C. 2016b. “Transparency and the Ethics of Belief.” Philosophical Studies 173 (5): 1191–201. Howard, C. 2018. “Fittingness.” Philosophy Compass 13 (11): https://doi.org/10. 1111/phc3.12542. Hurka, T. 2001. Virtue, Vice, and Value. Oxford: Oxford University Press. Jackson, F. 1997. From Metaphysics to Ethics. Oxford: Oxford University Press. Kelly, T. 2002. “The Rationality of Belief and Some Other Propositional Attitudes.” Philosophical Studies 110 (2): 163–96. Lynch, M. 2004. True to Life: Why Truth Matters. Cambridge, MA: MIT Press. McDowell, J. 1998. “Values and Secondary Qualities.” Reprinted in his Mind, Value and Reality. Cambridge, MA: Harvard University Press, chapter 7. McHugh, C. and J. Way. 2016. “Fittingness First.” Ethics 126 (3): 575–606. Markovits, J. 2011. “Internal Reasons and the Motivating Intuition.” In M. Brady (ed.), New Waves in Metaethics. Basingstoke: Palgrave Macmillan, 141–65. Moore, G. E. 1903. Principia Ethica. Cambridge: Cambridge University Press. Parfit, D. 2011. On What Matters. Oxford: Oxford University Press. Piller, C. 2006. “Content-Related and Attitude-Related Reasons for Preferences.” Royal Institute of Philosophy Supplement 81 (59): 155–81. Portmore, D. 2011. “The Teleological Conception of Practical Reasons.” Mind 120 (477): 117–53. Rabinowicz, W. and T. Rønnow-Rasmussen. 2004. “The Strike of the Demon: On Fitting Pro-Attitudes and Value.” Ethics 114 (3): 391–423. Raz, J. 1999. Engaging Reason. Oxford: Oxford University Press. Raz, J. 2009. “Reasons: Practical and Adaptive.” In D. Sobel and S. Wall (eds.), Reasons for Action. Cambridge: Cambridge University Press, 37–57. Reisner, A. 2009. “The Possibility of Pragmatic Reasons for Belief and the Wrong Kind of Reason Problem.” Philosophical Studies 145 (2): 257–72. Rosen, G. 2015a. “The Alethic Conception of Moral Responsibility.” In R. Clarke, M. McKenna, and A. M. Smith (eds.), The Nature of Moral Responsibility: New Essays. Oxford: Oxford University Press, 65–88. Rosen, G. 2015b. “Real Definition.” Analytic Philosophy 56 (3): 189–209. Ross, W. D. 1939. Foundations of Ethics. Oxford: Clarendon Press. Rowland, R. 2013. “Wrong Kind of Reasons and Consequences.” Utilitas 25 (3): 405–16. Rowland, R. 2017. “Reasons or Fittingness First?” Ethics 128 (1): 212–29. Scanlon, T. 1998. What We Owe to Each Other. Cambridge, MA: Harvard University Press. Schroeder, M. 2007. Slaves of the Passions. Oxford: Oxford University Press.

236

Christopher Howard

Schroeder, M. 2010. “Value and the Right Kind of Reason.” In R. Shafer-Landau (ed.), Oxford Studies in Metaethics, vol. 5. Oxford: Oxford University Press, 25–55. Schroeder, M. 2012. “The Ubiquity of State-Given Reasons.” Ethics 122 (3): 457–88. Shah, N. 2006. “A New Argument for Evidentialism.” Philosophical Quarterly 56 (225): 481–98. Sharadin, N. 2013. “Schroeder on the Wrong Kind of Reasons Problem for Attitudes.” Journal of Ethics and Social Philosophy 7 (3): 1–8. Sharadin, N. 2015. “Reasons Wrong and Right.” Pacific Philosophical Quarterly 96 (3): 1–29. Skorupski, J. 2010. The Domain of Reasons. Oxford: Oxford University Press. Stratton-Lake, P. 2005. “How to Deal with Evil Demons: Comment on Rabinowicz and Rønnow-Rasmussen.” Ethics 115 (4): 788–98. Svavarsdottir, S. 2014. “Having Value and Being Worth Valuing.” Journal of Philosophy 111 (2): 84–109. Sylvan, K. 2012. “Truth Monism Without Teleology.” Thought: A Journal of Philosophy 1 (3): 161–3. Thomson, J. J. 2008. Normativity. Chicago and La Salle, IL: Open Court. Way, J. 2012. “Transmission and the Wrong Kind of Reason.” Ethics 122 (3): 489–515. Way, J. 2013. “Value and Reasons to Favour.” In R. Shafer-Landau (ed.), Oxford Studies in Metaethics, vol. 8. Oxford: Oxford University Press, 27–49. Wedgwood, R. 2007. The Nature of Normativity. Oxford: Oxford University Press. Wiggins, D. 1987. “A Sensible Subjectivism?” In his Needs, Values, Truth: Essays in the Philosophy of Value. Oxford: Blackwell, 185–214. Zimmerman, M. 2011. “Partiality and Intrinsic Value.” Mind 120 (478): 447–83.

Index Note: Notes are indicated by ‘n’, respectively following the page number. accessibility condition 40 accommodation strategy, -ies 201 Accuracy Constraint 130–5, 139–41 Adams, Robert Merrihew 108n.32 additivism, simple versions of 98 additivists 98 affective connection 83–4 agent-centered constraint 136–7 in moral philosophy 136 in terms of the value of outcomes 122 agent-neutral act consequentialism 123 all reasons are 75–6 agent-relative and agent-neutral reasons, difference between 75–6 prudential reasons 76 reasons 74, 76–7 reasons for action, denial of 76 reasons for attitudes 76 agents and goodness facts, psychology of 117 agnosticism 35–6 alienness 96 Alston, William 225n.20 altruistic desires 133–5 Anscombe, Elizabeth 121, 121n.1 anti-realist conclusion ix, 25 arbitrariness 50–1 Argument from Conciliationism, The 27, 35–6 Argument from Disagreement in Ideal Conditions 40 Aristotelian eudaimonism 144 Aristotelianism 37 Arneson, Richard 105n.26 Asian moral community 56 attitudes, evaluative 101 attitudes, plausibility of prudential reasons for 80 auditory illusion, -s 132 authoritative conditions, ideally, notion of 114 authority 82–3 and the beginning of the regress 158–61

bad, basic intrinsic 103 Bader, Ralph 194, 208n.31 bads, intrinsic 98 Baker, Derek xii, 82n.35 Bayesian analyses of induction 205–6 Bayesian confirmation theory 205–6, 213–14 Bayesians 205–6 Beall, J. C. 210n.35 Bedke, Matthew 87n.46, 121, 168n.3, 187n.10 Bergmann, Michael 28 Berker, Selim 1–2, 5nn.13–14, 6–7, 11–19, 11n.25, 19n.46, 20n.49, 194nn.6–7, 212–13, 234n.31 case for going Humean regarding moral principles 16 conclusion about moral principles 19 Berker’s Challenge 1–2, 5–6, 9–10, 13–14, 19 Berker’s conclusion 16–17 Berker’s dilemma 14–16 Berker’s reasoning 6–7 Björnsson, Gunnar 47n.1, 49, 56n.14, 63, 63n.24, 64nn.28–9 Blackburn, Simon 50, 59 Bogardus, Tomas 26n.9, 29–30 Bolinger, Renee 64n.30 Booth, Anthony 43n.23 Brandt, Richard 229, 230n.27 Brentano, Franz 216, 229, 230n.27 Brewer, Talbot 125n.6 bridge-lawnon-naturalism 4n.7 Brink, David 24n.4, 25n.5, 95n.3, 105n.26 Broad, C. D. 229, 230n.27 Brogaard, Berit 47n.1, 63 Brunero, John xii, 225 buck-passer, -s xi, 168–9, 171–2, 174, 181–2, 185–6 buck-passing xi, 74n.15, 91n.52, 167–70, 173–4, 184 Burge, Tyler 131n.17 capital punishment 27 Capps, David 48n.4

238

Index

Carnap, Rudolph 193, 207n.27 Carroll, John W. 17n.41 Chappell, Richard 91n.54, 216, 229n.26, 230n.27, 231 Chierchia, Gennaro 126n.10 Chisholm, Roderick 230n.27 Christensen, David 26n.9, 28–9, 31–2, 43n.23 Climenhaga, Nevin 9n.20 cognitivism 24 vs non-cognitivism 86 coherentist reasoning 33 coherentists 34–5 comparability with determination 150–3 comparability without determination 150–1 competitive desires altruistic, cases of 135–6 cases of 137 making sense of 136 complementizer phrase (CP) 124, 127–8 attributions 127–8 clause 125 desire 124–5 concepts, tripartite division of 193 conciliationism 26–9, 35–6, 42–3 argument, -s from 26–32 as popular view 28 belief in 28 defense of 26n.9, 29–32 truth of 27–8 condition and modifier terminologies 194 conduct and etiquette, codes of 90n.51 conjunctions 16 Conjunctive strategy 207–9 consequentialists, traditional 128 constructivism 213–14 contextualism 47–8, 64–5; see also relativist versions of Contextualism, and subjectivist versions of Contextualism formulation of 47n.1 semantics of moral terms 56n.14 subjectivist versions of 47–8 contractualism 37 contrastivism 203 conversation, Griceannorms of 58 Cooperative Principle 58 Copp, David 82n.35, 144, 145n.1, 146, 164 counterfactuals 21 Cowie, Chris 80n.30, 91n.54 credibility 218n.4

Crisp, Roger 105n.25, 145n.1, 168n.2, 219n.8 cross-domain conflict, view for infinitism about 146 cross-domain reasons fundamentalism, epistemological component of 150 Cuneo, Terence 84n.41, 147 D’Arms, Justin 220–1, 220n.9, 224n.18, 230n.27 Dahmer, Jeffrey 127–8 Dancy, Jonathan 149, 168n.3, 192n.2, 194n.8 Danielsson, Sven 220–1, 229n.26, 230n.27 Darwall, Stephen 70nn.1–2, 84, 84n.42, 95n.3, 108n.32, 168nn.3–4 Dasgupta, Shamik 3n.5, 21n.52 Davidson, Donald 121 death penalty 40 decision theory 213–14 Decker, Jason 28, 43n.23 Default-Comparison strategy 202–3 deontic judgements 90–1 deontic prudential discourse 70 deontic prudential properties 89 deontic rationalism 149 deontic terms 73 deontic vocabulary 73 deontology, paradox of 136–9 dependence intuition 4n.6 DeScioli, Peter 59n.18 desiderative systems 135 desire, -s and representation of normative property or relation, uniform relationship between 123–4 and the guise of reasons 139–41 as context-sensitive 135 ascriptions of 124 ascriptions, non-finite 126 associated with a single normative property or relation 135 associated with judgment 123 associated with perception 123 competitive, making sense of 136 desirer-centeredness of 135 egoistic 132–4 fulfillment theories 101, 105–8 fulfilments 79 impersonal 132–4 is a relation toward properties 125–6 motivational character of 140

Index object of 124 propertitional 124–6 systematically associated with perceptions of impersonal good 135 systematically associated with perceptual states 130 theories are different from hedonism 105–6 Theory of Well-being 79, 86 theory, basic form of 106 determination, comparability with 150–3 determination, comparability without 150–1 determiner-phrase (DP) 124 desire ascription, -s 124–5 desire 124 Dietrich, Franz 198n.16 disagreement, -s and arbitrariness 49–51 arguments from 24 contextualist account of 56 in epistemology 34–5 in ideal conditions, arguments from 39–43 intractable 36–9 domain conflict 155–61 Dorsey, Dale 89n.49, 95n.2, 96n.6, 97n.7, 98n.11, 100n.15, 105n.26, 146 Dowell, Janice 54n.10, 55n.11, 56n.14, 84n.40, 229, 230n.27 doxastic attitudes 35 Dreier, James xii, 47nn.1–2, 48n.4, 63n.24 Dretske, Fred 130n.16, 131n.17 egoistic considerations 134 egoistic desires 132–4 Elga, Adam 26n.9, 30–2 Elliott, Aaron 16n.38 Enoch, David ix, 24n.4, 25nn.5,7, 28, 43n.22, 72, 72n.8, 77n.23, 145n.1, 212 epistemic coherentism 145–6 epistemic disagreement 34–5, 43 epistemic domain 144 and esthetic domain, conflict between 145 epistemic foundationalism 145–6 epistemic infinitism 145–6 epistemic justification 34–5 epistemic modals 51–3, 55–6 are propositional operators 51 contextualist accounts of 48–9

239

disagreement, and proposition clouds 51–6 epistemically ideal conditions 42 epistemological argument against non-skeptical realism 32 epistemological arguments 26 epistemological beliefs 35 epistemological challenge for non-skeptical realists 33 epistemological conclusions 24 epistemological debates 35 epistemological disagreements 35 epistemological inquiry 34 epistemological payoff 141 epistemological premise 25 epistemological principles 16 epistemology 26, 38–9 of morality 44 of peer disagreement 28 Equal-Weight View 30 ethical egoists 144 ethics 38–9 fundamental disagreement in 41 systematic account of 37 etiquette, domain of 144 evaluative and deontic vocabulary 73 evaluative attitudes 101 evaluative or deontic judgements 90–1 evaluative prudential discourse 70 Ewing, Alfred Cyril xii, 216, 229, 230n.27 existence internalism 95–6, 111 and prudential value 99–110 attempts to specify 96 concerned with relation between normative facts and human responsiveness 111 metaphysical thesis 110–11 testing 111 “Explanatory Ambitions of Moral Principles” 5n.11 externalism, dual form of 117 externalist view of pleasure 104 fact, -s non-normative 86–7 natural 3 natural and moral, relation between 4 non-moral 32 Feldman, Fred 97n.9 Feldman, Richard 26n.9, 43n.23 Fine, Kit 4n.10, 11n.23, 13–14, 14n.32, 18n.42, 19n.46, 217n.1

240

Index

Finlay, Stephen 47n.1, 49, 54n.10, 56n.14, 59n.19, 63, 63nn.24–7, 64nn.28–30, 66n.31, 70n.3, 71, 128n.14, 141n.25, 202, 225n.19, 226n.21 first-order factual disagreement 66 fit-related reasons are genuine reasons for attitudes 229 existence of 227 fit- and value-related reasons 218–19 fittingness 223 as fundamental 216 is normatively fundamental 216–17 normative relation of 216 wrong kind of 230 fittingness-first xii, 234 account of reasons 231–3 account of value 229–31 approach 216–17, 229–33 approach tonormativity 216 fittingness-value link (FVL) 227–8 has absurd implications 227 provides value-firsters with partial response to RKR problem 228–9 Fletcher, Guy x, 70n.3, 75n.18, 82n.33, 85n.44, 91n.53, 104n.19, 112n.34, 165n.7, 234n.31 Fogal, Daniel 21n.53, 183, 213n.44 Foot, Philippa 84n.41, 128, 154 Foot-Stomping accommodation strategy 198, 202 foundationalists 34–5 Frances, Bryan 26n.9 Franklin case, -s 190–9, 202, 205, 207–12 Franklinian conditions 208 fundamentalism—epistemological component, cross-domain reasons 149–50 fundamentalism—metaphysical component, cross-domain reasons 149 Generic Guise 122–4, 126, 129–30 genetic enhancement 27 Gibbons, John 221–2 Gillies, Anthony 48–9, 52–4 Glasgow, Joshua 98n.11 Goldman, Alvin 225n.20 good attributions 126–8 good choice, -s 99 claims about 110 good functions 126–7

good properties 123–4 good, attributive 126–7 good, basic intrinsic 103 good, -ness, impersonal 128, 134, 137 good, personal 127–9 good, potential accepting motivational externalism about 117 facts about 114 good, realized x, 105–6, 115, 117 and possible good, modal distinction between 98 and possible good, modal distinction between 98 and potential good, modal distinction between x goodness facts 115–16 intrinsic goodness 100 intrinsic prudential good 99–100, 102 good, the motivation to pursue 110–17 structure of 97–9, 118, 126–30 goodness facts 116–17 psychological outlook and 102–3 psychology of agents and 117 goodness non-moral 95 personal 134, 140 goods, intrinsic 107, 109 Gottlieb, Paula 211n.37 Greene, Joshua 122n.4 Gregory, Alex 91n.54, 122n.3, 123, 168n.2 Gregory, Richard 132nn.18,20 Grice, H. Paul 58 Griffin, James 101, 101n.16 grounding pluralism gambit 13–14 grounding principles 17–18 grounding relations 12–14 grounding work 7–10 legal 8 moral 10 guilt, companions in 38–9 Guise Impersonal 128–30, 132–4, 139, 141 guise of the good 126, 135–6, 140–1 difficulty for the 84–6 or the guise of reasons 121 plausible version of 128 theory 129 guise of the normative 122–4, 135, 141 can take different forms 121 epistemological applications of 122

Index Guise Personal 129–30, 132–4, 141 Guise Reasons 139–41 Guise theory 129–30 Haidt, Jonathan 122n.4 Hail Mary strategy 207–8 Hall, Alicia 111n.33 Hampton, Jean 87n.46 Hare, Richard M. 49n.5, 84n.40, 148–9 Harman, Gilbert 47n.3, 63n.24, 210n.35 Hawkins, Jennifer x, 98nn.10–11 Hayward, Max Khan 74n.17, 80n.29 Heathwood, Chris 88n.48, 95n.2, 96n.6, 100n.15, 102n.17, 104n.22, 106n.27 hedodism 103–5, 112 attraction of 104–5 classical 103 understanding the allure of 103 higher-order domains ground crossdomain overriding 156–8 higher-order domains, novel 149–55 Hobbes, Thomas 47n.3, 58 Hooker, Brad 70n.1, 168n.2 Horgan, Terence 84n.40 Horn, Justin 28n.15, 33–4 Horowitz, Sophie 43n.23, 201n.22, 214n.47 Horty, John F. 183, 189n.1 Howard, Christopher xii, 194–5, 212, 222nn.16–17, 230n.28 Howard, Ian 132n.19 Howard, Kristen 234n.31 Hume, David 80n.30 Humean 6, 16–17 and of non-Humean rules 17 generalizations 6–7 generalizations 8, 16, 18–19 laws 16–19 moral principles 16 picture 16 principles 17–18 theory of reasons 77–8 view 77–8 Humeanism 17n.41, 77–80 about Reasons 77–80 and PVM 78 and wellbeing 78 Hurka, Thomas 226n.22, 227n.23 IBE Argument from the History of Moral Philosophy, The 38–9 idealized reasoners 41–2

241

ideally authoritative conditions, notion of 114 imagination and sympathy, failures of 24–5 incomparability and comparability without determination 150–1 inductive theoretical reasoning 204–5 infinitism about cross-domain conflict 145–61 infinitists 34–5 information states 52 informed desire theory 107 internalism 97, 117 and theories of welfare 103–8 contrasted with judgment internalism 95–6 two-tier 114 internalist, -s account of pleasure 105 acknowledge existence of positive response in cases of pleasure 104 constraints on goodness facts 99 Interpersonal Uniqueness thesis (IUT) 42–3 intuition, dependence 4n.6 intuition, rational 30 “Is General Jurisprudence Interesting?” 18n.45 isolation test 178–80 Jackson, Frank 63n.24, 217n.1 Jacobson, Daniel 220–1, 220n.10, 224n.18, 230n.27 Jeffrey, Richard C. 197 Johnson King, Zoë xi, 214n.47 Joyce, Richard 58n.17, 89, 164 judgements, normative types of 81 judgment internalism 110–11 justice, theory of 185–6 justice, circumstances of 58n.16 Kagan, Shelly 97n.8, 108, 194n.9, 207n.28, 212n.41 Kantianism 37 Kaplan, David 47, 47n.3 Kearns, Stephen 149, 195n.12, 202 Kelly, Thomas 28, 43, 221–2 Kelsenian Grundnorm 8n.19 Killoren, David xi King, Nathan 28n.13 Klein, Peter 145–6 Kolodny, Niko 54 Kopec, Matt 43

242

Index

Kornblith, Hilary 26n.9 Korsgaard, Christine 97n.8 Kratzer, Angelika 48–9, 51 Kraut, Richard 70n.3 Kurzban, Robert 59n.18 language, philosophy of 38–9 Law, how law grounds 6–10 “Lawfully Grounded Non-Naturalism” 11n.25 legal and metaphysical grounding, relation between 14–15 legal fact, -s 10, 14 theory of 7 legalnorm, -s 6–10 metaphysically ground 7–8 legal rules, role of 1–2 Leiter, Brian 36–9 Leiter/Nietzsche view 39 Leiter’s hypothesis 38–9 Lewis, David 19, 51, 125n.8, 126, 212n.43 Lewis, Stephanie 212n.43 Lewis–Kratzer Semantics 51 Lin, Eden 77n.23, 78n.26, 91n.54 List, Christian 198n.16 Logic of Decision 197 logic, philosophy of 38–9 Lord, Errol 194n.8, 212 lying 40 Lynch, Michael 48n.4, 225n.20 MacFarlane, John 53n.9, 54 Mach, Ernst 125n.7 Mackie, John Leslie 36n.19, 86n.45 Madison Metaethics Workshop xii Maguire, Barry 194n.8, 212, 234n.31 Manne, Kate 77n.23 markers of normativity 81–6 Markovits, Julia 77n.23, 80n.30, 168–9, 186, 221, 222n.15 Massey, Daniel 48n.4 Matheson, Jonathan 26n.9, 27n.11, 43n.23 McDowell, John 229, 230n.27 McElwee, Brian 148n.3 McGrath, Sarah xii, 27n.11 McHugh, Conor 216, 222n.16, 224n.18, 229nn.25–6, 230n.27, 231 McLeod, Owen 154 McPherson, Tristram xii, 2n.2, 21n.53, 145n.1 Meacham, Christopher 43n.23

metaethical playoffs 19–21 metaethicists 73 metaethics 91 disagreement in 23 SRC-views in 54 metalinguistic negotiation 66 metalinguistic proposal 65–7 metanormative realism xii, 213–14 metanormative theories 91 metaphysical account of a property/ relation, notion of 217n.1 metaphysical arguments 26 metaphysical conclusions 24 metaphysical ground 10 metaphysical grounding 11 metaphysical hierarchy 171, 176, 179–81, 185–7 metaphysical hypothesis 36 metaphysical laws 17 metaphysical premises 25 metaphysics 16, 26, 38–9, 44 meta-prudential questions 71 meta-prudential theory 72–3 Millikan, Ruth 130n.16, 131n.17 Milona, Michael xi mind, philosophy of 130 modal terms 81 modal universe 13–14 Momtchiloff, Peter xii monadic property 128 Montaigne, Michel de 47n.3 Moore, George Edward 49n.5, 95n.1, 216 moral and metaphysical grounding 13 distinguishing between 10 relation between 15–16 moral and prudential judgements close fit between 82 disanalogy between 85–6 moral assertions, truth-evaluable contents for 49 moral beliefs 27, 33, 35–6, 38 moral case 9–10 moral claim, -s suspending judgment on 27 moral code, -s 47–8, 56–7, 61, 63, 65–6 moral communities 48, 50, 56–7, 60–1, 65 moral debates 35 moral disagreement, -s 23, 33–5, 41 and proposition clouds 56–62 as ordinary first-order factual disagreements 67

Index in society 24–5 intuitive 48 moral discourse 59, 82, 84, 88, 90 moral duties 10–11 moral epistemology 129, 141 moral existence internalism 96 moral expressions 54–5 moral fact, -s 3, 6, 10, 87 moral fixed points 147 moral ground 10 moral indeterminacy 24–5 moral inquiry 32, 36 method of 32–4 unreliability of 32–6 moral intuitions, conflicting 33 moral judgement, -s 24, 73, 81–3, 85, 89, 91 conceptual restrictions on 84n.41 dependent on affective states 122 moral on the non-moral, dependence of 87 moral particularists 4n.10 moral philosophers 27–8 moral philosophy 91 constraints in 136–7 disagreement about many debates in 41 moral principle, -s 1–2, 4–6, 10–12, 15–18, 32 and morally relevantnatural facts, conjunction of 14–15 Berker’s conclusion about 19 fundamental 19–20 grounding role for 16–17 role of 1–2 moral psychology xi moral realism 39 non-skeptical 24, 36, 38–40 moral reality, stance-independent 33 moral reasons 89–90 moral semantics ix moral skepticism 27–8, 34–6 moral standpoint 148–9 moral status 40 moral story 10–11 moral term, -s 65 semantics of 55–6 moral theory 37–8, 87 moral truths 36, 38, 40 moral utterances 49, 56, 66 moral value 95–6 moral views 27 morality 10, 82–3 and prudence, conflict between 145

243

and prudence, domains of 144 authoritative nature of 82 back to 10–12 epistemology of 44 is a normative domain 73 metaphysical conclusion about 36 non-normativity of, argument for the 87 troubling metaphysical conclusion about 39–40 trumps prudence 147 understanding of 39–40 motivation test, the 178–9 motivational externalism 113 Müller-Lyer illusion 131–2 multicultural societies 48 multiplenormative domains 144 Nagel, Thomas 74n.17, 75n.20, 98n.10, 154 Nair, Shyam 183, 190–3, 209, 211 Natural Robust Realist picture 2–4 naturalism, reductive 88 Neander, Karen 130n.16, 131n.17 Nietzsche, Friedrich 36–7, 39 Nisbett, Richard E. 196n.13 Noggle, Robert 105n.26 non-cognitive attitudes 24 non-cognitivism 72–3 non-competitive altruistic desires 140 non-finite ascription 124 non-finite desire 124–5 non-moral facts 32 non-naturalist meta-normative realism 213–14 non-nihilism 24 non-normative fact, -s 86–7 non-normativity of morality, argument for the 87 non-skeptical moral realism 24, 36, 38–40 non-skeptical realism, epistemological argument against 32 non-skepticism 24 normative domains 144–5 normative judgement, -s 71–3, 81 normative position, crazy 220–1 normative principle 19–20 normative properties 71 and normatively-relevant properties, distinction between 73 grounded in features of the agent 213–14

244 normative reasons 73–4, 89, 217 normative situations, assessment of 190 normative view 218–19, 229 normative weight 176 normativity, markers of 81–6, 90 normativity, ontology of xii, 216–17 Nozick, Robert 85 object-given/state-given theory 224n.18 objective list theories 108 Oddie, Graham 121, 121n.2, 134n.22 Olson, Jonas 72, 72n.10, 84n.42, 89n.49, 91n.54, 168n.2, 220–1, 229n.26, 230n.27 outweighing principle 152–3 overriding view, the xi paradigmatic non-normative fact 87 parallel phenomena 176–7, 183, 185–6 Parfit, Derek 73n.13, 75n.20, 102–3, 106n.27, 144, 168n.2, 184, 216, 221–2, 224n.18 parity strategy, the 73, 81–6 partial forgetting 179 Peels, Rik 43n.23 peer disagreement, epistemology of 28 perceptual experiences 131 perceptual illusions 131 Perry, John 125n.7 philosophical arguments 25 philosophical methodology xii philosophy of language 38–9 philosophy of logic 38–9 philosophy of mind 130 philosophy of science 38–9 Piller, Christian 224n.18 Pittard, John 26n.9 pleasure, nature of 104 Plunkett, David 64n.30 pluralism, grounding 1–2, 19–21 political philosophy 27, 38–9 Portmore, Douglas 137n.23, 221n.12 pragmatic fittingness, notion of 230–1 Prichard, Harold A. 54 probabilistic inductive reasoning 205–6 problem cases 132–6 promises, breaking 40 propertitional attitude 125–6 propertitional desire 124–6 proposition cloud, -s ix, 49, 55, 57, 59, 61, 64, 67 prostitution 27 prudence-plus-morality 154n.6

Index prudential and moral judgements 86 prudential considerations weighed against moral requirements 82–3 prudential discourse 71, 84–6, 90 evaluative 70 is normative 72 normative markers of 88 prudential existence internalism 96–7 prudential fact, -s 86–7 donot generate reasons 90 motivate individuals to act 110 types of, modal distinction between 96–7 prudential good, -s 97–8 beliefs about 111 goodness facts 107, 117 information theories of 116 intrinsic 97–9 prudential judgement, -s 71, 73, 80–4, 86, 89, 91 non-normative 72–3, 85–8, 91 Prudential Judgements are Normative (PJN) 72–3, 81, 85–6, 88, 91 encompasses evaluative and deontic judgements 90–1 objections to 86–91 rejecting 86 prudential normative properties 90–1 prudential normative reasons 80 prudential properties 71, 75, 79–80, 88 and judgements 86 and judgements are non-normative 87 judgements are non-normative 87 reductionism about 87–8 prudential reason, -s 71, 74–80, 89–90 prudential value 79, 96–7, 99 as distinct type of value 95 desire-based 86–7 facts 111 theory of 116 Prudential Value Matters (PVM) 71, 74–80, 91 psychological needs 37–8 psychological outlook and goodness facts 102–3 psychological responsiveness 96 psychology of agents and goodness facts 117 psychology, moral xi putative intrinsic good 108–9 quasi-realists 21 Quinn, Warren 121, 121n.1

Index Rabinowicz, Wlodek 20n.50, 168n.2, 220–1, 220n.9, 224n.18, 230n.27 Radford, Andrew 127n.13 radical prudential disagreement 88 Railton, Peter 70n.1, 96, 96n.5, 107nn.28–9 rational intuition 30 Rawls, John 58n.16, 185–6 Raz, Joseph 183, 217nn.2–3, 225n.19 “Reason-Giving and the Law” 18n.45 reasons 89–91 as agent-neutral 76 Reasons Contrastivism 198 reasons support properties 138–9 reasons-first approach 213–14, 216, 219–24 reasons-first program xii reasons-firsters 219–25 reasons-in-terms-of-fit 232 Reasons-in-terms-of-Value (RV) 225, 227–9 reductive naturalism 88 redundancy argument 169, 173–4 reflective equilibrium 32, 34 Reisner, Andrew 162, 222n.14 relativism in metaethics, traditional forms of 56 relativist versions of Contextualism 48 RC-views 48–50, 57 reliabilists 34–5 Reliability Argument from Disagreement (RAD) 33–6 response-dependence theorists 20 response-independence meta-normative views 19–20 Restall, Greg 210n.35 Ridge, Michael 64n.29, 75n.21, 91n.54 Right Kind of Reason (RKR) problem 225–6, 229 Right Kind of Reason (RKR) skepticism 226, 229 Robinson, Denis 64n.30 Robust Metaethical Realism 2–3 Robust Realism 2–4, 13, 16–17, 19–20 Robust Realist, -s 1–4, 6 Rock, Irvin 132n.18 Rønnow-Rasmussen, T. 20n.50, 168n.2, 220–1, 220n.9, 224n.18, 230n.27 Rosati, Connie 70nn.1,3, 91n.54, 95n.2, 96nn.4,6, 97, 105n.26, 108n.30, 111n.33, 113, 114nn.36–7, 115, 116n.39, 234n.31

245

Rosen, G. 4n.7, 9n.20, 11n.24, 14n.31, 15n.34, 16n.36, 18n.44, 217n.1, 220–1, 224n.18 Ross, W. David 224n.18 Rowland, Richard xii, 27n.11, 39–43, 91n.54, 155, 165n.7, 224n.18, 234n.31 Sampson, Eric ix, 29n.16, 32n.17, 199n.17, 214n.47 Sarch, Alexander 95n.2, 97n.7, 113n.35 Scanlon, T. M. 75n.18, 95n.1, 121, 123n.5, 168n.2, 170–1, 170n.6, 184, 194n.10, 216 Schafer, Karl 130n.15 Schoenfield, Miriam 43n.23 Schroeder, Mark xi, 19nn.47–8, 63, 71, 77n.23, 88n.47, 121n.2, 123n.5, 125n.9, 130n.15, 141n.25, 176n.7, 187n.10, 189n.1, 202, 214n.47, 216, 217n.1, 219n.7, 222n.15, 224n.18, 227–8, 230n.27 science, philosophy of 38–9 self-exempting strategy 30 self-interest, distorting influences of 24–5 self-undermining problem (SUP), the 28–32 semantic conclusions 24 semantic grounds, arbitrary 50–1 semantic value 47 semantics, moral ix Sepielli, Andrew xii Setiya, Kieran 192n.3, 194, 209n.33 Shafer-Landau, Russ 50–1, 56–7, 147 Shah, Nishi 221–2 Shanklin, Robert 127 Sharadin, Nathaniel 43, 224n.18, 230n.28 share-the-weight view 174, 176 Shoemaker, David 168–9, 186 Sidgwick, Henry 98n.12, 145, 145n.1, 154 Silk, Alex 21n.53, 54n.10, 64n.30, 66n.31 Silverstein, Matthew 197n.14 simple constraint (SC) 100–1 Sinhababu, Neil 125n.6 Sinnott-Armstrong, W. 27n.11 Skorupski, John 168n.2, 220, 222n.17

246

Index

Slote, Michael 98n.11 smaller moral communities 56 Smith, Michael 83n.38, 96n.4 Snedegar, Justin 91n.54, 176n.7, 198, 201n.21, 214n.47 Sobel, David 20n.49, 108n.30, 116n.39 Sorensen, Roy 189n.1, 207n.27, 214n.47 special-fact view 174 Standard Ordering Semantics 51–2, 54–5 Star, Daniel xii, 79n.27, 149, 195n.12, 202 Steadfast View, the 28 Stevenson, Charles 63n.24 Stratton-Lake, Philip 168nn.2–3, 170–1, 224n.18 Streumer, Bart 72, 72nn.9,11, 91n.54 Stroud, Sarah 89n.49, 91n.54 subjectivist versions of Contextualism (SC) 47–9, 100–3, 102n.17, 104n.19, 105–9, 113, 115–16, 118 SRC-framework 50 SRC-views 48–51, 54, 62–4, 67 Suikkanen, Jussi ix, 3n.4, 168n.2 Summary Fact response 200 Summary Fact strategy 199–203, 207–8 Sumner, Leonard Wayne 104, 105n.23, 106n.27 Sumner, William 47n.3 Sundell, Tim 64n.30 Svavarsdóttir, Sigrún xii, 83n.38, 230n.27 Swanson, Eric 176–7 Sylvan, Kurt 71n.5, 226n.22 Taking Morality Seriously 2n.1 Temkin, Larry 98n.11 Templeton, William 132n.19 Tenenbaum, Sergio 121, 121n.1, 214n.47 Thomson, Judith 128, 230n.27, 231 Tiberius, Valerie 111n.33 Tiffany, Evan 82n.35, 145n.1 Timmons, Mark 84n.40, 234n.31 Titelbaum, Michael G. xii, 28, 43, 213n.46 torture 40 Treatise of Human Nature 80n.30 two-tier internalism 97, 113–17 definitions 114 link between facts about potential individual good and subject’s current motivations 116 second requirement of 114–15

utilitarian principle 5–6 utilitarianism 3, 5–6, 11, 37, 75 utility 5–6 utility-suboptimality 5–6 value final 226–7 metaphysics of 96 theory of 226–7 value-first approach 216–17, 232–3 and fit-related reasons 224–9 value-firster, -s 224–6, 229, 231–3 value-in-terms-of-fit 229 value-in-terms-of-reasons (VR) 219 value-in-terms-of-RKRs 223 value-related reasons 218, 231–2 van Inwagen, P. 28 Van Roojen, Mark 96n.4 Vavova, Katia 26n.9, 28n.12 Väyrynen, Pekka 4n.6, 21n.53, 91n.54, 168n.2 vegetarianism 40 Velleman, J. David 95n.2, 96n.6, 97n.7, 98n.11 vocabulary, evaluative and deontic 73 von Fintel, Kai 48–9, 52–4 Wallace, Jay 197 Way, Jonathan 192, 192n.3, 208, 216, 220, 222n.16, 224n.18, 225, 229nn.25–6, 230n.27, 231, 234n.31 Weatherson, Brian 28, 51n.8, 187n.10 Wedgwood, Ralph 28, 141n.25, 217n.1 Weintraub, Ruth 28 Welfare and Rational Care 84 welfare, theory of 116–17 well-being, theories of 87, 97 Western moral community 56 White, Roger 43n.23 Why Be Moral? challenge 82n.34 Wiggins, David 229, 230n.27 Williams, Bernard 60n.21, 77n.23 Wilson, Timothy D. 196n.13 Wolf, Susan 145n.1 Wong, David 47n.1, 63n.24 Worsnip, Alex xii, 74n.17, 78n.24, 80n.29, 234n.31

Index wrong actions 10 Wrong Kind of Fittingness (WKF) problem 220–5, 230–1 Wrong Kind of Reason (WKR) problem 220–5, 230–1 Wrong Kind of Reason (WKR) skepticism 220–3 wrong, the semantics of 54–5

247

wrongness 10–11 metaphysical grounds of 21 Wygoda-Cohen, Shlomit 4n.10 Yablo, Stephen 176–7 Zimmerman, Michael J. 95n.1, 97n.8, 230n.27