Oxford Studies in Metaethics 13 0198823843, 9780198823841

Citation preview

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

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

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Oxford Studies in Metaethics Volume 13

Edited by RUSS SHAFER-LANDAU

1

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

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 2018 The moral rights of the authors have been asserted First Edition published in 2018 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: 2018933810 ISBN 978–0–19–882384–1 (hbk.) ISBN 978–0–19–882385–8 (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.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Contents List of Contributors Introduction 1. Why Believe in Normative Supervenience? Debbie Roberts 2. Non-Naturalism Gone Quasi: Explaining the Necessary Connections between the Natural and the Normative Teemu Toppinen 3. Non-Descriptive Relativism: Adding Options to the Expressivist Marketplace Matthew S. Bedke 4. How to Learn about Aesthetics and Morality through Acquaintance and Deference Errol Lord 5. Belief Pills and the Possibility of Moral Epistemology Neil Sinclair 6. The Puzzle of Pure Moral Motivation Adam Lerner 7. It Ain’t Necessarily So Nomy Arpaly 8. Moral Uncertainty and Value Comparison Amelia Hicks 9. What is (In)coherence? Alex Worsnip

vii ix 1

25

48

71 98 123 145 161 184

10. The Authority of Formality Jack Woods 11. Skepticism About Ought Simpliciter Derek Baker 12. Authoritatively Normative Concepts Tristram McPherson 13. The Rationality of Ends Sigrún Svavarsdóttir

207

Index

301

230 253 278

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

List of Contributors Nomy Arpaly is Professor of Philosophy, Brown University. Derek Baker is Associate Professor of Philosophy, Lingnan University. Matthew S. Bedke is Associate Professor of Philosophy, University of British Columbia. Amelia Hicks is Assistant Professor of Philosophy, Kansas State University. Adam Lerner is Assistant Professor and Faculty Fellow in the Center for Bioethics, New York University. Errol Lord is Assistant Professor of Philosophy, University of Pennsylvania. Tristram McPherson is Associate Professor of Philosophy, the Ohio State University. Debbie Roberts is Lecturer in Philosophy, University of Edinburgh. Neil Sinclair is Associate Professor in Philosophy, University of Nottingham. Sigrún Svavarsdóttir is Associate Professor, Tufts University. Teemu Toppinen is Post-Doctoral Researcher, University of Helsinki. Jack Woods is University Academic Fellow in Mathematical Philosophy, University of Leeds. Alex Worsnip is Assistant Professor of Philosophy, University of North Carolina at Chapel Hill.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Introduction Russ Shafer-Landau

This iteration of Oxford Studies in Metaethics starts off with a pair of essays that focus on the relations between the normative and the non-normative. Almost every philosopher who has written about these relations has endorsed one or more of the supervenience theses that seek to forge a relation between the two domains. Debbie Roberts, however, argues that such efforts are bound to misfire, at least so long as they are committed to this familiar thesis: (NS*) As a matter of conceptual necessity, whenever something has a normative property, it has a base property or collection of base properties that metaphysically necessitates the normative one. She reviews the most important arguments for this thesis and finds each of them wanting. Her efforts, if successful, should bring cheer especially to the hearts of non-naturalist realists, whose position has long been thought vulnerable as a result of its alleged inability to adequately explain the (presumed) truth of NS*. Teemu Toppinen notes that non-naturalist views need not be realist ones—all that’s required to enter the camp is the commitment to the notion, roughly, that normative properties and facts are sui generis. Since quasi-realists allow for the tenability of talk of such properties and facts, it is possible to develop a quasi-realist form of non-naturalism. This is what Toppinen seeks to do, using the supervenience of the normative on the natural as his case study. Toppinen helpfully discusses how the postulation of metaphysically necessary connections between the natural and the normative raise an explanatory challenge for realist non-naturalism. Many have thought that non-cognitivists have the upper hand here, but Toppinen also raises concerns for quasi-realist efforts to secure such an advantage. Still, he ends up encouraging us to understand the explanatory challenge in the light of a quasi-realist take on the relevant necessity judgments. Once we do so, he argues, we will see that this challenge takes the shape of a first-order normative issue—one that can be satisfactorily answered per quasi-realist strictures. Quasi-realist expressivism has dominated the field of non-descriptivist views—those that reject the realist’s contention that normative thought and discourse function primarily to describe aspects of our world. Matthew S. Bedke’s piece is meant to expand non-descriptivist metaethical options beyond the expressivist canon. He develops a form of relativism that begins

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

x

Introduction

by taking for granted various expressivist claims about the shortcomings of descriptivist views, while also seeking to sidestep some of the perennial worries that have beset expressivist positions. His entry identifies a novel family of relativistic metaethical theories that aim to explain the action-guiding qualities of normative thought and language. The general strategy is to consider different relations language might bear to a given content, where descriptivity (or its absence) is located in these relations, rather than in the expression of states of mind, or in a special kind of content. Bedke sketches the outline of one such view, which posits two different content-fixing cognitive roles— one descriptive, the other non-descriptive—for linguistic items. Our next two entries focus on issues in epistemology that have gained some prominence in the last decade. Errol Lord considers issues regarding the appropriateness of forming one’s moral beliefs by deferring to those held by others. Lord investigates this question by drawing connections with the parallel issue in aesthetics. In both areas, many have thought it infelicitous, at best, to form moral or aesthetic beliefs just by taking over the views of others. As Lord notes, a popular explanation of this disquietude in aesthetics appeals to the importance of aesthetic acquaintance. This kind of explanation has not been explored much in ethics. Lord defends a unified account of what is amiss about ethical and aesthetic deference by appealing to the claim that such deference does not allow us to come to possess the full range of reasons provided by the ethical and aesthetic facts. It has this feature because it doesn’t acquaint us with ethical and aesthetic facts. That said, Lord does not believe that there is any general obligation to refrain from engaging in such deference, and so concludes by developing a moderate optimism about the propriety of ethical and aesthetic deference. Neil Sinclair’s entry—winner of the 2016 Marc Sanders Prize in Metaethics—focuses on evolutionary debunking arguments, and seeks to show that such arguments are dialectically ineffective. Such arguments rely on the premise that moral judgments can be given evolutionary explanations that do not invoke their truth. The challenge for the debunker is to bridge the gap between this premise and the conclusion that moral judgments are unjustified. After discussing older attempts to bridge this gap, Sinclair focuses on Richard Joyce’s recent attempt, which claims that ‘we do not have a believable account of how moral facts could explain the mechanisms . . . which give rise to moral judgements’. Sinclair argues that whether there is such an account depends on what we may permissibly assume about moral truth. He further argues that it is reasonable to make assumptions that allow for the possibility of moral epistemologies that yield some epistemic success. The debunker must show that these assumptions are unreasonable, but Sinclair believes that the only ways to do so will render such debunking arguments superfluous.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Introduction

xi

Adam Lerner’s entry focuses on what he calls the Puzzle of Pure Moral Motivation. The puzzle is an explanatory one. According to Lerner, what needs explaining is the fact that ordinary people act rationally when they engage in pure moral inquiry, which occurs when one inquires into something’s moral features without also investigating its non-moral features. Lerner argues that each of the standard views in metaethics has trouble providing such an explanation. He argues that a metaethical view can provide such an explanation only if it meets two constraints. The first is that it allows ordinary moral inquirers to know the essences of moral properties. The second is that the essence of each moral property makes it rational to care for its own sake whether that property is instantiated. Nomy Arpaly’s contribution takes up one of the classic philosophical questions: what is the relationship between being good and living a flourishing life? Is virtue sufficient for such a life? Necessary? Neither? Arpaly focuses her attention on contemporary neo-Aristotelian virtue theories, which argue that the virtuous life constitutes a flourishing life, or is at least the primary element required for living such a life. Arpaly notes her agreement with Rosalind Hursthouse, who claims that a policy of being morally virtuous is a better bet than moral vice for the person who wants to flourish. Still, Arpaly argues that the neo-Aristotelian project is not ultimately successful, in that it cannot rule out the possibility that a morally mediocre life is better than a morally virtuous or a vicious one. Amelia Hicks takes up an issue that has occupied philosophers only relatively recently—that of how to respond rationally to cases of moral uncertainty. She employs as foils the views of several philosophers who argue that decision-theoretic frameworks for rational choice under risk fail to provide prescriptions for choice in cases of moral uncertainty. These philosophers conclude that there are no rational norms that are sensitive to a decision-maker’s moral uncertainty. Hicks disagrees. She argues that one sometimes has a rational obligation to take one’s moral uncertainty into account in the course of moral deliberation. To get to this conclusion, she first argues that one’s moral beliefs can affect what it’s rational for one to choose. She then addresses what she calls the problem of value comparison: the problem of being unable to determine the expected moral value of one’s actions. Some philosophers argue that the problem of value comparison entails that there are no rational norms governing choice under moral uncertainty. Hicks devotes the remainder of her chapter to arguing that this inference is mistaken. Alex Worsnip next considers the nature of coherence requirements. His focus is not on the content of such requirements, or with the question of whether agents necessarily have reason to comply with them. Rather, he takes a step back and focuses on the metanormative status of such requirements.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

xii

Introduction

His chapter focuses on both metaphysical and epistemological issues. Metaphysically: what is it for two or more mental states to be jointly incoherent, such that they are banned by a coherence requirement? In virtue of what are some putative requirements genuine and others not? Epistemologically: how are we to know which coherence requirements are genuine and which aren’t? Worsnip offers a dispositional account designed to answer these questions. On his view, the incoherence of a set of attitudes is a matter of its being constitutive of the attitudes in question that any agent who holds these attitudes jointly is disposed, when conditions of full transparency are met, to give at least one of them up. Jack Woods next takes up the ages-old problem of why we should do as morality requires. Ever since Foot’s seminal (1972) article, much of the work done on this question has sought to answer the question by reference to a comparison between the strictures of morality and etiquette. Woods follows suit. As he sees it, etiquette and other merely formal normative standards, such as legality, honor, and rules of games, are taken less seriously than they should be. While these standards aren’t intrinsically reasonproviding in the way morality is often taken to be, they do play an important role in our practical lives: we collectively treat them as important for assessing our behavior and as licensing particular forms of sanction for violations. Woods develops an account of the normativity of formal standards according to which the role they play in our practical lives explains a distinctive kind of reason to obey them. Etiquette (for instance) is important to us; therefore we have this special kind of reason to be polite. We also have this kind of reason to be moral because morality is important to us. As Woods see it, this parallel suggests that the importance we assign to morality is insufficient to justify its claim to be intrinsically reason-giving. Our next entries are a nicely paired duo that take diametrically opposed positions about the coherence and normative authority of the all-thingsconsidered ought. Derek Baker starts us off with a skeptical appraisal. As he notes, there are many different oughts—the moral one, a prudential ought, an epistemic ought, the ought of etiquette, and so on. These different oughts can prescribe incompatible actions; what one morally ought to do may be different from what one legally ought to do, for instance. It has seemed to most philosophers, and to most ordinary folk, that we can at least sometimes compare the normative force of these incompatible prescriptions and identify what one ought to do, simpliciter. To take an allegedly easy case: when a very weighty moral duty conflicts with a negligible prudential one, the moral duty wins out—one ought, all things considered, to comply with one’s moral duty in such a case. Baker disagrees. He argues that there is just one coherent notion of an ought simpliciter, and that it brings with it firstorder normative commitments that are so implausible as to make it more attractive to abandon the all-things-considered ought.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Introduction

xiii

Tristram McPherson disagrees. He seeks to provide an analysis of the concept of the authoritatively normative practical ought, a plausible candidate for what Baker dubs the ‘ought simpliciter’. His analysis appeals to the constitutive norms for the activity of non-arbitrary selection. McPherson argues that this analysis permits an attractive and substantive explanation of what the distinctive normative authority of this concept amounts to. He contrasts his account with more familiar constitutivist theories, and shows how it can answer ‘schmagency’-style objections to constitutivist explanations of normativity. That said, he takes his analysis to be metaethically ecumenical, and concludes by seeking to explain how his account can help realists, error theorists, and fictionalists address central challenges to their views. Sigrún Svavarsdóttir wraps up this volume by considering the longdisputed question of whether an agent’s selection of her ends can be rational. She answers in the affirmative, arguing that agents can be more or less rational when selecting even final ends. She does this by reference to a conception of practical rationality as an excellence in the exercise of cognitive capacities in one’s practical endeavors. Svavarsdóttir seeks to move beyond traditional fault lines in this debate by arguing that Humeans and anti-Humeans alike should accept her view. She seeks to refocus their disagreement on the question of whether an excellence in the exercise of cognitive capacities in one’s practical endeavors invariably yields a configuration of attitudes directed to ends that must make good sense to the agent. Along the way, she offers a new (albeit incomplete) account of ends. She also motivates her preferred cognitive excellence conception of practical rationality, contrasting it with conceptions of practical rationality as responsiveness to normative reasons or as coherence of attitudes and actions. The chapters collected here are, with one exception, much-revised versions of papers given at the 2nd annual Chapel Hill Metaethics Workshop in 2016. I’d like to thank the UNC Philosophy Department and the Parr Center for Ethics for generous support of the workshop, and to give a major shout-out to events coordinator extraordinaire Katie Bunyea, who handled the logistics of the three-day workshop so superbly. Over 110 submissions came in for eleven spots on the program. My sincere thanks to the terrific philosophers who helped to vet them: John Brunero, Matthew Chrisman, Dale Dorsey, Julia Markovits, Sarah McGrath, and David Plunkett. Once the workshop had concluded and our authors had the opportunity to revise their papers, Peter Momtchiloff, all-star editor at OUP, recruited two reviewers whose extremely detailed and insightful comments made these very good papers even better. One of these reviewers prefers anonymity, but Stephanie Leary has kindly allowed me to publicly acknowledge my gratitude for the tremendous help she provided our authors.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

1 Why Believe in Normative Supervenience? Debbie Roberts

According to many, that the normative supervenes on the non-normative is a truism of normative discourse. If one person is morally good and another not, they must differ in some other respect. If two paintings are qualitatively alike in all other respects they must be aesthetically alike. And if one person is justified in believing that p and another not, there must be some other difference between them. It’s commonly assumed that ‘other’ should be interpreted as non-moral, non-aesthetic, and non-epistemic, respectively.¹ To my knowledge there is only one epistemologist in the literature who doubts the epistemic supervenience claim, Keith Lehrer.² Those who have doubts about aesthetic supervenience have doubts not about the supervenience claim itself but only about how widely the supervenience base should be construed.³ We find more dissenters in metaethics, but still only a few.⁴ This makes it plausible to say, with Gideon Rosen, that the claim that the normative supervenes on the non-normative is the ‘least controversial’ thesis in metaethics.⁵ I argue that those committed to the more specific moral, aesthetic, and epistemic supervenience theses should also hold (NS*): (NS*): As a matter of conceptual necessity, whenever something has a normative property, it has a base property or collection of base properties that metaphysically necessitates the normative one.⁶

¹ I take that normative domains include ethics, epistemology, and aesthetics. This is not meant to be exhaustive. It is plausible, e.g., that the prudential is normative. I’m using ‘normative’ to include both the evaluative and the deontic. ² Lehrer (1997). ³ See, e.g., the exchange between Wicks (1988, 1992) and Zangwill (1992, 1994). ⁴ Griffin (1992), Dancy (1995), Raz (2000), Sturgeon (2009), Hills (2009), Hattiangadi (m.s.), and Rosen (m.s.). ⁵ Rosen (m.s) pp. 1–3. ⁶ A base property is one that is not normativity involving (McPherson 2012).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

2

Debbie Roberts

(NS*) is closely linked to a grounding claim: that normative facts obtain in virtue of non-normative facts. Many hold that this dependence relation explains (NS*).⁷ My main aim is to show that none of the available arguments establish (NS*), or indeed the relevant epistemic, aesthetic, and moral supervenience theses. (NS*) is not a conceptual truth.⁸ This has considerable dialectical importance. One interesting upshot is that it affords non-reductivists and nonnaturalists a novel way of resisting certain prominent supervenience-based objections to their views, including objections that formulate supervenience as a purely metaphysical thesis.⁹ I explain the relevant supervenience claim in Section 1.1 and outline the relation between this claim and a normative grounding claim in Section 1.2. In Section 1.3 I explain two possible views, radical holism and irreducible thickness, that have starring roles in my argument. Section 1.4 lays out the different arguments for supervenience, and my objections to them. I consider key implications in Section 1.5 and conclude in Section 1.6.

1.1. SUPERVENIENCE OF THE NORMATIVE The basic idea of supervenience is a modal one, captured by the slogan ‘no A difference without a B difference’. This can be made more precise by specifying how the modal strength is to be interpreted (metaphysical, conceptual, nomological) and what the thesis quantifies over (individuals, regions, entire possible worlds), and whether the distribution of A and B properties is restricted only within possible worlds (weak) or across possible worlds (strong). An initial statement of the relevant claim is the following individual, strong supervenience thesis:

⁷ See, e.g., Smith (2004), Sibley (1959), Van Cleve (1985). ⁸ My conclusion is similar to Sturgeon (2009). However, Sturgeon does not consider the arguments given for supervenience. His strategy is to argue that there is no supervenience thesis that is not parochial (i.e. acceptable from some metaethical view(s) but not others). (NS*) is similar to the thesis McPherson (2012) argues, in response to Sturgeon, should be common ground for at least metaethical realists. My version is one anti-realists would also be likely to accept, on the ascriptive understanding of it in the case of some expressivist views. ⁹ Non-naturalists are non-reductivists, but not necessarily the reverse. The objections are the ‘reduction’ objection, e.g. Jackson (1998), and the ‘explanation’ objection, which applies only to non-naturalism, e.g. Dreier (1992), McPherson (2012).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Why Believe in Normative Supervenience?

3

(NS): As a matter of conceptual necessity, whenever something has a normative property, it has a non-normative property or collection of properties that metaphysically necessitate(s) the normative one.

I refine (NS) in Section 1.1.1. Nonetheless, (NS) is useful for my first purpose in this section, which is to show that those committed to epistemic, moral, and aesthetic supervenience theses should also be committed to the normative supervenience thesis.

1.1.1. Epistemic, Aesthetic, Moral, and Normative Supervenience (ES): Necessarily, if an individual S has epistemic property E, then S has some non-epistemic property N such that, necessarily, any individual S* with N also has E.¹⁰ (AS): For any possible worlds, w and w*, and for any artworks x and y, if x in w has all of the same non-aesthetic properties as y in w*, then x in w is aesthetically indistinguishable from y in w*.¹¹ (MS): As a matter of conceptual necessity, whenever something has a moral property, it has a non-moral property that necessitates the moral one.¹²

Commonly, in both metaethics and in epistemology, supervenience is understood as individual and strong, holding with metaphysical necessity, i.e. the second ‘necessarily’ in (ES) and the ‘necessitates’ in (MS) are to be understood in terms of metaphysical necessity. The aesthetic case is less clear. It is common to understand aesthetic supervenience as an individual supervenience thesis. However, it is more controversial whether the modal strength of aesthetic supervenience is strong or weak (and often this is left unspecified).¹³ This controversy is connected to the main debate about aesthetic supervenience, namely which properties aesthetic properties supervene on. Once we include extrinsic, relational properties in the base, including human sensibilities, those attracted to the supervenience thesis are likely to accept the strong version, that is, (AS).

¹⁰ Kallestrup and Prichard (forthcoming) p. 1. ¹¹ Hick (2012) p. 309. ¹² Dreier (m.s.) p. 2. All these formulations state an ontological connection between properties, not a connection between types of judgments. However, they are relatively easily recast as ascriptive. Ascriptive version of the points I make later will hold against these versions (one of the arguments I consider is the main argument given for ascriptive supervenience). ¹³ See e.g. Levinson (1984), Bender (1987), Zangwill (1992, 1994, 2003), Fudge (2005), Hick (2012).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

4

Debbie Roberts

So (ES), (AS), (MS), and (NS) are all strong supervenience theses. They are individual supervenience theses about metaphysical necessities. If one holds that the moral (epistemic, aesthetic) strongly supervenes on the non-moral (non-epistemic, non-aesthetic) does one also then hold that the normative strongly supervenes on the non-normative? I take it that it is uncontroversial to hold that the moral is normative, and the same is true for the aesthetic. The epistemic case is more controversial, but it is still a widely held view. Moreover, the main argument for epistemic supervenience in the literature assumes that the epistemic is normative.¹⁴ Characterizing the base is trickier. ‘Non-normative’ is not appropriate for views, for example reductive views, that hold that normative properties are a subset of the base properties. I address this in Section 1.1.2. There is another issue here. The non-normative includes the non-moral, the non-epistemic, and the non-aesthetic. Putting aside the issue of the first necessity, (NS) thus entails each of (ES), (AS), and (MS). However, the reverse is not true since, for example, the non-moral doesn’t obviously include the non-epistemic or the non-aesthetic (or the non-prudential, non-intentional, and whatever other kinds of non-normative properties there are). In theory, a defender of (MS) could reject (NS) on the grounds that (MS) allows epistemic, aesthetic, and other kinds of normative but non-moral properties into the base whereas (NS) does not. Similarly for defenders of (AS) and (ES). However, proponents of these theses typically don’t have these sorts of views in mind. As well as using the terms ‘non-epistemic’, ‘non-aesthetic’, and ‘non-moral’, proponents of these theses typically use terms—for example, ‘descriptive’, ‘natural’, and ‘physical’—to refer to the supervenience base that make clear that the properties they have in mind are meant to contrast with the normative in general.¹⁵ There is one final matter to consider in resolving whether proponents of (ES), (AS), and (MS) should accept (NS), namely the nature of the first necessity. In metaethics, it’s widely held that an individual who fails to respect the supervenience thesis thereby reveals themselves to be an incompetent user of moral terms; the supervenience thesis operates as an a priori, conceptual constraint on moral judgments.¹⁶ The first necessity is thus widely held to be conceptual necessity. A proponent of (MS) should accept (NS). How to understand the first necessity isn’t discussed in epistemology or aesthetics. There is good reason to think, though, that those who hold the ¹⁴ See e.g. Sosa (1991) p. 192. ¹⁵ This contrast may not, ultimately, be a deep metaphysical one for it may turn out that normative properties are themselves descriptive, natural, or physical. ¹⁶ This is automatically the case if the thesis in question is ascriptive. Indeed, the ascriptive version is just supposed to be a deep truth about the nature of our normative judgments.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Why Believe in Normative Supervenience?

5

relevant supervenience theses would take the first necessity to be conceptual necessity. It is true, as in ethics, that which specific supervenience claims obtain may well be an a posteriori matter. However, it seems plausible that if you think that there can be no epistemic difference without a non-epistemic difference, or that two objects that share all of same non-aesthetic features share the same aesthetic features, you’d be likely to think of these as a priori conceptual truths. In support of this, note that proponents of aesthetic supervenience do not carefully distinguish this claim from a dependence thesis that is taken to be constitutive of aesthetic thought—someone who fails to respect it manifests incompetence with aesthetic concepts.¹⁷ Moreover, the main arguments given for epistemic and aesthetic supervenience are a priori.¹⁸ Thus, proponents of (AS) and (ES) should too accept (NS).¹⁹ I now turn to considering (NS) in more detail.

1.1.2. Normative Supervenience (NS) is in need of some refinement. These refinements primarily concern the nature of the supervenience base. The supervenience base is variously characterized, for example, as ‘the factual’, ‘the natural’, ‘the physical’, ‘the descriptive’, the ‘non-evaluative’, and the ‘non-normative’. Each of these, if Nicholas Sturgeon is correct, comes with certain costs to metanormative neutrality.²⁰ To avoid these issues I adapt a suggestion of Tristram McPherson’s and characterize a base property as any property that is not normativity involving.²¹ A normativity involving property is what might be called a non-natural property. It is to be understood ¹⁷ Zangwill (2014). Sibley (1959) is often credited as the origin of discussions of aesthetic supervenience. It is clear that what Sibley is concerned to elucidate is this sort of dependence relation. Many have gone on to characterize this relation as aesthetic supervenience. (In general, dependence and supervenience are not sharply distinguished in the aesthetic or epistemic literature.) ¹⁸ Van Cleve (1985) pp. 98–9, Sosa (1991) pp. 152, 179, 192. ¹⁹ Purely metaphysical versions of these theses are of course possible. My claim is that this is not how these theses are commonly understood. ²⁰ Sturgeon (2009). Sturgeon doesn’t consider ‘the non-normative’, but he is using non-evaluative to mean what I mean by non-normative. Sturgeon also doesn’t consider ‘the physical’. This isn’t a common construal of the base in metaethics. Unless physicalism is true a priori, the claim that the normative strongly supervenes on the physical is not a conceptual truth—its truth would require that certain substantive metaphysical, physicalist assumptions be true. ²¹ McPherson (2012). On his version base properties are those that are not ethically involving. Ridge (2007) suggests an alternative response to Sturgeon: that the base be characterized as either descriptive or non-normative. I agree with McPherson that Sturgeon’s argument puts Ridge’s proposal into doubt (2012 p. 213 n. 23). However, my arguments also work given this disjunctive construal of the base.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

6

Debbie Roberts

as either a sui generis normative property or one whose real definition ineliminably mentions such properties.²² We thus need to reformulate (NS): (NS*): As a matter of conceptual necessity, whenever something has a normative property, it has a base property or collection of base properties that metaphysically necessitate(s) the normative one.

(NS*) is an ontological thesis about properties, rather than an ascriptive thesis about judgments.²³ It should be common ground among realists, at least those who accept that supervenience of the normative is a conceptual truth.²⁴ The thesis could also be understood ascriptively, as a claim that two cases judged to be base-identical must be judged to be normatively identical as well.²⁵ Thus understood, anti-realists who accept normative supervenience but balk at talk of properties should accept (NS*). Error theorists should have no trouble accepting the thesis either. In what follows I shall mostly talk in terms of properties rather than judgments, though the points I make can be fairly easily recast in the ascriptive mode. (NS*) should be understood broadly rather than narrowly: on the broad understanding a supervenience relation obtaining does not rule out the supervening properties being part of the supervenience base, on the narrow understanding it does.²⁶ (NS*) understood broadly would be trivially true for naturalist realists, who hold that there are no sui generis normative

²² This is a potentially misleading term of art. There can be normative properties that are not normativity involving, as there would be if some version of normative naturalism were true. ²³ One might hold that the supervenience thesis should not be characterized in terms of base properties and normativity involving properties. A referee suggested to me that the intuitive thought that motivates the supervenience claim is that the facts that we describe with our normative language (whatever they are) necessarily covary with the facts that we describe with our non-normative language (whatever they are). This could be compatible with the claim that there is ‘normativity all the way down’, metaphysically speaking. So even if, as I go on to argue, there are some views about the thick that imply that reality might be normative all the way down, this does not show, as I claim it does, that the supervenience thesis is not a conceptual truth. However, the same arguments that I deploy later would work against this alternative construal of the supervenience thesis. Indeed, one way of putting my main point is that the conceptual possibility of (IT) means that we cannot claim that the facts that we describe with our normative language (which includes thick terms) do necessarily covary with the facts that we describe with our nonnormative language. Given (IT) there is no guarantee in any token case that we will be able to give a characterization in purely non-normative language of the features of the individual in virtue of which the thick concept applies. ²⁴ Those that don’t should accept the purely metaphysical version of (NS*). ²⁵ Cf. Klagge (1988), McPherson (2015). McPherson discusses a number of different ways that the ascriptive supervenience thesis, particularly the ‘must’, can be understood. ²⁶ Sturgeon (2009) p. 60.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Why Believe in Normative Supervenience?

7

properties. Normative properties are base properties, on these views, and so supervene, in effect, on themselves.²⁷ I formulate (NS*) as the claim that as a matter of conceptual necessity, whenever something has a normative property, it has a base property or collection of base properties that metaphysically necessitate(s) the normative one. The addition of ‘or collection of base properties’ is non-standard and makes this formulation different from (ES), (AS), or (MS). I include this phrase since all of these theses are supposed to be first-order view neutral. It is thus possible that what necessitates the normative property will be all the base properties (included as relational properties of the individual) of the entire world in which the relevant individual is situated.²⁸ It’s not clear that this is a property, even a very complex property.

1 . 2 . S U P E R V E N I E N C E A N D D EP E N D E N C E In general, the fact that A properties supervene on B properties does nothing to show that the A properties depend on the B properties. All that a supervenience thesis says is that the A and the B properties necessarily covary. The supervenience relation is purely modal and is strictly speaking nonsymmetric. Historically, normative supervenience is closely linked to a dependence claim: roughly, things are the way they are normatively speaking because of, or in virtue of, the way they are in terms of (some subset of the) base properties. Some have put this in terms of grounding: the normative is grounded in (some subset of the) base properties. This can be cast in the ascriptive mode as well. Reductive realists who defend an identity relation between particular base properties and normative properties should also accept this dependence claim. On this kind of view, some subset of the base properties is more fundamental with respect to the normative properties. For example, consider the view that the property of being painful just is the property of being bad. On this sort of view it is nonetheless correct to say the fact that an act is bad is grounded in the fact that it is painful. An act is bad if and only if, and because, it is painful.

²⁷ For Sturgeon, supervenience of the normative is only ever true in this trivial sense. That is why it matters so much how the base is construed, and why he thinks there is no non-parochial version of the thesis. ²⁸ See, for example, Dancy (2004) pp. 86–7. This point is just that (RH) could be true.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

8

Debbie Roberts

The grounding claim is often identified as the explanation for the supervenience claim (see Section 1.4). This is a structured picture of the metaphysics of the normative that places (a subset of ) the base properties at the more fundamental level.²⁹

1.3. RADICAL HOLISM AND IRREDUCIBLE THICKNESS In Section 1.4 I argue that none of the arguments for (NS*) succeed, largely because these arguments ignore two at least conceptually possible views: radical holism (RH) and irreducible thickness (IT). (RH) as I understand it is a view concerning the normative grounding relation. According to (RH) the base properties that make, for example, an act wrong in one case, may not be wrong-makers and may even make for a different normative property in a different case.³⁰ This holism is radical because, in addition, normative properties are shapeless with respect to the (subset of) base properties on which they depend:³¹ (RH) the commonality or real resemblance between different instances of a given normative property is irreducibly normative.³²

It will be helpful to have some background in place before formulating (IT). It is a view about the nature of thick concepts. Discussions about thick concepts are typically framed in terms of thick evaluative concepts that range over all normative domains. Moral and aesthetic examples are the most common. Whether there are thick epistemic concepts is a matter of substantive controversy.³³ The view that there are (which certain virtue epistemologists seem committed to) remains possible. (IT) holds that thick concepts and properties are inherently evaluative and shapeless with respect to those base properties on which they depend, but it is not merely an instance of radical holism.³⁴ On this view, thick concepts are just more specific evaluative concepts. Perhaps there are almost ²⁹ See Berker (forthcoming) p. 3 n. 4. ³⁰ This may be to hold that the normative grounding relation is not a necessitation relation. (Cf. Dancy’s view of resultance.) In metaphysics, grounding is usually taken to be a necessitation relation but there are some dissenters: see e.g. Leuenberger (2014), Skiles (2015), and Berker (forthcoming) p. 35 n. 59. ³¹ McDowell (1981), Dancy (1993), Roberts (2011). ³² Cf. Jackson et al. (2000) who put this as the claim that there is no descriptive pattern. ³³ See the papers collected in Philosophical Papers (2008) 37(3), especially Väyrynen (2008). ³⁴ Roberts (2011, 2013a, 2013b, 2017).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Why Believe in Normative Supervenience?

9

no restrictions on the things that can count as GOOD, and no restrictions on the kinds of features that can make things good; GOOD, we can say, operates in a relatively unrestricted domain. What makes KIND a more specific concept than GOOD is that it operates within a narrower domain. KIND applies to actions or people in virtue of those people or actions having features of a certain sort. According to (IT) we won’t be able to specify what sort of features kind people and actions must have, and perhaps not even what features they do have in a particular case, without using further, more specific evaluative terms. In the case of ‘kind’ these might be, for example, ‘considerate’, ‘sensitive’, or ‘thoughtful’. This last point is important: given the possibility that normative properties are normativity involving, on this view there is no guarantee that we will be able to give a purely base characterization of the features of the individual that the thick concept applies to. There is no guarantee, in other words, that we will be able to give a purely base account of the grounds of the thick property. To make it clearer why this is the case, consider the notion of ‘embedded evaluation’.³⁵ When it comes to thick terms and concepts, an evaluation is global if that evaluation applies to all the features that distinguish the things falling under that term or concept. We can contrast this with embedded evaluations, which are evaluations required to specify the very thing over which the global evaluation will take scope. Take ‘distributively just’ as an example, and assume that ‘x is distributively just’ means something like ‘x has features X, Y, and Z as a distribution and is good for having those features’. The ‘good’ that occurs in the analysis is a global evaluation. An embedded evaluation would be present if specifying the type of thing to which the global evaluation applies required evaluative information. In this case, if one or more of X, Y, or Z were an evaluative feature (perhaps X is ‘is the result of a fair procedure’) then ‘distributively just’ would contain an embedded evaluation. And ‘fair’ on this view would also contain an embedded evaluation or evaluations. We have no reason to assume in advance in any token case that we will be able to identify the base properties in virtue of which an institution, say, is just.³⁶ On the non-reductivist view of the thick, the paradigmatic cases of thick concepts contain embedded evaluation. Moreover, that embedded evaluation is itself thick, and there is no reason to assume that there will be a chain of embeddings that bottoms out, in some way, in base concepts ascribing base properties. The salient features for this discussion are as follows: (IT) thick concepts and properties are inherently evaluative and shapeless with respect to base properties. There is no guarantee, in any ³⁵ Elstein and Hurka (2009) p. 526.

³⁶ Cf. Hurley (1989) ch. 2.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

10

Debbie Roberts token case, that we will be able to give a purely base characterization of the features of the individual in virtue of which the thick concept applies.

Neither (RH) nor (IT) straightforwardly entail the denial of (NS*). (RH), as has been pointed out many times, is perfectly compatible with the supervenience of the normative on the base.³⁷ And it is consistent with (IT) that in each token case the properties that make the act kind, say, are base features, even if there is no guarantee that we can identify those features in base terms.

1 .4 . AR G UM E N T S A N D R E P L I E S Arguments for (NS)* (and for (ES), (AS), (MS)) are thin on the ground, perhaps because many have found the claim so obvious as to not need argument. Nonetheless, there are some to be found: a consistency argument, a grounding argument, and a conceivability argument.³⁸ In this section I set out, and raise problems for, each of these in turn.

1.4.1. The Consistency Argument This is an argument typically given by expressivists for ascriptive supervenience. Strictly speaking the conclusion is not (NS*) but an ascriptive version thereof (NS**).³⁹ P3: We judge objects to have normative properties because of the base properties they (we judge them to) have. P4: It is not possible to count as (correctly) engaging in normative thought if you judge that an object has a normative property because of certain base properties and at the same time do not judge that ³⁷ McDowell (1981), Dancy (1993 ch. 5). (RH) (and I think (IT)) are in principle compatible with any particular metaethical view, which is not to say that any naturalist or anti-realist has found (RH) or (IT) plausible. ³⁸ All three types appear in metaethics, the second in epistemology and the third in aesthetics. The main argument for (ES) is one that appeals to the claim that epistemic properties are normative properties. Kim (1988) p. 310, Sosa (1991) pp. 152, 179, 192. See also Turri (2010): P1: All normative properties supervene on base properties. P2: All epistemic properties are normative properties. C1: Therefore, all epistemic properties supervene on base properties. This is clearly valid, and I’ve already defended the assumption of P2. Its soundness thus rests on the plausibility of P1. ³⁹ Blackburn (1971) p. 122, (1984a) p. 186. See also n. 26.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Why Believe in Normative Supervenience?

11

another object exactly alike in its base properties has the same normative property; if you judge that two objects differ in normative properties you must judge that they differ in base properties. C: (NS**) Necessarily, whenever you judge that something has a normative property, you judge that it has a base property or collection of base properties that necessitate(s) the normative one.

The motivation for this argument in the moral case is the thought that morality requires consistency. One version of this thought is that morality would not be fit for purpose if there were no consistency, for morality is essentially a system for regulating behaviour. It thus needs to be regular so we can tell in advance what difference certain features make to how we should behave. Coupled with this view is the assumption that to be regular in this way, the features that we recommend or prohibit, or permit actions on the basis of, need to be non-normative, or the sort of features that we use ‘recognitional concepts’ to pick out.⁴⁰ This is to view morality as ‘something like a set of traffic regulations’.⁴¹ It might be nice if morality were like a set of traffic regulations, but it’s not clear that it isn’t more complex than that. Importantly in this context, the ‘traffic regulation view’ is a substantive view. If it is being assumed in an argument for supervenience as a conceptual truth it can safely be set aside.⁴² But there is another, more fundamental way to cash out the motivation behind this argument. Moral practice, and normative practice more generally, is rational; providing reasons and justifications for one’s normative judgments is an essential part of the practice. Rationality requires, at least, consistency. Dancy expresses this point as follows: It is often held [ . . . ] that the thesis of the supervenience of the moral on the natural is held in place by the concept of a reason; it is effectively an expression of the thought that where exactly the same reasons are in place, one must make the same moral judgment.⁴³

The conclusion, that if the same reasons are in place we must make the same moral judgment, is plausible for the normative in general. This is more or less P4 in different terms. That if the same reasons are in place we must make the same normative judgment is not something that I wish to dispute. I do, however, wish to dispute P3 and the thought that, necessarily, when we judge objects to have certain normative properties this is because of the base ⁴⁰ Gibbard (2003) pp. 102–6, 107–8. ⁴¹ Dancy (2004) p. 83. ⁴² Cf. the literature on ‘government house’ utilitarianism. ⁴³ Dancy (1995) p. 279 (my italics). The consistency argument is connected in interesting ways to the notion of universalizability, and also to the debate about the publicity of reasons (cf. Korsgaard 1993).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

12

Debbie Roberts

properties we take them to have. This is to dispute the thought that the reasons for our normative judgments must always be specifiable in base terms. P3 could be interpreted as an empirical/phenomenological claim, as a description of how we do in fact always make normative judgments. If it were such a claim I think it would be false.⁴⁴ More importantly, if it were such a claim and if it were true it would not be a conceptual truth. Thus P3 interpreted in this way has no place in an argument for (NS*). P3 is more plausible, and more plausible as a conceptual truth, interpreted normatively as a claim about the nature of the reasons, or justification, we must give for our normative judgments. But even on this interpretation P3 fails to be a conceptual truth. Often the reasons we point to in justifying a normative judgment will be other normative properties, thick properties for example. These function perfectly well as reasons. For example, the building is beautiful because of its elegant lines; the belief is unjustified because gullible; the action is wrong because cruel. And it is possible that normative properties are normativity involving, and not base. A common response at this point is ‘Aha! But these thick normative judgments are themselves judgments that must be justified by appeal to base features!’ I address this in Section 1.4.2. Note though that first, this is not to take P3 at face value, which does not make explicit that the focus is only ultimate reasons or grounds, and second, that this response ignores the possibility of (IT).

1.4.2. The Grounding Argument The grounding argument is offered as an explanation of why the supervenience relation holds. I set out two versions of this argument here. The first is Michael Smith’s argument for the supervenience of the evaluative, and the second is adapted from James Van Cleve’s argument for epistemic supervenience.⁴⁵ P5: It is a conceptual constraint on normative judgments that normative claims can’t be barely true; they must always be made true by

⁴⁴ Sometimes we make normative judgments without consciously first identifying the grounds. Sometimes the grounds we do identify are themselves normative. Then, even in the cases where we are seemingly consciously identifying base properties as grounds for our normative judgments, we take those to be reasons, which may be sufficient for them to count as normative (Section 1.4.2). ⁴⁵ Smith (2004), Van Cleve (1985). Smith’s argument is in terms of true ‘claims’. Van Cleve talks of ‘properties’. In the grounding literature the relata of the grounding relation are commonly construed as facts. In this section and in Section 1.5 I’ll talk of grounding of properties as shorthand. Similar grounding arguments are at least implicit in aesthetics, see e.g. Zangwill (2014).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Why Believe in Normative Supervenience?

13

other claims, ultimately by claims about base properties that make the normative claim appropriate. P6: (From P5) if an individual has a normative property, it has that normative property, ultimately in virtue of base properties that it has. P7: In a possible world that agrees in the truth of all the same claims about base properties, the same normative claim will be true. P8: (From P7) in a possible world where another individual has all of the same base properties, it will have the same normative property. C: (NS*) as a matter of conceptual necessity, whenever something has a normative property, it has a base property or collection of base properties that necessitate(s) the normative one.⁴⁶ P9: Either the normative supervenes on the base or there could be (e.g.) a justified belief with no base properties, or there could be a justified belief with some base properties but none that necessitates its normative properties. P10: There could not be (e.g.) a justified belief with no base properties. P11: There could not be (e.g.) a justified belief with some base properties but none that necessitates its normative properties. C: (NS*) as a matter of conceptual necessity, whenever something has a normative property, it has a base property or collection of base properties that necessitate(s) the normative one.⁴⁷

My targets in this section will be P5 and P11. It is crucial to keep in mind the possibility, which (NS*) leaves open, that normative properties might be normativity involving properties (i.e. not base properties). Consider P5. Smith claims that this is evident from our ordinary normative practice. But while it is plausible that ordinary practice establishes that normative claims can’t be barely true, it is not clear that it reveals (as a conceptual constraint) that normative claims must be made true, ultimately, by claims about base properties. The same points made in Section 1.4.1 are relevant here. Often the grounds we point to for our normative claims will themselves be normative—claims about thick properties for example. To think it is a conceptual truth that the grounds for claims about thick properties must ultimately be claims about base

⁴⁶ Smith (2004) pp. 225–9. My formulation. ⁴⁷ Van Cleve (1985) pp. 98–9. My formulation, following Turri (2010) pp. 4–5. Van Cleve’s argument trades on ‘grounding’ assumptions: about normative properties being unable to ‘float free’, and about what the nature of the grounds of normative properties must be. But he also appeals to conceivability. Strictly speaking this is best categorized as both a grounding and a conceivability argument.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

14

Debbie Roberts

properties is to ignore the possibility of (IT) and that thick properties are normativity involving. Why think that justification must bottom out in appeal to base properties? Indeed, why think that it is a conceptual truth that justification must bottom out in some foundation? As far as our concepts are concerned, there is no guarantee that we will be able to give a purely base characterization of the grounds of any thick property. To illustrate, imagine we set about uncovering the grounds for the claim that an action was wrong. These grounds must necessitate the action’s wrongness. It turns out the action was wrong because cruel, and cruel because it involved the gratuitous harming of an innocent person.⁴⁸ Why was it gratuitous? It was unjustified. What made it a harm? It involved a malicious deceit that negatively affected the individual’s welfare. Why was the person innocent? They hadn’t done anything wrong. It is perfectly possible that this process continues with even further appeal to normative (and normativity involving) features. Even in the cases where the grounds appear to be base properties, the fact that they function as reasons might be sufficient to make them normative, and thus possibly normativity involving (and so not base). Say the action was wrong because it caused pain. Pain here functions as a wrong-maker, the reason why the act was wrong. The thought now is that it couldn’t function as the grounds for the wrongness without being normative, or perhaps better normatively relevant, in this way. And the fact this it is normatively relevant is arguably a normative fact. Even in simple cases, then, it appears that it is a live, substantive issue whether or not we ‘bottom out’ in base features in the way that the grounding argument has it that we must.⁴⁹ This brings to light the tension between the assumption that normative claims must be fully grounded in claims about (a subset of ) base properties and the intuition that no normative claim can be fully explained (or justified) in (the relevant subset of ) base terms.⁵⁰ You might think that the tension here would be removed if a reductive account of the normative were correct. There are two points that need to be made about this. The first is that we cannot assume the truth of a reductive account in the context of an argument for (NS*) as a conceptual truth. The second is that even if a reductive account were true, it is not at all clear that the (full) grounds for normative facts would be only base facts. There is the statement of the identity which grounds the reduction itself. It is plausible that this would

⁴⁸ I’ve italicized all the normative terms. Perhaps ‘person’ is a normative term too, but I leave that aside. ⁴⁹ Cf. especially Väyrynen (2013), also Korsgaard (1996) and Schroeder (2005). ⁵⁰ See Väyrynen (2013) especially pp. 158–61.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Why Believe in Normative Supervenience?

15

count as a normative (thus potentially normativity involving) fact, if it were a fact. At the very least this is a substantive issue.⁵¹ The argument just given targets P11 as well. For the claim is now that our concepts and our normative practice leave open the possibility that no set of base properties could (fully) ground, for example, a justified belief, a beautiful painting, or a wrong action. There are now two replies to consider. The first is the claim that I am not being careful enough to distinguish conceptual and metaphysical necessities. It is compatible with the claim that certain base properties of an object ground, and thus metaphysically necessitate, its normative properties, that no base description will conceptually imply, just as a matter of the concepts involved, any particular normative conclusion. However, it is important that the argument for the metaphysical necessitation here is supposed to be an a priori argument that appeals to conceptual truths and ordinary normative practice. The point is precisely that as far as we can tell from our normative concepts, the possibility is left open that base properties do not metaphysically necessitate normative properties. The main point of this section can thus be put as follows: the core of the grounding argument for (NS*) is explicitly the claim that there are no brute normative truths and, crucially, that this is itself a conceptual truth. My contention is that it is not at all plausible that this is a conceptual truth. I have also given some reasons for doubting that it is true at all. The second reply to my argument abandons the Smith/Van Cleve-type grounding picture, but substitutes another, one that allows some brute normative truths but also yields (NS*): (NS*) is the consequence of the fact that whenever a particular individual has a normative property, there’s always a grounding explanation of this fact that cites only base properties of the individual together with a general (metaphysically necessary) normative principle. Normative properties are always fully grounded in base properties plus general normative principles. These general principles may themselves be grounded in further principles, but at bedrock there is a normative principle (or principles) that ground the particular normative facts and mid-level principles, but which are not themselves explained by citing base properties.⁵² ⁵¹ See Heathwood (2012) pp. 10–11. ⁵² See Enoch (2011), Scanlon (2014), Schroeder (2014), Skarsaune (2015), Leary (2017). Although it is commonly non-naturalist realists that defend this kind of account of the structure of the normative, it is a picture that is open to almost all metanormative positions, including reductive views. On reductive views the bedrock normative principle would be the identity statement of the reduction itself.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

16

Debbie Roberts

For this to constitute an argument for (NS*) it would have to be a conceptual truth that general normative principles are metaphysically necessary. This is at least a controversial claim. Fine and Rosen, for example, claim that normative principles are normatively, not metaphysically, necessary.⁵³ There are further reasons to doubt that this account of the structure of the normative is a conceptual truth. First, (RH) may be true. According to (RH) the base properties that make, for example, an act wrong in one case, may not be wrong-makers and may even make for a different normative property in a different case. Normative properties are shapeless with respect to (the relevant subset of ) base properties.⁵⁴ On this view the commonality or real resemblance between different instances of wrongness is normative. This leaves open the conceptual possibility that there is no pattern at the base grounding level. Principles (or at least ‘snappy’ principles) require that there be such a pattern.⁵⁵ One might reply that even if (RH) is true there will still be some grounding explanation of particular normative facts. This would cite only base properties of the relevant individual-in-a-world together with a general, perhaps infinitely complex, conditional, whether or not we want to call this conditional a (non-snappy) principle.⁵⁶ So the grounds for the wrongness of an action in a particular case would be all of the base properties of the action including all of the base properties of the entire world in which the action is situated, as relational properties of that action, plus a conditional stating that actions with these base properties are wrong.⁵⁷ However, given the possibility of (IT) it is not clear that we would even be able to formulate this base description of the individual-in-a-world plus a general, perhaps infinitely complex, conditional, in such a way that the instantiation of the normative property is indeed necessitated. For this would

⁵³ Fine (2002), Rosen (m.s.). ⁵⁴ McDowell (1981), Dancy (1993), Roberts (2011). ⁵⁵ See Rosen (m.s.) p. 2 n. 1, Jackson et al. (2000). ⁵⁶ If (RH) is the case, and we wanted the guarantee that a particular normative property would be instantiated by an individual then we would need to include every base property of the world in which the individual was situated, as relational properties of that individual, as part of the grounds. ⁵⁷ There is an interesting question here concerning whether this would amount to an explanation of the particular normative fact (grounding is after all supposed to be metaphysical explanation). This point is related to Dancy’s claim that the grounding relation, as far as the normative is concerned, is not a necessitation relation at all. The resultant base for a normative property does not necessitate that property, on his view. Dancy holds this because he holds (RH), or something close to it. And he thinks that to include all of the properties of the entire world in which the individual is situated in the base is to fail to single out the wrong-makers, or whatever the relevant normative property may be. On the face of it, this looks like a failure to explain what needs to be explained.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Why Believe in Normative Supervenience?

17

require that we could do precisely what (IT) says there is no guarantee that we can always do, i.e. that we be able to give an account purely in base terms of the grounds of any and all thick properties of this individual-in-a-world. If we can’t formulate it I can’t think what metaethically neutral reason we have to say that it must nonetheless be there.

1.4.3. The Conceivability Argument The conceivability argument is the most common type of argument for normative supervenience. It was first advanced by R. M. Hare, employing the example of St Francis.⁵⁸ Versions are put forward by, among others, Jamie Dreier and Mike Ridge.⁵⁹ I formulate it as follows: P12: If we conceive of another individual identical to X in terms of base properties then we conceive of an individual identical to X in terms of normative properties. We cannot conceive of another individual identical to X in respect of all base properties but different from X in respect of normative properties; we know a priori that there cannot be such an individual. P13: If we conceive of an individual different from X in terms of normative properties then we conceive of an individual different from X in terms of base properties. We cannot conceive of X being different in respect of normative properties without a difference in respect of base properties. C: (NS*) as a matter of conceptual necessity, whenever something has a normative property, it has a base property or collection of base properties that necessitate(s) the normative one.

It is worth emphasizing that this argument is supposed to be metaethically neutral. Crucially, it is supposed to be an argument that should move even the non-naturalist who thinks that normative properties are not base properties. From the perspective of the non-naturalist, we cannot then merely be engaging in the trivial activity of imagining another individual identical to X. For this argument to generate the conclusion (NS*) we have to be engaging in the exercise of conceiving of another individual identical to X only in respect of base properties that are uncontroversially base properties. In other words, we are not permitted to use any normative concepts in conceiving of this individual. If (RH) is true this could include all of the base properties of the entire world in which the individual is situated, as relational properties of the individual. Assume that conceiving of this is ⁵⁸ Hare (1952) p. 145.

⁵⁹ Dreier (m.s.) p. 3, Ridge (2007) p. 335.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

18

Debbie Roberts

possible. The argument requires that we conceive of only the (subset of ) base properties of X-in-a-world in such a way that necessitates that X* has the same normative property as X. Crucially though, since (IT) may be true, we’ve no guarantee that we will be able to conceive of all and only the base properties for any and all normative properties. According to (IT) there is no guarantee that we will be able to identify the base property grounds for the attribution of the thick property in every token case. We should conclude from this that it is possible that in conceiving of X*—another individual identical to X only in terms of base properties and without employing any normative concepts—we are not conceiving of something which necessitates that X* has the same normative property. If we do get the necessitation, it is possible that we have only done so by illegitimately employing a normative concept or two and thus allowing normativity involving properties into the supervenience base. As far as our concepts are concerned, it is possible to conceive of an individual identical to X in terms of base properties but different in terms of normative properties. Any resistance that is felt here is caused, I think, by the fact that most of the time when people perform this thought experiment they do not have either (RH) or (IT) in mind, and so they are potentially smuggling normativity involving properties into the supervenience base. The upshot is that it is not the case that, as a matter of conceptual necessity, whenever something has a normative property, it has a base property or collection of base properties that necessitate(s)the normative one. (NS*) is not a conceptual truth. It is not a conceptual truth, at least in part, because it can be rejected on a non-naturalist view that accepts both (RH) and (IT).

1 .5 . GO O D N E W S F O R N O N -R E D U C T I V I S T S A N D N O N - N AT UR A LI S T S ? The supervenience of the normative is used as a dialectical weapon in various ways, particularly against non-reductive and non-naturalist views. Any argument relying on (NS*) is undermined by my argument in Section 1.4, assuming that argument is correct.⁶⁰ This is indeed good news for defenders of these views. It is also important more generally. The notion that (NS*) or something close to it is a truism of normative discourse is extremely widely held and influential in various ways, not all of them ⁶⁰ Or the epistemic, aesthetic, or moral versions of (NS*).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Why Believe in Normative Supervenience?

19

as direct and obvious as the explicit employment of (NS*) in an objection to a view. Supervenience objections are nonetheless important, and not all supervenience objections employ (NS*). In this section I consider two prominent examples.

1.5.1. The ‘Reduction’ Objection The first is the argument for the claim that normative supervenience entails that normative properties are descriptive properties. For ease of exposition I use Frank Jackson’s terminology, though it should be clear that by ‘descriptive’ Jackson means (language that ascribes) base properties as defined in Section 1.1.2. I’m not going to rehearse Jackson’s argument here. Suffice to say that the purported conceptual truth of a supervenience thesis plays a crucial role: [T]he global supervenience of the ethical on the descriptive is special in that an unrestricted form, namely (S) For all w and w*, if w and w* are exactly alike descriptively then they are exactly alike normatively is both a priori true and necessary.⁶¹

There are already a number of objections to this argument, but none that cut it off, so to speak, at the source. If what I say in Section 1.4 is correct, then it fails at the first step. In general, though strong individual supervenience entails global supervenience, it is controversial whether the reverse is the case. Thus we cannot simply claim that (S) fails to be a conceptual truth because (NS*) does. However, at least some of the arguments I gave against the conceptual truth of (NS*) seem to tell against (S) as well. Consider the question of why we should believe (S)? The argument commonly given here is a global version of the conceivability argument, and Jackson means this argument to move at least all cognitivists including non-naturalists. The same points made against the individual version of that argument thus apply against the global version: given the possibility of (RH) and, crucially, (IT) how can we be certain that when we conceive of a world w* identical to w in all descriptive respects that we are also necessarily conceiving of a world identical to w in all normative respects? How can we be sure that we are not illegitimately smuggling normative respects into the base? ⁶¹ Jackson (1998) p. 119.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

20

Debbie Roberts

1.5.2. The ‘Explanation’ Objection The ‘explanation’ objection is a supervenience objection to non-naturalist realism in particular.⁶² I focus on a recent statement of the objection, by Tristram McPherson.⁶³ McPherson’s key claim is that non-naturalist realists are committed to the supervenience of the ethical being a metaphysically brute necessary connection between distinct properties, and that (given a modest version of Hume’s dictum) this is a significant theoretical cost. The supervenience claim that plays a crucial role in this argument is as follows: (M) No metaphysically possible world that is identical to a second world in all base respects can be different from the second world in its ethical respects.

This is a global supervenience thesis. McPherson explicitly states that his supervenience thesis is purely metaphysical.⁶⁴ However, this is a purely metaphysical thesis that is also supposed to be one that the non-naturalist realist is committed to, in McPherson’s view. The quick and dirty response to McPherson here is that the arguments in Section 1.4 establish that the non-naturalist who accepts (RH) and (IT) can reject (M). However, it is worth considering the reasons McPherson gives for holding (M): I take ethical supervenience theses to be best motivated by a two-part process. The first stage is to consider particular instances of what they rule out. For example, it seems impossible that another world might be identical to this one except that in the other world, a genocide otherwise identical to the actual Rwandan genocide differed solely in being ethically wonderful, rather than being an atrocity. Cases like this postulate a necessary connection: we seem to have discovered on reflection that the ethical features of the genocide cannot vary independently of its other features . . . The second stage of the process involves noticing that our views about these specific cases do not seem to rest on idiosyncrasies of the cases: analogous specific supervenience facts about everyday promising, for example, seem as compelling as such facts about radical evil like genocide. This point can encourage the inductive thought that it is impossible for there to be a case in which supervenience fails. After all, what would such a case look like?⁶⁵ ⁶² Simon Blackburn is the original source of this objection, however Blackburn’s version is significantly different. See Blackburn (1971, 1984a, 1984b), Miller (m.s.). There are versions of this objection that employ (NS*), and that are thus undermined by my arguments in Section 1.4. See Dreier (m.s.). ⁶³ McPherson (2012). ⁶⁴ McPherson (2012) pp. 215–16. ⁶⁵ McPherson (2012) pp. 211–12, my italics. In conversation, Jamie Dreier mentioned a similar motivation for believing the supervenience thesis: that we cannot think of a counterexample to it. Maybe we can’t but that could be because of the complexities

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Why Believe in Normative Supervenience?

21

McPherson appeals here to the same kind of conceivability argument that I argued against in Section 1.4.3. What my arguments in Section 1.4 were meant to call into question was precisely whether these kinds of reflections result in discovering, in a suitably metaethically neutral way, that normative features cannot vary independently of base features. It is true that McPherson uses the term ‘other’ here rather than ‘base’. This is important. It is implausible that a genocide could differ solely in being ethically wonderful rather than being an atrocity. But this does not establish the thesis of the supervenience of the normative on the base (or the ethical on the base), or even a very specific such supervenience claim for a non-naturalist realist who holds that normative properties are not base properties and who accepts (RH) and (IT). On such a view, to hold fixed (only) the base properties of the Rwandan genocide (R) would not be to try to conceive of the Rwandan genocide* (R*) with all of its properties except the property of being an atrocity. Again, we are not permitted to use any normative concepts in conceiving of (R*). Since we are taking out all of the ethical (normative) properties from the base, then, given (RH) and (IT) it is possible that what we are left with would not necessitate the property of being an atrocity.

1 .6 . C O N C L U D I N G R E M A R K S The core of my argument in this chapter is that supervenience of the normative on the base can be rejected by non-naturalist views that hold (RH) and (IT). This is what allows my responses in Section 1.5 and, though this is not the only reason I argued that (NS*) is not a conceptual truth, it is perhaps the main one. One might take my arguments to be a reductio of (RH) and (IT). But the dialectic here is important. There are metaphysical commitments that would support the reductio view, but these commitments are not metaethically neutral, and neither is (NS*). The prevalence of (NS*) and similar in metanormative theorizing obscures, and explains the neglect of, the possibility that there is normativity ‘all the way down’.⁶⁶ introduced by (RH) and (IT). In other words that we can’t think of one, if we can’t, doesn’t show there couldn’t be one. ⁶⁶ Earlier versions of this chapter were presented at: The Normativity: Epistemic and Practical Conference at University of Southampton; Dancy Day at University of Edinburgh; Philosophy Club at University of St Andrews; Normative Supervenience/ Grounding Workshop, Centre for the Study of Mind and Nature, University of Oslo; The Varieties of Normativity Conference, Uppsala University; University College Dublin;

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

22

Debbie Roberts

References Bender, John. (1987) ‘Supervenience and the Justification of Aesthetic Judgments’ Journal of Aesthetics and Art Criticism 46 (1): 31–40. Berker, Selim. (forthcoming) ‘The Unity of Grounding’ Mind. Blackburn, Simon. (1971) ‘Moral Realism’ in John Casey (ed.), Morality and Moral Reasoning, London: Methuen. Blackburn, Simon. (1984a) Spreading the Word, Oxford: Oxford University Press. Blackburn, Simon. (1984b) ‘Supervenience Revisited’ in Ian Hacking (ed.), Exercises in Analysis: Essays by Students of Casimir Lewy, Cambridge: Cambridge University Press. Dancy, Jonathan. (1993) Moral Reasons, Oxford: Blackwell. Dancy, Jonathan. (1995) ‘In Defense of Thick Concepts’ Midwest Studies on Philosophy 20 (1): 263–79. Dancy, Jonathan. (2004) Ethics Without Principles, Oxford: Oxford University Press. Dreier, James. (1992) ‘The Supervenience Argument Against Moral Realism’ Southern Journal of Philosophy 30 (3): 13–38. Dreier, James. (m.s.) ‘Is There a Supervenience Problem for Robust Moral Reliasm?’ Elstein, Daniel and Hurka, Thomas. (2009) ‘From Thick to Thin: Two Moral Reduction Plans’ Canadian Journal of Philosophy 39 (4): 515–35. Enoch, David. (2011) Taking Morality Seriously: A Defence of Moral Realism, Oxford: Oxford University Press. Fine, Kit. (2002) ‘Varieties of Necessity’ in Tamar Szabo Gendler and John Hawthorne (eds), Conceivability and Possibility, Oxford: Oxford University Press. Fudge, Robert. (2005) ‘A Vindication of Strong Aesthetic Supervenience’ Philosophical Papers 34 (2): 149–71. Gibbard, Alan. (2003) Thinking How to Live, Cambridge, MA: Harvard University Press. Griffin, James. (1992) ‘Values, Reduction, Supervenience, and Explanation by Ascent’ in David Charles and Kathleen Lennon (eds), Reduction, Explanation, and Realism, Oxford: Clarendon Press. Hare, R. M. (1952) The Language of Morals, Oxford: Clarendon Press. Hattiangadi, Anandi. (m.s.) ‘Nonnaturalism and Supervenience’. Heathwood, Chris. (2012) ‘Could Morality Have a Source?’ Journal of Ethics and Social Philosophy 6 (2): 1–19. Hick, D. H. (2012) ‘Aesthetic Supervenience Revisited’ British Journal of Aesthetics 52 (3): 301–16. Hills, Alison. (2009) ‘Supervenience and Moral Realism’ in Hieke Alexander and Leitgeb Hannes (eds), Reduction, Abstraction, Analysis, Heusenstamm: Ontos Verlag.

and the 2nd Chapel Hill Metaethics Workshop at University of North Carolina, Chapel Hill. I am grateful to all these audiences for helpful and enjoyable discussions. Thanks also to Jonathan Dancy, Guy Fletcher, Matthew Lutz, Pekka Väyrynen, and two anonymous referees for helpful written comments.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Why Believe in Normative Supervenience?

23

Hurley, Susan. (1989) Natural Reasons: Personality and Polity, Oxford: Oxford University Press. Jackson, Frank. (1998) From Metaphysics to Ethics: A Defence of Conceptual Analysis, Oxford: Oxford University Press. Jackson, Frank, Pettit, Philip, and Smith, Michael. (2000) ‘Ethical Particularism and Patterns’ in Brad Hooker and Margaret Olivia Little (eds), Moral Particularism, Oxford: Oxford University Press. Kallestrup, Jesper and Prichard, Duncan. (forthcoming) ‘Epistemic Supervenience, Anti-Individualism and Knowledge First Epistemology’ in J. Adam Carter, Emma Gordon, and Benjamin Jarvis (eds), Knowledge-First Epistemology, Oxford: Oxford University Press. Kim, Jaegwon. (1988) ‘Supervenience for Multiple Domains’ Philosophical Topics 16 (1): 129–50. Klagge, James C. (1988) ‘Supervenience: Ontological and Ascriptive’ Australasian Journal of Philosophy 66 (4): 461–70. Korsgaard, Christine. (1993) ‘The Reasons We Can Share: An Attack on the Distinction Between Agent-Relative and Agent-Neutral Values’ Social Philosophy and Policy 10 (1): 24–51. Korsgaard, Christine. (1996) The Sources of Normativity, Cambridge: Cambridge University Press. Leary, Stephanie. (2017) ‘Non-Naturalism and Normative Necessities’ in Russ Shafer-Landau (ed.), Oxford Studies in Metaethics, vol. 12, Oxford: Oxford University Press. Lehrer, Keith. (1997) Self-Trust: A Study of Reason, Knowledge, and Autonomy, Oxford: Oxford University Press. Leuenberger, Stephan. (2014) ‘Grounding and Necessity’ Inquiry 57 (2): 151–74. Levinson, Jerrold. (1984) ‘Aesthetic Supervenience’ Southern Journal of Philosophy 22 (S1): 93–110. McDowell, John. (1981) ‘Non-Cognitivism and Rule Following’ in S. Holtzman and Christopher M. Leich (eds), Wittgenstein: To Follow a Rule, London: Routledge. McPherson, Tristram. (2012) ‘Ethical Non-Naturalism and the Metaphysics of Supervenience’ in Russ Shafer-Landau (ed.), Oxford Studies in Metaethics, vol. 7, Oxford: Oxford University Press. McPherson, Tristram. (2015) ‘Supervenience in Ethics’ in Edward N. Zalta (ed.), The Stanford Encyclopedia of Philosophy, winter 2015 edition . Miller, Alex. (m.s.) ‘Moral Supervenience: A Defence of Blackburn’s Argument’. Raz, Joseph. (2000) ‘The Truth in Particularism’ in Brad Hooker and Margaret Olivia Little (eds), Moral Particularism, Oxford: Oxford University Press. Ridge, Michael. (2007) ‘Anti-Reductionism and Supervenience’ Journal of Moral Philosophy 4 (3): 330–48. Roberts, Debbie. (2011) ‘Shapelessness and the Thick’ Ethics 121 (3): 489–520. Roberts, Debbie. (2013a) ‘It’s Evaluation, Only Thicker’ in Simon Kirchin (ed.), Thick Concepts, Oxford: Oxford University Press. Roberts, Debbie. (2013b) ‘Thick Concepts’ Philosophy Compass 8 (8): 677–88.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

24

Debbie Roberts

Roberts, Debbie. (2017) ‘Thick Concepts’ in Tristram McPherson and David Plunkett (eds), The Routledge Handbook of Metaethics, London: Routledge. Rosen, Gideon. (m.s.) ‘Normative Necessity’. Scanlon, T. M. (2014) Being Realistic About Reasons, Oxford: Oxford University Press. Schroeder, Mark. (2005) ‘Realism and Reduction: The Quest for Robustness’ Philosophers’ Imprint 5 (1): 1–18. Schroeder, Mark. (2014) ‘The Price of Supervenience’ in Mark Schroeder, Explaining the Reasons We Share: Explanation and Expression in Ethics, Volume I, Oxford: Oxford University Press. Sibley, Frank. (1959) ‘Aesthetic Concepts’ Philosophical Review 68 (4): 421–50. Skarsaune, Knut Olav. (2015) ‘How to Be a Moral Platonist’ in Russ Shafer-Landau (ed.), Oxford Studies in Metaethics, vol. 10, Oxford: Oxford University Press. Skiles, Alexander. (2015) ‘Against Grounding Necessitarianism’ Erkenntnis 80 (4): 717–51. Smith, Michael. (2004) ‘Does the Evaluative Supervene on the Natural?’ in Michael Smith, Ethics and the A Priori: Selected Essays on Moral Psychology and Meta-Ethics, Cambridge: Cambridge University Press. Sosa, Ernest. (1991) Knowledge in Perspective: Selected Essays in Epistemology, Cambridge: Cambridge University Press. Sturgeon, Nicholas. (2009) ‘Doubts About the Supervenience of the Evaluative’ in Russ Shafer-Landau (ed.), Oxford Studies in Metaethics, vol. 4, Oxford: Oxford University Press. Turri, John. (2010) ‘Epistemic Supervenience’ in Jonathan Dancy, Ernest Sosa, and Matthias Steup (eds), Blackwell Companion to Epistemology, 2nd edition, Oxford: Blackwell. Van Cleve, James. (1985) ‘Epistemic Supervenience and the Circle of Belief ’ The Monist 68 (1): 90–104. Väyrynen, Pekka. (2008) ‘Slim Epistemology with a Thick Skin’ Philosophical Papers 37 (3): 389–412. Väyrynen, Pekka. (2013) ‘Grounding and Normative Explanation’ Aristotelian Society Supplementary Volume 87 (1): 155–78. Wicks, Roberts. (1988) ‘Supervenience and Aesthetic Judgment’ Journal of Aesthetics and Art Criticism 46 (4): 509–11. Wicks, Roberts. (1992) ‘Supervenience and the “Science of the Beautiful” ’ Journal of Aesthetics and Art Criticism 50 (4): 322–4. Zangwill, Nick. (1992) ‘Long Live Supervenience’ Journal of Aesthetics and Art Criticism 50 (4): 319–22. Zangwill, Nick. (1994) ‘Supervenience Unthwarted: Rejoinder to Wicks’ Journal of Aesthetics and Art Criticism 52 (4): 466–9. Zangwill, Nick. (2003) ‘Beauty’ in Jerrold Levinson (ed.), Oxford Companion to Aesthetics, Oxford: Oxford University Press. Zangwill, Nick. (2014) ‘Aesthetic Judgement’ in Edward N. Zalta (ed.), The Stanford Encyclopedia of Philosophy, fall 2014 edition .

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

2 Non-Naturalism Gone Quasi Explaining the Necessary Connections between the Natural and the Normative Teemu Toppinen

A few facts: capitalism is a source of unending degradation; a lot of what Donald Trump does is scummy; we ought to view the level of our greenhouse gas emissions with abhorrence; had I promised to smuggle six references to Finnish death metal into this paper, there would be some reason for me to do so; Rotten Sound is pretty awesome;¹ so is football; and Helsinki isn’t bad in the summer. Facts such as these—normative facts— seem quite unlike natural facts. Natural properties (e.g., being filled with water, being a parent, redness) and facts (e.g., the fact that I am a parent) are, roughly, properties and facts such that could figure in empirical regularities and that we can find out about by scientific or empirical means (cf. Smith 2000). Normative properties and facts seem very different. They seem to be something “over and above” the natural ones. I won’t say much about what this amounts to. Being filled with water plausibly isn’t anything over and above, but rather seems identical to, being filled with H₂O. My being a parent isn’t anything over and above, but rather fully consists in, my having a son. And an object’s being colored might be nothing over and above its being red, for being red is a way of being colored. However, that a certain type of action is wrong (say) doesn’t seem to be just a matter of how things are naturalistically speaking. Some natural features may make an action wrong, alright, but if so, the wrongness of the action nevertheless seems to be something over and above those natural features (see Enoch 2011, pp. 100–9; Parfit 2011, pp. 299–301). If that’s all correct—and I don’t expect

¹ That’s four or five, already (at least if deathgrind counts).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

26

Teemu Toppinen

to have made any new converts here—we should accept the following view, which I will call non-naturalism:² NON-NAT Normative properties and facts exist; they are sui generis, that is: different in kind from, and something over and above, natural and any other kind of properties and facts.

Non-naturalism is a fairly popular view (e.g., Enoch 2011; Parfit 2011, 2017; Scanlon 2014). Interestingly, and importantly, this view may be combined either with cognitivism or with non-cognitivism (cf. Dreier 2004, 2015): COG For any normative proposition, p, of the form “x is N ” (where “N ” stands for a normative property), (1) to think that p is to represent x as having the property of being N, and (2) this—the truth of (1)— doesn’t hold just in virtue of the fact that to think that p is to be in a certain kind of broadly desire-like state. NON-COG For any normative proposition, p, of the form “x is N,” (i) to think that p is to be in a certain broadly desire-like state of mind, and (ii) if it is also to represent x as having the property of being N, this is so just in virtue of the truth of (i).

The issue between cognitivism and non-cognitivism concerns, then, what it is in virtue of which some state of mind counts as having a normative belief, or a belief with a normative content. I take it that this comes at least pretty close to understanding the distinction between these views as a distinction in metasemantics—that is, as a distinction between different accounts of what it is in virtue of which certain meaningful items have the meanings that they have (see, e.g., Ridge 2014; Chrisman 2016). Combining non-naturalism with cognitivism gives us realist non-naturalism; the combination of nonnaturalism and non-cognitivism results in quasi-realist non-naturalism. Realist non-naturalism has been widely considered to have a hard time explaining the metaphysically necessary relations between the natural and the normative. In particular, many have worried that realist non-naturalism lacks the resources to explain the metaphysical necessities that figure in SUPER (see, e.g., McPherson 2012; Leary 2017; Rosen MS): SUPER Whenever something has a normative property, it also has some natural property such that it is metaphysically necessary that anything that has this natural property also has the normative property.

I argue that while realist non-naturalists do indeed seem to be in trouble with regard to explaining the metaphysically necessary links between the natural and the normative, accepting non-cognitivism makes all the difference.

² This formulation derives, with some changes, from the one given in Leary 2017.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Non-Naturalism Gone Quasi

27

Some may be inclined to suspect that quasi-realist non-naturalism represents a curious, badly motivated position in the metanormative landscape. I cannot address this issue in any detail here, but I do think that this suspicion would be misguided. The central arguments for non-naturalism—for instance, appeals to the intuition that normative properties are “just too different” from the natural ones (Parfit 2011), or to the idea that normative facts are indispensable for deliberation (Enoch 2011)—do not seem to point away from quasirealism. Likewise, many of the central arguments for non-cognitivism— those that do not simply presuppose some very austere form of naturalism; e.g., arguments from the practical character of normative thought—do not direct us away from non-naturalism, as long as non-naturalism is being understood in a quasi-realist friendly way. So, while there may be an element of surprise to the combination of non-naturalism and non-cognitivism, I take the resulting view to be not just plausible, but also a view that expressivists such as Blackburn (1998) and Gibbard (2012) and “non-metaphysical realists” such as Parfit (2011, 2017) and Scanlon (2014) would, or at least should, be sympathetic to. In Section 2.1, I very briefly explain how SUPER raises a challenge for nonnaturalism, and how—as has been recently pointed out by Jamie Dreier (2015)—it’s not at all obvious that quasi-realist non-naturalism offers a way of escaping the challenge. In Section 2.2, I briefly explore different kinds of accounts of what it is to have thoughts concerning metaphysical necessity. I then proceed to argue, in Section 2.3, that once we understand the explanatory challenge regarding SUPER in the light of a quasi-realist take on the relevant necessity judgments, this challenge takes the shape of a firstorder normative issue, and will then be answerable by the quasi-realists’ lights. When it comes to explaining the necessary connections between the normative and the natural, all will be fine, it seems, if non-naturalists just go a little quasi. Section 2.4 briefly concludes.

2.1. NON-NATURALISM AND THE SUPERVENIENCE CHALLENGE With many others—and following Dreier (1992, MS), in particular—I take the normative to strongly supervene on the natural: SUPER* ☐C(8F in α) (8x)[Fx ! (∃G in β) (Gx & ☐M(8y) (Gy ! Fy))]

This requires some explanation: α is the family of normative properties, β is the family of natural properties; ☐C stands for conceptual necessity, ☐M for

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

28

Teemu Toppinen

metaphysical necessity. And so what the thesis says is that it is conceptually necessary that the following is true: SUPER Whenever something has a normative property, it also has some natural property such that it is metaphysically necessary that anything that has this natural property also has the normative property.

This thesis is becoming increasingly controversial, these days, it seems (see Sturgeon 2009; Roberts this volume; Rosen MS), but I shall suppose that it’s true.³ What, then, is the challenge that this supervenience thesis raises for non-naturalists? Very briefly, the problem is as follows (this brief characterization of the problem owes a lot especially to Tristram McPherson’s (2012) treatment of the issue). Let us set the claim that SUPER is a conceptual truth to one side, and focus on asking: why is it that certain natural features metaphysically necessitate certain normative features? It isn’t plausible that such necessities are simply brute, or unexplainable. They seem to be in need of explanation. On some views, the relevant necessities are easily explained. If normative properties were natural properties, then the necessitation of the normative by the natural would be easily accounted for. However, according to the non-naturalist, normative properties are sui generis and radically discontinuous from the natural properties. The question then is: why are there necessary connections between these wholly discontinuous properties? Non-naturalism doesn’t seem to allow for any explanation for this, but rather seems to treat these necessary connections as brute. If that is correct, then this is plausibly a big minus for the view. If non-naturalism doesn’t have the resources to offer an explanation for the necessary connections between the natural and the normative, this may not constitute a refutation of the view, but it seems to give us considerable reason to reject it.⁴ Perhaps this is not quite the right way to push the challenge. Perhaps non-naturalists can explain the relevant necessities by appealing to the idea that certain natural properties just have the property of making things good, ³ For some responses to concerns raised in Sturgeon 2009 and Rosen MS, see, e.g., Ridge 2007; McPherson 2012; Dreier MS. ⁴ The problem, or something in its neighborhood, has been advanced by a number of philosophers. In addition to McPherson 2012, see, e.g., Blackburn 1971, 1984, 1985; Dreier 1992, MS; Horgan 1993; Wedgwood 1999; Schroeder 2007, 2014; and Väyrynen 2017. Non-naturalists have of course offered different responses to this problem. See, e.g., Shafer-Landau 2003; Stratton-Lake and Hooker 2006; Enoch 2011; Scanlon 2014; Leary 2017; and Rosen MS (although Shafer-Landau may not qualify as a non-naturalist by my lights, as on his view the natural exhaustively realizes the normative, just like, according to a non-reductive physicalist in philosophy of mind, certain physical states exhaustively realize the mental ones). For criticisms of these responses, see, e.g., Dreier 2015, MS; Leary 2017; and Toppinen forthcoming).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Non-Naturalism Gone Quasi

29

right, etc., or to the idea that the normative features of things result from their natural properties (Dancy 2004, pp. 88–9; Olson 2014, pp. 88–100; cf. Enoch 2011, ch. 6; Scanlon 2014, ch. 2). Let’s first suppose that a straightforward version of maximizing hedonistic utilitarianism is true. Why is it necessary that if an action fails to maximize happiness, it is morally wrong? Well, this is necessary because if an action has the property of failing to be happiness-maximizing, this makes it the case that the action has the further property of being morally wrong. Or suppose, alternatively, that the correct normative theory implies that the natural features that make one action wrong may fail to make another action with these same natural properties wrong, thanks to some relevant difference in the circumstances. Still, if we hold enough of the natural properties of the two actions fixed, both actions will be wrong, as we have now ruled out any possible differences that would explain why wrongness would result in one case but not in the other. And so, given enough similarity in natural respects, necessitation of wrongness by certain natural features is explained. These responses appear quite sensible, perhaps. Yet they also seem unhelpful. The challenge was to explain why having certain natural properties necessitates having certain normative properties, where the normative properties are something over and above the natural ones. It is perhaps fine to say that the necessitation relations in question are explained by the fact that if something—for example, an action—has certain natural properties, this makes it wrong (or right, or whatever). But now the question is: how should we understand the making-relation here, given that the normative is something over and above the natural. We might say that if some substance is H₂O, this makes it water. Or that if I have a son, this makes me a father. Or that if a chair is red, this makes the chair in question colored. However, we get to say these things because certain intimate metaphysical relations hold between the relevant properties—because having one of the properties (being water, being a parent, being colored) is nothing over and above having the other (being H₂O, having a son, being red). If normative properties are sui generis, wholly discontinuous from the natural ones, then how is it that, for example, an action’s having a natural property can make it have a normative property, and thereby necessitate its being wrong? It seems that critics of non-naturalism may insist that even if non-naturalists do have an explanation to give—they can appeal to the making-relation (or to some more complex resultance-relation)—the explanation that they have to offer is a bad, metaphysically queer one.⁵ ⁵ Jonas Olson (2014) suggests that non-naturalists can provide a satisfying response to the supervenience challenge by appealing to the making-relation, and that if one insists on this being a metaphysically mysterious move, the worry concerning non-naturalism

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

30

Teemu Toppinen

It is often supposed that non-cognitivists easily escape the kind of challenge that I have outlined here. The thought is that for a noncognitivist, SUPER doesn’t even make any sense, strictly speaking (as there aren’t any normative properties, really), or should, in any case, be understood as merely articulating a certain kind of restriction on our normative concepts (e.g., Klagge 1988). The restriction would be that we treat possible scenarios, events, etc., that are exactly alike in all natural respects, the exact same way in our normative thinking. Moreover, it is sometimes suggested, non-cognitivists are well positioned to give an explanation for why this kind of restriction is in place. The thought is that noncognitivism may have promising resources for explaining why it is that the normative supervenes on the natural as a matter of conceptual fact. Here is Blackburn’s (1971, p. 122) early articulation of the basic idea (see also, e.g., Hare 1952, ch. 10; Blackburn 1984, pp. 185–6; Smith 1994, pp. 24–5; Gibbard 2003, ch. 5): There can be no question that we often choose, admire, commend, or desire, objects because of their naturalistic properties. Now it is not possible to hold an attitude to a thing because of its possessing certain properties and, at the same time, not hold that attitude to another thing that is believed to have the same properties. The nonexistence of the attitude in the second case shows that it is not because of the shared properties that I hold it in the first case. Now, moral attitudes are to be held towards things because of their naturalistic properties. Therefore it is not possible to hold a moral attitude to one thing, believe a second to be exactly alike, yet at the same time not hold the same attitude to the second thing. Anybody who appears to do this is convicted of misidentifying a caprice as a moral opinion.

This may need some revising. We should probably allow for the possibility of violating the supervenience constraint. So, perhaps it is possible to have an attitude of the relevant kind toward one thing on the basis of its having certain naturalistic properties, to believe that another thing has the same naturalistic properties, and yet to fail to have the relevant attitude toward the latter thing. It’s just that this is incoherent. Having this combination of attitudes fails to make sense. This seems, indeed, quite plausible. Suppose that, having heard a recording of a song by the band Convulse from 1992, I judge the recording in question to be wonderful, thanks to its instantiating musical murkiness and brutality in certain most appealing ways. Now, suppose that, miraculously, I then instantly come across another recording collapses to a worry about the sheer intuitive queerness of normative properties. But this doesn’t seem right, as the worry may still concern primarily the queerness of the explanation of necessary relations between the natural and the normative properties. Normative properties end up seeming queer because there’s no good explanation for the relevant necessitation relations (see Toppinen 2016).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Non-Naturalism Gone Quasi

31

from 1992, which is sonically indistinguishable from the first one.⁶ And suppose, moreover, that I have no other (possibly relevant) knowledge about the properties of these recordings. If I now deem this latter recording to be completely non-wonderful, I perhaps hold a possible combination of attitudes, but this combination of attitudes fails to make any sense. As an evaluator, I am bound to find things wonderful on the basis of their natural properties, and so, if I form judgments about the wonderfulness of the two recordings on the basis of the evidence that is available to me, I am committed to deeming them equally wonderful. This is compatible with the idea that, had the recordings been sonically distinguishable, it might have made sense for me to judge them differently despite being unable to articulate any reason for why this is so (“Yeah, this, too, is murky and brutal—makes your ears bleed, huh—but, I don’t know, there’s something missing here”). And of course, given additional knowledge, I might be sensitive to various other features of the recordings. For instance, a miraculously sonically indistinguishable recording, which I know to be from a retro band in 2016, might not sound so exciting to me. Now, it is controversial whether any non-cognitivist explanation of broadly this sort, for the status of SUPER as a conceptual truth, is successful (see Zangwill 1995, p. 246; Shafer-Landau 2003, pp. 88–9; Sturgeon 2009, pp. 82–7). But suppose that some such explanation can be made to work. Still, as has been recently observed by Dreier (2015), this would not mean that the quasi-realist non-naturalist would have any edge over the realist non-naturalist when it comes to explaining the particular metaphysical necessities that SUPER commits us to. Or it is not, in any case, instantly clear why the quasi-realist would be in any happier position. This is so because we need to distinguish between two different explanatory challenges. One is to explain why SUPER is a conceptual truth. The other one is to explain why certain necessary connections between the natural and the normative—those that SUPER commits us to—obtain. I have already outlined a quasi-realist answer to the former question. But, as is nicely explained by Dreier (2015, pp. 289–90), this answer does not seem to help us, at all, with regard to answering the latter challenge. SUPER tells us that the normative features of things are metaphysically necessitated by their natural features. Let us suppose that one of the relevant metaphysical necessities is this: that happiness-maximizing actions are morally right. Non-naturalists—both realist and quasi-realist—may accept that this is so. They may both admit that some actions really have the property of being morally right,

⁶ Some of those who are not connoisseurs of death metal may fail to find this scenario very miraculous.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

32

Teemu Toppinen

thanks to their being happiness-maximizing, or hedonically optimal. The puzzling question then is, again, why is it metaphysically necessary that whatever is hedonically optimal is right? Why the necessary connection between these (by the non-naturalists’ lights) completely distinct, very different, properties? As already noted, the realist non-naturalist doesn’t seem to have an easy response to this question. But nothing in the quasirealist’s explanation for why SUPER is a conceptual truth seems to directly suggest an answer either. And so it might very well seem that the quasirealist is in the exact same boat with the realist with regard to this challenge. At least it remains to be seen why this would not be so.⁷

2 . 2. M O D A L RE A LI SM A N D Q U A S I - R E A L I S M There may be a good, non-naturalism-friendly way of responding to the explanatory challenge that SUPER gives rise to. This can be seen once we consider what kind of shape the explanatory challenge takes on the assumption that non-cognitivism is true. The key is to realize that the challenge of explaining the necessary relations between the natural and the normative, itself, plausibly takes a different shape depending on how we understand thought about normative matters, and on how we understand thought about what is metaphysically necessary. Given realist non-naturalism, the challenge emerges as difficult to answer; given quasi-realist assumptions, it turns out to be more tractable. Or that is what I will argue for in this section and in Section 2.3. ⁷ Although I shall not try to press the point here, I will just note that if quasi-realism has the resources to explain the status of SUPER as a conceptual truth, this might be an important point in its favor. It is not clear, after all, that the realist non-naturalist can muster any satisfying explanation here. It is sometimes suggested that this kind of explanation is easily available. Olson (2014, p. 90; see also Dreier 1992, p. 21; Enoch 2011, pp. 148–50) writes that “non-naturalists can appeal to what I claim would be a natural reaction to a speaker who flouts [this thesis], namely that such a person has failed to grasp our moral concepts adequately.” This may be a good way of arguing that SUPER is a conceptual truth, but one could still reasonably demand an explanation for why this is so. On Olson’s (2014, p. 90) view, we realize the truth of SUPER by reflecting on the nature of moral properties “just as when we reflect on the property of being a rectangle we realize that rectangles can have no more and no less than four corners.” However, the cases seem importantly different. Whereas rectangles’ having four corners flows straightforwardly from the content of our rectangle-beliefs, the truth of SUPER doesn’t seem to flow in a similar way from our moral concepts—at least given that non-naturalism is true, and that our basic normative concepts are unanalyzable. Finally, even if no explanation is needed for the status of SUPER as a conceptual truth, having one might nevertheless be a neat bonus (see Shafer-Landau 2005, p. 328).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Non-Naturalism Gone Quasi

33

In order to see how the explanatory challenge at issue may take different shapes depending on our choice of metanormative theory, we should look into what it is to think that such-and-such is metaphysically necessary. Just as in metaethics, with metaphysical modality, too, we may distinguish between representationalist and non-representationalist views: ☐M-REP For any proposition, p, concerning some proposition, q, being metaphysically necessary, (1) to think that p is to represent modal reality as being a certain way, and (2) this—the truth of (1)— doesn’t hold just in virtue of the fact that to think that p is to be in a certain kind of non-representational state. ☐M-N-R For any proposition, p, concerning some proposition, q, being metaphysically necessary, (i) to think that p is to be in a certain non-representational state of mind, and (ii) if it is also to represent modal reality as being a certain way, this is so just in virtue of the truth of (i).

We are only interested, in the present context, in views according to which there really are facts about what is necessary, and these facts are suitably mind-independent. With these assumptions in place, the representationalist view amounts to modal realism. There is no need to consider in any great detail the variety of forms that realist views about metaphysical necessity might take. On some such views, claims about metaphysical necessity represent the modal properties of the actual world. We might call such views, with Amie Thomasson (2007, MS), forms of heavyweight modal realism. On other modal realist views, modal claims rather represent modal reality as being a certain way by representing the distribution of non-modal features also in non-actual worlds. Thomasson calls such a view possible worlds realism or Lewisian realism.⁸ According to modal realist views, thinking that having one property, Fness, necessitates having another, Gness, is usefully understood as representing Fness as having the property of necessitating Gness, or as representing Fness as being accompanied by Gness in all possible worlds. I am admittedly being very sparse on details here, but the differences between the different realist views shouldn’t matter much in what follows (for discussion of different varieties of modal realism, see Divers 2002). However, we must give a somewhat closer look at the nonrepresentationalist or quasi-realist option. The basic idea behind many of the non-representationalist approaches would be that when someone thinks, for instance, that all bachelors must be males, this is a matter of endorsing a rule for thought (Thomasson 2007, 2009, 2013, MS; cf. Ryle 1950; Sellars 1958; Blackburn 1984, pp. 213–17, 1987; Brandom 2008). On Amie ⁸ Thomasson is not claiming that David Lewis’s (1986) view would necessarily be Lewisian in this sense.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

34

Teemu Toppinen

Thomasson’s (2007, MS) view, for example, the function of claims about metaphysical necessity is to convey, or to express, constitutive semantic rules while remaining in the object-language. One of the constitutive semantic rules governing the term “bachelor” may be stated in the imperative mood, in metalanguage, as follows: “Apply ‘bachelor’ only to things to which ‘man’ also applies.” But we may also conveniently convey, or express, this rule in the object-language by saying “Necessarily, bachelors are men.” Or that is, very roughly, Thomasson’s idea. We could then say that, on this view, thinking that it is necessary that bachelors are men is, roughly, a matter of endorsing—in virtue of mastering the constitutive meaning rules for “bachelor”—thinking of something, x, as a bachelor only if x is a man. Or perhaps we could say that this would be a matter of having—in virtue of mastering the constitutive meaning rules for the term—a certain kind of disposition or a “plan” (Gibbard 2012) to think of someone as a bachelor only if he is a man. I should add that I will understand the relevant sort of endorsement (or disposition) as involving an endorsement of moving (or a disposition to move), if the question arises, from thinking that someone, x, is a bachelor to thinking that x is a man. This is a little rough, but perhaps sufficient for our purposes here. Metaphysical necessities are often a posteriori. How would a view along these lines make sense of claims concerning such necessities? When someone claims that, necessarily, water is H₂O, this cannot be a matter of her conveying a constitutive meaning rule governing our use of “water.” The meaning rules for “water” don’t involve H₂O, after all; one can fully master the rules for the proper use of “water” without having the resources to consider the possibility of being composed of H₂O molecules. Thomasson’s view captures a posteriori necessities as involving schematic and worlddeferential constitutive semantic rules, which are then filled in with suitable empirical facts (Thomasson MS, ch. 4). The idea is that when someone accepts that, necessarily, water is H₂O, this is, indeed, a matter of her conveying a rule that says to apply “water” only to whatever “H₂O” applies to, but this rule is derived from a more fundamental rule together with empirical information. The more fundamental rule is a constitutive semantic rule governing “water,” which says to apply “water” only to whatever has a certain kind of microstructure—roughly, to whatever has the microstructure that the actual watery stuff of our acquaintance has. This rule, together with the a posteriori knowable fact that the actual watery stuff of our acquaintance has the H₂O-microstructure, gives us, then, the derived rule that tells to apply “water” only to H₂O. So, we could say that when someone thinks that, necessarily, water is H₂O, this is a matter of her endorsing—in virtue of mastering the semantic rules for “water”—applying “water” only to stuff of a certain kind, and of her believing that that stuff is made of

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Non-Naturalism Gone Quasi

35

H₂O. Or perhaps it is a matter of her endorsing the derivative rule on the basis of being in these states of mind.⁹ Thomasson doesn’t discuss normative necessities, but we may consider how a view along the lines that she proposes would handle such cases. Suppose that we wish to know what it is to think the following: that it is metaphysically necessary, that if an action is happiness-maximizing, it is morally right. As with water and H₂O, this cannot be a matter of endorsing any movement of mind simply in virtue of having mastery of the semantic rules governing the relevant terms. At least according to the non-naturalists, the semantic rules for “morally right” won’t link proper applications of the term to any particular natural properties in the way that the rules for “bachelor” do—but also not in the way that the rules for “water” (perhaps) do. However, we could extend Thomasson’s view to the normative case by saying, for example, that thinking that it is necessary that happiness-maximizing acts are right amounts to endorsing, if the question arises, thinking of any act that is happiness-maximizing, that it is also morally right. Perhaps we could also say that here, too, one conveys what follows from the semantic rules governing “morally right” with certain further assumptions. A quasi-realist, at least, could say that the relevant semantic rule tells us, roughly, to apply “morally right” only to those actions that one approves of.¹⁰ This rule would then be supplemented by one’s standards for approval (e.g., yay for happiness-maximizing!). It is not clear to me how this would work with realist non-naturalism. Here is one possibility: a realist could say that the semantic rule for “morally right” completely uselessly tells us to apply this term to the morally right things. This rule would then be supplemented by the deliverances of a priori intuition telling us that happiness-maximizing actions are right (say). Another recent non-representationalism-friendly account of necessity thought has been articulated by John Divers (in collaboration with others; see Divers and Elstein 2012; Divers and Gonzáles-Varela 2013).¹¹ Divers et al. seek a functional characterization of necessity thought—or of “boxbelief,” as they put it—by offering an account of the “proper” (or normal) ways of forming box-beliefs, as well as of the proper ways of manifesting them. They do not take a stance on the issue of modal realism vs. quasi-realism, ⁹ For discussion of other sorts of cases, such as the necessity of Hilary Clinton’s having originated from Dorothy Howell and Hugh Rodham, see Thomasson MS. ¹⁰ Or better, the rule would tell us to apply “morally right” to those actions that have dthat[the property the instantiation of which you approve of]. The idea is not that the expression “morally right” stands for this property (as a speaker-relativist à la Dreier 1992 would have it), but rather that it is to be predicated of something, x, only when one thinks that x has this property. ¹¹ Thanks to Pekka Väyrynen for pointing me toward Divers and Elstein 2012.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

36

Teemu Toppinen

but rather remain neutral on whether being a box-belief just is a matter of playing the relevant functional role, or whether playing this role is instead to be explained with reference to an account of box-belief as representing the modal reality. Nevertheless, their account is intended to be “at least consistent with various anti-realistic theories of modality” (Divers and Gonzáles-Varela 2013, p. 361, n. 2). Roughly, on this account, someone (properly) believes that some proposition, q, is necessary, if and only if she is unable to make anything of the supposition that not-q, and if she is, moreover, prepared to appeal to q as a premise under any supposition.¹² A posteriori necessities raise a prima facie problem also for this kind of non-representationalist account. I think that it is necessary that water contains hydrogen. However, it still seems that I might be able to make something of a supposition that water does not contain hydrogen. Also, I might not be prepared to appeal to the proposition that water contains hydrogen in reasoning under any supposition—for instance under the supposition that water actually does not contain hydrogen, or under the supposition that, more generally, modern chemistry is largely mistaken. Divers and Gonzáles-Varela might then seem to fail in articulating necessary conditions for my having the relevant box-belief. The solution that Divers and Gonzáles-Varela offer is to make use of a distinction between two ways in which we may suppose that something is the case. To A-suppose that p is (roughly) to suppose-as-actual that p, whereas to C-suppose that p is (roughly) to suppose-as-counterfactual that p. These attitudes are quite different. It is one thing to A-suppose that Socrates is a robot, that is, to suppose that Socrates actually was a robot. It is a wholly different business to C-suppose that Socrates is a robot, that is, to suppose that Socrates had been a robot, even though he in fact wasn’t (Divers and Gonzáles-Varela 2013, pp. 362–6). With this distinction at hand, Divers and Gonzáles-Varela (2013, pp. 366, 381) propose an account of box-belief that p in terms of its acquisition and manifestation conditions: ACQ (i) X believes that p and (ii) X finds herself [able to sustain the A-supposition that p, but (iii) unable to sustain under that A-supposition, the C-supposition that not-p]. MAN (i) X believes that p and (ii) for all S, such that X finds herself [(iii) able to A-suppose p and (iv) subsequently to C-suppose that S] X is prepared to add p as a premise in reasoning from the C-supposition that S.¹³ ¹² The idea that a box-belief in some proposition is a matter of being unable to make anything of the contrary proposition is reminiscent of Blackburn’s (1993 [1986]) work, to which Divers and Gonzáles-Varela give due credit. ¹³ I have changed the mistaken numbering ((i), (ii), (ii), (iii)) in the original.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Non-Naturalism Gone Quasi

37

This would seem to take care of the problem with a posteriori necessities. Even though I am able to suppose, in a sense, that water doesn’t contain any hydrogen (I am able to suppose that I am surrounded by watery stuff with no hydrogen in it), I am not able to believe that it does contain hydrogen, to sustain a supposition that this is so, and, under that supposition, to suppose that it might not have contained any hydrogen. Also, I am prepared to add the supposition that water contains hydrogen to reasoning under any counterfactual supposition that I am able to entertain under the supposition that (as I believe) water actually does contain hydrogen. Although it is true that I am not prepared to add the supposition that water contains hydrogen to reasoning under the supposition that modern chemistry is actually entirely mistaken, this is, by MAN, wholly irrelevant to my treating water’s containing hydrogen as necessary. On this kind of account, to think (in the paradigmatic, proper, way) that it is necessary that whatever maximizes happiness is also right is to think, in a way spiced up by certain attitudes toward certain suppositions, that happiness-maximizing acts are right. Roughly, it is to think this while being unable to suppose that a happiness-maximizing action might not have been right, and with readiness to appeal to the belief that happinessmaximizing actions are right in reasoning under any counterfactual supposition.

2 . 3. T H E S U P E R V E N I E N C E C HA L L E N G E REVISITED We are now in a position to consider the prospects of answering our explanatory challenge under the different guises that it might take, depending on how we understand both thought concerning normativity and thought concerning metaphysical necessity. We might, in principle, try combining both realist non-naturalism and quasi-realist non-naturalism in metaethics with either a realist or a quasi-realist account of metaphysical modal judgment. In this section, I suggest that while the examination of these possible combinations of views offers no comfort for the realist nonnaturalist, quasi-realist non-naturalism turns out to provide promising results under either way of understanding metaphysical necessity thought. In his discussion of the supervenience challenge and quasi-realism, Dreier (2015, pp. 291–4) suggests that while all non-naturalists perhaps fail to explain the necessary relations between the natural and the normative, quasi-realists, by contrast with realists, might be able to explain why no explanation is needed. Dreier (2015, p. 294) finds this idea suggestive, “but

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

38

Teemu Toppinen

not really satisfying,” as he cannot quite see why the quasi-realist should, in the end, be better placed in this way. The following discussion can be read as an attempt to vindicate Dreier’s idea—as an attempt to explain why, given quasi-realism, we can explain why no explanation for the necessary connections between the natural and the normative is needed. For a realist non-naturalist in metaethics, it seems natural to take a realist, representationalist approach also to metaphysical necessity. The strengths and weaknesses of a representationalist view will be partly similar in both cases. For instance, if the appeal to the idea that a certain class of judgments purports to represent a certain realm of properties and facts helps in providing a neat explanation for the meaning of the relevant judgments, then this advantage is presumably available in relation to both normative and modal thought. And if there are metaphysical or epistemological costs to positing facts to be represented by our normative or modal claims, similar costs may pertain to both cases. Moreover, there is considerable pressure to give a unified account of alethic modal thought and of at least parts of the paradigmatically normative thought—both dealing in musts, oughts, and mays (see, e.g., Chrisman 2016; Wodak 2017). Let us again suppose that our explanatory challenge is that of explaining why it is metaphysically necessary that hedonically optimal actions are morally right. How should we understand this explanatory challenge, given adoption of a realist view regarding both normative thought and necessity thought? Given this combination of views, to think that it is necessary that hedonically optimal actions are morally right is to represent (having) the property of being hedonically optimal as necessitating (having) the property of being morally right, or to represent all possible worlds as being such that whatever maximizes happiness is also morally right. Our requesting an explanation for why it is that being hedonically optimal necessitates being morally right amounts, then, to our asking: CHAL 1 Why does the property of being happiness-maximizing have the property of necessitating having the property of being morally right? CHAL 2 Why is it that, in all worlds, anything that has the property of being happiness-maximizing also has the property of being morally right?

I have already briefly explained why answering this kind of explanatory challenge seems difficult for a non-naturalist. Casting the challenge in an explicitly realist-seeming light offers no help. An appeal may be made, again, to the idea that the property of being happiness-maximizing is a rightmaking property (see Section 2.1). But again, this seems like a mysterious, bad explanation—at least when offered within the realist framework. What if we adopt a non-representational, quasi-realist account of metaphysical modal thought? Would that be helpful for the realist non-naturalist?

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Non-Naturalism Gone Quasi

39

I don’t think so. I have already suggested that there would be some reason for a realist non-naturalist to stick with a representationalist, or a realist, account of modal thought, too. But let us set that to one side. Let us suppose, first, that something in the neighborhood of Thomasson’s account of metaphysical necessity claims is correct. The challenge to explain why it is that happinessmaximizing actions are morally right would now seem to take roughly the following form: CHAL 3 Why think (partly in virtue of accepting the constitutive semantic rules for the relevant terms), of all actions that one takes to be happiness-maximizing, that they are morally right?

This may seem like a change of topic (“Weren’t we talking about necessity, rather than about what to think?”), but if judgments about metaphysical necessity are understood roughly in terms of being for or against certain kinds of movements of mind, our question “Why is it metaphysically necessary that happiness-maximizing acts are morally right?” would seem to take something like this form. Given this kind of non-representationalist take on judgments about metaphysical necessity, we may, by the realist nonnaturalist’s lights, understand this question as follows: CHAL 3R Why represent (partly in virtue of accepting the constitutive semantic rules for the relevant terms) all actions that one takes to be happiness-maximizing as being morally right?

This doesn’t seem helpful. According to the realist non-naturalist, the constitutive semantic rules for “morally right” do not allow us to link happiness-maximizing to rightness. Perhaps we somehow intuit that happiness-maximizing is actually morally right. But assuming that being happiness-maximizing and being morally right are wholly distinct properties, it just isn’t clear why we should restrict our thinking so as to be committed to always representing happiness-maximizing actions as right (if the question arises). It is worth emphasizing, perhaps, that one cannot answer this question by showing why it would be desirable, say, to think in certain ways. The answer must engage with the perspective provided by the relevant constitutive semantic rules. How does our explanatory challenge look if we understand box-belief along the lines proposed by Divers et al.? Given this kind of view, our explanatory challenge would seem to take roughly the following sort of shape: CHAL 4 Why, given that happiness-maximizing actions are right, find oneself unable to suppose that they might not have been right? And why rely on this being true under any counterfactual supposition?

Again, given a realist non-naturalist reading, we seem to lack answers to these questions. Given that the properties of being happiness-maximizing

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Teemu Toppinen

40

and being right are completely distinct, why should one’s supposings be restricted in the relevant ways? Things start to look importantly different, however, once we take on board a quasi-realist account of thought concerning what is right. Could a quasi-realist non-naturalist adopt a representationalist, realist, account of claims concerning metaphysical necessity? As already noted, there’s some pressure to adopt the same kind of explanation of all thought that deals in musts, oughts, and mays. However, a quasi-realist might propose that we nevertheless accept a potentially somewhat disunified account of modal thought and treat metaphysical necessity thought as representational. Judgments about normative (e.g., moral) necessities, such as the judgment that it is metaphysically necessary that happiness-maximizing actions are morally right, are normative judgments of sorts. As such, these judgments, as a whole, receive a non-representational treatment from a quasi-realist. But the quasi-realist could hold that the contribution of the necessity operator to such judgments nevertheless is a representational one.¹⁴ We may look again at the following formulation of our question, understanding it now in a quasi-realist vein: CHAL 2 Why is it that, in all worlds, anything that has the property of being happiness-maximizing also has the property of being morally right?

Let us suppose—implausibly, but for the sake of simplicity—that to think that an action is morally right just is to plan to perform the action in question. If to think that an action is morally right is to plan to perform it, then, plausibly, asking why an action is right is something like asking why perform it. Our challenge now takes the form of answering this: CHAL 2Q-R happiness?

Why perform, in all worlds, actions that maximize

This question seems drastically different from the questions that realist nonnaturalism gave rise to. This is just a practical question concerning how to live. We have, in effect, eliminated from our explanatory challenge any reference to the property of being morally right. Of course the challenge still concerns the property of being right—the question before us still concerns the necessary relation between hedonic optimality and rightness. Or so the quasi-realist non-naturalist will agree. But she gives us an explanation of what this question amounts to, or a story about what it is to ask this question, in which rightness does not figure. Tough metaphysical questions dissipate; the question about rightness takes the form of a practical question concerning how to live. ¹⁴ Thanks to Mark Schroeder for a very helpful discussion here.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Non-Naturalism Gone Quasi

41

This is not to say that practical questions concerning how to live are not tough. It is extremely difficult to correctly answer questions regarding how to live. But we know what to do with such questions. We know how, in principle, to go about answering them. Questions of what to do end somewhere. They ultimately receive their answers with some ultimate plans or decisions. According to our toy quasi-realist view, a classical utilitarian ultimately plans to always maximize happiness. If we plan the way she does, we will think that the challenge to explain why it is necessary that hedonically optimal actions are right requires no answer—that there is no need to explain this. It is plausible, independently of the plausibility of quasi-realism, that in first-order normative theory, too, we hit a point, somewhere, where we are happy to reject the demand for further explanations or justifications. Why maximize happiness? Well, because that maximizes happiness. That’s just the thing to do. This may of course be the wrong ending point for the chain of justifications. Many think that the question “Why maximize happiness?” mistakenly presupposes that we are to maximize happiness. Some others might think that although we are to maximize happiness, we can say more about why this is so. Perhaps they think that we are to maximize happiness because we are to obey God’s commands, and God commands us to maximize happiness, or whatever. But the answer “Maximizing happiness maximizes happiness; it’s just the thing to do” seems like the kind of answer that a classical utilitarian can sensibly offer. Now, which questions about the necessary connections between the natural and the normative are relevant, in the present context, depends on what the correct normative theory is. If utilitarianism is correct, then one of the relevant questions is: why is it necessary that any action that is hedonically optimal is also morally right? If we are assuming that classical utilitarianism is right, then, given quasi-realist non-naturalism (of the simple sort that we’re toying with), this question is, roughly, a question of why maximize happiness. And given the assumption of classical utilitarianism, it seems perfectly fine to answer this question with “That’s just the thing to do”—that is, with a plan to always maximize happiness. And so, once we hit the point, in normative theorizing, where we need no further justifications, we can—given our toy quasi-realist view—also rest content regarding the corresponding, seemingly “metanormative” challenge to explain the necessary connection between wildly discontinuous properties. What first seemed like a puzzling metaethical question turns out to be, by the quasi-realist’s lights, a first-order normative issue that doesn’t give rise to metaphysical mysteries. We might also consider a slightly more complex toy version of quasirealist non-naturalism, a simple form of a relational view, according to which to think that an action is morally right is to plan to perform actions with some (possibly rather complex) natural property, and to represent the action in question as having that property (see Schroeder 2013; Toppinen 2013;

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

42

Teemu Toppinen

Ridge 2014). The relevant property may vary from one thinker to another, as long as their plans and representations are related in the same way. So, consider two thinkers—call them “Disgrace” and “Archgoat”. Disgrace plans to maximize misery and represents a certain action, φ-ing, as hedonically pessimal (as one might put it), while Archgoat plans to maximize heteronomy and represents φ-ing as best conducive to this goal. Disgrace and Archgoat then differently realize the kind of massively multiply realizable relational state that they are both in, and that constitutes thinking that φ-ing would be morally right. According to this kind of relational view, to wonder why an action is morally right amounts to this: being in the relational state of planning, for some property, to perform actions with this property, and of wondering why the action has this property. On this kind of view, our challenge becomes one that a classical utilitarian should answer by answering a question along the following lines: CHAL 2Q-R* (I shall perform happiness-maximizing actions! +) Why is it that, in all worlds, anything that is happiness-maximizing is happiness-maximizing?

The conative element of the question (“I shall perform happiness-maximizing actions!”) is, of course, crucial here. On this kind of view, asking why it is necessary that happiness-maximizing actions are right is not simply to ask: why is it necessary that happiness-maximizing actions are happiness-maximizing? However, for a classical utilitarian, the first question is answered by answering the second. Just as with the first toy quasi-realist account, any problematic reference to the property of rightness dissipates here. And again, we come to see that the challenge doesn’t require much of an answer. Why is it necessary that happiness-maximizing actions are right? It just is; happiness-maximizing actions are what they are. Finally, we may try combining quasi-realist non-naturalism with the nonrepresentationalist renditions of our explanatory challenge. Let us begin with the Thomassonian view. Our challenge then is, again, to answer this: CHAL 3 Why think (partly in virtue of accepting the constitutive semantic rules for the relevant terms), of all actions that one takes to be happiness-maximizing, that they are morally right?

With our “pure”non-cognitivist view, this transforms into something like: CHAL 3Q-R Why plan (partly in virtue of accepting the constitutive semantic rules for the relevant terms) to always perform actions that maximize happiness?

This comes very close to CHAL 2Q-R. Here, too, the question is not: why is it good to plan to perform actions that maximize happiness? Or: why see to

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Non-Naturalism Gone Quasi

43

it that one plans to perform actions that maximize happiness? Rather, this is a question the answer to which also works in response to: why maximize happiness? Our metaphysical question has again taken the form of a practical question concerning how to plan, or how to live. And assuming the correctness of hedonistic utilitarianism plus the quasi-realist view, the question doesn’t really seem to need any further answer. Given acceptance of the relevant semantic rules, we apply “morally right” to the sorts of actions that we plan to perform. Given utilitarian plans, one is then committed to applying this term to, and to planning to perform, whatever actions one thinks are hedonically optimal. With our toy version of relational expressivism, the challenge—when given a Thomassonian form—would be one that a classical utilitarian should answer by answering a question along the following lines: CHAL 3Q-R* Why, given the constitutive semantic rules for the relevant terms, plan to always perform actions that are happiness-maximizing, and represent happiness-maximizing actions as being like that (that is, as being happiness-maximizing)?

Given the truth of classical utilitarianism and our toy relational view, this needs no answer. This is the way to plan; happiness-maximizing actions are what they are. I find it tricky to try to articulate a candidate for a Diversian way of understanding our explanatory challenge in terms of the pure noncognitivist toy theory. The Diversian account makes heavy use of the attitude of supposition, and it is not at all obvious how a defender of pure non-cognitivism should understand the attitude of supposing that an action is right. This problem is an instance of a more general problem concerning how to account, within a non-cognitivist framework, for various types of nonbelief attitudes with normative content (see Schroeder 2010, pp. 83–4). It is not obvious, at all, how the pure non-cognitivist should endeavor to make sense, for instance, of wondering whether an action is right, or of hoping that it would be. Simon Blackburn (1998, p. 70) suggests that we could understand wondering whether an action is right in terms of wondering what to do. Similarly, we might say: to suppose that an action might not have been right is to suppose that the action might not have been the one to perform. If this is fine, we could then take the explanatory challenge, when understood in the Diversian way, to be that of answering something like this: CHAL 4Q-R Why, given that happiness-maximizing actions are to be performed, find oneself unable to suppose that they might not have been such as to be performed? And why rely on happiness-maximizing actions being such as to be performed under any counterfactual supposition?

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

44

Teemu Toppinen

But now it seems that we might, just as well, identify believing that happiness-maximizing actions are right with believing that they are such as to be performed, and it’s not clear that this affords us much illumination, really, regarding the nature of a belief concerning what’s right. We have replaced the property of being morally right with the property of being such as to be performed. If the former property and its connections to the natural look problematic in certain ways, then presumably this is so also when it comes to the latter property. For a relational view, making sense of normative supposing seems easier, as we may simply say that to suppose that so-and-so is right is to be in a relational state that is realized by planning to perform actions with a certain property, and by supposing that so-and-so has that property (see Ridge 2014). For the relational expressivist, the challenge, when understood along the Diversian lines, would be to answer something like this: CHAL 4Q-R* (I shall perform happiness-maximizing actions! +) Why, given that happiness-maximizing actions are happiness-maximizing, find oneself unable to suppose that happiness-maximizing actions might not have been happiness-maximizing? And why rely on this (the fact that happiness-maximizing actions are happiness-maximizing) being true under any counterfactual supposition?

These questions don’t seem pressing. We cannot conceive of a scenario where what maximizes happiness fails to maximize happiness. And that happinessmaximizing actions are happiness-maximizing is something we can safely rely on under any counterfactual assumption. Once again, from a classical utilitarian perspective, it seems that we can just shrug off the challenge. 2.4. CONCLUSION I have argued that once we take a look at the supervenience challenge through properly quasi-realist lenses, the challenge is likely to emerge as tractable. Of course, I have only discussed very simple, toy versions of quasirealism. But the conclusions seem likely to generalize. Given quasi-realism, the relevant explanatory challenge takes shape as a first-order normative challenge, or vanishes given the relevant normative perspective. And so it doesn’t give rise to any metaphysical mysteries. If that’s right, the quasi-realist non-naturalist is not in the same boat with the realist non-naturalist, after all. This would be a weighty reason—to be balanced with other considerations, of course—for someone with non-naturalist sympathies to adopt a quasi-realist position.¹⁵ ¹⁵ I thank Matt Bedke, Jamie Dreier, Eric Hubble, Stephanie Leary, Mike Ridge, Geoff Sayre-McCord, Josh Schechter, Mark Schroeder, Jussi Suikkanen, Mark van Roojen, Jack Woods, and the rest of the audiences at the 2nd CHill Meta and at the

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Non-Naturalism Gone Quasi

45

References Blackburn, S., 1971. “Moral Realism,” in Essays in Quasi-Realism. New York: Oxford University Press, 1993. Blackburn, S., 1984. Spreading the Word: Groundings in the Philosophy of Language. Oxford: Oxford University Press. Blackburn, S., 1985. “Supervenience Revisited,” in Essays in Quasi-Realism. New York: Oxford University Press, 1993. Blackburn, S., 1987. “Morals and Modals,” in Essays in Quasi-Realism. New York: Oxford University Press, 1993. Blackburn, S., 1993 [1986]. “Morals and Modals,” in Essays in Quasi-Realism. New York: Oxford University Press. Blackburn, S., 1998. Ruling Passions. Oxford: Oxford University Press. Brandom, R., 2008. Between Saying and Doing: Towards an Analytic Pragmatism. Oxford: Oxford University Press. Chrisman, M., 2016. The Meaning of “Ought”: Beyond Descriptivism and Expressivism in Metaethics. Oxford: Oxford University Press. Dancy, J., 2004. Ethics without Principles. Oxford: Oxford University Press. Divers, J., 2002. Possible Worlds. London: Routledge. Divers, J. and Elstein, D. Y., 2012. “Manifesting Belief in Absolute Necessity,” Philosophical Studies 158: 109–30. Divers, J. and Gonzáles-Varela, J. E., 2013. “Belief in Absolute Necessity,” Philosophy and Phenomenological Research 87: 358–91. Dreier, J., 1992. “The Supervenience Argument against Moral Realism,” Southern Journal of Philosophy 30: 13–38. Dreier, J., 2004. “Meta-Ethics and the Problem of Creeping Minimalism,” Philosophical Perspectives 18: 23–44. Dreier, J., 2015. “Explaining the Quasi-Real,” in R. Shafer-Landau (ed.), Oxford Studies in Metaethics, Vol. 10. Oxford: Oxford University Press. Dreier, J., MS. “Is There a Supervenience Problem for Robust Moral Realism?” Enoch, D., 2011. Taking Morality Seriously. New York: Oxford University Press. Gibbard, A., 2003. Thinking How to Live. Cambridge, MA: Harvard University Press. Gibbard, A., 2012. Meaning and Normativity. Oxford: Oxford University Press. Hare, R. M., 1952. The Language of Morals. Oxford: Oxford University Press.

New Directions for Expressivism conference in Sheffield for helpful feedback. I’m also grateful to Russ Shafer-Landau, and to Graham Bex-Priestley and Stephen Ingram, for organizing these events. Some relevant material was presented also in Stockholm and Uppsala, and at the 2016 colloquium of the Philosophical Society of Finland. Thanks to the audiences on these occasions; I should especially single out my excellent commentators Olle Risberg and Daniel Fogal, as well as Krister Bykvist, Matti Eklund, Anandi Hattiangadi, and Jonas Olson. Pekka Väyrynen provided useful feedback on a related piece qua a referee. Finally, I should also thank the two anonymous reviewers for Oxford University Press for very helpful comments, and the Helsinki Metaethics Reading Group, and especially Vilma Venesmaa, for discussions on related topics. The work on this chapter was made possible by the funding from the Academy of Finland.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

46

Teemu Toppinen

Horgan, T., 1993. “From Supervenience to Superdupervenience: Meeting the Demands of a Material World,” Mind 102: 555–86. Klagge, J., 1988. “Supervenience: Ontological and Ascriptive,” Australasian Journal of Philosophy 66: 461–70. Leary, S., 2017. “Non-Naturalism and Normative Necessities,” in R. Shafer-Landau (ed.), Oxford Studies in Metaethics, Vol. 12. Oxford: Oxford University Press. Lewis, D., 1986. On the Plurality of Worlds. Oxford: Blackwell Publishers. McPherson, T., 2012. “Ethical Non-Naturalism and the Metaphysics of Supervenience,” in R. Shafer-Landau (ed.), Oxford Studies in Metaethics, Vol. 7. Oxford: Oxford University Press. Olson, J., 2014. Moral Error Theory. Oxford: Oxford University Press. Parfit, D., 2011. On What Matters, Vol. 2. Oxford: Oxford University Press. Parfit, D., 2017. On What Matters, Vol. 3. Oxford: Oxford University Press. Ridge, M., 2007. “Anti-Reductionism and Supervenience,” Journal of Moral Philosophy 4: 330–48. Ridge, M., 2014. Impassioned Belief. Oxford: Oxford University Press. Rosen, G., MS. “What is Normative Necessity?” Ryle, G., 1950. “ ‘If ’, ‘So’, and ‘Because’,” in Collected Papers, Vol. 2. Bristol: Thoemmes, 1971. Scanlon, T. M., 2014. Being Realistic about Reasons. Oxford: Oxford University Press. Schroeder, M., 2007. Slaves of the Passions. Oxford: Oxford University Press. Schroeder, M., 2010. Noncognitivism in Ethics. Abingdon: Routledge. Schroeder, M., 2013. “Tempered Expressivism,” in R. Shafer-Landau (ed.), Oxford Studies in Metaethics, Vol. 8. Oxford: Oxford University Press. Schroeder, M., 2014. “The Price of Supervenience,” in Explaining the Reasons We Share: Explanation and Expression in Ethics, Vol. 1. Oxford: Oxford University Press. Sellars, W., 1958. “Counterfactuals, Dispositions, and the Causal Modalities,” in H. Feigl, M. Scriven, and G. Maxwell (eds.), Minnesota Studies in Philosophy of Science, Vol. 2: Concepts, Theories, and the Mind-Body Problem. Minneapolis: University of Minnesota Press. Shafer-Landau, R., 2003. Moral Realism: A Defence. Oxford: Oxford University Press. Shafer-Landau, R., 2005. “Replies to Critics,” Philosophical Studies 126: 313–29. Smith, M., 1994. The Moral Problem. Oxford: Blackwell. Smith, M., 2000. “Does the Evaluative Supervene on the Natural,” reprinted in Ethics and the A Priori. Cambridge: Cambridge University Press, 2004. Stratton-Lake, P. and Hooker, B., 2006. “Scanlon versus Moore on Goodness,” in T. Horgan and M. Timmons (eds.), Metaethics after Moore. Oxford: Oxford University Press. Sturgeon, N., 2009. “Doubts about the Supervenience of the Evaluative,” in R. Shafer-Landau (ed.), Oxford Studies in Metaethics, Vol. 4. Oxford: Oxford University Press. Thomasson, A., 2007. “Modal Expressivism and the Methods of Metaphysics,” Philosophical Topics 35: 135–60. Thomasson, A., 2009. “Non-Descriptivism about Modality: A Brief History and Revival,” The Baltic International Yearbook of Cognition, Logic and Communication 4: 1–26.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Non-Naturalism Gone Quasi

47

Thomasson, A., 2013. “The Nancy D. Simco Lecture: Norms and Necessity,” Southern Journal of Philosophy 51: 143–60. Thomasson, A., MS. “Norms and Necessity.” Toppinen, T., 2013. “Believing in Expressivism,” in R. Shafer-Landau (ed.), Oxford Studies in Metaethics, Vol. 8. Oxford: Oxford University Press. Toppinen, T., 2016. “Is Irreducible Normativity Impossibly Queer?” Journal of Moral Philosophy 13: 338–61. Toppinen, T., forthcoming. “Essentially Grounded Non-Naturalism and Normative Supervenience,” Topoi. Väyrynen, P., 2017. “The Supervenience Challenge to Non-Naturalism,” in D. Plunkett and T. McPherson (eds.), The Routledge Handbook of Metaethics. London: Routledge. Wedgwood, R., 1999. “The Price of Non-Reductive Moral Realism,” Ethical Theory and Moral Practice 2: 199–215. Wodak, D., 2017. “Expressivism and Varieties of Normativity,” in R. Shafer-Landau (ed.), Oxford Studies in Metaethics, Vol. 12. Oxford: Oxford University Press. Zangwill, N., 1995. “Moral Supervenience,” Midwest Studies in Philosophy 20: 240–62.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

3 Non-Descriptive Relativism Adding Options to the Expressivist Marketplace Matthew S. Bedke

3.1. INTRODUCTION Suppose you are in the market for a metaethical theory and two features are high on your wish list. First, you do not want to assign any descriptive or representational content to thin normative terms, arguably terms like “ought to,” “is wrong,” “is good,” and “is a reason to,” at least when these terms implicate authoritative force or weight.¹ Second, you do want to explain the motivational and endorsement roles of normative thought and language, at least when these terms implicate authoritative force or weight. It will not take long to discover that one view has virtually cornered the market: expressivism.² Though expressivism certainly delivers when it comes to non-descriptivity and action guidance, there are some contested issues. To name a few: it needs to explain why authoritative force or weight appears to be independent of us and our attitudes, it needs an account of normative disagreement, and it needs to accommodate the bundle of logical, semantic, and epistemic roles highlighted by Frege–Geach concerns. My aim here is to add some competition into this marketplace by introducing a family of views I call Non-Descriptive Relativism. Taking for granted the evidence suggesting that normative terms do not purport to represent ways of the world, neither natural nor sui generis, and taking for granted evidence that normative thoughts have a motivational role, and normative assertions an endorsement role, that typical beliefs and assertions lack, I try to forge an alternative to expressivism. I briefly address other issues

¹ I think this will amount to normative thought and language that is reasonimplicating (Bedke 2010; Parfit 2011: 38). ² I will not address hybrid theories that sully expressivism with descriptive content.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Non-Descriptive Relativism

49

along the way, such as the appearance of authoritative weight, normative disagreement, and Frege–Geach roles, for the hope is that some version of Non-Descriptive Relativism will be part of the best explanation of all the metaethical data. Of course, that is a larger project that only begins here.

3 . 2 . T H E CO R E ID EA I can convey the core idea behind Non-Descriptive Relativism by contrasting it with two closely related views: speaker subjectivism and expressivism. Let me start with a crude version of speaker subjectivism, which says that normative thoughts represent the thinker’s conative attitudes, and normative assertions report on or describe the speaker’s conative attitudes. On this view, someone who asserts “Stepping on gouty toes is wrong” is describing her own psychology in much the same way as if she had said “I morally oppose stepping on gouty toes,” where moral opposition is a conative attitude. This is not the best version of speaker subjectivism for at least two reasons. First, it is too self-regarding. When one says that some act is wrong the claim seems to be about worldly features that are independent of the claimant’s psychological makeup. Second, it entails some repugnant counterfactuals, like this: “If I didn’t morally oppose stepping on gouty toes, then stepping on gouty toes would not be wrong.” After all, a sentence that is meant to have a very similar meaning, “If I didn’t morally oppose stepping on gouty toes, then I wouldn’t morally oppose stepping on gouty toes,” is true. Fortunately, there is a better, sophisticated, version of speaker subjectivism, one that gives the speaker’s attitudes some role in determining exactly which worldly features the speaker’s claims are about without locating her attitudes in the content of what is said or thought. One illustrative way to run the view is to say that S’s assertion that stepping on gouty toes is wrong has the content , where S morally opposes F in action. My assertions would be about those properties I morally oppose in action and your assertions would be about those properties you morally oppose in action. These assertions would be about the objects of our attitudes, not the attitudes themselves. So sophisticated subjectivism avoids the kind of selfregard we find in crude subjectivism. There is also a way of formulating the view to avoid repugnant counterfactuals. We simply need to make normative predicates rigidly express those properties the speaker actually morally opposes. I prefer doing this by using Kaplan’s “dthat” operator, where we say that a sentence of the form “X is wrong” expresses the proposition that x exhibits dthat(a property the speaker morally opposes in action). While this

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

50

Matthew S. Bedke

suggestion breaks with Kaplan’s use of “dthat” in allowing complements that are not definite descriptions, a subjectivist can still capitalize on the operator to avoid repugnant counterfactuals. The antecedent of “If I didn’t morally oppose stepping on gouty toes, then stepping on gouty toes would not be wrong” takes me to worlds where I do not morally oppose stepping on gouty toes, but the consequent considers whether the action has features that I actually morally oppose.³ What my benighted counterfactual self opposes has nothing to do with it. Another nice feature of the operator is that it allows us to use the parenthetical to capture a mode of presentation of the content it helps to fix. That is, while sophisticated subjectivism says that speaker attitudes are not part of content, one’s attitudes could color the mode of presentation of the content they help to fix (more on this in Section 3.7). Sophisticated subjectivism has other things going for it. It explains plausible data concerning action guidance and it delivers on most Frege– Geach roles. Nevertheless, (pure) expressivists are unhappy with any descriptive content for thin normative terms.⁴ Typically focusing on the crude version, they suggest that we distinguish describing our attitudes from expressing those attitudes and they maintain that uses of normative language express but do not report on the speaker’s attitudes. Formulated as a bit of semantic theory the idea is that all declarative sentences are in the business of expressing states of mind, and unlike sentences that express beliefs, basic normative sentences express conative attitudes (see, e.g., Gibbard 2003, Rosen 1998: 387, and Schroeder 2008a: 33).⁵ When metaethicists then try to craft compositional and truth functional semantic contents for the view, they typically distinguish two kinds of content. Bits of language that express beliefs have traditional descriptive contents, such as sets of possible worlds or structured propositions or otherwise ways-things-might-be contents. Bits of language that express conative attitudes have non-traditional, non-descriptive contents, such as Gibbard’s hyperplans (2003), or Horgan and Timmons’ ought commitments (2006).

³ Cf. Humberstone and Davies (1980). ⁴ I think the main concern here is that speaker-variant descriptive content does not correctly categorize clear cases of disagreement. But see Jackson (2008). ⁵ Alternatively, relational expressivism says that normative assertions express relations between beliefs and conative attitudes (Ridge 2014; Schroeder 2013; Toppinen 2013). I should also note that expressivism need not be located in semantic theory. One alternative is to locate it in meta-semantic theory, where sentences get their semantic values (whatever they are) in virtue of expressing states of mind (see, e.g., Chrisman 2012: 323–30, and Ridge 2014: 102–11, 124–31). A second alternative is to develop expressivism as a radical pragmatic theory of meaning (see, e.g., the essays in Price et al. 2013).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Non-Descriptive Relativism

51

That is certainly one way to depart from speaker subjectivism’s descriptivity while explaining action guidance,⁶ but I think there is an alternative point of departure that might prove more fruitful. Instead of moving the field of play to a semantic theory built on expressing states of mind, and instead of positing a novel kind of non-descriptive content within that theory, we might instead consider different relations language might bear to standard contents, one relation being descriptive and another nondescriptive. That is, perhaps we should not be so quick to assume that if declarative sentence D has way-the-world-might-be content C, then use of D is normally in the business of describing things in the C way.⁷ Indeed, perhaps it is possible for D to bear a non-descriptive relation to way-thingsmight-be C, so that declarative uses of D are not normally putting C at issue as an accurate description of how things are. This is the core idea of the alternative approach pursued here. We might still speak of non-descriptive contents on this alternative approach, but this would not refer to a novel kind of content that is not inherently descriptive (like a set of hyperplans as opposed to a set of worlds); instead, it would refer to the fact that the language with the content in question does not bear a descriptive relation to that content. I grant that this is prima facie puzzling. How is it possible for a sentence to have a way-the-world-might-be content—content the sentence is about— but for declarative uses of that sentence to lack descriptive purport? To resolve this into a sensible metaethical view I am going to sketch a particular species of it. What follows as an example of how to make the core move— that of drawing a distinction between a descriptive and a non-descriptive relation some bit of language might bear to a given content. Though I am certainly open to other ways of making that move, I will try to show that the version of Non-Descriptive Relativism that follows has enough explanatory power to have a seat at the table for metaethical theories that aim for nondescriptivity, action guidance, and other metaethical desiderata.

⁶ Of course, expressivism would predict a stronger motivational profile for normative judgments, as they would be constituted by conative attitudes, whereas subjectivism makes the presence of conative attitude a condition of semantic correctness. I think the weaker profile predicted by subjectivism is a better fit with the data, but that is another paper. ⁷ On some ways of doing semantics there is no presumption that semantic content is that which language is about. Instead, contents (or semantic values) are simply theoretical posits whose only virtue is helping to model compositionality and truth functionality. However, I am interested in aboutness, and I want to leave room for what we might call non-descriptive (or non-representational) aboutness.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

52

Matthew S. Bedke 3 .3 . C O G N I T I O N A N D D E S C R I P T I ON

Let us step back for a moment and ask ourselves how language gets its content. The approach I will take says linguistic contents are fixed (at least in part) by some cognitive roles (or Kaplanian characters) associated with the conventional meanings of language. We can formulate these cognitive roles in terms of rules—rules encoding conventional meanings that help to fix semantic contents, and rules that we grasp and know how to follow when we grasp the conventional meanings of terms. These rules are analogues of classical conceptual analyses. So if the conceptual analysis for “water” is “X is a sample of water iff x has the same explanatory nature as dthat(the clear, potable stuff that falls from the sky, fills our lakes and streams . . . ),” we might formulate the cognitive rule like this: “Predicate ‘is water’ of x iff x has the same explanatory nature as dthat(the clear, potable stuff that falls from the sky, fills our lakes and streams . . . ).”⁸ To clarify, these rules are not intended to require predication or forbid it whenever the conditions they identify obtain. They are intended to identify the conditions under which one correctly predicates, when one does predicate, where the correctness here has a semantic flavor.⁹ What would a rule for “is wrong” look like under sophisticated subjectivism? Here is one plausible candidate: First Pass: Predicate “is wrong” of an action iff it exhibits dthat(a property the speaker morally opposes in action).

This is fine as far as it goes. But I think it simply fails to capture an important aspect of speaker subjectivism, namely, that “x is wrong” is in the business of describing x in some way. All First Pass delivers are conditions for correct predication, and that is not to identify conditions that are described or represented through predication. In other words, “iff ” does not encode that the term on the left-hand side is purporting to represent or otherwise stand for the conditions identified on the right-hand side. To do that, I suggest we formulate a use rule like this:

⁸ This tradition needs the usual qualifications. First, just as there might be good reason to drop necessary conditions from classical conceptual analyses, there might be good reason to drop them from our cognitive rules. Second, different people might have different but interrelated cognitive rules that ensure they are trying to refer to the same thing. Third, any explicit formulation of a cognitive rule will be somewhat crude and unlikely to fully capture the cognitive roles associated with conventional meanings, but the formulations (and counterexamples and corrections) can be illuminating nonetheless. ⁹ I do not think this sense of “correct” is reason-implicating. But that is another story.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Non-Descriptive Relativism

53

“Wrong” F-Rule: Predicate “is wrong” of an action for (and only for) exhibiting dthat(a property the speaker morally opposes in action).

The point of switching from “if ” to “for” is to encode in our cognitive rule the speaker subjectivist proposal that the left-hand side is to bear a representation relation to the right-hand side. “For” signals that there is something about the cognitive role of the term that makes the predicate stand for something. It not only makes the predicate applicable on the basis of something, but it makes uses of the predicate describe that basis of application. So understood, “Wrong” F-Rule (i) makes it so the sentence “X is wrong” has a character that is sensitive to speaker attitude; (ii) takes us from contexts of use where the speaker morally opposes F1, F2, and F3 in action to the semantic content ; and (iii) makes it so the sentence “X is wrong” bears a descriptive relation to its content. (iii) is important for our discussion. Formulated within the cognitive rule framework, sophisticated subjectivism says that “wrong” is governed by “Wrong” F-Rule. But suppose we side with expressivism and reject descriptive content for thin normative terms. Our framework makes room to do so without moving the field of play to a semantics of attitude expression and adopting a novel kind of semantic content. Instead, we can swap out the “for” locution found in “Wrong” F-Rule with a different locution that encodes that the term on the left-hand side is not used to describe its content. I suggest that we posit a cognitive rule that says when to use a predicate along the following lines:¹⁰ “Wrong” W-Rule: Predicate “is wrong” of an action when (and only when)¹¹ it exhibits dthat(a property the speaker morally opposes in action).

This is subtly but importantly different from the previous rule. Unlike “for,” “when” encodes a non-descriptive role for the language. This key distinction between descriptive purport (“for”) and lack of it (“when”) is what “if ” fails to capture. Note that this approach retains the content suggested by “Wrong” F-Rule—the sets of worlds picked out by the right-hand side are the same whether the rule employs “for” or “when”—and it retains the conditions for correct predication. All it does is drop the descriptive relation to the language. Crucially, the difference is not between truth conditions and assertability conditions, as though f-rules introduce truth conditions and ¹⁰ Cf. Smith and Stoljar (2003) and Jackson and Pettit (1998). ¹¹ If there are no necessary conditions for falling under a concept, we can drop these “only for/when” clauses, as we would drop “only if” in traditional conceptual analyses. I will not take a stand on this issue, but for ease of exposition I will henceforth omit the “only for/when” clause.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

54

Matthew S. Bedke

w-rules introduce assertability conditions. We simply have two ways in which semantic content can be fixed, either by w-rules (securing a nondescriptive relation to content) or by f-rules (securing a descriptive relation to content). F-rules and w-rules would generate separate dimensions to semantics. F-rules would take us from contexts to descriptive contents. Descriptive content would be evaluated relative to a circumstance of evaluation (a possible world) to get an extension. I suggest descriptive accuracy and descriptive inaccuracy (and descriptive indeterminacy) as the possible extensions for contents generated by f-rules, pushing to one side for the time being the question of truth values (which I take up in Section 3.6). Obviously, this retains familiar players from compositional semantics: character, context, content, circumstance of evaluation, and extension. We simply interpret some of these values in a way that makes it transparent that we are modeling a descriptive dimension to meaning. W-rules would have a structurally parallel dimension to meaning. They would take us from contexts of use to non-descriptive contents, and these contents would be evaluated relative to a circumstance of evaluation (again, this could be a possible world) to get a non-descriptive extension. For full generality we could say the extensions are non-descriptive accuracy, and nondescriptive inaccuracy (and non-descriptive indeterminacy), but since my examples concern contents fixed by evaluative conative attitudes it makes sense in the present context to consider evaluative accuracy and evaluative inaccuracy (and evaluative indeterminacy) as the possible extensions for w-rule fixed contents.¹² The upshot is that there is more than one way for a non-descriptivist to part company with speaker subjectivism. One way is to embrace expressivism (or one of its progenitors). But an alternative is to posit a distinction between kinds of relations a sentence can bear to its content, descriptive and non-descriptive. I have sketched this idea using f-rules and w-rules. Within this framework, the version of Non-Descriptive Relativism I want to consider is this: thin normative terms are governed only by w-rules like the one in “Wrong” W-Rule.¹³ If so, “X is wrong” would simply fail to rule out any way x might be and we could model this by saying that the descriptive

¹² The approach here has some affinities with use conditional approaches to meaning (see Gutzmann and Gärtner 2013). It also has some affinities with Kaplan’s notion of information delimitation (unpublished). ¹³ I leave open whether there are non-normative terms governed by w-rules. One good candidate might be “is whack,” as when Whitney Houston said “Crack is whack.” Modals might also be good candidates, though presumably their w-rules would not mention conative attitudes.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Non-Descriptive Relativism

55

content of the sentence is the set of all possible worlds, which is the same as the descriptive content of “X is not wrong.” Thin terms would be maximally vague with respect to descriptive distinctions.¹⁴ No matter what we are considering—taking candy from a baby, torturing children at random, donating blood, dedicating one’s life to helping the poor and powerless— one can call it wrong (or not wrong, or predicate any other thin term of it) without misdescribing it. As for action guidance, it is clear that this view captures action guidance as well as sophisticated (descriptive) subjectivism. It does so by positing a semantic connection between normative judgments and having certain conative attitudes (more on this in Section 3.7). By way of contrast, we can hypothesize that a purely descriptive sentence like “Stepping on gouty toes is uncommon” would only be governed by f-rules. It would be empty along the non-descriptive dimension in the sense that it rules out no possible non-descriptive contents. As a result, these sentences will be vacuously evaluatively accurate. Within this framework it is quite possible to have a predicate that is meaningful on both dimensions. For it is possible that a term is governed both by f-rules and w-rules. This is a good option for thick normative terms and their sentences. Consider “is cruel,” for which we can posit an f-rule along the following lines: “Cruel” F-Rule: Predicate “cruel” of someone for exhibiting indifference to the pain or suffering of others.

We might also posit a w-rule to capture its non-descriptive content: “Cruel” W-Rule: Predicate “cruel” of someone when she exhibits indifference to the pain or suffering of others, which the speaker actually morally opposes in character.

On this view “X is cruel,” when said by S, will be descriptively accurate just in case x is a person who is indifferent to the pain or suffering of others. The sentence said by S will be evaluatively accurate just in case x is a person who is indifferent to the pain or suffering of others and S actually morally opposes those qualities in character. ¹⁴ The approach has some pedigree. See Joshua Gert (2007: 92–5; 2012: 51–6), Lenhart Åqvist (1964), and Stephen Schiffer (2002). Giulia Pravato (unpublished) has an approach to maximal vagueness in her dissertation, what she calls “irresolvable vagueness.” Also, though I’d like to say thin normative terms have no descriptive content, I can acknowledge that there is a close cousin to descriptive content in the neighborhood. For somehow category mistakes are ruled out, such as a lamp shade being wrong, or the number 7 being wrong, or really any non-action being wrong. I think this kind of domain restriction is prior to any restriction done by the content-fixing f- and w-rules, and for that reason I do not include in descriptive content the ruling out of category mistakes.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

56

Matthew S. Bedke 3.4. NON-DESCRIPTIVE CONTENTS: SETS OF WORLD-STANCE PAIRS

I have just offered a way of distinguishing descriptive content and nondescriptive content that does not make use of an expressivist semantic framework or novel kinds of contents. But we might wish to introduce contents that are not simply ways-things-might-be for other reasons. For example, we might wish to avoid speaker variant contents for a given thin normative predicate because they make it hard to accommodate certain cases of disagreement. It does not matter whether we consider “Wrong” F-Rule or “Wrong” W-Rule. If I oppose actions for exhibiting properties F1, F2, or F3, and you oppose actions for exhibiting properties G1, G2, or G3, my assertion “Stepping on gouty toes is wrong” would be accurate if the action exhibits some F property, while your assertion “Stepping on gouty toes is not wrong” would be accurate if the action fails to exhibit any G properties. Whence the disagreement? We can start to fix this problem by adjusting the w-rule for “is wrong.” Though the discussion of disagreement won’t come full circle until Section 3.8, the first piece of the puzzle is to modify the w-rule for “is wrong” to drop reference to what the speaker morally opposes. Consider the following: Better “Wrong” W-Rule: Predicate “is wrong” of an action when it exhibits a property that is morally opposed.

Instead of referring to the speaker’s attitudes, this rule refers to moral opposition without saying whose opposition is in play (cf. Wiggins 1987: 187). We can articulate the content given by this w-rule as the set of worldmoral-opposition-stance pairs () such that the object of the evaluation in w has some property that is an object of the stance of moral opposition.¹⁵ Instead of focusing on what the speaker morally opposes, we consider all possible stances of moral opposition, where, as before, moral opposition is a conative attitude. Some possible stances will morally oppose breaking promises, some causing pain, etc. For “Stepping on gouty toes is wrong” the non-descriptive content would be the set of world-moral-oppositionstance pairs such that stepping on gouty toes in the world has some property that the stance morally opposes. For example, a world-stance pair would ¹⁵ Cf. Lasersohn (2005: 663). The content could be modeled as Lewisian de se content, or a set of centered worlds where the centers of worlds include agents with normative stances (Egan 2012), though I worry that the second option is objectionably self-regarding.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Non-Descriptive Relativism

57

make it into the set if stepping on gouty toes in the world is a non-optimific action and the stance morally opposes non-optimific actions. The content posited here ensures that each speaker considers the same non-descriptive content for a given normative sentence. And the content is similar to Gibbard’s world-hyperplan pairs (Gibbard 2003: ch. 3). As I understand his view, the content of the sentence “Stepping on gouty toes is wrong” would be the set of world-hyperplan pairs that includes all possible worlds and those maximally decided hyperplans that include the plan to not step on gouty toes. The most important difference between his approach and my own is that Gibbard works with one dimension to semantics and models all content in terms of sets of world-hyperplan pairs. Non-Descriptive Relativism, on the other hand, posits two dimensions to semantics, each dimension with its own content. Descriptive content is modeled with sets of worlds and non-descriptive content is modeled with sets of world-stance pairs. Further, the worlds featured in the non-descriptive content need not be the same as the worlds featured in the descriptive content. Predicating “is wrong,” for example, excludes no worlds whatsoever in the descriptive content, but its non-descriptive content excludes world-stance pairs where the item evaluated lacks any property picked out by the paired stance. Second, Gibbard cashes out all normativity in terms of planning states. The current approach associates “is wrong” with stances of moral opposition, understood as conative attitudes, and it is open to us to associate different normative terms with different conative attitudes. The stances associated with “is a reason to” might be given by conative perspectives of pro tanto favoring, for example.

3. 5 . C I R C U M S T A N C E S O F E V AL U A T I O N : WORLDS AND WORLD-STANCE PAIRS We now need to evaluate these worlds for descriptive accuracy, and evaluate world-stance pairs for evaluative accuracy. We do not need to say anything unorthodox about the descriptive dimension. Let the circumstances of evaluation be the worlds of the contexts of use. As for the non-descriptive dimension, the view that will ultimately help us explain competent use of normative language and disagreement is this: the circumstances of evaluation are the worlds of the context of use paired with evaluator stances. The evaluator is not necessarily the speaker, but whoever is assessing the content (or whatever standard is being used to assess the content). In this way, evaluative accuracy will be sensitive to what MacFarlane calls the context of

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

58

Matthew S. Bedke

assessment, not just sensitive to the context of use, whereas we leave open the possibility that descriptive accuracy is only sensitive to the world of the context of use. Within this setup, descriptive accuracy is familiar: if the world of the context of use is within the set of worlds that makeup descriptive content, the sentence is descriptively accurate; otherwise it is not. Evaluative accuracy is unfamiliar but predictable: if the world-stance pair of the circumstance of evaluation is within the set of world-stance pairs that make up nondescriptive content, the sentence is evaluatively accurate; otherwise it is not. In effect, each person evaluating the sentence or its content will be supplying their own stances of evaluation, at least as a default.¹⁶ When I evaluate “X is wrong” I supply my own stance of opposition and the content will be evaluatively accurate relative to my stance iff it exhibits some property I morally oppose in action. When you evaluate the same sentence in context, or its content, you supply your stance of opposition and the content will be evaluatively accurate relative to your stance just in case it exhibits some property you morally oppose in action. And so on and so forth. As mentioned, this move is similar to semantic views that make extensions relative to a non-standard parameter in the circumstance of evaluation.¹⁷ Many truth relativists discuss taste predicates, like “is tasty,” where the content of such a predicate has an extension only relative to some standard of taste. Vegemite, for example, would be tasty relative to some standards of taste, and not tasty relative to other standards of taste. Non-Descriptive Relativism is similar, though the field of play is a non-descriptive dimension to meaning. One significant problem for truth relativism is making sense of contents that are true or false only relative to a parameter other than the world and what it is like. On a straightforward reading, truth relativists are asking us to think of some properties as having relative extensions. Applied to normativity, there would be a property of being wrong, which would have an extension only relative to a standard. This has some strange consequences. Take actions like feeding the hungry, and securing peace in the Middle East. It looks like these actions will be within the extension of the property of being wrong relative to a certain assessor standard, and so something like “Feeding the hungry is wrong” will accurately describe the world relative to certain standards. Then again, “Feeding the hungry is not wrong” will also ¹⁶ The default stance can be shifted with operators. If I say “According to act utilitarianism, agonizing torture is justifiable so long as it prevents sufficiently many mild itches” the stance of evaluation is not given by my conative attitudes, but act utilitarianism. ¹⁷ See, e.g., Kölbel (2003, 2004), Kolodny and MacFarlane (2010), Lasersohn (2005), and MacFarlane (2007, 2014).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Non-Descriptive Relativism

59

accurately describe the world relative to certain other standards. On such a view, at least some of the ways things might be appear to be deeply perspectival. I have a hard time understanding the metaphysics of such extensionrelative properties and perspectival facts. Of course, it is possible for the formal semanticist to demur on questions of metaphysics. Perhaps the elements of the model are only posited for their formal properties without concern for issues of aboutness, in which case the contents of these predicates are not properties as the metaphysicist understands them. But if we want our semantic theory to model aboutness—which I do—and if we want it to have the structure to distinguish descriptive discourse from non-descriptive discourse—which I do—it is hard to see how to sensibly do this within one semantic dimension. That said, it is debatable whether relativism is committed to deeply perspectival ways the world might be. But I think we avoid the difficulties entirely when we place the standard sensitivity of extensions within a nondescriptive dimension to meaning. Let us not say that some properties have extensions only relative to standards, and let us not say we are describing how things are in a perspectival way. Our non-descriptive contents for “is wrong” are not properties in any robust metaphysical sense. They are, roughly, all the ways in which one might be morally opposed to some features in actions. Further, a sentence like “Stepping on gouty toes is wrong” is not evaluated directly in terms of truth relative to a perspective, but in terms of descriptive accuracy and evaluative accuracy. This division of labor is exactly what we need to make sense of extensions relative to standards, and to distinguish descriptive from non-descriptive discourse. To be sure, Non-Descriptive Relativism maintains that a sentence’s evaluative accuracy is stance relative, and by default each evaluator supplies her stance. That is the key similarity with semantic relativism. So “Feeding the hungry is wrong” is going to be evaluatively accurate relative to certain stances. But once we drop the idea of properties with relative extensions this relativity is an asset rather than a liability. Like versions of semantic relativism, we capture the fact that semantically competent use of normative terms depends on one’s attitudes. If there is nothing about stepping on gouty toes that one morally opposes, and one knows this, it would be extremely odd for them to call it wrong. This is akin to the oddity of someone calling a food disgusting even though they themselves find nothing gustatorily or aesthetically unpleasant about it (cf. Lasersohn 2005: 655). In both cases the oddity is nicely captured by saying that there are conventional meanings for these terms that make semantically competent acceptance of such a claim sensitive to the thinker’s attitudes. Further, when we allow the default stance to be that of the evaluator we solve the problem of repugnant counterfactuals without needing a rigid

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

60

Matthew S. Bedke

designator in the cognitive rule. The antecedent of “If I didn’t disapprove of stepping on gouty toes, then stepping on gouty toes would not be wrong” takes me to worlds where I do not disapprove of stepping on gouty toes, but when assessing the consequent the stance is supplied by me, the evaluator, not by my counterfactual, benighted self. Applying my stance of moral opposition to those counterfactual worlds, stepping on gouty toes still exhibits some property I morally oppose. So the view predicts that I competently reject the counterfactual. And that is the correct result.

3 .6 . T HE T R U T H I N V A L UE S It is time to recover truth. One option is to equate truth with descriptive accuracy. But the following cognitive rule for “is true” captures an appealing alternative: “True” W-Rule: Predicate “is true” of content C (or sentence S with content C in some context) when (and only when) C is descriptively and evaluatively accurate.¹⁸

Because purely descriptive sentences are vacuously evaluatively accurate, satisfying the rule for “is true” hangs on descriptive accuracy for those sentences. What of other sentences, like “X is wrong”? We have some idea of the conditions under which the content of “X is wrong” is descriptively and evaluatively accurate. It is vacuously descriptively accurate and it is evaluatively accurate relative to stance s when x has some property that is the object of s. Turning to truth, and given “True” W-Rule, it is semantically correct for me to say “It is true that x is wrong” just in case x has some property that I morally oppose in action. More generally, semantically competent ascriptions of truth are sensitive to the evaluator’s attitudes, at least when normativity is involved. Just as the stance sensitivity of evaluative accuracy helped to account for semantically competent use of normative terms, this treatment of truth helps to capture semantically competent use of the truth predicate. We wouldn’t want the semantically competent assertion of “Stepping on gouty toes is wrong” to come apart from the semantically competent assertion “It is true that stepping on gouty toes is wrong.” And as with the basic sentence, ¹⁸ For full generality, we might say “descriptively and non-descriptively accurate.” Also, this ignores many complications concerning truth, but the main point here is to make it sensitive to something other than descriptive accuracy. And, again, the rule should be interpreted as supplying the conditions for semantically correct deployment of “true” (see note 8 and surrounding text).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Non-Descriptive Relativism

61

there is something very odd about someone acknowledging that it is true that stepping on gouty toes is wrong when one thinks that there is nothing about it one morally opposes. If you have nothing against stepping on gouty toes it is semantically correct for you to say that it is false that stepping on gouty toes is wrong, just as it is semantically correct for you to say that it is not wrong. In making these claims about semantic propriety it is important that I am not thereby agreeing with them or accepting what they say or saying it is true. I am merely acknowledging a semantic fact.

3.7. PSYCHOLOGY I have been speaking of the meanings of normative language. But what of normative thought? What is it to think that stepping on gouty toes is wrong? I am not at all confident about what to say here, though I am inclined to say something a bit surprising: Non-Descriptive Relativism should say the same thing as sophisticated subjectivism. That is, whether a content is fixed by f-rules or w-rules makes no difference when it comes to psychology. So, for S to think that stepping on gouty toes is wrong is, roughly, for S to believe that stepping on gouty toes has some property she (S) morally opposes in action. Granted, the non-descriptive content of the sentence “Stepping on gouty toes is wrong” consists in all those world-stance pairs such that stepping on gouty toes has some property that the stance morally opposes in action, which has nothing in particular to do with S’s or anyone else’s stance of opposition. However, to accept this content one must affirm it or think it true. On our semantic picture the content is true only relative to a world-stance pair. When S is making the judgment she is an assessor and her stances of opposition are plugged in. For S to think that stepping on gouty toes is wrong, then, is roughly for S to believe that stepping on gouty toes exhibits some property she morally opposes.¹⁹ I say “roughly” for three reasons. First, the content of the thought does not include anything about S’s psychology. That would be objectionably self-regarding. The content of the thought is , where S morally opposes F. Second, the thought that stepping on gouty toes is wrong does not have the same cognitive significance as the thought that stepping on gouty toes has some property F (where S morally opposes F), even though they might have the same content. To see ¹⁹ This mixes non-descriptivity in language with representational belief in thought. The combination has been overlooked, largely because expressivism offers such a tight connection between a non-representational normative psychology and non-descriptive normative language. See Bedke (2018).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

62

Matthew S. Bedke

this, suppose someone morally opposes actions insofar as they exhibit properties F1, F2, or F3. Her thought that stepping on gouty toes is wrong will be a belief that stepping on gouty toes exhibits F1, F2, or F3. But there can be more than one way of conceiving of properties F1, F2, or F3, and indeed we distinguish normative ways of thinking about them from non-normative ways of thinking about them. The key to the normative mode of conception is to note that the evaluator’s relevant conative attitudes—here, moral opposition—help to populate the content of the normative thought. In this manner the content is being thought of under the guise of one’s normative attitudes. The role of “wrong” is to instruct the cognitive system to consult the thinker’s attitudes of moral opposition to determine the content of the thought. Whatever that content turns out to be, there are going to be other ways of thinking of the same content, ones that are not looking at them through the lens of one’s conative attitudes, and that is what makes the thoughts non-normative.²⁰ Third, for normative ways of thinking one might worry that the view is still objectionably selfregarding insofar as one’s attitudes are involved in that way of thinking. But the attitudes are meant to be backgrounded in thought. S is in no way thinking about her attitudes even though her attitudes are doing some work in populating the content of her thoughts. Because her thoughts are focused on the various properties picked out by her attitudes, not the attitudes themselves, S’s normative thoughts are no more self-regarding than her visual appearances are lens-regarding when she is wearing glasses.²¹ I think this view has the resources to explain the authoritative force that appears in normative thought and language, a force that can appear independent of us and our attitudes. The rough idea is that since normative concepts represent worldly features conceived under the guise of our most important conative attitudes, we get a mix of attitude-independence (whereby attitudes do not show up in content) that is colored by a conative-attitude-infused mode of presentation. If we have attitudes of opposition, or being against, or favoring or disfavoring that color our various modes of conception, this can be easily interpreted as presenting contents that include worldly opposition relations, or not-to-be-doneness relations, or relations whereby some considerations count in favor of some responses

²⁰ Cf. Dreier (1990: 19–20; 1999: 567–8). Egan (2012: 564–5) worries about the lack of common subject matter for value-laden thought. ²¹ While the approach here is similar to Egan’s (2012) de se approach to value judgments, he thinks that accepting a value sentence is to self-ascribe a property. I worry that his view is objectionably self-regarding, and I think it is better to characterize the acceptance as a belief about worldly properties (albeit ones fixed by one’s conative attitudes).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Non-Descriptive Relativism

63

whereas other considerations count against. That is, it can appear as though the objects of our attitudes are themselves weighing against or counting in favor of various responses. Whether the mode of presentation is thereby erroneous is a difficult question. I am inclined to think not, but further discussion is for another time. I have more straightforward comments about action guidance. NonDescriptive Relativism predicts that normative judgments will have the same special motivational profile predicted by sophisticated (descriptive) subjectivism. The psychology on offer is the same. For a judgment of the form x is wrong we predict the motivational profile roughly akin to the judgment that x exhibits some property I (the speaker) morally oppose in action. This is not the place to unpack the prediction, or examine how well it fits the motivational data. But at least we can see that Non-Descriptive Relativism delivers a special connection to motivation that standard nonnormative beliefs lack, albeit a motivational connection that is weaker than the sort of profile we expect for conative states. We can also see how the assertion “X is wrong” would play an endorsement role that many assertions lack, for opposing one of x’s properties would be a condition on the semantic correctness of the sentence—a condition encoded in the cognitive role of “is wrong” by anyone who fully grasps its conventional meaning. Because the psychology is cognitivist, Non-Descriptive Relativism avoids some of the Frege–Geach problems that dog versions of expressivism.²² Consider a version of the negation problem that boils down to accounting for the logical inconsistency of the following two thoughts: (a) Bill’s thought that stepping on gouty toes is wrong. (b) Bill’s thought that stepping on gouty toes is not wrong. If (a) is Bill’s opposition to stepping on gouty toes, while (b) is something like Bill’s toleration of stepping on gouty toes, it is hard to see how these thoughts logically conflict. Attitudes of opposition and toleration toward one and the same thing are in some kind of conflict, but it does not appear to be the right kind of conflict (Schroeder 2008a: ch. 3). Non-Descriptive Relativism simply avoids the problem by having a cognitivist psychology akin to descriptive speaker subjectivism. It says that (a) is Bill’s belief that stepping on gouty toes has some property he morally opposes in action, and (b) is Bill’s belief that stepping on gouty toes does not ²² For the initial Frege–Geach problems, see Geach (1960, 1965) and Searle (1969: 136–41). For some nice discussions see Dreier (1996), van Roojen (1996), and Schroeder (2008b). See also Baker and Woods (2015), Blackburn (1984: ch. 6, 1998: ch. 3), Charlow (2014), Gibbard (2003: chs. 3–4), Horgan and Timmons (2006, 2009), Schroeder (2008a), and Schwartz and Hom (2015).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

64

Matthew S. Bedke

have some property he morally opposes in action. Those are clearly logically inconsistent thoughts.

3 .8 . D I S A G R E E M E N T Let me finally return to the issue of disagreement. Expressivists have struggled with the phenomenon of disagreement for some time. The general strategy has been to posit the possibility of disagreement in attitude, which is distinct from disagreement in fact or in belief. If that can be made to work, Non-Descriptive Relativism might be a beneficiary (cf. Jackson 2008). But here is a simpler alternative. Why don’t we say that what it is for two people to disagree is for one to accept some truth-apt content that the other rejects. This requires us to drop any presumption that what it is to accept some content that another rejects is to describe or represent the world in incompatible ways. That might happen in some cases that concern descriptive contents. However, according to the package of views on offer here people can disagree over normative matters by considering one and the same truthapt non-descriptive content (a set of world-stance pairs) and each can evaluate it relative to her own stance to come to different conclusions. If Able accepts that stepping on gouty toes is wrong she is evaluating a certain set of world-stance pairs—those worlds where stepping on gouty toes exhibits some property that is the object of the paired stance of moral opposition—relative to her stance of moral opposition, and it could come out as evaluatively accurate, and it could be semantically correct for her to think that stepping on gouty toes is wrong and that it is true that it is wrong. For Bea to reject that same truth-apt content she evaluates that same set of world-stance pairs relative to her own evaluative stances and that content could come out as evaluatively inaccurate, and it could be semantically correct for her to think that stepping on gouty toes is not wrong and that it is false that it is wrong. Psychologically, Able thereby believes that the action exhibits some property (which she morally opposes in action), while Bea thereby believes that the action does not exhibit some property (which she morally opposes in action). These beliefs do not represent the world in incompatible ways. But the proposal is that so long as one is accepting truthapt content that the other rejects they disagree. Importantly, the account does not overgeneralize. It is not the case that any difference in attitude is a disagreement. We need truth-apt content—content that is assessable in ways that show up in “True” W-Rule. I take it that the main controversy here is not the characterization of disagreement in terms of accepting and rejecting the same truth-apt content.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Non-Descriptive Relativism

65

The main controversy is the idea that there are w-rules governing normative terms that generate truth-apt contents. The only way to settle this is to see if the best semantic theory posits w-rules and makes competent use of the truth predicate sensitive to evaluative accuracy. If so, we have good reason to believe that the non-descriptive content generated by w-rules is truth apt. Perhaps a deeper worry is whether my discussion of w-rules and the rest is enough to show that a predicate so governed is meaningful, and particularly meaningful in the ways in which normative terms seem to be meaningful (they are compositional and can feature in valid arguments, etc.). Dreier (1996) nicely illustrates how to fall short. He considers a semantic “theory” for “is hiyo” that says only this: “Is hiyo” is syntactically a predicate and it is used to accost, so “Bob is hiyo” is used to accost Bob. Dreier then complains, rightly, that though the “theory” says what the term is used to do, that is not enough to show that we have a meaningful predicate on our hands, particularly one that figures in truth-apt sentences that play certain Frege–Geach roles. We do not have any idea, for example, what the complex “Either Bob is hiyo or a dingo is near” means, or how to reason with such a premise. Does a semantics with w-rules do any better? Well, if I were to cook up a predicate “is hiyo” using a w-rule, it might be this: “Predicate ‘is hiyo’ of x when you accost x” or maybe “Predicate ‘is hiyo’ of x when you want to accost x.” These w-rules are unlike the “theory” that only discusses what the term is used to do. They actually give semantic correctness conditions, and we can add to the story by filling in whether the contents here are sensitive to certain parameters of the circumstances of evaluation, and so on. If we do all that, I think we would thereby introduce a sensible predicate. Having said that, it is hard to imagine what it would be like to use such a term, for our language doesn’t have such a term and it is hard to imagine why some language would develop such a term. There is no need for a non-descriptive predicate “is hiyo.” All our communicative needs in this neighborhood are adequately served by the descriptive predicates “is accosted by me” or “is someone I want to accost.”²³ On the other hand, it is easy to imagine why language would develop predicates that allow us to reason about the objects of our most important ²³ Notice that if there were a term “is hiyo” governed by w-rules, the term would not bake in any performative character. Arguably, the language does need inherently performative terms, which is what the original “hiyo” was meant to do. If I were to try to capture the cognitive role for the original “hiyo,” used for accosting, I might try this: Say “hiyo” to x to accost x. This t-rule bakes in the term’s performative character. But unlike our f-rules and w-rules it does not deliver the kinds of contents that are assessable for truth. The thought is that language has no use for truth predicates that are sensitive t-rule “content.”

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

66

Matthew S. Bedke

conative attitudes, and so indirectly to reason about which attitudes to have toward which objects. It is nice to be able to use modus ponens, for example, to help decide whether to morally oppose one action in light of moral opposition to another (cf. Gibbard 2003: 13). We also might like a way of communicating our attitudes to others, to reason together about what attitudes to have, and to contest the attitudes of others without contesting what they say for descriptive inaccuracy. Whether descriptive content and descriptive accuracy alone can meet these needs (perhaps combined with other semantic and pragmatic machinery, like implicatures) is a good question. But language could do an excellent job meeting these needs if only it could avail itself of semantically nondescriptive terms that are connected to our attitudes but also sensibly play the predicate role, deliver truth aptitude for its sentences, enter into inferential reasoning, etc. But how? The suggestion here is that a semantic dimension with w-rules and a truth predicate sensitive to the nondescriptive content it generates would do the job nicely. Granted, it is one thing to see how useful it would be for a language to develop w-rules concerning objects of our conations, and a truth predicate sensitive to the non-descriptive content they deliver. It is another matter entirely to show that our language is like this. Fulfilling this second project depends on whether the Non-Descriptive Relativist hypotheses offered here feature in the best explanations of all the metaethical data we would like to explain, just as we should believe that all declaratives express states of mind if that features in the best explanation of all the data we would like to explain. Sadly, this chapter only engages with this bigger project in a piecemeal way.

3.9. MORE FREGE–GEACH ROLES In light of that, let me say a bit more about composition and validity, since the two types of content here might raise some questions about how this goes. Consider the following sentences: (A) Stepping on gouty toes is wrong. (B) Apple’s stock will rise. (C) Either stepping on gouty toes is wrong or Apple’s stock will rise. We can say that the non-descriptive content of (A) is the set of world-stance pairs where stepping on gouty toes in the world has some property the paired stance morally opposes, and no descriptive content is ruled out. (B)’s

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Non-Descriptive Relativism

67

descriptive content is the set of worlds where Apple’s stock will rise, and it rules out no non-descriptive world-stance pairs. But what about the complex sentence in (C)? We should say that “or” operates on the descriptive and nondescriptive contents jointly. So (C) puts the following content in play: (I) Those (non-descriptive) world-stance pairs where stepping on gouty toes has some property morally opposed by the stance, plus all (descriptive) worlds, S

(II) All (non-descriptive) world-stance pairs, plus those (descriptive) worlds where Apple’s stock rises.

In effect, the descriptive and non-descriptive contents for each disjunct are conjoined and the “or” operates on the conjoined contents. Each disjunct is assessable for descriptive and evaluative accuracy relative to a (descriptive) world of evaluation and a (non-descriptive) world-stance pair of evaluation, and truth is assessable for each disjunct. Finally, (C) is true if and only if either of the disjuncts is true. In this way we have standard compositionality of truth conditions, it’s just that truth conditions have been enriched to include non-descriptive contents in addition to descriptive contents, and there are two different circumstances of evaluation for each dimension of content corresponding to the extensions of descriptive accuracy and evaluative accuracy. What is it to accept the disjunction in thought? We should say that to accept the disjunction is to assess it positively relative to the relevant circumstance of evaluation. Because there is normative content here, the relevant non-descriptive circumstance of evaluation will include the evaluator’s stances. For me to accept the disjunction boils down to my believing disjunctive content: either Apple’s stock will continue to rise or stepping on gouty toes has some property F (where I morally oppose F in action). For validity, consider the following argument: (1) If tormenting the cat is wrong, then tormenting the dog is wrong. (2) Tormenting the cat is wrong. (C) Tormenting the dog is wrong.

The classic statement of validity holds. If the premises are true the conclusion must be true. More carefully, for all contexts and circumstances of evaluation, if the premises are true relative to some context and circumstance, then the conclusion is true relative to that same context and circumstance. Beneath the surface of this truth-conditional approach we have descriptive and non-descriptive dimensions at play. But they have no effect on our characterization of validity.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

68

Matthew S. Bedke

I hope it is clear that these points generalize to other arguments. Let me briefly consider one that is especially difficult for expressivism: (1r) Either dogs do not feel pain, or kicking the dog is wrong. (2r) Dogs feel pain. (Cr) Kicking the dog is wrong.

Expressivism has a hard time here because the conclusion is said to express a conative attitude toward kicking the dog, and yet we do not have any premises that express a conative attitude in any straightforward sense. That makes it puzzling how the argument generates a conclusion that does express a conative attitude. How do we get such attitude expression out of premises that do not express attitude? Non-Descriptive Relativism avoids the puzzle. First, attitude expression does no theoretical work for us. Instead, when I accept the conclusion I form a belief—roughly, the belief that kicking the dog has some property (which I morally oppose in action). Second, this acceptance is foisted upon me if I accept both premises. By accepting (1r), I believe that either the dog does not feel pain, or kicking the dog is an action that has some property F (where I morally oppose F in action). By accepting (2r) I believe that dogs feel pain. So I must now believe that kicking the dog is an action that has some property F (where I morally oppose F in action), on pain of logical error of a familiar kind.

3. 10 . CONC L U S ION I have been assuming that non-descriptivism and action-guidance are worth defending, and that it would be desirable to have a metaethical alternative to expressivism. I have tried to show how a version of Non-Descriptive Relativism fills the bill. The version I have developed posits w-rules for some bits of language, rules that operate right alongside descriptive semantics. Whether we should believe it depends on whether it is part of the best explanation of all the metaethical data we would like to explain. Inevitably, I have only begun to do that explanatory work here and frankly I am much more confident that we should be exploring the non-descriptivist and action-guiding terrain, and generating alternatives to expressivism, than I am in the details of the view sketched here.²⁴

²⁴ For comments or helpful discussion on previous versions of this chapter, thanks to Giulia Pravato, Michael Ridge, Max Kolbel, Tristram McPherson, Mark Schroeder, Andrew Alwood, two referees for Oxford University Press, the audience at the 2016 CHillMeta Workshop, and all those I am forgetting.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Non-Descriptive Relativism

69

References Åqvist, L. (1964). Vagueness and value. Ratio, 6(2), 121–7. Baker, D. and Woods, J. (2015). How expressivists can and should explain inconsistency. Ethics, 125(2), 391–424. Bedke, M. (2010). Might all normativity be queer? Australasian Journal of Philosophy, 88(1), 41–58. Bedke, M. (2018). Cognitivism and non-cognitivism. In T. McPherson and D. Plunkett (eds.), The Routledge Handbook of Metaethics. London: Routledge, 292–307. Blackburn, S. (1984). Spreading the Word: Groundings in the Philosophy of Language. Oxford and New York: Oxford University Press. Blackburn, S. (1998). Ruling Passions: A Theory of Practical Reasoning. Oxford and New York: Oxford University Press. Charlow, N. (2014). The problem with the Frege–Geach problem. Philosophical Studies, 167(3), 635–65. Chrisman, M. (2012). On the meaning of “ought.” In R. Shafer-Landau (ed.), Oxford Studies in Metaethics, Volume 7. Oxford: Oxford University Press, 304–32. Dreier, J. (1990). Internalism and speaker relativism. Ethics, 101(1), 6–26. Dreier, J. (1996). Expressivist embeddings and minimalist truth. Philosophical Studies, 83(1), 29–51. Dreier, J. (1999). Transforming expressivism. Noûs, 33(4), 558–72. Egan, A. (2012). Relativist dispositional theories of value. The Southern Journal of Philosophy, 50(4), 557–82. Geach, P. (1960). Ascriptivism. The Philosophical Review, 69(2), 221–5. Geach, P. (1965). Assertion. The Philosophical Review, 74(4), 449–65. Gert, J. (2007). Cognitivism, expressivism, and agreement in response. In R. ShaferLandau (ed.), Oxford Studies in Metaethics, Volume 2. Oxford: Oxford University Press, 77–110. Gert, J. (2012). Basic normative terms. In Normative Bedrock. Oxford: Oxford University Press, 34–70. Gibbard, A. (2003). Thinking How to Live. Cambridge, MA: Harvard University Press. Gutzmann, D. and Gärtner, H.-M. (eds.) (2013). Beyond Expressives. Explorations in Use-Conditional Meaning. Current Research in the Semantics Pragmatics-Interface, Vol. 28. Leiden: Brill. Horgan, T. and Timmons, M. (2006). Cognitivist expressivism. In T. Horgan and M. Timmons (eds.), Metaethics After Moore. Oxford: Oxford University Press, 255–98. Horgan, T. and Timmons, M. (2009). Expressivism and contrary-forming negation. Noûs-Supplement: Philosophical Issues, 19, 92–112. Humberstone, I. L. and Davies, M. (1980). Two notions of necessity. Philosophical Studies 38(1), 1–30. Jackson, F. (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.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

70

Matthew S. Bedke

Jackson, F. and Pettit, P. (1998). A problem for expressivism. Analysis, 58(4), 239–51. Kaplan, D. (unpublished). The meaning of “ouch” and “oops”: explorations in the theory of meaning as use. Kölbel, M. (2003). Faultless disagreement. Proceedings of the Aristotelian Society, 104 (1), 53–73. Kölbel, M. (2004). Indexical relativism versus genuine relativism. International Journal of Philosophical Studies, 12(3), 297–313. Kolodny, N. and MacFarlane, J. (2010). Ifs and oughts. The Journal of Philosophy, 107(3), 115–43. Lasersohn, P. (2005). Context dependence, disagreement, and predicates of personal taste. Linguistics and Philosophy, 28, 643–86. MacFarlane, J. (2007). Relativism and disagreement. Philosophical Studies, 132, 17–31. MacFarlane, J. (2014). Assessment Sensitivity: Relative Truth and Its Applications. Oxford: Oxford University Press. Parfit, D. (2011). On What Matters, Volume 1. Oxford: Oxford University Press. Pravato, G. (unpublished). The meaning and use of normative language. Price, H., Blackburn, S., Brandom, R., Horwich, P., and Williams, M. (2013). Expressivism, Pragmatism and Representationalism. Cambridge: Cambridge University Press. Ridge, M. (2014). Impassioned Belief. Oxford: Oxford University Press. Rosen, G. (1998). Blackburn’s Essays in Quasi-Realism. New York: Oxford University Press. Schiffer, S. (2002). Moral realism and indeterminacy. Philosophical Issues, 12(1), 286–304. Schroeder, M. (2008a). Being For: Evaluating the Semantic Program of Expressivism. Oxford and New York: Oxford University Press. Schroeder, M. (2008b). What is the Frege–Geach problem? Philosophy Compass, 3(4), 703–20. Schroeder, M. (2013). Tempered expressivism. In R. Shafer-Landau (ed.), Oxford Studies in Metaethics, Volume 8. Oxford: Oxford University Press, 283–313. Schwartz, J. and Hom, C. (2015). Why the negation problem is not a problem for expressivism. Noûs, 49(4), 824–45. Searle, J. R. (1969). Speech Acts. Cambridge: Cambridge University Press. Smith, M. and Stoljar, D. (2003). Is there a Lockean argument against expressivism? Analysis, 63(1), 76–86. Toppinen, T. (2013). Believing in expressivism. In R. Shafer-Landau (ed.), Oxford Studies in Metaethics, Volume 8. Oxford: Oxford University Press, 252–82. van Roojen, M. (1996). Expressivism and irrationality. Philosophical Review, 105(3), 311–35. Wiggins, D. (1987). A sensible subjectivism? In Needs, Values, Truth: Essays in the Philosophy of Value. Oxford: Blackwell, 185–214.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

4 How to Learn about Aesthetics and Morality through Acquaintance and Deference Errol Lord

There are parallel debates in aesthetics and metaethics about the epistemic merits of deference.¹ Most hold that there is something amiss with purely deferential beliefs. The similarities between the two debates dry up quickly, though. The main controversy in aesthetics is whether it is even possible to acquire deferential aesthetic knowledge. In metaethics, on the other hand, most agree that deferential moral knowledge is possible.² The trick is explaining why, despite this, there is something fishy about deferential moral beliefs. A plausible hypothesis about why the two literatures diverge is that it is widely accepted in aesthetics that acquaintance with things that have aesthetic value is necessary to have paradigmatic aesthetic knowledge.³ Since one doesn’t become acquainted with things that have aesthetic value when one defers, many have thought that deferential aesthetic knowledge is impossible. In metaethics, on the other hand, relatively little has been said about the role of acquaintance in the acquisition of moral knowledge and how this

¹ As has been pointed out in the literature (e.g., McGrath, 2009; Howell, 2014), there is a difference between testimony and deference, and the puzzle is about deference. Relying on the testimony of others in certain ways is a way of deferring, but there are other ways of finding out what others believe. The cases I will focus on are cases of testimony given how central testimony is to common ways of thinking, but I will also often speak of deference and intend the conclusions to apply to all sorts of deference. ² On the aesthetics side, see Robson (2012) and the citations therein. On the metaethics side, see Hills (2013) and the citations therein. ³ In what follows I will explain what I mean by paradigmatic aesthetic knowledge. I use the fudge term now because it is controversial which epistemic state acquaintance is required for.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

72

Errol Lord

might bear on the debate about deference.⁴ In this chapter, I argue that it’s plausible that acquaintance does play a prominent role in the acquisition of paradigmatic moral knowledge and that this helps explain what’s amiss with moral deference. In fact, I will defend a general theory of what is going on in both morality and aesthetics. According to this theory, acquaintance enables the possession of certain facts as reasons. It does this by putting one in a position to exercise a certain type of know-how. By being acquainted with certain normatively relevant facts, one is put in a position to exercise knowledge about how to use those facts in various ways. Deference cannot acquaint us with the full range of normatively relevant properties. Thus, deference cannot enable possession of the full range of reasons. This limits what we can do with the normative information. This is what is amiss with deferential ways of thinking about morality and aesthetics. The plan is as follows. In Section 4.1 I will introduce the puzzles of moral and aesthetic deference. In Section 4.2 I will defend a story about what is amiss with aesthetic deference. This story relies on the claim that acquaintance is required for what I call appreciative aesthetic knowledge. In Section 4.3 I will sketch a way to generalize this story to the moral case. While I take the generalization to have serious appeal, a full defense requires that I further defend the claim that acquaintance is important for the acquisition of moral knowledge. In Section 4.4 I will argue that it is important by responding to two objections. Section 4.5 wraps up.

4. 1. TH E P U Z ZLE O F M ORA L A N D AESTHETIC DEFERENCE Deference—with testimony being the exemplar—is an easy and efficient way to extend our knowledge. With this fact in mind, it is surprising that it is intuitive that there is something amiss with deferential ways of thinking about aesthetics and morality. We can see this by reflecting on particular cases. Consider the following two cases. Nefertiti: Hanna just returned from a trip to Berlin during which she saw Nefertiti’s Bust. Hanna’s sister Clara asks her about the museums. Hanna tells her that Nefertiti’s Bust was especially beautiful. Clara comes to believe that Nefertiti’s Bust is beautiful solely on the basis of Hanna’s word.

⁴ Fletcher (2016) expresses sympathy for the thought that acquaintance is required in both the aesthetic and the moral case. He does not pursue the details, though.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Acquaintance and Deference

73

Newbie: Edmund is a new Anglo police officer in the 1920s Burmese police force. As is the tradition for Anglo police officers in Burma, Edmund is immediately scheduled to guard over a hanging of a Burmese prisoner the next morning. Edmund has no opinion about the moral status of hanging and has not thought about it much in his short life (although he knows what hanging is). One of his fellow police officers, George, tells him that hanging is wrong. Edmund comes to believe that hanging is wrong solely on the basis of George’s word.

Both Clara and Edmund defer. There is something odd with Clara and Edmund relying solely on others about which aesthetic and moral views to adopt. And this oddness does not go away when we stipulate that Clara and Edmund know Hanna and George are reliable. It has seemed to many that when it comes to aesthetics and morality, one should think through the issues for oneself. One should not farm out one’s views to others. It is important to be clear about the boundaries of this particular intuition. First, I will focus on cases of direct deference about a pure assertion that makes a thin normative evaluation.⁵ Someone’s deferential belief is direct when that belief is solely based upon the testimony of someone else. An assertion is pure when the content does not include any descriptive or explanatory information. An evaluation is thin when it only ascribes thin normative properties. I won’t take a stand on the nature of thick/thin distinction. I’ll stick to beauty and wrongness. Cases with these restrictions are the best ones for isolating the epistemic merits of deference. The direct deference restriction helps by ensuring that the relevant characters don’t hold their beliefs for other reasons. For example, in a different version of Hanna and Clara’s case, Clara partially believes Nefertiti’s Bust is beautiful because of photographs of the bust she has seen. In that case Clara might know that Nefertiti’s Bust is beautiful independently of Hanna’s testimony even if she partially believes it is beautiful on the basis of the testimony. Thus, allowing one to believe for other reasons presents a confounding factor for the epistemic evaluation of Hanna’s testimony. The restriction to pure assertions of thin evaluations helps for similar reasons. Impure assertions mention a purported ground for the claim asserted. This information can help the deferer learn the normative fact on other grounds. For example, if George said that hanging is wrong because it tramples upon human dignity, Edmund might come to know it is wrong by inferring that it is wrong from his antecedent knowledge that acts that trample on human dignity are wrong. In that case George’s testimony

⁵ This terminology follows Fletcher (2016).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

74

Errol Lord

about the wrongness is at best sharing the epistemic work. His testimony about the grounds plus Edmund’s antecedent knowledge is what is doing the full epistemic work. Thick evaluative facts entail certain descriptive facts. Thus, one can learn about some of the descriptive facts via testimony about thick evaluations. One might then use antecedent knowledge about the aesthetic or moral upshot of those descriptive facts to come to know the aesthetic or moral fact via inference. In this the antecedent knowledge is doing important epistemic work. A more precise statement of the intuitions I am interested in is that it is odd to directly defer to someone else about a pure normative claim about a thin normative property. Herein I will say that one defers only when someone does this. This, however, is still too broad. This is because everyone agrees that there are circumstances where it is not only permissible for one to defer but required that one defer. Some examples: one must make a decision before one can think the issue through, one is morally stunted to the point that one cannot think it through, one knows that one is extremely unreliable when one thinks things through.⁶ In these cases, one should defer. Nevertheless, deferring is still odd in the subset of cases that involve healthy and mature moral and aesthetic agents that have time to reflect and have good reason to think that such reflection will be successful. Once one morally and aesthetically matures, it is odd to farm out one’s views, at least when one has the time to think things through. Here’s the rub: this oddness is itself puzzling. After all, it is more than okay to farm out one’s views on nearly all other subjects, at least if those to whom you defer know what they are talking about. No one is puzzled about deference to accountants about tax matters, doctors about health, or chemists about molecular bonding. So it looks like there is an asymmetry between the merits of deferring on moral and aesthetic matters and the merits of deferring on all other matters. This asymmetry needs to be explained. Many in both aesthetics and metaethics are pessimists about deference. Pessimists think that we ought not defer. There are two different kinds of pessimists. Some—call them epistemic pessimists—hold that we ought not defer because deference is not a source of moral or aesthetic knowledge. Epistemic pessimism has been a popular view in aesthetics since at least Kant. The foremost reason for this is that it is plausible that acquaintance is required for well-formed aesthetic beliefs. Deference cannot acquaint one with the aesthetically relevant features. Thus, if one has a strong acquaintance requirement, epistemic pessimism follows.⁷

⁶ See Jones (1999), Hills (2009), Sliwa (2012), Enoch (2014). ⁷ Cf. Budd (2003), Livingston (2003), Robson (2012).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Acquaintance and Deference

75

Epistemic pessimism is not popular in metaethics.⁸ Most agree that one can gain moral knowledge through deference. The trick in metaethics has been to explain what is amiss with deference despite this. Thus, most pessimists in metaethics (and some in aesthetics) are non-epistemic pessimists. The most popular kind of non-epistemic pessimism holds that deference is morally or aesthetically forbidden.⁹ Optimists about testimony—i.e., those who think that deference is often (epistemically, aesthetically, morally) permitted—have an explanatory burden. What this burden is will depend on whether they think there is still something amiss with deference. If they do, then they have to explain why there is still something amiss with deference even though we are often permitted to defer. If they don’t think that there is something amiss when deference is permitted, they have to explain why so many have thought otherwise. I am an optimist that thinks that there is still something amiss with deference. The goal of Section 4.2 and Section 4.3 is to (i) explain what is amiss with aesthetic and moral deference and (ii) explain why optimism is nevertheless true.

4 .2 . AC Q U A I N T A N C E , AP P R E C I A T I V E K N O W L E D G E , AN D T H E P O S S E S S I O N O F R E A S O N S

4.2.1. The Perks of Aesthetic Acquaintance It is doubted by no one that aesthetic acquaintance has certain intellectual perks. To see what some of these perks are, let’s think more about Hanna: Examining Nefertiti: Hanna is visiting Berlin for the first time. She rushes to the Neues Museum to see Nefertiti’s Bust. She is overwhelmed by the beauty of the bust. She studies its features for an hour. She not only passively looks at the bust, but also actively thinks about how the various features of the bust interact.

Hanna has a host of reactions to Nefertiti’s Bust. She believes that it is beautiful, awesome, graceful. She is impressed by it and in awe of it. She desires its preservation, intends to promote it, and hopes it persists forever. ⁸ There are some arguments that raise epistemic problems for moral deference, most prominently in Jones (1999), Driver (2006), McGrath (2009). ⁹ Alison Hills is a prominent defender of non-epistemic pessimism in metaethics (see especially Hills, 2009). See also McGrath (2011c), Howell (2014), Hopkins (2007). Robert Hopkins is the primary defender of non-epistemic pessimism in aesthetics (see especially Hopkins, 2011).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

76

Errol Lord

Given her interaction with the work, it’s plausible that all of the beliefs cited above constitute knowledge. Further, it’s plausible that the other reactions are fitting as well. There is an important connection between Hanna’s affective and conative reactions being fitting and Hanna knowing the important aesthetic features of Nefertiti’s Bust. Hanna’s conative and affective reactions are fitting in part because she knows the aesthetic facts. Those facts themselves make the affective and conative reactions fitting—e.g., the fact that Nefertiti’s Bust is awesome makes awe fitting. Finally, it’s plausible that the way in which Hanna acquired her aesthetic knowledge is important. Hanna is put in a particularly good position to have fitting attitudes in response to the work. Cases like Hanna’s involve what I call appreciative aesthetic knowledge. Appreciative aesthetic knowledge is the sort of knowledge that allows one to fittingly have the full range of affective and conative reactions.¹⁰ In short, it is the kind of knowledge that enables appreciation. This kind of knowledge is central to our lives as aesthetic agents. Further, explaining how it is that we acquire such knowledge has been a central task of the epistemology of aesthetics.

4.2.2. Acquaintance and the Possession of Reasons So far, this should be plausible to a very wide range of theorists. Now I want to offer an explanation of why Hanna’s knowledge allows her to rationally appreciate the bust. To start, notice that the fact that Nefertiti’s Bust is beautiful provides a reason to appreciate the bust. That is, the fact that Nefertiti’s Bust is beautiful recommends that Hanna appreciate Nefertiti’s Bust. It recommends being moved by the bust, wanting to protect the bust, intending to look at the bust, caring about the bust, etc. I take this to be an entrenched piece of common sense about the normative upshot of beauty. This doesn’t by itself help much, since a reason plausibly doesn’t justify one’s reactions if one is in the dark about the existence of that reason. For example, before Hanna encounters the bust, the fact that it is beautiful does not make it rational for Hanna to appreciate the bust. For this reason, it’s plausible that in order for a reason to justify or rationalize a reaction, one must possess that reason. When one possesses a reason, the fact that provides that reason is in one’s ken as a reason. Before explaining Hanna’s case more, let me say some general things about possession.¹¹ ¹⁰ To be clear, I am not claiming that appreciative knowledge is literally a different kind of knowledge. It might be (and I am attracted to this) identical in kind to all knowledge even though it is differentiated from some knowledge by its additional rational powers. ¹¹ For much more, see Lord (2010, 2014) and especially (Lord, FCa, chs. 3–4). See also Schroeder (2008, 2011), Sylvan (2015), Whiting (2014).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Acquaintance and Deference

77

The least controversial feature of possession is that when you possess a reason, you stand in a special epistemic relationship with that reason. If the reason is not in your epistemic ken, you cannot possess that reason. It is controversial which epistemic relation is the relevant relation. Fortunately, for my purposes it suffices to assume that all of the characters in the relevant cases know the facts that provide the reasons, for all plausible views of possession hold that knowledge is sufficient for meeting the epistemic condition. We needn’t adjudicate the controversy about the epistemic condition because it is the other aspect of possession that is relevant here. Let’s first see why meeting the epistemic condition is not sufficient for possessing some fact as a reason for a particular reaction. Consider Watson. Watson was just told by Sherlock that the boot print in the snow was made by a size 9 Iron Ranger. Although it is clear that Sherlock has inferred who did it from this information, Watson stares back blankly. It is only after Sherlock informs Watson that the cabby wears size 9 Iron Rangers that Watson gets it. Before Sherlock fills Watson in on the cabby’s choice of boots, Watson fails to possess the fact that the print was made by a size 9 Iron Ranger as a reason to believe the cabby did it.¹² This is plausible because two other claims are plausible. First, that when one possesses a reason to have some reaction, that reason affects the rational case for having that reaction. Second, the rational case for Watson believing the cabby did it is not affected by merely finding out that the boot print was made by a size 9 Iron Ranger. After all, if it were affected, it seems like Watson would be irrational for not drawing the inference. But he isn’t irrational. Nevertheless, he knows that the boot print was created by a size 9 Iron Ranger—he meets the epistemic condition. Yet, Watson doesn’t possess that fact as a reason to believe the cabby did it. So meeting the epistemic condition is insufficient. On my view, what’s missing is that, given his epistemic situation, Watson cannot manifest a certain kind of know-how. Watson cannot manifest knowledge about how to use that fact as the reason it is to believe the cabby did it. If he were to start inferring claims about who did it from that fact, he would be shooting in the dark. He wouldn’t be using knowledge about how to use that fact as the reason it is. This is why, I claim, he’s not in a position to form a rational belief about who did it on the basis of that fact. Things change once Sherlock informs him about the cabby’s choice of boot. Once Watson finds this out, he is in a position to manifest knowledge ¹² I put this qualification in to avoid lending the impression that possession of a fact as a reason is an all or nothing affair. One can possess a fact as a reason to ϕ but not possess it as a reason to ψ even though it is a reason to ψ. As I go on to argue (cf. Lord, 2016), this makes a big difference in the current debate.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

78

Errol Lord

about how to use that fact to believe the cabby did it. He does this by immediately inferring that the cabby did it by using a bit of his know-how. His resulting belief is a rational achievement. This is why the resulting belief is rational.¹³ Now back to Hanna. My view is that, in Hanna’s case, her acquaintance with Nefertiti’s Bust enables her to meet both conditions on possession. She comes to know that Nefertiti’s Bust is beautiful and she comes to be in a position to manifest knowledge about how to fully react to that fact. By coming to know that Nefertiti’s Bust is beautiful via acquaintance, Hanna is put in a position to manifest knowledge about how to react affectively and conatively to the bust’s beauty. While I take this story to be intuitive, a full defense requires an explanation of why acquaintance plays the role I am claiming it does. The short answer is that the reactions that require acquaintance are reactions to the way in which Nefertiti’s Bust is beautiful. Budd (2003, p. 392) puts the point well when he writes ‘appreciation of a work is not a matter of knowing what its aesthetic properties are, but of perceiving them as realized in the work’. We cannot display the right sensitivities to the particular way the bust is beautiful just by hearing a description of the bust. However, we can display the relevant sensitivities when we have direct access to the features of the bust. This is why acquaintance is needed in order for us to manifest knowledge about how to fully react to all of the reasons provided by the aesthetic features of any particular work. The fundamental rational significance of acquaintance with aesthetic properties, then, is that it puts us in a position to rationally appreciate. It does this by enabling us to possess certain facts as reasons for appreciation. And it does this by putting us in a position to manifest knowledge about how to use those facts as reasons for appreciation.

4.2.3. The Downsides (and Upsides) of Aesthetic Deference We are now in a position to explain what is amiss with aesthetic deference. The primary downside is that deference cannot put us in a position to gain appreciative knowledge. This is because it does not put us in a position to be acquainted with the specific ways in which the aesthetic features are realized. This is a serious failing given the centrality of appreciation to our aesthetic lives. Most of the time we are not particularly concerned with having mere knowledge of the aesthetic facts; instead, we want to gain appreciation of the

¹³ For much more on this, see Lord (FCa) and Lord (FCb).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Acquaintance and Deference

79

aesthetic objects that interest us. This is one reason why we dedicate so many resources to engaging with things with aesthetic value. This by itself should not drive us to pessimism. The failure of deference to deliver appreciative knowledge does not entail that we ought not defer. We have good reason to think that we can gain knowledge by deferring. One piece of evidence is the very common intuition that aesthetic deference yields knowledge.¹⁴ A weightier reason is provided by the fact that it’s plausible that by deferring we can come to possess reasons for some reactions that are provided by the aesthetic facts.¹⁵ On the basis of Hanna’s testimony Clara can come to possess the reason to go to the Neues Museum provided by the fact that Nefertiti’s Bust is beautiful. Possession of this reason does not depend on one being acquainted with the particular way in which the bust is beautiful. On my view, this is because one can display a sensitivity to this reason without being acquainted with the bust’s particular beauty—Clara could rationally go to the Neues on the basis of the fact that the bust is beautiful. It is worth stressing that listening to trusted informants also seems to be a central aspect of our aesthetic inquiries. There are a lot of things in the world to aesthetically engage with. Given various pressures on our time, attention, and stamina, it is important for many of us to be pointed in the right direction by trusted informants. And, I think, most of us often engage in this practice. We do defer to friends and experts when planning about what to engage with. When things go well, we come to know various aesthetic truths and act on that knowledge. Thus, I think it is plausible that we can gain deferential aesthetic knowledge. It is thus implausible to think that we epistemically ought not defer. Epistemic pessimism should be rejected. This leaves non-epistemic pessimism. I will focus on the most popular (and plausible) version, which holds that we aesthetically ought not defer. I’ll consider the basic idea driving Hopkins’ (2011) discussion of the thought that we aesthetically ought to make up our own minds about the aesthetic merits of things. One way he fleshes this out is that in order to correctly make up one’s mind, one must be acquainted with the relevant aesthetic properties.¹⁶ ¹⁴ See, e.g., Meskin (2004), Meskin and Robson (2015), Robson (2013), Hopkins (2011), Budd (2003), Driver (2006). ¹⁵ Whiting (2015) argues that since Clara cannot possess a reason for appreciation via Hanna’s testimony, she cannot possess a reason for action via Hanna’s testimony. In Lord (2016) I show that Whiting’s argument relies on implausible claims about possession. ¹⁶ He also considers the claim that in order to correctly make up one’s own mind, one must grasp the grounds for the aesthetic facts. This is an aesthetic version of Hills’ understanding view. I think what I go on to say against Hills’ view applies to this view as well.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

80

Errol Lord

On a flat-footed way of understanding the basic idea, it seems much too strong to say that we aesthetically ought to make up our own minds. This is because, as we just saw, it also seems central to our aesthetic lives that we defer to trusted informants, at least when it comes to planning about what to engage with. In these cases it is implausible that we are required in any way to make up our own minds. Hopkins agrees with this, but he thinks his form of non-epistemic pessimism can account for these cases. This is because he maintains that the cases where one is permitted to defer are all cases where one can’t be acquainted with the relevant properties. Since ‘ought’ implies ‘can’, it follows that the relevant aesthetic obligation doesn’t apply. The best test case for this is a case where one has interacted with the work and is unsure what to think about it on the basis of this interaction. Suppose that Clara goes to see Nefertiti’s Bust and is unsure whether it is beautiful. Hopkins argues that if optimism is true, then Clara should be able to permissibly use Hanna’s testimony to rationally believe the bust is beautiful. Hopkins reports a strong intuition that this is not permitted. This provides evidence that there is an aesthetic obligation to make up one’s own mind when one can be acquainted with the relevant properties. While this reply has obvious appeal, I don’t think it works. Let’s start with Hopkins’ focus on cases like Clara’s. The devil will lie in the details of these cases. In many of them it’s right that the protagonist cannot permissibly defer. However, in at least many cases this will be for epistemic reasons. After all, if it is rational for Clara to be unsure about the bust’s beauty after seeing it, she will possess strong reasons not to believe it is beautiful. Given that it is plausible that the evidence acquired through interaction with the bust will generally be of higher quality than the evidence provided by Hanna’s testimony, it is plausible that the reasons not to believe the bust is beautiful will outweigh the reason to believe it is beautiful provided by Hanna’s testimony. But then it won’t be epistemically permissible to defer. Other cases look different. Suppose Akshai is in the same position as Clara, with one twist. After he interacts with the bust and gets Hanna’s testimonial evidence, he finds out that he was slipped a pill that makes him unreliable at evaluating art. In this case it seems permissible to defer to Hanna about the bust’s beauty even though he was acquainted with the work.¹⁷

¹⁷ As a referee pointed out to me, upon reflection it is not obvious that Akshai meets a plausible version of the acquaintance requirement in this case. It will depend on what we are required by the acquaintance requirement to be acquainted with. Most in the literature hold that one needs to be acquainted with the work itself (or a so-called ‘adequate surrogate’). Akshai does meet this requirement. In other work I actually argue against the common view and hold instead that one needs to be acquainted with the

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Acquaintance and Deference

81

Notice that Akshai’s case casts doubt on Hopkins’ claim that the only cases where deference is permitted are cases where one is not in a position to make up one’s own mind. Akshai is in the position to make up his own mind insofar as he has been acquainted with the bust. Thus, Hopkins’ view predicts he aesthetically ought not defer. It’s plausible that this is the wrong prediction. The upshot is that despite Hopkins’ arguments, it still seems like cases of permissibly deferring to trusted informants is a problem for pessimism of all kinds, including Hopkins’ version of non-epistemic pessimism. Further, my optimistic view gives a plausible explanation of what’s amiss with deference while also having a nice explanation of why deference is often epistemically and aesthetically permissible. I see this as strong reason to accept my moderately optimistic view about the rational power of aesthetic deference.

4 . 3. G E N E R A LI Z I N G T O T H E M O R A L C A S E So far I have argued that there is a plausible story to tell about the merits and demerits of aesthetic deference that crucially appeals to acquaintance. Given the similarities between the puzzle of aesthetic deference and the puzzle of moral deference, it is worth seriously considering whether such a story can be generalized to the moral case. In the rest of the chapter I will lay out a package of views that delivers on such a generalization. The goal is not to prove that such a package is true. Rather, it is make a preliminary case for the package of views.¹⁸

4.3.1. Becoming Acquainted with Moral Acquaintance Although not discussed as often, I think that there are cases of moral learning that closely resemble cases like Hanna’s.¹⁹ Let’s fill in the properties of the work (see Lord, MS). It’s not clear that Akshai is acquainted with the relevant properties. Still, I don’t see any reason why this isn’t possible. The easiest way to see this is to imagine that he wasn’t in fact slipped the pill but he is told he was by a reliable source. I think this misleading higher-order reason makes it permissible to defer even though we can stipulate that he in fact meets the acquaintance requirement (whatever that turns out to be). ¹⁸ Note that there are two crucial elements of my view. The first is the claim that deference lacks certain rational powers because it doesn’t provide appreciative knowledge. The second element is the claim that deference doesn’t provide appreciative knowledge because it doesn’t acquaint us with the normative properties. One could—and I would— accept the first part of the view even if the second part is false. Thus, even if acquaintance isn’t required for appreciative moral knowledge, it could still turn out that deference is defective because it doesn’t provide appreciative moral knowledge. ¹⁹ This is remarked upon in one way or another by Oddie (2005), Johnston (2001), and McGrath (2011b).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

82

Errol Lord

background of George, who is both the protagonist of George Orwell’s ‘A Hanging’ (Orwell, 2000) and Edmund’s informant in Section 4.1.²⁰ A Hanging: George is an Anglo police officer in 1920s Burma. As a sort of initiation, he is required to guard over the execution of a Burmese man for an unknown crime. George participates in the march from the holding area to the gallows. He watches as a rope is fastened around the man’s neck, followed by a potato sack over the man’s head. He and his companions then wait for the superintendent to give the required order. Minutes pass slowly, during which the man prays with a simple refrain of ‘Ram, Ram, Ram, Ram’. Finally the superintendent makes his decision and the man is killed.

In Orwell’s telling of the story—which is plausibly based on his own experiences as a police officer in Burma during the 1920s—George’s firsthand experience of the execution has a profound impact on him. Orwell’s George describes his experience thusly: It is curious, but till that moment I had never realized what it means to destroy a healthy, conscious man. When I saw the prisoner step aside to avoid the puddle, I saw the mystery, the unspeakable wrongness, of cutting a life short when it is in full tide. This man was not dying, he was alive just as we were alive. All the organs of his body were working—bowels digesting food, skin renewing itself, nails growing, tissues forming—all toiling away in solemn foolery. His nails would still be growing when he stood on the drop, when he was falling through the air with a tenth of a second to live. His eyes saw the yellow gravel and the grey walls, and his brain still remembered, foresaw, reasoned—reasoned even about puddles. He and we were a party of men walking together, seeing, hearing, feeling, understanding the same world; and in two minutes, with a sudden snap, one of us would be gone—one mind less, one world less. (Orwell, 2000, p. 45)

George learns that the hanging is wrong. Further, he also seems to come to know that hanging in general is wrong.²¹ Although it is not focused on extensively, it is clear that George has a host of conative and affective reactions to the scene in front of him. We can easily imagine George feeling disgust, repulsion, sadness. We can also imagine that he desires that the man’s life be spared, intends to speak out against hanging and capital punishment (as Orwell went on to do), hopes for the abolishment of hanging and capital punishment, and more besides. George’s knowledge that the hanging is wrong is similar to Hanna’s knowledge that Nefertiti’s Bust is beautiful. This is because there is a tight ²⁰ Orwell’s story is also discussed in McGrath (2011b). ²¹ It seems all George claims to come to know is the ‘unspeakable wrongness of cutting a life short in full tide’. That’s fine by me. For reasons that will be articulated, I will focus mostly on the knowledge that the particular hanging is wrong.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Acquaintance and Deference

83

connection between his knowledge and the fittingness of his conative and affective attitudes. The facts that he knows are themselves strong reasons to have those reactions—the fact that the execution is wrong is a reason to be repulsed and a reason to desire that the man’s life be spared. Finally, the way in which he acquired the set of attitudes seems important. His acquaintance with the particular hanging puts him in an especially good position to not only gain the moral knowledge he gains, but also to have the fitting conative and affective reactions he ends up having. George’s knowledge is appreciative knowledge. Just as it is in the aesthetic case, this appreciation is centered on the particular. In George’s case it is the particular wrongness of the hanging. It should be said that appreciation of this particular moral fact plays an important role in the development of more general moral sensitivities. By experiencing the particular features of the hanging, George is better equipped to be sensitive to similar features in other situations. Further, he is better equipped to appreciate the force of a certain type of consideration—e.g., the great weight of the reason not to kill provided by the capacities of agents like the Burmese prisoner.

4.3.2. The Downsides (and Upsides) of Moral Deference The preceding story about appreciative knowledge paves the way for an explanation of what is amiss with moral deference. Deference does not put us in a position to be acquainted with the morally relevant properties and thus does not put us in a position to gain appreciative knowledge.²² This is because without acquaintance, we do not come to possess the moral facts as reasons for appreciation. Given the importance of appreciative knowledge to our moral lives, this is a serious disadvantage of deference. This, I claim, is what is amiss with direct deference as a way of thinking.²³

²² Fletcher’s (2016) argument also appeals to the sort of states involved in appreciation to offer an explanation of what is amiss with deference. His view, though, is designed not to appeal to normative differences between moral and aesthetic deference and other sorts of deference. Rather, he wants to offer a ‘psychological’ explanation that appeals to the nature of the mental states involved (this mirrors an old and popular view in aesthetics, cf. Gorodeisky, 2010). Fletcher’s goals are slightly orthogonal to mine. As I see it, he is not primarily interested in the pessimism/optimism divide. While he shows that a certain form of realist optimism is compatible with what he says, the realist optimism involved is a strange view. Further, the sort of explanation that such a theorist would give of what is amiss with deference is also odd. This is not a knock on Fletcher’s paper (I think he would see this as reason to give up realism). Nevertheless, I think realists (like me) should look for other explanations. My explanation is compatible with all stripes of realism. I see that as a virtue. ²³ The last bit is important. On my view, the location of the problem is deference as a thinking type. The problem—at least not in all of the cases—is not located in the agent

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

84

Errol Lord

This, however, doesn’t mean that we ought not defer. Given space constraints, I will focus on the most popular pessimistic view in metaethics, which is a form of non-epistemic pessimism.²⁴ According to this view, what is amiss with moral deference is that one cannot gain understanding from deference.²⁵ This is morally bad because understanding is required for virtue and for acting in a morally worthy way. This moral badness provides reasons not to defer that often ground a moral requirement not to defer. Thus, on this view, one is often morally required not to defer. My framework provides a precise way of thinking about virtue and acting in a morally worthy way. Elsewhere I have defended the view that to ϕ in a morally worthy way is to manifest knowledge about how to use moral reasons as reasons to ϕ.²⁶ I have also argued that to possess a reason r to ϕ is to be in a position to manifest knowledge about how to use r as the reason it is to ϕ.²⁷ Putting these together, we get the view that when we possess a fact as a (moral) reason to ϕ, we are in a position to ϕ in a morally worthy way. The fact that hanging is generally wrong provides a reason not to hang someone. The question is this: Can one possess this reason as a reason not to hang someone without understanding why hanging is wrong? It seems very plausible to me that one can. Suppose the warden decides to try to harden Edmund even more. He does this by leaving it up to Edmund whether to execute the day’s prisoner. It is clear that the warden expects him to execute, but it is also clear that it is up to Edmund. Suppose he decides not to have the prisoner executed. Could this decision be based on the fact that hanging is wrong? It seems to me that it could. Since Edmund is a decent person, he has the background competence to refrain from doing things because they are wrong. Once he learns that hanging is wrong from George (which even Hills admits he can), he is in a position to manifest his competence to refrain from doing something because it is wrong. Given this, I find Hills’ view connecting moral worth and understanding wanting.²⁸ One avenue of response is to insist that to act in a morally worthy way is to act for the reasons that make the act right. The fact that hanging is wrong (pace Howell, 2014), the belief formed (pace McGrath, 2009; Driver, 2006), or the act of deferring (pace Hills, 2009). For more on why all of these three views are problematic, see Lewis (MS). ²⁴ Discussion of epistemic pessimism can be found in Jones (1999), Driver (2006), and McGrath (2009, 2011c). For what I take to be persuasive arguments against epistemic pessimism, see Hopkins (2007, 2011), Howell (2014), and Fletcher (2016). ²⁵ The most prominent defender of this view is Hills (2009). See also Driver (2006), McGrath (2011c), and Nickel (2001). ²⁶ See Lord (FCb) and Lord (FCa, ch. 5). ²⁷ See Lord (FCa, chs. 3–4). ²⁸ Cf. Howell (2014).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Acquaintance and Deference

85

plausibly doesn’t make refraining from hanging the prisoner the right action. Thus, this view of moral worth challenges my argument. I have two replies. First, despite the initial plausibility of the ‘making’ talk, I don’t think this view can be maintained (at least given how it is usually understood). The right view holds that morally worthy actions are performed for moral reasons that sufficiently recommend those actions. In order for this to yield Hills’ view, it needs to be that facts about what’s wrong do not themselves provide moral reasons. But this is false. Facts about what’s wrong do provide moral reasons. They intuitively count in favor of certain reactions, they can be defeated, and they can be rationally acted upon.29,³⁰ These are the hallmarks of normative reasons. The second reply is that I think there is a way in which the fact that hanging is wrong can make not hanging the prisoner the right thing to do. This is because the fact that hanging is wrong can make it the case that Edmund ought not hang the prisoner.³¹ This is because Edmund’s reason to refrain from hanging provided by the fact that hanging is wrong is decisive. This allows us to salvage a version of Hills’ view. On this view, actions performed for decisive moral reasons are morally worthy. Edmund, I think, does refrain from hanging for a decisive moral reason. Notice that this mirrors what I said about the aesthetics case. There I said that we come to possess the aesthetic facts as reasons for action when we defer even though we do not come to possess them as reasons to appreciate. This is also what happens in the moral case. We come to possess the moral facts as reasons to act in various ways. This is why we come to be in a position to act in a morally worthy way when we learn moral facts through deference. We don’t, however, come to possess those facts as reasons for appreciation. Matters are more complicated when it comes to virtue. I agree with Hills and others that learning about morality through deference is not a good way to cultivate many of the fine-grained virtues. This is because fine-grained virtues are sensitivities to particular types of reasons. When we defer we fail to come into the right kind of contact with these reasons. This is a bad feature of deference. There are three points to make about this. First, this by itself does not provide an alternative explanation of what is amiss with moral deference. My view elegantly explains why we don’t gain the fine-grained

²⁹ Further, popular double-counting arguments against the idea that facts about what’s wrong provide reasons that are problematic for reasons articulated by Schroeder (2009) and Väyrynen (2006). ³⁰ For a lengthier discussion of this argument, see Lord (FCc) and Kiesewetter (2017). ³¹ This is most plausible if the sort of ought that we are dealing with is perspectival. For a defense of the importance of this sort of ought, see Lord (2015).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

86

Errol Lord

virtues through deference: appreciative knowledge is crucial to developing the kind of fine-grained sensitivities that are involved in the more finegrained virtues. Now this is to disagree with Hills about what is required for the finegrained virtues. She holds that understanding is required. I just think appreciative knowledge is required. Appreciative knowledge needn’t involve understanding since understanding requires intellectual skills that outstrip the skills required for appreciation. While I cannot settle the debate here, let me note that the issue comes down to whether one needs to be able to think about morality in certain ways.³² Hills thinks that it does, as she thinks that understanding requires that one have the ability to explain why the moral facts obtain. I, on the other hand, hold that being able to correctly act and conatively and affectively react to the world is all that is needed. The second point to make is that the aesthetic analogue of Hills’ view is far from obvious.³³ It is not plausible that in order to be aesthetically virtuous one needs to understand why the aesthetic facts obtain. Such understanding is an intellectual good. However, when it comes to one’s aesthetic character, appreciation seems more central. This provides some reason to think that this is also the case when it comes to one’s moral character. The third point is the most important. On my view, when Edmund refrains from hanging the prisoner, he manifests knowledge about how to use the fact that hanging is wrong as a reason not to have the prisoner hanged. This know-how is a virtue. This is why, I claim, Edmund’s decision is morally worthy—it is the manifestation of virtue. It is right that given how he learned that hanging is wrong, Edmund is not in a position to cultivate a different virtue that is a sensitivity to the features of hanging that make it wrong. This, though, does not show that Edmund fails to manifest virtue when he decides not to hang the prisoner. This is important because it shows that virtue can still play an important role in these cases. This, in turn, seriously undermines any attempt to show that the fact that deference precludes the cultivation of fine-grained virtues provides decisive moral reason not to defer. It is simply not plausible to think that the cultivation of fine-grained virtues is that important. If by deferring one comes to be in a position to do the right thing in a way that manifests a virtue, it is implausible to think that one ought not defer simply because learning the particular moral fact in that way precludes one from cultivating a fine-grained virtue.

³² For arguments for my side, see Lord (FCb) and Howell (2014). ³³ Fletcher (2016) makes a similar point. A case for the analogue is made in Hills (MS).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Acquaintance and Deference

87

The upshot, I think, is that the most popular versions of both sorts of pessimism look like overgeneralizations. While there is something defective about deference, the defect does not make it the case that we generally ought not defer. For this reason, pessimism fails. If this is right, then we should conclude that deference is often permissible. That said, generalizing my view to the moral case is not so straightforward. This is because it is natural to worry that an appeal to acquaintance is innocuous in the aesthetic case but problematic in the moral case. Thus, in order to adequately defend the generalization, I need to argue that my appeal to acquaintance is on solid ground. The rest of the chapter will provide such an argument.

4 .4 . A C C E S S A N D A C Q UA I N T A N CE I take it as a datum that George’s particular experiences play an important role in his coming to acquire appreciative knowledge that the hanging was wrong. The question now is whether the moral case is like the aesthetic case in that acquaintance is required for appreciative knowledge.³⁴ In this section I will explicate and respond to two challenges to this more ambitious thought. There are two natural ways of resisting the ambitious thought. First, by claiming that we do not have perceptual access to facts of moral relevance even though we do have such access to objects of aesthetic relevance. I will call this the Lack of Perceptual Moral Access objection. Second, by claiming that we have armchair access to facts of moral relevance in a way that does not involve acquaintance. We lack this kind of access to facts of aesthetic relevance. I will call this the Lack of Armchair Access objection. Both of these objections have serious merit for at least two reasons. First, the ideas at their heart seem to point out important differences between the way we think about aesthetic matters and the way we think about moral matters. Second, they both are anchored in views that are at the heart of orthodox theories in both aesthetics and metaethics. That said, the remainder of the chapter will argue that such views can plausibly be rejected. There is a plausible package of views that fit very nicely with the claim that acquaintance is required for appreciative moral knowledge.

³⁴ As I said in note 18, I could maintain that what is amiss with moral deference is that it doesn’t provide appreciative knowledge even if one could gain appreciative moral knowledge without acquaintance. So the core of my explanation of what’s amiss with deference remains even if this section fails.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

88

Errol Lord

4.4.1. Ground Clearing To start, let me clarify what one is required to be acquainted with. I have consistently said that what we are required to be acquainted with are the aesthetically and morally relevant properties. This might mean that we are required to be acquainted with the aesthetic or moral properties themselves. But it might just mean that we are required to be acquainted with the non-aesthetic and non-moral properties that the aesthetic and moral properties depend upon. The resolution of this ambiguity makes a major difference to how we tackle the Lack of Perceptual Access objection. I will focus on the more ambitious claim that acquaintance is required with the moral and aesthetic properties themselves. Although I lack the space to argue for this here, I think this is the most promising version of the idea. A second bit of ground clearing is about how I am understanding acquaintance. I understand acquaintance in a broad way to include any kind of direct access to an object, property, or fact. This follows the precedent set by many using the notion—see, e.g., Johnston (2001, 2004), McGrath (2011b), and Chudnoff (2013a). In particular, I maintain that if you see that p, then you are acquainted with p.³⁵ As we will see, I think that some affective responses also give us direct access to moral and aesthetic properties.³⁶

4.4.2. Accessing Moral Properties Directly The Lack of Perceptual Moral Access objection maintains that the difference between the aesthetics case and the moral case is that in the aesthetics case we have perceptual acquaintance with the aesthetic properties themselves, whereas in the moral case we do not. It is a common trope in metaethics that we lack perceptual access to the moral facts.³⁷ Such skepticism has not

³⁵ On some views of seeing that p, it is a stretch to think that seeing that p amounts to acquaintance with p. This is because many views of perception hold that perception only yields a kind of indirect access to p. This is a source of many objections that so-called direct realists level at these accounts (cf. Johnston, 2004; Brewer, 2011). Since I am attracted to this picture, I will assume that seeing is roughly how the direct realist conceives of it. ³⁶ There is a tradition in epistemology that does not understand acquaintance in this way. According to this tradition, acquaintance is a kind of direct access, but it is nonconceptual and basic (see Hasan and Fumerton, 2014, and the citations therein). This tradition is strongly internalist and insists that the only things we can be acquainted with are our own internal states like pain. This notion will not help much in the epistemologies of aesthetics and morality. Fortunately, there is the broader notion connected to seeing and related states. ³⁷ See, e.g., Harman’s (1977) classic discussion of the burning cat.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Acquaintance and Deference

89

played a prominent role in aesthetics.³⁸ It is common for aestheticians to assume that we perceive aesthetic properties—we see the painting’s beauty, hear the symphony’s grace, taste the dish’s delectability. There are two lines of defense to this objection. While these are sometimes seen as competitors (e.g., in Milona, FCb), I see them as complementary. The first line of defense insists that we can access the moral facts via the traditional five senses. This view has been defended with increasing sophistication in the last decade.³⁹ The argumentative strategies designed to show that we see some non-normative higher-level properties—e.g., kind membership—can be applied to show that we see normative properties as well.⁴⁰ These argumentative strategies rely on the hypothesis that there is cognitive penetration of perception. This means, roughly, that one’s background attitudes can affect which properties one perceives. When things go well (they don’t always), one’s background attitudes can put one in a position to perceive the actual moral properties. While I am sympathetic to this view—which I will call normative perceptualism—it has two important problems. The first problem casts doubt on the foundationalist bona fides of the view.⁴¹ The problem is that if the content of one’s perceptions are influenced by one’s background attitudes, the rational power of those perceptions will be held hostage to the rational status of those background attitudes. So, if George sees that the hanging is wrong because of his more general compassion, the rational power of that particular perception is partially generated by his compassion. Similarly, if one perceives black faces as dangerous because one’s irrational background belief (or alief )⁴² that African Americans are more dangerous than the average citizen, the perception lacks full rational power. This is a problem if one wants normative perceptual beliefs to be one’s foundations in a foundationalist normative epistemology. Foundational beliefs are foundational in virtue of the fact that their rational status does not depend on the rational status of any of the subject’s other attitudes. It’s plausible that this is not so for cognitively penetrated perceptions of normative facts. The second problem is particularly severe in our context. It looks like we need more than the five senses to account for the full range of cases involving acquaintance.⁴³ To see this, notice that it’s plausible that we can gain

³⁸ See Stokes (2014) for relevant discussion. Stokes defends the view that we see higher-level aesthetic properties. ³⁹ See, e.g., Audi (2013), Werner (2016). ⁴⁰ See, e.g., Werner (2016) on the moral case and Stokes (2014) on the aesthetic case. ⁴¹ Cf. Silins (2016), Väyrynen (FC), Milona (FCb), Siegel (2012). ⁴² See Gendler (2010). ⁴³ This sort of problem is raised in Milona (FCb).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

90

Errol Lord

appreciative aesthetic knowledge via the imagination. For example, musicians can come to gain appreciative knowledge of the aesthetic features of a work by imagining what the work sounds like. The imaginative processes do not involve actual auditory perception. So it doesn’t look like normative perceptions can do all of the work. These problems motivate a search for a different kind of direct access. Fortunately, certain affective or conative states seem to fit the bill. According to one popular view, we have certain attraction and repulsion states that represent particulars as having certain normative features.⁴⁴ These states are perceptions of the normative. So, for example, George has a basic repulsion to the concrete scene before him. This state presents the hanging as wrong. It provides George with basic knowledge that the hanging is wrong. This sort of sentimentalist perceptualism holds promise when it comes to the objections to normative perceptualism. There is reason to think that they fail to fall prey to the rational encroachment objection. This is because, roughly, the function of the relevant affective states is to track the normative properties in the same way that the function of visual perception is to track low-level properties.⁴⁵ Further, we can have these reactions in response to imagined works or scenarios. When we do, we have a kind of direct access to the properties of those imagined scenarios.⁴⁶ These experiences can provide us with basic knowledge of the moral features of those scenarios. While sentimentalist perceptualism can do helpful work for me, this is not sufficient to show that it is true. I lack the space to mount a full defense, but here’s a sketch of a prominent and, I think, compelling argument for the view.⁴⁷ The anchoring claim of the argument is that certain affective/ conative states put us in a position to have fitting reactions. For example, George is in a position to fittingly condemn the warden because of his repulsion.⁴⁸ The key question is what best explains the rational power of ⁴⁴ See Johnston (2001) and Schafer (2013) for defense of this particular kind of sentimentalist perceptualism. See Milona (FCa, FCb), Oddie (2005), and Döring (2007) for defenses of other sorts of sentimentalist perceptualism. ⁴⁵ On the tracking point, see Oddie (2005). On the function point, see Milona (MS). ⁴⁶ Or, more conservatively, we have direct access to the properties the scenarios would have if they were actual. Cf. Milona (MS). ⁴⁷ See Döring (2007), Milona (MS), Johnston (2001). ⁴⁸ To be clear, the idea is that the facts about the warden are what provide the reasons to condemn. George’s repulsion makes those features manifest. Analogously, facts about the external world provide reasons to believe that our perceptual capacities make manifest. Of course, the sentimentalist perceptualist maintains that we can have non-veridical sentimental perceptions (to use a referee’s example, the bigot might feel repulsion towards homosexual sex). Those experiences will not make any part of the actual world manifest. They might not even provide any justification (although I think the sentimental

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Acquaintance and Deference

91

the affective state. The best explanation, according to this argument, is that the affective state gives George direct access to the property of wrongness. This is usually shown via a process of elimination. It cannot be that George’s affect is itself a belief about wrongness (cf. Schafer, 2013; Milona, MS). Nor can it be that George is merely projecting wrongness on to the hanging (cf. Johnston, 2001). Rather, the affect gains the power to justify downstream reactions by perceptually representing the normative property.⁴⁹ The upshot of this section is that there are interesting resources for arguing that we are acquainted with the moral facts. First, it’s plausible that we have direct access to the moral facts through our traditional sense modalities. While this is helpful for me, it doesn’t do everything one might want. Fortunately, it is also plausible that we have a sort of direct access to the moral facts through our affective states. This sort of access looks like it can deliver the epistemic goods that I am after.

4.4.3. The Lack of Armchair Access Objection The Lack of Perceptual Moral Access objection contends that we have a kind of access—perceptual access—to the aesthetic facts that we lack when it comes to the moral facts. The Lack of Armchair Access objection contends that we have a kind of access to the moral facts—armchair access—that we lack when it comes to the aesthetic facts. This objection is also a natural one to make if you hold textbook accounts of moral and aesthetic epistemology. Aestheticians have been very suspicious of any kind of armchair aesthetic knowledge.⁵⁰ On the other hand, many metaethicists see armchair moral knowledge as paradigmatic.⁵¹ This on its own doesn’t challenge my picture. It does once we add the claim that armchair moral knowledge doesn’t involve acquaintance. This looks plausible for at least some purported armchair knowledge.⁵² Further, one way to put pressure on the sort of perceptualist views sketched

perceptualist will have to maintain that it is possible for them to provide justification). Thanks to a referee for pushing me to say more here. ⁴⁹ For other arguments for the view, see Schafer (2013), Oddie (2005), and Milona (FCa, MS). ⁵⁰ See Sibley (2001) for a classic discussion. See Dorsch (2014) for a nice overview. ⁵¹ See, e.g., Smith (2000), Shafer-Landau (2003), Sayre-McCord (1996). ⁵² This on its own doesn’t count against the sort of perceptualism defended in Section 4.4.2. Pace McGrath (2011a), there is no tension between experience playing a role in the acquisition of token moral knowledge and the armchair knowability of moral truths. Nearly all armchair-knowable facts are also knowable a posteriori. What is truistic is that the moral truths are armchair knowable, not that whenever we know them we know them through armchair methods.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

92

Errol Lord

in Section 4.4.2 is to argue that armchair thinking is both central to the acquisition of moral knowledge and radically different from the sort of thinking the perceptualist appeals to. One way to frame the challenge is by thinking about the modal status of the truths learned.⁵³ Some moral facts are necessarily true—e.g., the (purported) fact that murder is wrong. Some seem to only be contingently true—e.g., the fact that hanging the Burmese prisoner is wrong.⁵⁴ Aestheticians are not very interested in necessary aesthetic truths. Indeed, many are extremely skeptical that there are any such truths. On the other hand, nearly everyone is interested in at least some contingent aesthetic truths. Things are different in ethics. Investigating necessary moral truths has always been part of ethical theorizing. Indeed, it has monopolized most theoretical discussions. The pursuit of contingent ethical truths is not popular among moral philosophers (even applied ethicists). Our theoretical priorities, then, suggest that in ethics the necessary truths take precedence. How we learn about them is a matter of great controversy. It is not popular, however, to think that we learn about them via acquaintance.⁵⁵ If this is right, then it’s easy to see how armchair knowledge can threaten my project. The problem is that our theoretical priorities give a misleading picture of how most moral learning works. It is implausible to think that knowledge of necessary moral truths plays a prominent role in the average person’s acquisition of new moral knowledge. I doubt this is true even of moral philosophers. I myself do not flat-out believe many necessary moral claims. A large part of the theoretical interest in such claims is that they are hard to learn! Thus, I don’t think we should allow moral philosophers’ theoretical priorities to shape how we think of moral epistemology.⁵⁶ Still, a challenge remains. This is because it’s plausible that we acquire much of our moral knowledge of contingent moral truths through armchair ways of thinking. We often acquire new moral knowledge by thinking about what to do in the future or what to do were the world to turn out a certain way. Much of the time we gain appreciative knowledge through these armchair ways of thinking. Thus, it is still on me to explain how it is that acquaintance plays a prominent role in these ways of thinking. ⁵³ This way of thinking about things is greatly indebted to Milona (MS). ⁵⁴ One needn’t be a whole-hog consequentialist to think that it wouldn’t be wrong to hang the prisoner if the lives of 100,000 children were on the line. ⁵⁵ One important exception to this is Elijah Chudnoff, who does think that veridical intuitions are generated by awareness of abstracta. See Chudnoff (2013b). I lack space to fully discuss Chudnoff ’s view, but I am sympathetic to it as a further generalization of my picture. ⁵⁶ For a similar point, see Milona (MS).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Acquaintance and Deference

93

Fortunately, the outline of the story has already been given. The most common way we acquire armchair knowledge of contingent truths is by thinking about cases. When we do this, we imagine the cases in at least some detail.⁵⁷ The circumstances that we imagine have certain morally relevant properties. We become acquainted with these properties via these imaginative experiences. Indeed, I think it’s plausible that we often have affective perceptual experiences of these properties. Thus, I think that ultimately the way we acquire knowledge of contingent truths via thinking about cases is very similar to the way we acquire knowledge of contingent truths by encountering concrete situations in the actual world.⁵⁸ It’s important to stress that a similar story is needed in aesthetics. As we saw earlier, there are cases where one gains appreciative aesthetic knowledge via the imagination. An easy example is someone—a composer, say— imagining what a certain string of notes sounds like. On my view, these cases are analogous to cases where we imaginatively think about cases. In both kinds of case we gain a sort of direct access to the normative features the objects of imagination would have if they were to obtain. The upshot is that once again there is a plausible story to tell about how acquaintance enables appreciative armchair knowledge of contingent truths. Further, this story is structurally identical to the most plausible story about how the imagination yields appreciative aesthetic knowledge. While more work needs to be done, it’s far from clear that armchair knowledge is going to drive a wedge between the moral case and the aesthetic case.

4.5. CONCLUSION The main goal of this chapter was to make a preliminary case for a general picture of two methods of acquiring moral and aesthetic knowledge. At the heart of this picture is the thought that acquaintance plays an important role in the acquisition of an important kind of normative knowledge. This is what I call appreciative knowledge. This knowledge is the sort of knowledge that enables appreciation of the aesthetic and moral facts. Such appreciation is plausibly central to being a good moral and aesthetic agent. ⁵⁷ This cuts against some of the literature. For example, McGrath (2011b), taking inspiration from Kagan (2001), claims we evaluate cases just by thinking about the descriptions of the cases we are given. This is of course possible, but I conjecture that most of us instead have imaginative representations of the cases that are described to us. ⁵⁸ There are, of course, details to be worked out. One important detail is how we gain access to the relevant properties of the imagined circumstances via the imagination. For a nice story about this, see Milona (MS). For a different sort of story that perhaps fits better with my direct realist leanings, see Johnston (2001, 2004).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

94

Errol Lord

This view of appreciative knowledge paves the way for a view about what is defective about deference. Deference is defective insofar as it cannot put one in a position to be acquainted with normative properties. Deference thus cannot put one in a position to gain appreciative knowledge. This is a shortcoming of deference. Nevertheless, I argued that this does not mean that we shouldn’t defer—the defect of deference does not ground an obligation not to defer. This left us with a moderately optimistic view about the rational power of deference. To end, let me return to the nature of the puzzle. As the literature has grown so has the number of interpretations of what the puzzle is and what kind of view can explain it. I have focused on one way of thinking that there is a deep asymmetry between normative deference and non-normative deference. This is the view that there is a general obligation not to defer about aesthetic and moral matters. I don’t think there are strong reasons to think that this is so. It’s important to be clear eyed about what this conclusion shows. Many in the literature are not focused on pessimism and optimism as I understand them. Some have much mushier conceptions of the asymmetry. So, for example, one might think that the mere fact that there is something generally defective with normative deference is enough to show that there is an important asymmetry between normative deference and nonnormative deference. Since I do think there is something generally defective with normative deference, my view actually vindicates this asymmetry. I am happy with this result. A truth lingering in many pessimistic views is that our normative views play a central role in shaping our conceptions of the world. This is in contrast to our views about the tax code, the rise in temperatures, and molecular bonding (at least for those of us who aren’t accountants, climate scientists, and chemists). This asymmetry should be accounted for. My view accounts for this asymmetry while maintaining a role for deference. This, like baby bear’s porridge, seems just right.⁵⁹

References Audi, R. (2013). Moral Perception. Princeton University Press. Brewer, B. (2011). Perception and Its Objects. Oxford University Press. Budd, M. (2003). The acquaintance principle. British Journal of Aesthetics, 43(4), 386–92. ⁵⁹ Thanks to audiences at Princeton, Northwestern, Notre Dame, Washington University in St. Louis, and ChillMeta for helpful discussion, especially Tom Kelly, Adam Lerner, Dan Singer, Daniel Wodak, Sukaina Hirji, Earl Conee, Baron Reed, Kyla Ebels-Duggan, Stephen White, Michael Smith, Carl Craver, Jon Kvanvig, Roy Sorensen, Fred Rush, Robert Audi, Daniel Nolan, Fritz Warfield, and Nathan Howard. Thanks also to Andrew Huddleston, Keren Gorodeisky, Alison Hills, Max Lewis, and two anonymous referees.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Acquaintance and Deference

95

Chudnoff, E. (2013a). Awareness of abstract objects. Noûs, 47(4), 706–26. Chudnoff, E. (2013b). Intuition. Oxford University Press. Döring, S. A. (2007). Seeing what to do: affective perception and rational motivation. Dialectica, 61(3), 363–94. Dorsch, F. (2014). The limits of aesthetic empiricism. In G. Currie, M. Kieran, A. Meskin, and J. Robson (Eds.), Aesthetics and the Sciences of Mind (pp. 75–100). Oxford University Press. Driver, J. (2006). Autonomy and the asymmetry problem for moral expertise. Philosophical Studies, 128(3), 619–44. Enoch, D. (2014). A defense of moral deference. Journal of Philosophy, 111, 1–30. Fletcher, G. (2016). Moral testimony: once more with feeling. In R. Shafer-Landau (Ed.), Oxford Studies in Metaethics, volume 11. Oxford University Press. Gendler, T. S. (2010). Intuition, Imagination, and Philosophical Methodology. Oxford University Press. Gorodeisky, K. (2010). A new look at Kant’s view of aesthetic testimony. British Journal of Aesthetics, 50(1), 53–70. Harman, G. (1977). The Nature of Morality: An Introduction to Ethics. Oxford University Press. Hasan, A. and Fumerton, R. (2014). Knowledge by acquaintance vs. description. In Zalta (Ed.), Stanford Encyclopedia of Philosophy . Hills, A. (2009). Moral testimony and moral epistemology. Ethics, 120(1), 94–127. Hills, A. (2013). Moral testimony. Philosophy Compass, 8(6), 552–9. Hills, A. (MS). Testimony, understanding, and virtue in ethics and aesthetics. Manuscript, University of Oxford. Hopkins, R. (2007). What is wrong with moral testimony? Philosophy and Phenomenological Research, 74(3), 611–34. Hopkins, R. (2011). How to be a pessimist about aesthetic testimony. Journal of Philosophy, 108(3), 138–57. Howell, R. J. (2014). Google morals, virtue, and the asymmetry of deference. Noûs, 48(3), 389–415. Johnston, M. (2001). The authority of affect. Philosophy and Phenomenological Research, 63, 181–214. Johnston, M. (2004). The obscure object of hallucination. Philosophical Studies, 120 (1–3), 113–83. Jones, K. (1999). Second-hand moral knowledge. Journal of Philosophy, 96(2), 55. Kagan, S. (2001). Thinking about cases. Social Philosophy & Policy, 18(2), 44–63. Kiesewetter, B. (2017). The Normativity of Rationality. Oxford University Press. Lewis, M. (MS). The goldilocks dilemma for pessimists about moral deference. Manuscript, University of Pennsylvania. Livingston, P. (2003). On an apparent truism in aesthetics. British Journal of Aesthetics, 43(3), 260–78. Lord, E. (2010). Having reasons and the factoring account. Philosophical Studies, 149(3), 283–96. Lord, E. (2014). The coherent and the rational. Analytic Philosophy, 55(2), 151–75. Lord, E. (2015). Acting for the right reasons, abilities, and obligation. In R. Shafer-Landau (Ed.), Oxford Studies in Metaethics, volume 10 (pp. 26–52). Oxford University Press.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

96

Errol Lord

Lord, E. (2016). On the rational power of aesthetic testimony. British Journal of Aesthetics, 56(1), 1–13. Lord, E. (FCa). The Importance of Being Rational. Oxford University Press. Lord, E. (FCb). On the intellectual conditions for responsibility: acting for the right reasons, conceptualization, and credit. Philosophy and Phenomenological Research. Lord, E. (FCc). What you’re rationally required to do and what you ought to do (are the same thing!). Mind. Lord, E. (MS). Knowing what it’s like: how to save the acquaintance principle from photographs, the imagination, and abstracta. Manuscript, University of Pennsylvania. McGrath, S. (2009). The puzzle of pure moral deference. Philosophical Perspectives, 23, 321–44. McGrath, S. (2011a). Moral knowledge and experience. Oxford Studies in Metaethics, 6, 107–27. McGrath, S. (2011b). Normative ethics, conversion, and pictures as tools of moral persuasion. Oxford Studies in Normative Ethics, 1, 268–94. McGrath, S. (2011c). Skepticism about moral expertise as a puzzle for moral realism. Journal of Philosophy, 108(3), 111–37. Meskin, A. (2004). Aesthetic testimony: what can we learn from others about beauty and art? Philosophy and Phenomenological Research, 69(1), 65–91. Meskin, A. and Robson, J. (2015). Taste and acquaintance. Journal of Aesthetics and Art Criticism, 73(2), 127–39. Milona, M. (FCa). Intellect vs. affect: finding leverage in an old debate. Philosophical Studies. Milona, M. (FCb). On the epistemological significance of value perception. In A. Bergqvist and R. Cowan (Eds.), Evaluative Perception: Aesthetic, Ethical, and Normative. Oxford University Press. Milona, M. (MS). Armchair moral knowledge and the perceptual analogy. Manuscript, Cornell University. Nickel, P. (2001). Moral testimony and its authority. Ethical Theory and Moral Practice, 4(3), 253–66. Oddie, G. (2005). Value, Reality, and Desire. Oxford University Press. Orwell, G. (2000). A hanging. In S. Orwell and I. Angus (Eds.), George Orwell: An Age Like This. Nonpareil Books. Robson, J. (2012). Aesthetic testimony. Philosophy Compass, 7(1), 1–10. Robson, J. (2013). Appreciating the acquaintance principle: a reply to Konigsberg. British Journal of Aesthetics, 53, 237–45. Sayre-McCord, G. (1996). Coherentist epistemology and moral theory. In W. Sinnott-Armstrong and M. Timmons (Eds.), Moral Knowledge? New Readings in Moral Epistemology. Oxford University Press. Schafer, K. (2013). Perception and the rational force of desire. Journal of Philosophy, 110(5), 258–81. Schroeder, M. (2008). Having reasons. Philosophical Studies, 139(1), 57–71. Schroeder, M. (2009). Buck passers’ negative thesis. Philosophical Explorations, 12(3), 341–7.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Acquaintance and Deference

97

Schroeder, M. (2011). What does it take to ‘have’ a reason? In A. Reisner and A. Steglich-Petersen (Eds.), Reasons for Belief. Cambridge University Press. Shafer-Landau, R. (2003). Moral Realism: A Defence. Oxford University Press. Sibley, F. (2001). Approach to Aesthetics: Collected Papers on Philosophical Aesthetics. Oxford University Press. Siegel, S. (2012). Cognitive penetrability and perceptual justification. Noûs, 46(2), 201–22. Silins, N. (2016). Cognitive penetration and the epistemology of perception. Blackwell Compass, 11(1), 24–42. Sliwa, P. (2012). In defense of moral testimony. Philosophical Studies, 158(2), 175–95. Smith, M. (2000). Moral realism. In H. Lafollette (Ed.), The Blackwell Guide to Ethical Theory (pp. 15–37). Blackwell Publishers. Stokes, D. (2014). Cognitive penetration and the perception of art (winner of 2012 Dialectica essay prize). Dialectica, 68(1), 1–34. Sylvan, K. (2015). What apparent reasons appear to be. Philosophical Studies, 172(3), 587–606. Väyrynen, P. (2006). Moral generalism: enjoy in moderation. Ethics, 116, 707–41. Väyrynen, P. (FC). Doubts about moral perception. In A. Bergqvist and R. Cowan (Eds.), Evaluative Perception. Oxford University Press. Werner, P. J. (2016). Moral perception and the contents of experience. Journal of Moral Philosophy, 13(3), 294–317. Whiting, D. (2014). Keep things in perspective: reasons, rationality, and the a priori. Journal of Ethics and Social Philosophy, 8, 1–22. Whiting, D. (2015). The glass is half empty: a new argument for pessimism about aesthetic testimony. British Journal of Aesthetics, 55, 91–107.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

5 Belief Pills and the Possibility of Moral Epistemology Neil Sinclair

5.1. INTRODUCTION How should we respond to empirical discoveries about the evolutionary origins of our moral sensibilities? Broadly speaking, there are two types of answer. Optimists hold that substantive moral claims, or theories of the nature of moral practice, or our continued engagement in moral practice, are vindicated by an evolutionary account of morality (for the first, see Casebeer 2003; Mikhail 2011; for the second see Harms 2000; for the third see Campbell 1996). Pessimists take moral practice—or some of its features— to be undermined or debunked by evolutionary understanding. Debunking arguments begin from a premise concerning the evolution of some aspect of moral practice, and can be distinguished by their target conclusion. Some urge that substantive moral theories are false or unjustified (Greene 2008; Singer 2005). Others target metaethical theories. Street, for example, targets moral realism, according to which ‘there are at least some evaluative facts . . . that hold independently of all our evaluative attitudes’ (2006: 110). On Street’s view, realists are ‘unable to accommodate the fact that Darwinian forces have deeply influenced the content of human values’ (2006: 109). Joyce (2014) calls this ‘theory debunking’. Two further types of debunking argument target the moral judgements of ordinary folk. The first claims that we can move from evolutionary premises to a conclusion regarding the falsity of such judgements. Joyce (2014) calls this ‘truth debunking’ (see Ruse 1986: 254). The second aims for a conclusion regarding the epistemic status of moral judgements: that they are unjustified, not knowledge, or otherwise epistemically deficient. Joyce calls this ‘justification debunking’. Such debunking itself comes in many flavours. For example, some arguments allow that some moral judgements

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Belief Pills and the Possibility of Moral Epistemology

99

might be initially justified and take evolutionary discoveries to undermine this justification. Others allow only that some moral judgements appear justified and urge that evolutionary considerations show these appearances to be deceptive (White 2010: 575). An orthogonal distinction concerns whether the supposed debunking proceeds from the claim that (some) moral judgements rest on ignorance or false belief about the nature of morality or simply from the claim that moral judgements fail to meet some necessary condition for justification (Joyce 2006: 180; Lillehammer 2003a: 571–3). In addition, truth and justification debunking arguments can be local or global, depending on whether they target some or all moral judgements. Joyce is perhaps the foremost proponent of a global justification debunking argument for moral judgements. He writes: We have an empirically confirmed theory about where our moral judgments come from . . . This theory doesn’t state or imply that they are true, it doesn’t have as a background assumption that they are true, and, importantly, their truth is not surreptitiously buried in the theory by virtue of any form of moral naturalism. This amounts to the discovery that our moral beliefs are the products of a process that is entirely independent of their truth, which forces the recognition that we have no grounds one way or another for maintaining these beliefs. (2006: 211)

According to Joyce, this parallels the following argument: Suppose that there were a pill that makes you believe that Napoleon won Waterloo, and another that makes you believe that he lost . . . [I]magine that you are proceeding through life happily believing that Napoleon lost Waterloo . . . and then you discover that at some point in your past someone slipped you a ‘Napoleon lost Waterloo’ belief pill . . . Should this undermine your faith in your belief that Napoleon lost Waterloo? Of course it should. It doesn’t show that your belief is false . . . but this knowledge is certainly sufficient to place this belief on the dubious list. (2006: 179)

In this chapter I evaluate global justification debunking arguments such as Joyce’s. I understand such arguments as aiming to establish their conclusion independently of truth-debunking arguments. So understood, I argue they are dialectically ineffective. One point is worth mentioning at the outset. Many take justification debunking arguments in morality to deploy substantive metaethical assumptions of realism or cognitivism. According to the former, the epistemic credentials of moral judgements are threatened on the assumptions that (a) those judgements are cognitive and (b) that the realm of moral facts which such judgements seek to cognize exists independently of the minds and languages of human beings (Clarke-Doane 2012: 315–17; Kahane 2011: 112;

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

100

Neil Sinclair

Street 2006: 110). According to the latter the epistemic credentials of moral judgements are threatened simply on the assumption of cognitivism (Handfield 2016; Joyce 2016: 126–9; Mason 2010: 775). These assumptions have done much to shape discussion of such arguments. For example, those holding the former view tend to present these arguments as parts of arguments against moral realism—the thought being that if a global moral scepticism results from an assumption of moral realism, we should reject that assumption (Street 2006: 109). But are these assumptions necessary for debunking arguments? Arguably not. Such arguments target the idea that moral judgements are epistemically justified. The minimal assumption required, therefore, is simply that moral judgements are the sorts of thing that can be, or might have been, epistemically justified. But it is well known that sophisticated versions of non-cognitivism hold that moral judgements can be so justified. Quasirealists hold that there are available notions of truth and justification that can warrantedly apply to moral judgements, even though such judgements are, at root, non-representational (Blackburn 1993). Since such views assume that there are conditions on epistemic justification which some moral judgements meet, they are potentially vulnerable to justification debunking. Hence the following discussion assumes only that moral judgements are the sorts of things that can, or might have been, epistemically justified (Lillehammer 2010: 361).

5 .2 . E V O L U T I O N A R Y EX PL A N A T I O N S OF MORAL JUDGEMENTS Global justification debunking arguments such as Joyce’s start from the premise that moral practice can be given an evolutionary explanation, that is, an explanation in terms of that practice’s ability to increase the relative reproductive fitness of some ancestral population. But what, precisely, do such explanations explain? One possible explanandum is our possession of moral concepts—generating the ‘capacity hypothesis’. Another is our tendency to deploy moral concepts in particular ways—generating the ‘tendency hypothesis’. Joyce defends the former (2006: 108–42), Street the latter (2006: 114–15; see FitzPatrick 2015 for the distinction). But note that neither is a complete explanation of all moral judgements. The capacity hypothesis by itself does not explain any moral judgements (only our capacity to make them); the tendency hypothesis only explains (the tendency to make) some moral judgements. This incompleteness is one source of vulnerability in debunking arguments.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Belief Pills and the Possibility of Moral Epistemology

101

It is worth contrasting evolutionary explanations of moral practice with evolutionary explanations of other practices (Fraser 2014: 463; Joyce 2006: 182–4). Consider, for example, dry-goods practice: the practice of thinking and talking about medium-sized dry-goods, such as boulders. The capacity to make dry-goods judgements can be given an evolutionary explanation. But there is crucial asymmetry with the moral case. Whereas the capacity to make dry-goods judgements plausibly evolved in part as a mechanism to track mind-independent facts about the location, number, and movements of independently existing dry-goods, the capacity to make moral judgements evolved (according to Joyce) as a mechanism to promote prosocial behaviour and help sustain cooperative social frameworks. Thus whereas the capacity to make dry-goods judgements was selected partly on the basis of its possession enabling the possessor to track mindindependent facts, the capacity to make moral judgements was (according to Joyce) selected purely on the basis of its possession enabling (or making more likely) certain behaviours. Thus it seems that whereas the dry-goods faculty was selected partly on the basis of its upstream tracking abilities, the moral faculty was selected purely on the basis of its downstream motivational effects. (The same point applies, mutatis mutandis, for the tendency hypothesis.) This highlights the fact that what matters in assessing justification debunking arguments is not whether our moral faculties and judgements can be given some kind of evolutionary explanation but whether the type of explanation offered undermines justification (Joyce 2006: 212). 5 . 3 . A G E N E R I C D E B U N K I N G AR G U M E N T Gathering the foregoing threads together, consider the following argument: (1)

All (actual, human) moral judgements can be given plausible evolutionary explanations.

(2)

The process referred to in these explanations does not track moral truth.

Hence (C)

All (actual, human) moral judgements are unjustified.

This generic argument is not formally valid. Turning it into a valid argument is the task for Sections 5.4–5.6. Consider (2)—the ‘epistemic premise’ (Kahane 2011: 106). What does ‘does not track moral truth’ mean? Synonyms include: ‘an off-track process’ (Kahane 2011: 106), ‘tracking failure’ (Lillehammer 2010: 365), ‘truth-mooting

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Neil Sinclair

102

genealogy’ (Mason 2010: 773). Different answers to this question generate different debunking arguments. But there seems to be considerable agreement on the following, minimal, account: the process referred to in these explanations does not track moral truth insofar as these explanations do not invoke or assume the truth of the judgements concerned (Fraser 2014: 459; Joyce 2006: 183–4; Street 2006: 155). (1) is the ‘causal premise’. Neither the capacity nor tendency hypothesis provides a complete explanation of all moral judgements. Hence (1) does not follow from either. Given the capacity hypothesis particular moral judgements will be partly the result of the evolutionary history of the moral capacity, partly the result of more proximal environmental stimuli. There is thus the potential for debunking arguments based on this hypothesis to fall at the first hurdle (Toner 2011). I will not explore the details of this issue here. Instead I note that debunkers usually overcome this problem by assuming that whatever non-evolutionary factors are involved in the explanation of specific moral judgements, these are also truth-mooting with respect to moral truth (Kahane 2011: 106; Mason 2010: 774). Joyce, for example, responds to an analogous worry about his belief-pill argument by modifying the scenario to that of a ‘concept pill’—one which ‘makes you form beliefs about Napoleon in general’. He suggests a case in which, knowing one has been slipped such a pill, one’s Napoleon beliefs are determined ‘randomly’ by ‘certain environmental triggers’, presumably not involving Napoleon himself. In such a case, Joyce suggests, the resulting beliefs are still undermined (2006: 181). To incorporate this into the argument, call an ‘evolutionary+’ explanation one that involves the capacity or tendency hypothesis together with additional morally truth-mooting factors. Together with the minimal account of ‘does not track moral truth’, this generates the following: (1*) (2*)

All (actual, human) moral judgements can be given plausible evolutionary+ explanations. These explanations do not invoke or assume the truth of such judgements.

Hence (3)

All (actual, human) moral judgements can be plausibly explained without invoking or assuming their truth.

(C)

All (actual, human) moral judgements are unjustified.

Hence (3) may also be supported by non-evolutionary premises, concerning, for example, sociological explanations of moral judgements. For now though, my concern is validating the move from (3) to (C).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Belief Pills and the Possibility of Moral Epistemology

103

5 .4 . R E D U N D AN C Y What additional premises might bridge this gap? Consider: (3)

All (actual, human) moral judgements can be plausibly explained without invoking or assuming their truth. (4a) If a judgement can be plausibly explained without invoking or assuming its truth then it is not justified.

Hence (C)

All (actual, human) moral judgements are unjustified.¹

This is valid and plausible. After all, what reason could there be to believe in a set of truths if reference to those truths was not required to explain our judgements about them? (Lillehammer 2003a: 577; Nichols 2014: 729.) But despite this, Redundancy is unsuccessful. There are at least two problems with (4a). First, the possibility of reduction. Suppose that property A (such as being water) is reducible to property B (such as being H₂O). This entails, among other things, that A is identical to B. Suppose a particular judgement concerning A (e.g. Jo’s belief that the water in the pipes has frozen) can be explained by citing an instance of B (e.g. ‘Jo believes that the water in the pipes has frozen because the H₂O in the pipes is frozen’). Then it seems that the judgement can be explained without invoking or assuming its truth, but can still be justified. The general point is that an explanation of a judgement in terms of a level of properties not invoked or assumed by the judgement is consistent with (but does not require) those properties being the reductive basis of the properties ascribed by the judgement, and hence consistent with the judgement being responsive to the properties it ascribes (Lillehammer 2003a: 577; Quinn 1986: 537–9). Of course, this problem with (4a) may not be a problem if reduction is unavailable. Joyce (2006: 190–209) argues that this is so in the moral case. But on reflection the same problem arises with weaker metaphysical relations. Suppose that Jo’s judgement that x is C is explained by x’s being D. This explanation does not invoke or assume the truth that x is C. But it is perfectly consistent with this explanation that C supervenes on D and hence that Jo’s judgement is counterfactually dependent on x’s possession of C. Such sensitivity is in turn consistent with (and suggestive of) the judgement

¹ Throughout I understand debunking arguments to utilize externalist conditions on justification, such as the antecedent of (4a). But my criticisms of debunking arguments apply equally to internalist versions.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

104

Neil Sinclair

being justified. Again the point is general: an explanation of a judgement in terms of one level of properties is consistent with (but does not require) those properties being metaphysically related to the property ascribed in the judgement in such a way as to render the judgement justified. Another reply to the reduction problem urges that the relevant judgements are explained neither by the properties referred to by the judgement itself, nor by any plausible reductive or supervenience base for those properties. But though this response delivers a more plausible version of (4a) it renders (3) implausible. To secure validity, (3) would need to be modified to the claim that all moral judgements can be plausibly explained without invoking or assuming their truth, or any reductive or subvening base for that truth. But such a claim is unsupported by our evolutionary hypotheses. Because such hypotheses are relevantly incomplete, it is consistent with them to suppose that some moral judgements are explained in part by invoking a plausible reductive or subvening base for the properties they refer to. The second problem with (4a) is the existence of counterexamples. I judge that the sun will rise tomorrow and that all men are mortal, and both judgements are justified. But in neither case does the relevant truth play a role in explaining my judgement (Dancy 1985: 34).

5. 5 . F U R T HE R A R GU M E N T S Redundancy employs the minimal sense of tracking failure noted earlier. In light of its failure, the would-be debunker may argue that there is a further— robust—sense in which moral judgements are ‘off-track’. This sense may be supported either by the minimal sense itself or by the considerations which support it. This section outlines one argument of this type and considers prospects for others.

5.5.1. Sensitivity The thought behind the sensitivity argument is that given evolutionary explanations such as the capacity hypothesis, we would make the same moral judgements that we do now, even if the moral truths had been different (Clarke-Doane 2012; Joyce 2016: 129; Lillehammer 2010: 365; White 2010: 581; Wielenberg 2010: 454–6).² More formally:

² Bedke (2014) discusses a related condition—that of ‘obliviousness’—but his resulting debunking argument targets only specific types of cognitivism.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Belief Pills and the Possibility of Moral Epistemology (3)

All (actual, human) moral judgements can be plausibly explained without invoking or assuming their truth.

(4b)

All (actual, human) moral judgements are insensitive to moral truth.

(5b)

If a judgement is not sensitive to the truth of its content then it is not justified.

105

Hence

Hence (C) All (actual, human) moral judgements are unjustified.

To say that a judgement is sensitive to the truth of its content is to say that were that content false, the judgement would not be made. (Strictly speaking, sensitivity needs to be relativized to a method. See Nozick 1981: 179.) One problem here is that (4b) is unsupported. It follows neither directly from (3) nor from the evolutionary premises (1*) and (2*) and it is independently implausible (Clarke-Doane 2012; Wielenberg 2010: 454–6). All three can be demonstrated with the same example. Consider my judgement that Jones’ torturing is wrong, which I possess partly because Jones’ torturing causes unwanted agony. Grant that we can explain this judgement without invoking or assuming its truth. For example we can say that I make this judgement partly because evolution has given me a moral capacity and partly because of the proximal fact that Jones’ torturing causes unwanted agony. Still, my judgement can be sensitive to moral truth so long as the wrongness of Jones’ act supervenes on (or is reducible to) its causing unwanted agony. And that seems to be the case here. Were Jones’ act not to be wrong, it wouldn’t be an act of causing unwanted agony, and hence I wouldn’t judge it wrong. In this case, my moral judgement is sensitive to moral truth even though it can be explained without invoking that truth, and even though it can be given an evolutionary+ explanation which does not invoke or assume that truth. Thus (4b) follows neither from (3) nor from (1*) and (2*), nor is it independently plausible. There is something unsettling about this reply. It relies on two substantial moral assumptions: that Jones’ torturing is wrong and that its wrongness supervenes on its causing unwanted agony. But, in the current context, this is unproblematic. For the burden of proof here is with the would-be debunker. It is sufficient, to resist the argument, to show that (4b) does not follow from the premises offered in its support. To show this, all that is necessary is to show that there is a possible situation where (3) holds and (4b) does not. My judgement concerning Jones is such a case. There is a difference, then, between assuming a moral truth for the sake of explaining moral judgements, and assuming such a truth for the sake of testing

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

106

Neil Sinclair

whether an evolutionary+ explanation of those judgements shows them to be insensitive. The former may be question-begging, but the latter is not (see FitzPatrick 2015: 897). A deeper problem for Sensitivity concerns (5b). Sensitivity does not seem necessary for justification, since false judgements can be justified, but are insensitive (Dancy 1985: 39; White 2010: 580). Further, Sensitivity inherits well-known problems about the assessment of the counterfactual conditionals it employs (Joyce 2016: 130–2).

5.5.2. Other Arguments Sensitivity holds that moral judgements exemplify a robust sense of ‘tracking failure’. It claims that this robust sense is supported by the minimal sense and that the absence of such tracking failure is a necessary condition for justification. The argument fails because the robust sense of tracking failure does not follow from the minimal sense and because the putative necessary condition is implausible. There are several further robust senses of ‘tracking failure’ which can be used to construct similar arguments. For instance, one might hold that minimal tracking failure shows moral judgements to be not safe, contingent, probably false, or unreliable. But all of these arguments are vulnerable in a similar way. First these more robust senses of tracking failure don’t follow from the minimal sense. Second, the putative necessary conditions are contentious (Ichikawa and Steup 2012). Most of these arguments have received extensive discussion elsewhere. (For arguments from safety see Clarke-Doane 2012; Handfield 2016; Joyce 2016; for contingency see Lillehammer 2010; Setiya 2012: 91–2; Wielenberg 2010; for ‘probably false’ see Brosnan 2011; Street 2006; for unreliability see Clarke-Doane 2016; Wielenberg 2010.) So here I will examine a different type of argument, suggested by Joyce’s pill cases.

5 . 6 . L AC K O F M E T H O D One natural response to my dismissal of Sensitivity is as follows. It might be that some moral judgements are sensitive and it might be that one can provide particular examples of moral judgements which satisfy this condition. But this is not to provide any general account of the mechanisms by which humans form moral judgements such that those judgements might end up being in good epistemic standing. In particular, it is not to provide any general account of how the mechanisms for forming moral judgements which we actually employ might constitute ways of getting into contact with

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Belief Pills and the Possibility of Moral Epistemology

107

moral truth. According to this argument, evolutionary explanations of our moral judgements debunk their justification insofar as they emphasize the absence of a positive epistemological story for those judgements (Joyce 2006: 135; Setiya 2012: 96–115). This argument might be expressed thus: (3)

All (actual, human) moral judgements can be plausibly explained without invoking or assuming their truth. (4c) There is no plausible general account of how moral truths could help explain (actual, human) moral judgements. (5c) If there is no plausible general account of how a set of truths could help explain judgements concerning them and we can plausibly explain those judgements without invoking or assuming their truth, then those judgements are unjustified.

Hence (C)

All (actual, human) moral judgements are unjustified.

The antecedent of (5c) is not met in the case of dry-goods judgements, since even in a time before human beings studied their own perceptual systems, they did not have an explanation of their dry-goods judgements that did not invoke their truth. But this condition is seemingly met in the case of moral judgements. For not only do we lack a general account of the mechanism whereby our moral judgements might hook up with moral truth, we also possess an (evolutionary+) account of those judgements that nowhere invokes or assumes their truth. How might the opponent respond? I think they can reasonably reject (4c). In Section 5.6.1, I consider one unsuccessful way of doing this. In Section 5.6.2, I outline a better way.

5.6.1. Pure Epistemology The field of moral epistemology is not empty. There are many extant accounts of the putative mechanisms by which our moral beliefs are epistemically justified. Many of these accounts are motivated independently of the results of debates about the ontological or substantive nature of moral truth. As such they might be labelled ‘pure epistemologies’ (Joyce 2006: 211). According to the current reply, the availability of these epistemologies allows reasonable rejection of (4c). Three popular ‘pure’ moral epistemologies are conservativism, coherentism, and foundationalism. According to conservativism, moral judgements are prima facie justified in virtue of being held (Joyce 2006: 216). According to coherentism, moral judgements are justified insofar as they are part of a

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Neil Sinclair

108

coherent system of judgements (Brink 1989: 100–43). According to foundationalism, moral judgements are justified insofar as they are selfevident (Ross 1930) or suitably related to foundational ‘seeming states’ (Huemer 2005). Unfortunately for the anti-debunker, none of these epistemologies provide, by themselves, sufficient reason to reject (4c). For none goes any way to explaining why the judgements formed as a result of their method might be explained by moral truth. For example, the conservativist position, by itself, gives no reason to think that any of the judgements we happen to find ourselves with are explained by moral truth. Likewise, it is a well-worn criticism of coherentist views that a perfectly coherent set of beliefs can be false. The problem is that these epistemologies are too pure to provide any positive reason for thinking that the judgements they produce are connected to moral truth. They presuppose this claim, rather than independently supporting it (Joyce 2016: 139–43).

5.6.2. An Alternative Reply One diagnosis of this failure is that it is difficult to provide an account of the mechanisms whereby some moral judgements are explained by moral truth without saying anything about the nature of that truth. Just as it seems impossible to give an account of how we know the meaning of words without assuming that they have meaning, it seems impossible to give an account of how we can know moral truths without assuming that there are such truths to be known. The pertinent question, then, is this: What, if anything, is it legitimate to assume about the nature of moral truth in order to construct a general account of how some of our moral judgements might be explained by it?³ My answer to this question will be: enough to make rejecting (4c) reasonable.

5.6.2.1. General Principle Consider a set of judgements concerning a putative subject matter (S-judgements), a set of truths which constitute that subject matter (S-truths),

³ This is an instance of a general question: What, if anything, is it legitimate to assume about the nature of moral truth in order to answer any sceptical challenge? Lillehammer (2010: 374) and Vavova (2014) note that it is impossible for the debunker to establish that our moral belief-forming mechanisms are off-track with respect to moral truth without making assumptions about that truth. However, the Lack of Method argument does not seek to show that our moral belief-forming mechanisms are off-track, only that we lack an account of how they might be on-track. Therefore it remains, at this point, an open question whether assumptions about moral truth are legitimate. FitzPatrick (2015) assumes that they are.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Belief Pills and the Possibility of Moral Epistemology

109

and a general account of how such truths could help explain some of those judgements (an S-epistemology). The following principle appears undeniable: in order to be justified in claiming that there is no plausible S-epistemology we need to make non-trivial assumptions about the nature of S-truth. This can be seen using a variant of the concept-pill scenario. Suppose we are told that X-judgements are (partly) the result of being slipped a pill which gives us X-concepts. Does this show that there is no plausible X-epistemology? Hardly. For unless we are told something about the nature of X-truth, we cannot rule out that it features in the explanation of some of our X-judgements. X-truth might, for example, be truth concerning the composition of the pill we ingested. Returning to Joyce’s pill cases, it is only because we make tacit non-trivial assumptions about the nature of Napoleon-truth—viz. that Napoleon is a physical being who exists independently of our own minds and who has nothing to do with the pill we were slipped—that we take the scenarios described to show that there is no plausible Napoleon-epistemology. Having established the general principle, it follows that in order to assess whether there is a plausible moral epistemology—i.e. whether (4c) is true— we need to make non-trivial assumptions about moral truth.

5.6.2.2. Formal vs. Substantive Assumptions But which non-trivial assumptions are legitimate? It is possible to distinguish two types of answer. According to the formal view, in assessing whether there is a plausible S-epistemology we are permitted only to make non-trivial assumptions about S-truth which do not (by themselves) entail any substantive S-claims. For example, a non-trivial yet formal assumption about Napoleon-truth is that it is truth concerning a physical being. A non-trivial yet formal assumption about moral truth is that it is mind-independent in Street’s sense (Section 5.1). Two further non-trivial formal assumptions about moral truth are suggested by Joyce (2006: 57–64): that it concerns requirements that are inescapable (i.e. do not depend on the agents to whom they apply having relevant motives) and that have rational authority (i.e. necessarily entail practical reasons). These assumptions are formal insofar as they do not entail that the Napoleon-related or moral properties are ever instantiated: they merely tell you what those properties would be like, were they to be instantiated. According to the substantive view, in assessing whether there is a plausible S-epistemology we are permitted to make non-trivial assumptions about S-truth which entail substantive S-claims. In other words, according to the substantive view, it is permissible, in constructing an S-epistemology, to assume the approximate truth of some of our positive S-judgements.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

110

Neil Sinclair

Which view should we prefer? The formal view seems prima facie legitimate, at least in some cases. Suppose a friend claims that they can communicate with ghosts—non-physical beings not bound by physical laws. In assessing whether there is a plausible general account of how ghost-truths could explain some of our friend’s ghost-judgements we need to make some assumptions about ghost-truth (e.g. that it is truth concerning non-physical beings) but we surely grant our friend too much if we permit them to assume that auntie’s ghost lives in the attic. Yet in other cases the formal view seems inappropriate. Consider drygoods judgements. Suppose that in assessing whether there is a plausible dry-goods epistemology we are permitted only to make formal assumptions about dry-goods truth—for example that it is truth concerning mindindependent material entities. On such a slender basis, constructing an epistemology is impossible. Unless we begin with the substantive assumption that we ourselves are dry-goods, possessed with sense organs which provide some type of connection to other dry-goods, and inhabiting a drygoods world somewhat like our initial dry-goods judgements take it to be, we cannot begin to investigate in detail how some of our dry-goods judgements might be explained by dry-goods truth. But once we make such assumptions the beginning of a dry-goods epistemology (involving, for example, details about the functioning of our sense organs) is relatively easy to construct. This partial epistemology can in turn help illuminate how some (perhaps most) of our dry-goods are well-explained by dry-goods truth, while others (including some of those on which we initially relied) are not so explained, and should be disowned. From here, further processes of refinement, such as eliminating inconsistencies and arbitrariness, adding further beliefs that increase overall coherence, and weeding out judgements based on unjustified assumptions, can help create a more complete theory of the dry-goods world (perhaps even including details about the evolutionary history of certain humanoid dry-goods and their perceptual capacities). This in turn can help us provide a more complete epistemology, and clearer ideas of when our dry-goods judgements are in error. Thus a gradual process of refinement emerges, with initial judgements helping to construct a tentative epistemology, which may in turn (together with processes of reasoning and reflection) help us refine our theory of the dry-goods world, which in turn helps refine our epistemology, and so on. The success of such an account is neither trivial nor inevitable (Shogenji 2000), but seems impossible to get off the ground without assuming, at the outset, the defeasible justifiedness of some substantive dry-goods judgements (Field 1989: 25–30; see Boyd 1988: §3.2). Blackburn makes a similar point for the case of physics. He asks what kind of theory might vindicate our confidence that, if we exercise our

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Belief Pills and the Possibility of Moral Epistemology

111

sensory and cognitive faculties well and end up believing that p, then p. He continues: ‘If p is a thesis from basic physical theory, only the theory itself . . . Any attempt at a background, an underwriting of the conditional from outside the theory, is certain to be bogus’ (1993: 167). If so, in assessing whether there is a plausible general account of how S-truths could explain some of our S-judgements, it is sometimes legitimate to make substantive assumptions about the nature of S-truth. But is it legitimate in the moral case?

5.6.2.3. Epistemology for Mind-Dependent Conceptions of Moral Truth Suppose it is not. On this formal view, in assessing whether there is a plausible moral epistemology we are not permitted to make substantive moral assumptions. Does it follow that there is no plausible moral epistemology? Not necessarily. We saw in the case of ghost-judgements and drygoods judgements that it was impossible to construct a plausible epistemology while only making formal assumptions about the relevant truths. In both cases, the formal assumptions included the assumption that the relevant truth is (or would be) mind-independent. But suppose that the assumptions in the moral case include the thought that moral truth is mind-dependent.⁴ This is to say that moral truth concerns the instantiation of properties which satisfy the following schema: Object, x, possesses property M iff x tends to elicit R from P in C,

where R is a reaction (e.g. an emotion), P a set of persons, and C a set of circumstances (such as conditions of imaginative acquaintance). To capture genuine mind-dependence, the conditions on the right-hand side must also be specifiable independently of the extension of M (Johnston 1989). There are many possible versions of moral mind-dependence. But the present concern is just whether such an account can help construct a moral epistemology. And it seems that it can. As Lewis notes: ‘In general, to find out whether something is disposed to give response R in conditions C, you can put it in C and find out whether you get R’ (1989: 116). Hence, assuming that moral properties are mind-dependent, moral judgements formed as a result of people placing themselves in (or approximating) circumstances C and seeing whether, in those circumstances, they have ⁴ Another possibility is to deploy the formal assumption that moral terms are rigid designators, and therefore that moral truth is whatever natural truth is revealed by a privileged subset of our uses of those terms (see Boyd 1988). On this view, moral epistemology is straightforward: some instances of moral properties are causally responsible for some of our moral judgements (Setiya 2012: 119–20; Lillehammer 2016: 117).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Neil Sinclair

112

reaction R, will be partly explicable in terms of moral truth. (Perhaps this is just the method we employ when using thought experiments in ethics; note also that it is consistent with the possibility of error.) Hence, even if we are not permitted to make substantive assumptions about moral truth in order to construct a moral epistemology, the purely formal assumption that moral truth is mind-dependent is enough.⁵ In response, debunkers might reject the assumption that moral truth is minddependent. Joyce argues that construing moral truth as mind-dependent fails to accommodate the fact that moral properties, when instantiated, provide requirements which are inescapable and rationally authoritative (2006: 190–209). But two points are worth mentioning. First, if one accepts Joyce’s approach, then one must hold that (4c) is only reasonable to those who have independent reasons to reject moral mind-dependence. These reasons will be metaethical. Hence, ‘[this] debunking argument has teeth only if certain metaethical arguments succeed’ (Joyce 2016: 136). Second, one might think that the mere fact that the debunking argument would succeed were one to reject moral mind-dependence casts doubt on that rejection (compare Setiya 2012: 115; Bedke 2014: 123).⁶ That is, one might tollens the following ponens: (i) If moral truth is not mind-dependent, then (by the Lack of Method argument) all (actual, human) moral judgements are unjustified. (ii) Moral truth is not mind-dependent.

Hence (iii)

All (actual, human) moral judgements are unjustified.

Dialectical stalemate looms. Joyce accepts (i), (ii), and (iii). The alternative accepts (i), not-(iii), and not-(ii). Joyce may insist that his position is preferable, since (ii) is independently more plausible than not-(iii). But in fact the very same types of reasons which Joyce cites in support of (ii) support the denial of (iii). One type of argument Joyce gives is a version of the argument from moral appearances (Glassen 1959; Joyce 2001: 12–16; see Sinclair 2012b). According to this argument the way in which ordinary moralizers use moral language reveals the linguistic conventions which govern moral discourse and (thus) delineates our conception of moral truth. Further, when we look at this usage, we see that ordinary moralizers deploy moral language in a way that confirms the assumption that moral truth is (conceived as) mind-independent. Unfortunately, in the current ⁵ For a different attempt to construct a moral epistemology from purely formal assumptions, see Behrends (2013). ⁶ Given footnote 4, the following argument oversimplifies a little.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Belief Pills and the Possibility of Moral Epistemology

113

context a similar argument supports the view that some moral judgements are justified. For when we look at the way people use moral language and concepts, we see that this use reveals a conception of moral truth such that some of our current judgements are justified (if approximate) intimations of it. One potential lesson here is that actual usage reveals that everyday morality exemplifies no coherent conception of moral truth (Lillehammer 2003b). But it doesn’t follow that the best conception of moral truth is the Joycean version (which eschews epistemological optimism in favour of mind-independence) rather than the alternative (which eschews mindindependence in favour of epistemological optimism). The second type of argument Joyce gives for (ii) is that views that take moral truth to be mind-dependent cannot accommodate the inescapability and rational authority of moral truth (2006: 190–9). These are ‘nonnegotiable’ features of moral practice: ideas about the nature of moral truth such that, if they were not assented to by a population of speakers, that population would not be interpreted as engaging in moral practice at all (2006: 199–209). Here again the problem is that if the argument goes through in the case of the assumptions of inescapability and rational authority, it also goes through in the case of the assumption that some of our moral judgements are justified. For a population of speakers and thinkers otherwise like us, but who took none of their moral judgements to be justified, could scarcely be considered to be engaging in moral practice. Even if the arguments of the last two paragraphs are unsuccessful, the previous point stands. Given our current assumptions about what it is permissible to assume when constructing a moral epistemology, it is reasonable to accept (4c) only if one has sufficient independent reason to reject the claim that moral truth is mind-dependent. The set of philosophers who deny that there is such reason is not empty. It includes, for example, Brower (1993), Lewis (1989), Railton (1986), Smith (1989), and Street (2006).

5.6.2.4. Epistemology for Mind-Independent Conceptions of Moral Truth Section 5.6.2.3 considered the formal view applied to the moral case. On the alternative view, in assessing whether there is a plausible general account of how moral truth could explain some of our moral judgements it is legitimate to make substantive assumptions about moral truth. There are at least two versions of this approach. According to the first— reductive—version the relevant assumptions include claims regarding the identity of moral with natural properties. For example, one might hold that the property of moral rightness is identical with the property of being

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

114

Neil Sinclair

maximally pleasure-producing (it does not follow that the two phrases are synonymous). According to the second—non-reductive—version the relevant assumptions include both formal claims about the nature of moral properties (e.g. that they are mind-independent) and substantive claims about the distribution of their instances. According to both versions moral truth is relevantly mind-independent. But non-reductive views are compatible with taking the relevant substantive assumptions to be partial accounts of moral truth—that is, assumptions which do not specify necessary and sufficient conditions for an action to be right or a state of affairs good (for example), but which nevertheless include substantive claims. If it is legitimate to make substantive moral assumptions in the process of assessing whether there is a plausible general account of how moral truth could explain some of our moral judgements then, it seems, it is relatively easy to construct a moral epistemology. For example, assuming utilitarianism, all judgements of rightness formed after a process which is sensitive to facts about which actions maximize pleasure would be explicable in terms of moral truth (see Setiya 2012: 112). Or consider the view that ‘what’s morally good has to do with behaviors that promote rather than hinder wellbeing’ (Brosnan 2011: 62; Raz 1986: 194) or which answer to fundamental human needs (Boyd 1988: §4.3). Again, judgements of goodness formed by a process sensitive to facts about which actions promote wellbeing or fundamental needs will be good candidates for judgements that are explicable by moral truth. Or consider the view that what’s just for a society has to do with which social arrangements best satisfy or respect the deeply held concerns of its members. Then judgements of justice formed after a process that is sensitive to such facts are good candidates for being explicable by moral truth. In these latter cases, anti-debunkers can even co-opt elements of Joyce’s evolutionary account into their epistemology, since if, as Joyce’s theory claims, one role of moral judgements is to facilitate productive schemes of social cooperation, we can expect such judgements to be somewhat sensitive to facts about human needs and concerns, and about which schemes of cooperation best promote these (see Boyd 1988: §4.4; FitzPatrick 2015: 894; Sinclair 2012a). Insofar as these latter cases also begin with substantive assumptions about moral truth that are partial, they are compatible with a process of refinement that mirrors that described in the dry-goods case (Section 5.6.2.2), whereby initial assumptions about moral truth help generate a tentative moral epistemology, which in turn helps refine our sense of which of the initial assumptions are, after all, trustworthy, which in turn helps further our theory of moral truth, which allows us in turn to refine the epistemology, and so on, until ultimately a complete conception of moral truth and the nature of our contact with it is reached. This, of course, is just the

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Belief Pills and the Possibility of Moral Epistemology

115

method of broad reflective equilibrium in ethics (Boyd 1988: §4.2; Brink 1989: 122–33).⁷ It seems, therefore, that constructing an epistemology for mind-independent conceptions of moral truth is perfectly possible, provided we are permitted substantial assumptions about where that truth lies. The debunker will reply that it is not legitimate to rely on such assumptions. In fact there are two distinct replies here, which are worth distinguishing: (a) it is illegitimate, in the moral case, to begin the task of epistemology by making any substantive assumptions; (b) it is illegitimate to begin this task with any of the particular substantive assumptions mentioned earlier. I’ll take these in turn. Why might it be illegitimate, in the moral case, to make any substantive assumptions when constructing an epistemology? I argued in Section 5.6.2.2 that it is not universally illegitimate. What particular reason might there be for thinking that the formal approach is the right one to take in the moral case? I consider four putative arguments.⁸ ⁷ Note that the accounts outlined here are not identical to so-called ‘third-factor’ views. According to Enoch (2010), for example, the substantive assumption that survival is good explains why there is a pre-established harmony between some evolutionarily selected moral beliefs and moral truth, but particular moral truths are not involved in the explanation of particular beliefs (for variations and criticism see Bedke 2014; Setiya 2012: 111–16). The accounts suggested here are distinct insofar as, if successful, particular moral beliefs can be explained in terms of (and are sensitive to) their content. This is one place where the relevant incompleteness of the evolutionary hypotheses is key (see FitzPatrick 2015 for one version of this view). Note also that this method moves beyond simply assuming particular moral truths for the sake of explaining our beliefs in them (see Vavova 2014: 86). Rather it begins with defeasible assumptions about the nature of that truth and uses them to attempt to construct a more detailed yet general moral epistemology (Bedke (2014: 113) calls this ‘probationary epistemology’). As in the perceptual case, the assumptions with which the process begins (e.g. that the world is roughly like initial perception presents it as being, that goodness has to do with fundamental needs) will not generally be explanans in the resulting explanations (e.g. the table in front of me explains my belief in it, the goodness of this action explains why I believe it to be good) but rather parts of the general picture within which such explanations are justified. ⁸ Four further arguments can be quickly dismissed. The first holds that the fact that moral judgements can be explained without citing moral truth shows that they are subject to distorting influences (Fraser 2014: 469; Street 2006: 121). This argument fails since the distortion metaphor only makes sense given substantive assumptions about moral truth (Lillehammer 2010: 375; White 2010: 590). The second suggests that were we ‘allowed to rely on some substantive moral claims in assessing the reliability of moral faculties, then we would be effectively immunizing a class of moral beliefs from genealogical doubts’ (Shafer-Landau 2012: 21; see Vavova 2014: 81). But according to the methodology I have outlined, no judgement is completely immune from doubt—rather, certain judgements are presumed to be justified for the purposes of constructing an epistemology. Both the failure to construct such an epistemology, and its details, may be sufficient to undermine this presumptive justification (this applies in both the perceptual and the moral cases; see Bedke 2014: 113–14, Wielenberg 2016: 93). The third holds that any substantive assumptions we make are illegitimate because contingent, i.e. we could have

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

116

Neil Sinclair

First, one might appeal to the analogy of the poisoned well, or the principle of junk-in/junk-out (Street 2006: 123–4, 140). The thought is that epistemology cannot begin by making assumptions whose epistemic credentials are themselves in doubt. But sawn of rhetorical imagery, this argument is no more than the reassertion of the applicability of the formal view to the moral case. Furthermore, the current context is one of attempting to provide independent support for a premise in a global justification debunking argument. This is not a context where the epistemic credentials of all moral judgements are as yet in doubt. Second, one might think that the level of disagreement in moral judgements counts against assuming even the approximate truth of any (positive) moral judgement when constructing a moral epistemology. However, on this view, the strength of the debunking argument depends entirely on premises about the scope and nature of moral disagreement. These premises are both highly contentious and have been taken to feature in distinct arguments for moral scepticism (see e.g. Decker and Groll 2013). On this view the evolutionary debunking argument is at best redundant, at worst unsound. Third, one might hold that insofar as one of the tasks of metaethics is to locate moral practice within the world view which begins with dry-goods judgements and ends with a fully developed natural science, we should not begin this task by assuming the actual instantiation of any moral properties (by contrast, there is no ‘location’ problem for science; Blackburn 1993: 166). However, on this view the debunking argument is no threat to those who deny the naturalistic methodological assumption that the task of metaethics includes a placing of moral properties within the natural realm (McDowell 1985). Furthermore, even methodological naturalists may

easily begun with different assumptions, making any belief formed using the methods of the resulting epistemology at best accidently true. But this point is too general since it would equally rule out, for example, relying on substantive assumptions in the dry-goods case—we are equally lucky that we are not brains-in-vats. Given these assumptions, and the fact that it is possible to construct an epistemology on the basis of them, it is no accident that any belief well formed by the methods of that epistemology is true. Neither will it be an accident that such methods are themselves reliable (contra Setiya 2012: 112–14) since the process of reflective equilibrium, if successful, will have refined them in light of their earlier reliability. (For discussion of contingency see Boyd 1988: §§3.3, 4.4; Lillehammer 2010: 375–7. An alternative response would be to begin with necessary substantive assumptions—see Cuneo and Shafer-Landau 2014, discussed in Vavova 2015.) Fourth, it might be illegitimate to make substantive moral assumptions because, unlike in the case of dry-goods judgements, we have no way of testing or validating (e.g. in terms of predictive power) the theories of the moral domain that may result (see Boyd 1988: §4.2). But this is question-begging in the current context, since it amounts to the demand that moral theory be assimilated to scientific theory.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Belief Pills and the Possibility of Moral Epistemology

117

question whether their naturalism precludes them from assuming the approximate truth of some moral judgements when constructing a moral epistemology. After all, one notable feature of the natural phenomenon of moral practice is that practically all its practitioners are convinced of the truth of some moral judgements; a plausible explanation of this is that some of those judgements are (approximately) true. Finally, the debunker might argue that reflection on the nature of moral properties supports the error-theoretic view that such properties cannot be instantiated and hence no moral judgement can be relied on when constructing a moral epistemology. However, this debunking argument is entirely dependent on a distinct truth-debunking argument (see Section 5.1). Furthermore, one might question the latter: it seems that the very same sorts of considerations which support the error-theoretic claim that moral judgements are cognitive also support that claim that some of those judgements are approximately true (Kirchin 2010). Hence it seems that there is no reason as yet to think that it is illegitimate to begin the task of constructing a moral epistemology by making substantive moral assumptions. Still, the debunker might reply—this is point (b) mentioned earlier—that none of the particular substantive assumptions with which anti-debunkers begin this task are plausible. This seems to be Joyce’s view: I’ll go out on a limb and assert . . . that we do not have a believable account of how moral facts could help explain the mechanisms and forces that give rise to moral judgments . . . . Some people think that they already have believable accounts, but they’re mistaken . . . Utilitarians, for example, believe that they can explain how moral facts relate to moral judgments, but anyone who isn’t a utilitarian . . . thinks that they’re mistaken. Kantians . . . believe that they can explain how moral facts relate to moral judgments, but anyone who isn’t a Kantian . . . thinks that they’re mistaken. And so on. It doesn’t really add anything shocking to this . . . picture to accept the view that all people who think that they can explain how moral facts relate to moral judgments are mistaken. (2016: 135)

Note the assumption of this passage: that in order to construct an account of how moral facts could explain some of our moral judgements, we need a complete account of moral truth, such as utilitarianism. Joyce may be right that this is one way of proceeding, but it is not the only way. Given the analogy with dry-goods judgements, all that is required to get the epistemological story off the ground is the assumption of the approximate truth of some judgements: the content of these judgements needn’t constitute a complete account of moral truth. The upshot is as follows. We have seen no particular reason to consider it illegitimate, for the task of constructing a moral epistemology, to begin by

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

118

Neil Sinclair

assuming the approximate truth of some (positive) moral judgements. Given this, it is possible to begin to construct a plausible account of how moral truth might be involved in the explanation of some of our moral judgements, and thus reject (4c).

5.6.2.5. Summary and Objection Here, then, is my response to the Lack of Method argument. Whether there is a plausible general account of how moral truths could help explain some of our moral judgements depends on what it is permissible to assume about moral truth in order to construct such an account. If we make the formal assumption that moral truth is mind-dependent then such an account is available and (4c) can be rejected.⁹ Distinct moral epistemologies are available if we assume that moral truth is mind-independent, but add to this substantive assumptions about the distribution of moral properties. Some such assumptions are complete accounts of moral truth, and I agree with Joyce that many such assumptions are implausible. But other assumptions are merely partial and, crucially, these are enough to begin to construct a moral epistemology, and hence reject (4c). Further, we have seen no particular reason to think that, in the moral case, making such substantive assumptions is illegitimate. (At least, no reason that is compatible with evolutionary justification debunking arguments maintaining independent force.) One final reply from the would-be debunker runs as follows: ‘I accept that there are many types of metaethical view which can reject (4c). In particular (4c) can be rejected by those who accept a mind-dependent conception of moral truth, by those who take moral properties to be reducible to particular natural properties, and by those who believe that moral truth is mindindependent, non-reducible, but supervenient on particular aspects of natural truth. But all these views are implausible for independent reasons. Both mind-dependent and reductionist views fail to capture the non-negotiable elements of inescapability and rational authority, and the non-reducible view is ruled out by Ockham’s razor’ (see Joyce 2006: 190–211). One problem here is the possibility of a quasi-realist view which provides expressivist-friendly interpretations of the claims that moral truth is mindindependent and non-reducible, and which doesn’t violate Ockham’s razor. But this aside, even if the above reply is accepted the evolutionary debunking argument would be redundant. For if we have good reason to think that moral truth is neither reducible to any natural truth nor existent but irreducible, then we have a much more direct argument for thinking that none of our moral judgements are justified. For if we know that moral truth does not exist, we know that none of our (positive) judgements about it are justified. ⁹ Other formal assumptions may do the same job—see footnote 4.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Belief Pills and the Possibility of Moral Epistemology

119

5.7. CONCLUSION Here is a tentative hypothesis to diagnose the fact that some find justification debunking arguments compelling, while others are unmoved. The former consider it illegitimate, for the task of answering sceptical challenges, to assume even the approximate truth of any substantive moral judgement. Such people find premises like (4c) plausible. Conversely, the latter consider it legitimate, for the same task, to assume the approximate truth of some moral judgements. One way of understanding the latter group is that they hold that, as in the case of physics, the epistemic validation of our moral judgements can only come from a position within the relevant practice. If this is right, progress in debunking debates will only be made by addressing the underlying question of what it is legitimate to assume (about relevant truth) in answering sceptical challenges. This will involve, in turn, a clear grasp of the intended import, independence, and generality of each sceptical challenge. More particularly, in the case of evolutionary debunking arguments of morality the challenge for the justification debunker will be to give independent arguments for their constraints on constructing moral epistemologies which do not also render their debunking argument superfluous.¹⁰

References Bedke, M. 2014. ‘No Coincidence?’ Oxford Studies in Metaethics 9: 102–25. Behrends, J. 2013. ‘Meta-Normative Realism, Evolution and Our Reasons to Survive’ Pacific Philosophical Quarterly 94: 486–502. Blackburn, S. 1993. Essays in Quasi-Realism. Oxford: Oxford University Press. Boyd, R. 1988. ‘How to be Moral Realist’ in G. Sayre-McCord, ed., Essays on Moral Realism. Ithaca, NY: Cornell University Press, 181–228. Brink, D. 1989. Moral Realism and the Foundations of Ethics. Cambridge: Cambridge University Press. Brosnan, K. 2011. ‘Do the Evolutionary Origins of Our Moral Beliefs Undermine Moral Knowledge?’ Biology and Philosophy 26: 51–64. Brower, B. 1993. ‘Dispositional Ethical Realism’ Ethics 103: 221–49.

¹⁰ This work was supported by the Arts and Humanities Research Council (grant number AH/J006394/1). For helpful discussion at conferences in UNC and FSU, thanks to: Andrew Alwood, Matt Bedke, Jeff Behrends, Justin Clarke-Doane, William FitzPatrick, Benjamin Fraser, Justin Horn, Zoë Johnson King, Hallvard Lillehammer, Kathryn Lindeman, Don Loeb, Matthew Lutz, Jeffrey O’Connell, Herman Philipse, Debbie Roberts, Michael Ruse, Bart Streumer, Christine Tiefensee, Michael Vlerick, and Silvan Wittwer. Special thanks to Uri Leibowitz, Russ Shafer-Landau, Barry Lam, the Marc Sanders Foundation, Ruth Chang, Stephen Finlay, Chris Heathwood, and an anonymous referee for this volume.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

120

Neil Sinclair

Campbell, R. 1996. ‘Can Biology Make Ethics Objective?’ Biology and Philosophy 11: 21–31. Casebeer, W. D. 2003. Natural Ethical Facts: Evolution, Connectionism, and Moral Cognition. Cambridge, MA: MIT Press. Clarke-Doane, J. 2012. ‘Morality and Mathematics: The Evolutionary Challenge’ Ethics 122(2): 313–40. Clarke-Doane, J. 2016. ‘Debunking and Dispensability’ in Leibowitz and Sinclair 2016. Cuneo, T. and Shafer-Landau, R. 2014. ‘The Moral Fixed Points’ Philosophical Studies 171(3): 399–443. Dancy, J. 1985. Introduction to Contemporary Epistemology. Oxford: Basil Blackwell. 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. Enoch, D. 2010. ‘The Epistemological Challenge to Metanormative Realism: How Best to Understand it, and How to Cope with it’ Philosophical Studies 148: 412–38. Field, H. 1989. Realism, Mathematics and Modality. Oxford: Blackwell. FitzPatrick, W. 2015. ‘Debunking Evolutionary Debunking of Ethical Realism’ Philosophical Studies 172: 883–904. Fraser, B. 2014. ‘Evolutionary Debunking Arguments and the Reliability of Moral Cognition’ Philosophical Studies 168: 457–73. Glassen, P. 1959. ‘The Cognitivity of Moral Judgments’ Mind 68(269): 57–72. Greene, J. 2008. ‘The Secret Joke of Kant’s Soul’ in W. Sinnot-Armstrong, ed., Moral Psychology, Volume 3: The Neuroscience of Morality: Emotion, Brain Disorders and Development. Cambridge, MA: MIT Press. Handfield, T. 2016. ‘Genealogical Explanations of Change and Morals’ in Leibowitz and Sinclair 2016. Harms, W. F. 2000. ‘Adaption and Moral Realism’ Biology and Philosophy 15: 699–712. Huemer, M. 2005. Ethical Intuitionism. Basingstoke: Palgrave Macmillan. Ichikawa, J. and Steup, M. 2012. ‘The Analysis of Knowledge’ The Stanford Encyclopedia of Philosophy (spring 2014 edition), Edward N. Zalta, ed., . Johnston, M. 1989. ‘Dispositional Theories of Value’ Proceedings of the Aristotelian Society, supplementary volume 63: 139–74. Joyce, R. 2001. The Myth of Morality. Cambridge: Cambridge University Press. Joyce, R. 2006. The Evolution of Morality. Cambridge, MA: MIT Press. Joyce, R. 2014. ‘The Evolutionary Debunking of Morality’ in J. Feinberg and R. Shafer-Landau, eds, Reason and Responsibility: Readings in Some Basic Problems of Philosophy, 15th edition. Belmont, CA: Wadsworth/Thomson Learning. Joyce, R. 2016. ‘Reply: Confessions of a Modest Debunker’ in Leibowitz and Sinclair 2016. Kahane, G. 2011. ‘Evolutionary Debunking Arguments’ Noûs 45(1): 103–25. Kirchin, S. 2010. ‘A Tension in the Moral Error Theory’ in R. Joyce and S. Kirchin, eds, A World Without Values: Essays on John Mackie’s Moral Error Theory. London: Springer.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Belief Pills and the Possibility of Moral Epistemology

121

Leibowitz, U. and Sinclair N. 2016. Explanation in Ethics and Mathematics: Debunking and Dispensability. Oxford: Oxford University Press. Lewis, D. 1989. ‘Dispositional Theories of Value’ Proceedings of the Aristotelian Society, supplementary volume 63: 113–37. Lillehammer, H. 2003a. ‘Debunking Morality: Evolutionary Naturalism and Moral Error Theory’, Biology and Philosophy 18: 567–81. Lillehammer, H. 2003b. ‘The Idea of a Normative Reason’ in P. Schaber and R. Huntelmann, eds, Grundlagen der Ethik. Frankfurt: Ontos Verlag. Lillehammer, H. 2010. ‘Methods of Ethics and the Descent of Man: Darwin and Sidgwick on Ethics and Evolution’ Biology and Philosophy 25: 361–78. Lillehammer, H. 2016. ‘ “An Assumption of Extreme Significance”: Moore, Ross and Spencer on Ethics and Evolution’ in Leibowitz and Sinclair 2016. Mason, K. 2010. ‘Debunking Arguments and the Genealogy of Religion and Morality’ Philosophy Compass 5(9): 770–8. McDowell, J. 1985. ‘Values and Secondary Properties’ in T. Honderich, ed., Morality and Objectivity. London: Routledge & Kegan Paul, 110–29. Mikhail, J. 2011. Elements of Moral Cognition. Oxford: Oxford University Press. Nichols, S. 2014. ‘Process Debunking and Ethics’ Ethics 124: 727–49. Nozick, R. 1981. Philosophical Explanations. Oxford: Oxford University Press. Quinn, W. 1986. ‘Truth and Explanation in Ethics’ Ethics 96(3): 524–44. Railton, P. 1986. ‘Moral Realism’ Philosophical Review 95: 163–207. Raz, J. 1986. The Morality of Freedom. Oxford: Clarendon Press. Ross, W. D. 1930. The Right and the Good. Oxford: Oxford University Press. Ruse, M. 1986. Taking Darwin Seriously. Oxford: Basil Blackwell. Setiya, K. 2012. Knowing Right from Wrong. Oxford: Oxford University Press. Shafer-Landau, R. 2012. ‘Evolutionary Debunking, Moral Realism and Moral Knowledge’ Journal of Ethics and Social Philosophy 7(1): 311–29. Shogenji, T. 2000. ‘Self-Dependent Justification Without Circularity’ British Journal for the Philosophy of Science 51: 287–98. Sinclair, N. 2012a. ‘Metaethics, Teleosemantics and the Function of Moral Judgements’ Biology and Philosophy 27(5): 639–62. Sinclair, N. 2012b. ‘Moral Realism, Face-Values and Presumptions’ Analytic Philosophy 53(2): 158–79. Singer, P. 2005. ‘Ethics and Intuitions’ The Journal of Ethics 9: 331–52. Smith, M. 1989. ‘Dispositional Theories of Value’ Proceedings of the Aristotelian Society, supplementary volume 63: 89–111. Street, S. 2006. ‘A Darwinian Dilemma for Realist Theories of Value’ Philosophical Studies 127: 109–66. Toner, C. 2011. ‘Evolution, Naturalism, and the Worthwhile: A Critique of Richard Joyce’s Evolutionary Debunking of Morality’ Metaphilosophy 42(4): 520–46. Vavova, K. 2014. ‘Debunking Evolutionary Debunking’ Oxford Studies in Metaethics 9: 76–101. Vavova, K. 2015. ‘Evolutionary Debunking of Moral Realism’ Philosophy Compass 10(2): 104–16.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

122

Neil Sinclair

White, R. 2010. ‘You Just Believe that Because . . .’ Philosophical Perspectives 24: 573–615. Wielenberg, E. 2010. ‘On the Evolutionary Debunking of Morality’ Ethics 120(3): 441–64. Wielenberg, E. 2016. ‘Evolutionary Debunking Arguments in Religion and Morality’ in Leibowitz and Sinclair 2016.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

6 The Puzzle of Pure Moral Motivation Adam Lerner

Some people care about doing the right thing as such. They want to do the right thing under that description. These people have what has come to be known as a de dicto desire to do what is right.¹ Some people who have a desire to do what’s right as such desire to do what’s right only as a means. They want to do the right thing in order to avoid punishment, or as a way of doing what God or their mother would approve of, or because it would be a sign that they are predestined for salvation. But many care about doing the right thing for its own sake: because it’s the right thing. They have a final desire to do what’s right as such. This desire is an instance of what I will call pure moral motivation (PMM).² Other instances of PMM include the desire to avoid doing what’s wrong as such, the desire to promote what’s good as such, and the desire to prevent what’s bad as such. Contemporary views in metaethics have been built to satisfy familiar desiderata: accommodating Moorean open questions, explaining the connection between moral judgment and moral motivation, accounting for the supervenience of the moral on the non-moral, and many others. In this chapter, I introduce a new desideratum: vindicating the rationality of PMM. I argue that PMM is often rational, and that any adequate view must accommodate this fact. The puzzle of PMM is to explain how that can be done. I argue that solving the puzzle poses a serious challenge for the standard views in metaethics.³ In Section 6.1, I argue that PMM is at least sometimes rational. In Section 6.2, I explain why non-cognitivism has trouble accommodating

¹ The terminology originates with Smith (1994, pp. 74–6). ² McGrath (2009) provides the inspiration for both the terminology and the chapter title. ³ While I focus on moral motivation in this chapter, I believe the puzzle arises for every kind of pure normative motivation.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

124

Adam Lerner

the rationality of PMM. In Section 6.3, I introduce two principles regarding the conditions under which final desires and preferences are rational. In Sections 6.4 and 6.5, I draw on these principles to argue that synthetic naturalism and non-naturalism both have trouble accommodating the rationality of PMM. In Section 6.6, I explain why analytic naturalism has trouble accommodating the rationality of PMM. Section 6.7 concludes.

6.1. THE RATIONALITY OF PMM At the heart of moral practice lies moral inquiry. And at the heart of moral inquiry lies the constitutive aim of discovering the moral truth; one cannot engage in moral inquiry without having a desire to discover the moral truth. People regularly engage in moral inquiry. So they must have a desire to discover the moral truth. Why would anyone want to discover the moral truth? Any number of things might motivate someone to seek the moral truth. They might get paid to do it. They might want to predict what virtuous people will do in certain circumstances so they may exploit their virtuous character. They might just be curious. But this is not why most people seek the moral truth. Most people seek the moral truth because they want to act in accordance with the moral truth. They want to do the right thing as such.⁴ And they want to do it for its own sake. They have a final desire to do the right thing as such.⁵ They exhibit PMM. In this section, I argue that it can be rational to have PMM.⁶ I offer two types of cases in which it’s rational to choose to engage in moral inquiry and in which the best explanation of why ordinary people do this is that they ⁴ In desiring to do the right thing as such, one conceives of the thing one wants to do as the right thing. One desires it under that description, the thing that has the property of being right. When one has a merely de re desire to do the right thing, one conceives of the thing one wants to do under some alternative guise—say, as the thing that saves the drowning child. A de re desire to do the right thing counts as a desire to do the right thing because the object of the desire is right, not because it’s conceived to be. ⁵ We can understand final desires by contrasting them with instrumental desires. S’s desire that P is an instrumental desire just in case S has that desire only because (a) S desires something else Q, and (b) S believes that P raises the probability of Q either by causing, realizing, or signifying Q. (When I say that S has that desire only because of these further facts, I mean these further facts together constitute S’s sole rationale for that desire.) S’s desire is a final desire just in case S’s rationale for their desire does not consist only in pairs of claims like (a) and (b) (cf. Harman, 2000; McDaniel and Bradley, 2008). ⁶ For purposes of this chapter, a motivation is rational if and only if the person who has the motivation possesses at least some good reason of the right kind to have that motivation and it’s their possession of this reason that sustains that motivation. For someone to possess a reason to have some motivation, it isn’t enough that there be some reason for them to have it: that reason must be within their ken (cf. Lord, 2015).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

The Puzzle of Pure Moral Motivation

125

have PMM. I argue that because in both types of cases the choice to engage in moral inquiry is rational, the PMM that lies behind this choice must itself be rational.

6.1.1. Pure Moral Inquiry with Full Information Delilah owns a factory farm. She knows everything there is to know about what goes on inside a factory farm. She also knows everything there is to know about how factory farms impact the environment, the economy, and people’s gustatory experiences. But Delilah has never thought about whether it might be wrong to run a factory farm. Or maybe she has given it some thought, but she hasn’t come to a settled view. She now faces a decision: should she sink time and effort into trying to figure out whether running a factory farm is right? Suppose Delilah chooses to deliberate. She seems perfectly rational in making this choice. Why would Delilah make this choice? Clearly, she has a desire to know what’s morally right in this case. But presumably she doesn’t want this bit of moral knowledge for its own sake. She wants it because she wants to do what’s right.⁷ Importantly, this desire to do what’s right must be a desire to do what’s right as such. If Delilah merely had a de re desire to do what’s right, then she would desire to do it under some non-moral description. But if she wanted to do what’s right under some non-moral description, she would already know what act she would have to perform in order to satisfy her desire to do what’s right—after all, she already knows all of the morally relevant nonmoral facts about the case. So she wouldn’t deliberate. So when Delilah chooses to deliberate, she must be acting on a desire to do what’s right as such.⁸ Now suppose she doesn’t want to do the right thing in order to avoid punishment, or as a way of doing what her mother wants, or to confirm that she’s bound for heaven. She wants it for its own sake. Her desire to do what’s right is a final desire. This shows that in choosing to deliberate, Delilah is acting on PMM. Is it rational for Delilah to have this PMM? Yes. Why? Because Delilah acts rationally in choosing to deliberate. And if an agent rationally chooses to ⁷ Or at least, she wants to avoid doing what’s wrong. ⁸ Alternatively, she may be acting on a second-order final (de dicto) desire to do whatever final desires constitute (de re) desires to do what’s right (Dreier, 2000, p. 632). Or she may be acting on a desire to do what’s just or benevolent, as long as justice and benevolence each are irreducibly moral or have an irreducibly moral component. Since these alternative desires essentially have moral content, they are still instances of PMM. So what I go on to say about the de dicto desire to do what’s right applies just as much to these desires.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

126

Adam Lerner

φ, then the desires she acts on in making that choice are rational.⁹ Thus, we have a case in which PMM is rational.¹⁰ A Humean might object that this merely shows Delilah isn’t irrational in acting on her PMM. They might hold that PMM is a deep-seated final desire that cannot itself be subject to rational assessment. I disagree. It seems it would be rational for Delilah to act on PMM even if she knew that doing so would likely defeat an even more deep-seated desire of hers (e.g., getting rich). What this shows is that when we say Delilah’s PMM is rational, we’re not just saying it’s immune to rational criticism. We’re saying that it’s positively rational—that she has some good reason to have that desire. Delilah’s case may sound strange. Does any actual person ever take themselves to know all of the relevant non-moral facts and still engage in moral inquiry by deliberating or seeking out pure moral testimony? Indeed, pure moral inquiry of this kind is perfectly familiar. When people find themselves puzzling over whether it’s okay to get an abortion, whether they’re giving enough to charity, or whether they should stop eating meat, they aren’t always just wondering about how the world is in non-moral respects. They often take themselves to know all of the relevant non-moral facts. Likewise, people often go to an impartial third party to tell them how to resolve a dispute, even when the third party knows no more about the dispute than they do. In such cases, people rationally choose to engage in moral inquiry, and the only plausible explanation of why they do so is that they have PMM. So Delilah’s case is not all that unusual. PMM is common and often rational.

6.1.2. Pure Moral Inquiry with Conceptual Limitations Painfree has never felt pain before. He has just come across an opportunity to torture some puppies, and he thinks this would be fun. But he stops to ponder whether it would be right to do this. As he reflects, the village’s moral expert swoops in and informs him that there is an experience he has never had—pain—whose nature bears on whether it’s right to torture puppies. (Suppose Painfree already has excellent reason to believe the village’s moral expert is a moral expert.) The moral expert cannot tell Painfree anything about the intrinsic nature of the experience. But the ⁹ While cases where it’s rational to desire that one act (or be disposed to act) irrationally are familiar from the literature on rational irrationality, I know of no cases in which someone acts rationally (i.e., responds adequately to their reasons) but in so doing acts on an irrational desire. ¹⁰ Although my argument does not depend on it, I also believe Delilah is praiseworthy (cf. Smith, 1994; Arpaly, 2002; Weatherson, 2014).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

The Puzzle of Pure Moral Motivation

127

expert does tell Painfree that what the experience is like bears on whether it’s right to torture puppies. Unfortunately, just before the expert can tell Painfree how it bears on whether it’s right, the expert gets hit by a bus. Suppose then that the village sadist—having overheard Painfree’s conversation with the expert—walks up and offers Painfree the opportunity to learn what pain feels like, free of charge. After deliberating, Painfree decides to accept the sadist’s offer. Why would Painfree do this? He isn’t trying to figure out whether “pain” refers to something he antecedently cares about— he doesn’t know anything about pain, so he hasn’t any opinion about it. In accepting the sadist’s offer, then, he must be trying to figure out whether “pain” refers to something that it’s wrong to cause. Why does he want to know this? He must want to know this because he wants to do what’s right. Now suppose that Painfree does not think that doing what’s right as such would promote the satisfaction of any of his other desires. It follows that his desire must be a final desire. Because a final desire to do what’s right is an instance of PMM, this shows that Painfree must have PMM. It also shows that Painfree’s PMM is rational. Why? Because Painfree acts rationally in accepting the sadist’s offer. And because Painfree acts on PMM in accepting the sadist’s offer, his PMM must be rational. So we have another case where PMM is rational. While Delilah’s situation seems unusual, Painfree’s situation might appear fantastical. He is not merely uncertain which of various non-normative ways the world might be is actual. There are certain non-normative ways the world might be that he cannot even conceive. His ignorance is not empirical; it’s conceptual. Are we ever in Painfree’s position? Yes. This occurs whenever we know there is some morally relevant non-moral fact whose nature we don’t know. For example, one might think that how morally responsible an addict is for the wrongs they commit in order to obtain drugs is partly a function of what it’s like to be an addict, and that one cannot know this without having been an addict. Likewise, perhaps there are morally relevant facts about living through war or being oppressed that one cannot even imagine until one has been to war or been oppressed. In cases like these, we regularly defer to individuals who know morally relevant facts that we cannot even conceive. We defer to recovering addicts on the question of how to treat people who are in the grip of an addiction. We defer to war veterans on how we should treat prisoners of war. And we defer to people who are oppressed on how they should be treated. In so doing, we are acting on a final desire to do what’s right as such—we are acting out of PMM. And just as Painfree is rational to defer on the basis of PMM, so are we. At this point, I take myself to have provided compelling evidence that there are circumstances in which it’s rational to have PMM. I assume in

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

128

Adam Lerner

what follows that any adequate metaethics ought to account for this fact— and I argue that it’s surprisingly difficult to see how any of the standard views in metaethics could. Those who remain unconvinced by the arguments in this section can interpret what follows as an argument for a conditional claim: if PMM is rational, then it’s hard to see how any of the standard views in metaethics could accommodate that fact.

6.2. NON-COGNITIVISM According to cognitivism, moral judgments are (genuine, full-fledged) beliefs. They’re in the business of describing reality, of attributing moral properties to actions, people, and states of affairs. They can be true or false in a non-deflationary, non-minimalist sense. Non-cognitivism is the denial of cognitivism. Non-cognitivism comes in many flavors. But all pure versions of noncognitivism share one feature in virtue of which they are incompatible with the rationality of PMM: they deny that we have moral concepts that represent genuine moral properties.¹¹ This shared assumption makes it impossible for non-cognitivism to explain how it could be rational to have PMM since it implies that it’s not possible to have PMM in the first place.¹² The reason why is that PMM is a final desire to (e.g.) do what’s right as such—to do the right thing under that description—and it is thereby partly constituted by a moral concept that represents a genuine moral property. Since non-cognitivism denies the existence of such concepts, it must deny the existence of PMM. This shows non-cognitivism cannot solve the puzzle of PMM, narrowly construed. But it does not show that it cannot solve the puzzle of PMM, broadly construed. The puzzle of PMM, broadly construed, is to explain how it can be rational to engage in (pure) moral inquiry. I argued that we must assume people act on PMM because this provides the best explanation of why people engage in moral inquiry. The non-cognitivist denies the existence of PMM. In its place, they must find some other mental state— let us call it non-cognitivist PMM or NCPMM—that it’s rational to have and ¹¹ Even if the non-cognitivist can “earn the right” to speak of such concepts, a commitment to these concepts does not figure in their ground-level account of moral thought and talk (Blackburn, 1998). ¹² For purposes of this chapter, I assimilate hybrid versions of non-cognitivism (e.g., Ridge, 2014; Toppinen, 2013; Schroeder, 2013) to the naturalistic views I go on to discuss. While they may be able to account for the possibility of PMM, it’s unclear whether they can account for the rationality of PMM. Thanks to Mike Ridge for discussion.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

The Puzzle of Pure Moral Motivation

129

that can play the same role as PMM in explaining why people engage in moral inquiry. Only by identifying such a state can they solve the puzzle of PMM, broadly construed. In order to evaluate whether NCPMM satisfies these two desiderata, we need a better idea of what it could be. The non-cognitivist has two options. They can either identity NCPMM with a desire that has exclusively nonmoral content, or else they can identify it with a motivational state that lacks representational content altogether. I will consider each strategy in turn. I will argue that while both can explain why people engage in moral inquiry, neither can explain why people are rational in doing so. On the first strategy, the non-cognitivist must find some final desire whose content doesn’t involve a moral property, and this desire must be able to explain why we engage in moral inquiry. I submit that the only motivational state that satisfies this job description is a final desire to engage in moral inquiry. If the non-cognitivist attempted to identify NCPMM with any other desire (e.g., the desire to do what one would want to do after engaging in moral inquiry or to do what one’s true self wants to do), there would be overwhelming pressure to identify the content of this desire with doing what’s right. And this would just be to abandon non-cognitivism in favor of some form of naturalism. So the only desire the non-cognitivist can identify with NCPMM is a final desire to engage in moral inquiry. If the non-cognitivist chooses to identify NCPMM with a final desire to engage in moral inquiry, then they will have shown that NCPMM can explain why people engage in moral inquiry. But in order to solve the puzzle of PMM, broadly construed, they must show that it’s rational to have this state. And in fact, it doesn’t seem rational to have this state. While it might be rational to have an instrumental desire to engage in moral inquiry—as a way of doing something intellectually stimulating, or as a way of increasing the probability of doing what’s right—it isn’t rational to have a final desire to engage in moral inquiry. Now consider the second strategy. On this strategy, the non-cognitivist identifies NCPMM with a motivational state that has no representational content whatsoever and yet can still explain why people engage in moral inquiry.¹³ Such a state would have to be a brute disposition to φ when one has the belief that φ-ing will increase the probability that one does what’s right.¹⁴ If the non-cognitivist chooses to identify PMM with such a brute disposition, they will have shown that NCPMM can explain why people

¹³ Thanks to Eric Hubble for offering this line of response in conversation. ¹⁴ Or the categorical basis of such a disposition.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

130

Adam Lerner

engage in moral inquiry. But again, in order to solve the puzzle of PMM they must show that it’s rational to have this state. And it doesn’t seem rational to have this state, because it doesn’t seem that this brute disposition is the kind of state that it can be rational or irrational to have in the first place. While some brute dispositions might be rational to have, this doesn’t seem to be one of them. Perhaps it’s rational to have the brute disposition to believe P when you believe Q and Q is good evidence for P. But the brute disposition to engage in moral inquiry when you believe that doing so will lead you to do what’s right is not like a disposition to respond appropriately to one’s reasons. It’s more like a pure association, a mere tendency to pass from one mental state to another. While it might be rational to try to cultivate in oneself brute dispositions of this sort, the dispositions themselves are not rational or irrational. There is no right kind of reason to have states like these. But suppose for the sake of argument that it’s rational to have this brute disposition. If this were true, then the non-cognitivist would have found a state that it’s rational to have and that can explain why people engage in moral inquiry. But that is not yet to solve the puzzle of PMM, broadly construed. In order to do that, the non-cognitivist must show not only that having this brute disposition but acting on it—i.e., engaging in moral inquiry—is rational. I will now argue that they cannot do this. If we grant that it’s rational to have the disposition in question, what obstacle is there to concluding that it’s rational to act on this disposition? Recall that the disposition in question is a disposition to φ when one has the belief that φ-ing will increase the probability that one does what’s right. I will argue that once we have a better understanding of what this belief is according to non-cognitivism, we will see that acting on this brute disposition is not rational. I have called the state in question a “belief.” But this belief appears to have moral content. As such, the non-cognitivist cannot take it at face value. They must reduce it either to a genuine belief with non-moral content, or else to a pro-attitude of some kind. On either option, the moral inquirer comes out looking irrational. First, consider the possibility that the “belief ” in question is a proattitude. As a pro-attitude, it must be a pro-attitude toward something or other. And it seems the only thing that could be the object of this proattitude is the act, whatever it is, that one would perform after engaging in moral inquiry. This makes the pro-attitude into a final de dicto desire to do what one would do after engaging in moral inquiry. Unfortunately, a final de dicto desire to do what one would do after engaging in moral inquiry and other desires like this are not desires it’s rational to have. As I argue in Section 6.6 against a version of analytic naturalism, it can often make sense

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

The Puzzle of Pure Moral Motivation

131

to have an instrumental desire to do what one would do after engaging in moral inquiry, but it’s bizarre to want to do whatever one would do after engaging in moral inquiry for its own sake. So if the non-cognitivist reduces the belief in question to a pro-attitude, they cannot vindicate the rationality of engaging in moral inquiry, and so fail to solve the puzzle of PMM, broadly construed. Now consider the possibility that the belief in question is a genuine belief with non-moral content. If so, it would have to be the belief that engaging in moral inquiry increases the likelihood that one will give money to charity, or help an elderly person across the street, or . . . , where each disjunct is an action, naturalistically described, toward which one already takes the proattitude required by the non-cognitivist to count as judging that action to be right. The problem with this proposal is that it cannot explain why any rational person would bother to engage in moral inquiry. Why? Because it implies that people who engage in moral inquiry already have a settled view about which of their options they ought to take. But this makes people who engage in moral inquiry look irrational, for it means they already have an answer to the question they pursue in moral inquiry. Genuinely inquiring whether P when one is already certain that P is, if not impossible, irrational. So again, the non-cognitivist cannot vindicate the rationality of engaging in moral inquiry, and so fails to solve the puzzle of PMM, broadly construed.

6 . 3 . A N E P I S T E M I C C O N S T R AI N T In Sections 6.4 and 6.5, I evaluate synthetic naturalism and non-naturalism. In evaluating these views, I draw on the following principle: The Epistemic Constraint on Rational Final Desire. For any property P, if you do not know P’s essence, it is not rational to finally desire that P be instantiated as such.¹⁵

Why endorse this Epistemic Constraint? Suppose that I have never before experienced the taste of Vegemite. In fact, I have only just now learned that Vegemite exists. Upon learning that it exists, I find myself overcome with a ¹⁵ What is the essence of a property? For these purposes, the essence of a property is what makes that property the property it is. It is the property’s complete intrinsic nature. The complete intrinsic nature of a simple property is a matter of what it is in itself. The complete intrinsic nature of a complex property is a matter of what simple properties and relations it’s made up of, and what those properties and relations are in themselves. Importantly, necessary features of a property may not be part of its essence. It may be a necessary feature of goodness that God loves it without God loving it being part of its essence (cf. Fine, 1994).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

132

Adam Lerner

desire to taste Vegemite as such.¹⁶ And this isn’t just any desire. It’s a final desire. My rationale for wanting to taste Vegemite isn’t that doing so will increase the probability that some other desire of mine will be satisfied. My rationale is that it’s tasting Vegemite. Is it rational for me to have the Vegemite Desire? It isn’t. Why? Because I don’t know enough about what it’s like to taste Vegemite—I don’t know enough about the intrinsic nature of the experience. All I know about it is that it’s a taste experience. I need to know more—I need to know something about the Vegemite taste—before I can rationally desire it for its own sake. How much more? Suppose I learn a bit more about the intrinsic nature of Vegemite’s taste without learning its essence—I learn that Vegemite’s taste is, among other things, salty. Can I rationally have a final desire to taste Vegemite as such once I know this much? No. At best, I could rationally have a final desire to taste something salty as such. And this final desire is not the same as a final desire to taste Vegemite as such. To see why I’m only in a position to rationally have a final desire to taste something salty as such, consider how I would go about explaining why I desire to taste Vegemite. If asked why I desire to taste Vegemite, I could not rationally say that it’s because it’s Vegemite. The only rational answer I could give is that I desire to taste Vegemite because it tastes salty. And this reveals that my desire to taste Vegemite is not a final desire. It’s an instrumental desire based on my final desire to taste something salty and my means-end belief that tasting Vegemite is a way of tasting something salty. This same line of reasoning can be repeated for any version of the case in which my knowledge of how Vegemite tastes falls short of knowing its essence. Furthermore, it can be repeated for any case in which I desire something whose essence I don’t know. There’s nothing special here about the nature of subjective experience. Whenever I lack knowledge of an object’s essence, my desire for that object can only be instrumental if it is to be rational.¹⁷ It may seem that the Epistemic Constraint on Rational Final Desire makes rational final desire hard to come by. For example, many think that if it’s rational to have a final desire for anything as such, it’s rational to have a ¹⁶ Assume my concept of Vegemite directly refers to Vegemite. It’s in virtue of deploying this concept that I manage to have a desire to taste Vegemite as such rather than, e.g., a desire to taste what everyone actually refers to using the term “Vegemite.” ¹⁷ Experiencing the taste of Vegemite can allow one to know the essence of that experience only if property dualism is true. Even if property dualism is false, the Epistemic Constraint is compatible with the existence of a rational final desire in the vicinity: the final desire to have whatever experience I know under such-and-such a mode of presentation (i.e., the phenomenal character of tasting Vegemite).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

The Puzzle of Pure Moral Motivation

133

final desire for knowledge as such. But the lesson of the literature responding to Gettier’s seminal (1963) article is that we don’t know the essence of knowledge. So if the Epistemic Constraint on Rational Final Desire is correct, final desires for knowledge as such are not rational. Since they are rational, the Epistemic Constraint on Rational Final Desire faces an apparent counterexample. Whether this objection succeeds depends on whether we know the essence of what we finally desire. Take the case of knowledge. The failure of the Gettier literature doesn’t show that we don’t know the essence of knowledge. People have strong intuitions about whether knowledge obtains in Gettier cases. If this ability to determine whether knowledge obtains extends to all possible cases, then we seem to have at least tacit knowledge of knowledge’s essence. And this may be all that’s required to make final desires for knowledge rational. But even if our ability to determine whether knowledge obtains does not extend to all possible cases, we may still know the essence of knowledge in some other way, so long as its essence does not consist in what falls within its extension. If we don’t—if the essence of knowledge consists in what falls within its extension and we don’t know what falls within its extension— then the Epistemic Constraint seems to get the right result: it isn’t rational to finally desire knowledge. If it were, it would be rational for someone to finally desire to have knowledge in some case, to not know whether they have knowledge in that case, and to know all of the other facts about that case. But clearly someone who knew all the facts that could ground their having knowledge (whether their belief was true, reliably formed, etc.) and yet did not know whether they have knowledge could not rationally have a final desire that they have knowledge. This person would irrationally desire they know-not-what. Even so, there are other cases where it clearly seems rational to have a final desire for P as such even though one clearly does not know P’s essence. For example, most people want Mom to be happy as such. (Suppose “Mom” directly refers to one’s own mother.) Even supposing people know the essence of happiness, no one knows the essence of Mom. And yet people have a final desire that Mom is happy as such, and they seem to be rational in having this desire. This shows there is a class of exceptions to the Epistemic Constraint. What unifies this class? Notice that the rationale for the Mom Desire must involve facts about Mom’s extrinsic nature. If asked why one wants Mom to be happy, one’s reply must be something like “Mom gave birth to me” or “Mom raised me” or “Mom and I are pretty close.” These replies all invoke facts about Mom’s extrinsic properties. This suggests that you can be rational in having a final desire for an object when you don’t know

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

134

Adam Lerner

the essence of that object so long as you know the essence of that object’s extrinsic nature and the object’s extrinsic nature makes the object worth caring about for its own sake. Having noticed that there is this class of exceptions to the Epistemic Constraint, we can safely ignore it. In what follows, I argue that synthetic naturalism and non-naturalism cannot vindicate the rationality of PMM because the Epistemic Constraint is true and these views imply that we don’t know the essence of any moral property. Even though the Epistemic Constraint has exceptions, it isn’t plausible to resist this argument by saying that PMM constitutes such an exception. This is because facts about the extrinsic natures of moral properties play no part in any virtuous person’s rationale for having PMM. If asked why they desire to do the right thing, neither Delilah nor Painfree would cite anything about the extrinsic nature of rightness—they wouldn’t say “Because my mother wants me to do the right thing.” They would say “Because it’s right.” So even if there are exceptions to the Epistemic Constraint, PMM isn’t one of them. Although the arguments in Sections 6.4 and 6.5 are stated most succinctly using this principle, I will now offer a second, less controversial principle that can do the same work. This is the Epistemic Constraint on Rational Final Preference: for any two properties X and Y, if you cannot distinguish the intrinsic nature of X from the intrinsic nature of Y, you cannot remain rational while finally desiring the instantiation of X as such more than you finally desire the instantiation of Y as such. So if I have just learned not only that Vegemite exists, but that there is another substance— Marmite—that also exists, and all I know about the intrinsic nature of each is that they’re salty, then I cannot rationally desire to taste one more than the other. This result is eminently plausible, and its plausibility does not turn on anything specific about taste sensations. So the Epistemic Constraint on Rational Final Preference poses an equally severe problem for the standard views in metaethics: the problem of explaining why it’s rational to want to do what’s right more than what’s not right. 6. 4. SYNT HE TIC N A T U RA L ISM According to synthetic naturalism, moral properties are identical to natural properties, and statements identifying moral properties with natural properties are synthetic, a posteriori truths. The most prominent version of synthetic naturalism is Cornell Realism, and the most prominent defender of Cornell Realism is Richard Boyd. I will argue that our Epistemic Constraints show that synthetic naturalist views such as Boyd’s cannot easily account for the rationality of PMM.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

The Puzzle of Pure Moral Motivation

135

Boyd models his metasemantics for moral concepts on the externalist metasemantics for natural kind concepts championed by Kripke and Putnam. On this view, a concept has the referent that it does in virtue of standing in an appropriate causal relation to its referent. A key upshot of his view is that, just as we can have the concept of “water” without knowing that the essence of water is H₂O, we can possess a concept that refers to the property of being right without knowing the essence of that property: “we can and do refer to things such that we certainly don’t intend to refer to them under anything like the descriptions which in fact identify their true natures” (Boyd, 2003, p. 549). Moreover, we will not know the essence of the property of being right until we reach the end of moral inquiry: “The question of just which properties and mechanisms belong in the definition of [the property of being right] is an a posteriori question—often a difficult theoretical one” (Boyd, 1988, p. 197). Boyd takes this to be a positive feature of the view, for it enables it to accommodate Moorean open questions. As Moore observed, for any natural property N, it appears to be an open question whether anything that has that property is good. This is allegedly problematic for a naturalist view on which the identity of goodness with N is analytic. If the identity of goodness with N is settled by the meaning of “goodness” and “N,” how could people who are competent with the concepts expressed by these terms question the identity of goodness and N? Open questions are no embarrassment for Cornell Realism, for Cornell Realism predicts that competent speakers would find even true identity claims to have the appearance of being open. On this view, the identity of N with goodness is a synthetic, a posteriori truth, not a conceptual truth: “If the good is defined by a homeostatic phenomenon the details of which we still do not entirely know, then it is a paradigm case of a property whose ‘essence’ is given by a natural rather than a stipulative definition” (Boyd, 1988, p. 210). Because one can refer to the property of being right (or in this case, the property of being good) without knowing its essence, someone who hasn’t reached the end of inquiry can competently doubt that any given N constitutes its essence, even if it does. The problem for Cornell Realism should by now be obvious. The feature of Cornell Realism that allows it to accommodate Moorean open questions is ultimately a bug. It’s a bug because, together with the Epistemic Constraint on Rational Final Desire, it falsely implies that PMM is not rational. The argument is simple. According to the Epistemic Constraint on Rational Final Desire, it’s never rational to have a final desire for the instantiation of some property P as such if you don’t know the essence of P. According to Cornell Realism, no one knows the essence of the property of being right. (And no one will, until they reach the end of moral inquiry.)

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

136

Adam Lerner

Therefore, if Cornell Realism is true, it’s never rational for ordinary people engaged in moral inquiry to have a final desire to do the right thing as such. Since it’s rational for ordinary people engaged in moral inquiry to have this desire, this poses a serious problem for Cornell Realism. I have just argued against Cornell Realism relying on the Epistemic Constraint on Final Desire. But the Epistemic Constraint on Rational Final Preference poses an equally severe problem for Cornell Realism: that of explaining why it’s rational to want more strongly to do what’s right as such than to do what’s not right as such. To see why, consider that whatever the property of being right is, there will be a very similar property being right* which is just like being right, except that its essence doesn’t include φ-ing. Suppose we have an agent who is in a “hard case” and is thus unsure whether φ-ing is right. Suppose this person wants to do what’s right as such for its own sake. Now we can ask this person: why not desire to do what’s right* for its own sake instead? After all, doing what’s right* and doing what’s right appear to you to be indistinguishable. You may know that φ-ing lies within the essence of one and not the other, but you don’t know whether it lies within the essence of doing what’s right or the essence of doing what’s right*. So why, then, finally prefer to do the right thing rather than the right* thing? If the Epistemic Constraint on Rational Final Preference is true, what rationality requires in this case is indifference. But since it does seem rational to finally prefer doing what’s right over what’s right*, this poses a serious problem for Cornell Realism.

6.5. NON-NATURALISM Non-naturalism is a form of cognitivism, but it denies that moral properties are identical with natural properties. Instead, it takes moral properties to be a species of non-natural properties. Is non-naturalism compatible with the rationality of PMM? If non-naturalism is to allow for the rationality of PMM, then non-naturalism must avoid the problem faced by synthetic naturalism: it must allow everyday people to know the essences of moral properties. But it must also avoid a familiar objection to non-naturalism: that non-natural moral properties are not worth caring about for their own sake. These two desiderata must be satisfied if nonnaturalism is to solve the puzzle of PMM. Consider the second desideratum first. Some have thought that nonnatural properties cannot be worth caring about for their own sake. Here is a characteristic remark from Frank Jackson: Are we supposed to take seriously someone who says, “I see that this action will kill many and save no-one, but that is not enough to justify my not doing it; what really

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

The Puzzle of Pure Moral Motivation

137

matters is that the action has an extra property that only ethical terms are suited to pick out”? In short, the extra properties would [be] ethical “idlers”. (1998, p. 127)

Some non-naturalists reply to Jackson’s challenge by claiming that they never took non-natural moral properties to be worth caring about for their own sake in the first place.¹⁸ This reply, however, simply concedes that nonnaturalism cannot solve the puzzle of PMM. Fortunately, we can find nonnaturalists who believe non-natural moral properties are worth caring about. For example, in On What Matters Derek Parfit repeatedly claims that if there were no non-natural moral properties, “Sidgwick, Ross, and I, and others would have wasted much of our lives” (2011, pp. 12, 304, 367). But why think non-natural moral properties have essences that make them worth caring about? At this point, the best bet for the non-naturalist is to give up on describing the essence of being right and to just claim that the essence of being right, ineffable as it is, makes it worth caring about for its own sake. Just as someone pressed to explain why they desire pleasure for its own sake can do no more than point to the nature of the experience itself, someone pressed to explain why they desire to do what’s right for its own sake may be able to do no more than point to the nature of being right.¹⁹ “To know it is to love it,” they might say. However plausible this move is, it will not allow the non-naturalist to solve the PMM. And that is because non-naturalism fails the first desideratum considered above: it cannot explain how ordinary moral inquirers could know the essence of a non-natural moral property. In order to show that this desideratum is satisfied, the non-naturalist must show that it’s possible to stand in a special relation of acquaintance with that property, a relation that delivers knowledge of its essence.²⁰ Moreover, they must show that ordinary moral inquirers actually stand in this relation to non-natural moral properties. I will argue that standard versions of non-naturalism cannot show this. What would an essence-revealing relation of acquaintance look like? The most vivid example of such a relation comes from visual experience. Articulating a view he attributes to Bertrand Russell, Mark Johnston writes that “one naturally does take and should take one’s visual experience as of, e.g., a canary yellow surface, as completely revealing the intrinsic nature of canary yellow, so that canary yellow is counted as having just those intrinsic and

¹⁸ See, e.g., FitzPatrick (2014, p. 564). ¹⁹ Cf. Moore (1903, p. 7). ²⁰ The kind of acquaintance I have in mind corresponds to what Dasgupta (2015, p. 464) has in mind: “Let us say that a subject S is acquainted with x iff the nature of x is directly presented or revealed to S (this is just a label, not an analysis).”

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

138

Adam Lerner

essential features which are evident in an experience as of canary yellow” (1992, p. 223). Likewise, it’s plausible to think that the essence of Vegemite’s taste is completely revealed to someone who tastes Vegemite. The case of canary yellow provides a model for how the essence of a property can be revealed through experience. But standard versions of non-naturalism reject the idea that moral properties are the kinds of properties we can be acquainted with in conscious experience.²¹ So what other option is there? Perhaps standing in a special causal relation to an entity can give one knowledge of its essence, even if that causal relation doesn’t give rise to any conscious experience of the object’s nature. Unfortunately, standard versions of non-naturalism hold that moral properties are causally inert.²² So this proposal is also a non-starter for standard versions of non-naturalism. It would be too quick to conclude at this point that we don’t know the essence of non-natural moral properties. After all, it’s plausible to think we know the essence of certain logical relations and mathematical entities, but it isn’t plausible to think we’re causally connected to or immediately acquainted with them (Dasgupta, 2015, pp. 464–5; Chalmers, 2012, p. 404). So how does one come to know the essence of a logical or mathematical entity? A natural proposal is that we have this knowledge by having logical and mathematical concepts that play an appropriate role in our cognitive economy (Chalmers, 2012, p. 466). Can a similar proposal show that we know the essences of non-natural moral properties? Whether such a story could be told depends on whether an appropriate conceptual role can be found for our moral concepts. I will consider two proposals. On the first proposal, the appropriate conceptual role consists in being able to apply the concept in exactly those cases in which it’s correctly applied. This proposal holds that one might know a moral property’s essence by knowing the complete truth about what lies in its extension across all possible worlds. Unfortunately, this proposal is a non-starter, for the cases where PMM appears most clearly rational—cases of people engaging in moral inquiry—are precisely cases where people lack knowledge of what lies in the property’s extension. So non-naturalists who accept this proposal cannot deliver the verdict that PMM is rational.

²¹ For possible exceptions, see Oddie (2005), Atiq (unpublished manuscript), and Johnston (2001). ²² For exceptions, see Shafer-Landau (2012), Wedgwood (2007), Cuneo (2006), and Oddie (2005). Although these views allow for causal efficacy, more work needs to be done to show that this causal efficacy allows us to know the essence of any moral entity (cf. Langton, 1998; Lewis, 2009; Locke, 2009; and Dasgupta, 2015).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

The Puzzle of Pure Moral Motivation

139

On the second proposal, the appropriate conceptual role has less to do with our disposition to apply the concept and more to do with how we behave once we’ve applied it. Indeed, Ralph Wedgwood and David Enoch have offered such accounts in an attempt to explain how we can come to think about non-natural moral properties in the first place (Wedgwood, 2007; Enoch, 2011). Can such accounts also explain how we could know a moral property’s essence? No. On these views, a mental representation refers to a non-natural moral property in virtue of playing a certain “downstream” conceptual role. For example, on Wedgwood’s view, I count as competent with the concept of being what I ought to do just in case believing that I ought to φ commits me (in the relevant sense) to φ-ing. This is a very easy condition to satisfy. Indeed, it doesn’t require that I have any true beliefs about what I ought to do. I could believe the only thing I ever ought to do is torture puppies for fun and yet still count as competent with the concept of being what I ought to do, so long as in believing that I ought to torture puppies I commit myself (in the relevant sense) to torturing puppies (cf. Schroeter and Schroeter, 2009, pp. 4–9; Gibbard, 2003, pp. 28–9, 149–50). Even if Wedgwood’s view is correct as an account of how our thoughts manage to latch onto the non-natural moral property of being what I ought to do, the present case shows that this account cannot also show that we know that property’s essence. This is because it’s implausible to think that I could believe that the only thing I ever ought to do is torture puppies while at the same time knowing the essence of being what I ought to do. Although it’s plausible to think that knowing the essence of being what I ought to do is compatible with not knowing what has this property in hard cases, it isn’t at all plausible to think that someone could know that property’s essence while simultaneously believing that all they ought to do is torture puppies. Knowing a moral property’s essence can’t come that far apart from knowledge of what lies in its extension. I have argued that non-naturalism cannot vindicate the rationality of PMM because it cannot show how ordinary moral inquirers could know any moral property’s essence. This argument goes through only if the Epistemic Constraint on Rational Final Desire is true. But the same conclusion can be reached using the Epistemic Constraint on Rational Final Preference. The problem is that non-naturalism has a hard time explaining how we could learn how the essence of being right differed from, say, the essence of being right*. The typical routes by which we would come to know this difference are closed off: we aren’t acquainted with them in experience, and we don’t causally interact with them. And it’s hard to see how having a concept that plays the role specified by Wedgwood could tell us anything about the essence of being right—or at least anything that wasn’t also shared by being

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

140

Adam Lerner

right*. Lastly, we might know that the action we are inquiring about falls into the extension of the one but not the extension of the other, but we don’t know which—if we did, we wouldn’t have to deliberate. So even if we reject the Epistemic Constraint on Rational Final Desire, the Epistemic Constraint on Rational Final Preference creates an equally serious problem for standard versions of non-naturalism.

6 . 6. A N A LY T I C N A T U R A L I S M According to analytic naturalism, moral properties are identical to natural properties, and the relevant identity statements are analytic truths. Different versions of the view identify different moral properties with different natural properties, but all versions agree that the correct identities are analytic truths. The leading variant of analytic naturalism is moral functionalism, defended by Frank Jackson and Philip Pettit (Jackson, 1998; Jackson and Pettit, 1995). According to moral functionalism, the property of being right (e.g.) is the property of having the property that plays the rightness role in mature folk morality, and this is an analytic truth.²³ Unlike Cornell Realism and non-naturalism, moral functionalism allows that everyday people who are competent with moral concepts know the essence of the properties these concepts refer to. They may not be able to articulate a set of necessary and sufficient conditions that captures this essence, but they nevertheless have tacit knowledge of it. So moral functionalism succeeds where Cornell Realism and non-naturalism fail. But moral functionalism faces a problem of its own: although it allows ordinary inquirers to know the essences of moral properties, it takes the essences of these properties to be such that we have no reason to care about them for their own sake. This is because it identifies, e.g., the property of being right with the property of having the property that plays the rightness role in mature folk morality. According to Jackson, folk morality is “the network of moral opinions, intuitions, principles and concepts whose mastery is part and parcel of having a sense of what’s right and wrong, and of being able to engage in ²³ In their initial statement of the view, Jackson and Pettit remain neutral on whether rightness should be identified with the property of having the property that plays the rightness role (i.e., the property of having the realizer property) or with the property that plays the rightness role (i.e., the realizer property itself ). If we understand rightness to be the property of being right, then the latter interpretation makes moral functionalism into a version of synthetic naturalism, a version that fails for the same reason that Cornell Realism fails: it implies that ordinary moral inquirers don’t know the essence of being right.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

The Puzzle of Pure Moral Motivation

141

meaningful debate about what ought to be done” (Jackson, 1998, p. 130). Mature folk morality is “where folk morality will end up after it has been exposed to debate and critical reflection (or would end up, should we keep at it consistently and not become extinct too soon)” (Jackson, 1998, p. 133). So, roughly, the property of being right is the property we would treat as right if we took our ordinary sense of right and wrong and cleaned it up by thinking and arguing for a long time. Unfortunately, whether an act instantiates this property doesn’t seem to matter for its own sake. Why should we care for its own sake about doing what we would treat as right under such conditions? A satisfactory answer to this question would have to come by way of a richer description of what it would be to treat something as right under such conditions, which Jackson supplies. According to Jackson, “to believe that something is right is to believe in part that it is what we would in ideal circumstances desire,” and being what we would in ideal circumstances desire is part of what it is to have a property that plays the rightness role in mature folk morality (1998, p. 159). This suggests that to desire that we do the right thing is to desire in part that we do what we would desire if we were fully informed, coherent, etc. as such.²⁴ And doesn’t it make sense to care for its own sake about whether you’re doing what you would desire under such ideal conditions? No. It may make sense to care instrumentally about the fact that an act would be what we would in ideal circumstances desire—after all, this fact strongly suggests that the act has a distinct property that’s worth caring about for its own sake. But it makes little sense to care for its own sake about the fact that φ-ing would be what we would in ideal circumstances desire. At least, it doesn’t make sense to care as strongly about this fact as it does to care about whether we do what’s right. Caring so strongly about whether our act satisfies the desires we would have in ideal conditions for its own sake seems fetishistic. So if moral functionalism identifies our strong desire to do what’s right with a strong desire to do what we would in ideal circumstances desire, it can’t explain why it would be rational for us to engage in moral inquiry as frequently as we rationally engage in moral inquiry.

6.7. CONCLUSION I have argued that the standard views in metaethics all have trouble vindicating the rationality of PMM. Although the prospects for non-cognitivism

²⁴ A move like this makes moral functionalism into a type of ideal observer or ideal advisor theory (e.g., Firth, 1952; Smith, 1994).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

142

Adam Lerner

and synthetic naturalism are bleak, some plausible form of analytic naturalism or non-naturalism may yet be developed that can solve the puzzle of PMM. While it isn’t clear what that view will look like in all its details, we know in advance that it will have two features that no standard view has both of. First, unlike standard versions of synthetic naturalism and non-naturalism, it will have to allow that everyday moral inquirers know the essences of moral properties despite not knowing whether the acts they inquire about lie in their extensions. Second, unlike standard versions of analytic naturalism, the essences it assigns to moral properties must make those properties worth caring about for their own sake. It remains an open question whether there is an adequate view that can satisfy both of these constraints. But we have reason to be optimistic. Since each constraint is satisfied by at least one of the standard views, we know that each constraint can be satisfied in principle. And we have no reason to doubt they can both be satisfied by a single view. Suppose it turns out that these constraints can’t both be satisfied by a single view. We then have a choice. We can either conclude these are not genuine constraints, or we can take the fact that no view satisfies them as a new argument for error theory about morality. I myself would be inclined toward error theory. But that is an argument for another day.²⁵

References Arpaly, N. (2002). Moral Worth. Journal of Philosophy, 99 (5), 223–45. Atiq, E. H. (unpublished manuscript). Intuitionism, Humeanism, and the Metaphysics of Desire. Blackburn, S. (1998). Ruling Passions. Oxford: Oxford University Press. Boyd, R. N. (1988). How to Be a Moral Realist. In G. Sayre-McCord (Ed.), Essays on Moral Realism (pp. 181–228). Ithaca, NY: Cornell University Press. Boyd, R. N. (2003). Finite Beings, Finite Goods: The Semantics, Metaphysics and Ethics of Naturalist Consequentialism, Part I. Philosophy and Phenomenological Research, 66 (3), 505–53. Chalmers, D. (2012). Constructing the World. Oxford: Oxford University Press. Cuneo, T. (2006). Moral Facts as Configuring Causes. Pacific Philosophical Quarterly, 87 (2), 141–62.

²⁵ Thanks to audiences at Princeton University, the University of Michigan, the Central European University, and the 2016 Chapel Hill Metaethics Workshop. Special thanks to David Heering, Eric Hubble, Mark Johnston, Sarah McGrath, Joseph Moore, Mike Ridge, Gideon Rosen, Thomas Schmidt, Michael Smith, Nat Tabris, and two anonymous reviewers for extensive discussion and written feedback.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

The Puzzle of Pure Moral Motivation

143

Dasgupta, S. (2015). Inexpressible Ignorance. Philosophical Review, 124 (4), 441–80. Dreier, J. (2000). Dispositions and Fetishes: Externalist Models of Moral Motivation. Philosophy and Phenomenological Research, 61 (3), 619–38. Enoch, D. (2011). Taking Morality Seriously: A Defense of Robust Realism. Oxford: Oxford University Press. Fine, K. (1994). Essence and Modality. Philosophical Perspectives, 8, 1–16. Firth, R. (1952). Ethical Absolutism and the Ideal Observer. Philosophy and Phenomenological Research, 12 (3), 317–45. FitzPatrick, W. (2014). Skepticism about Naturalizing Normativity: In Defense of Ethical Non-Naturalism. Res Philosophica, 91 (4), 559–88. Gettier, E. (1963). Is Justified True Belief Knowledge? Analysis, 23 (6), 121–3. Gibbard, A. (2003). Thinking How to Live. Cambridge, MA: Harvard University Press. Harman, G. (2000). Desired Desires. In Explaining Value (pp. 117–36). Oxford: Oxford University Press. Jackson, F. (1998). From Metaphysics to Ethics. Oxford: Oxford University Press. Jackson, F. and Pettit, P. (1995). Moral Functionalism and Moral Motivation. Philosophical Quarterly, 45 (178), 20–40. Johnston, M. (1992). How to Speak of the Colors. Philosophical Studies, 68 (3), 221–63. Johnston, M. (2001). The Authority of Affect. Philosophy and Phenomenological Research, 63 (1), 181–214. Langton, R. (1998). Kantian Humility: Our Ignorance of Things in Themselves. Oxford: Oxford University Press. Lewis, D. (2009). Ramseyan Humility. In D. Braddon-Mitchell and R. Nola (Eds.), Conceptual Analysis and Philosophical Naturalism (pp. 203–22). Cambridge, MA: MIT Press. Locke, D. (2009). A Partial Defense of Ramseyan Humility. In D. BraddonMitchell and R. Nola (Eds.), Conceptual Analysis and Philosophical Naturalism (pp. 223–41). Cambridge, MA: MIT Press. Lord, E. (2015). Acting for the Right Reasons, Abilities, and Obligation. In R. Shafer-Landau (Ed.), Oxford Studies in Metaethics, vol. 10 (pp. 26–52). Oxford: Oxford University Press. McDaniel, K., and Bradley, B. (2008). Desires. Mind, 117 (466), 267–302. McGrath, S. (2009). The Puzzle of Pure Moral Deference. Philosophical Perspectives, 23 (1), 321–44. Moore, G. E. (1903). Principia Ethica. Cambridge: Cambridge University Press. Oddie, G. (2005). Value, Reality, and Desire. Oxford: Oxford University Press. Parfit, D. (2011). On What Matters: Volume Two. Oxford: Oxford University Press. Ridge, M. (2014). Impassioned Belief. Oxford: Oxford University Press. Schroeder, M. (2013). Tempered Expressivism. In R. Shafer-Landau (Ed.), Oxford Studies in Metaethics, vol. 8 (pp. 283–313). Oxford: Oxford University Press. Schroeter, L., and Schroeter, F. (2009). A Third Way in Metaethics. Noûs, 43 (1), 1–30. Shafer-Landau, R. (2012). Evolutionary Debunking, Moral Realism, and Moral Knowledge. Journal of Ethics and Social Philosophy, 7 (1), 1–37.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

144

Adam Lerner

Smith, M. (1994). The Moral Problem. Oxford: Blackwell. Toppinen, T. (2013). Believing in Expressivism. In R. Shafer-Landau (Ed.), Oxford Studies in Metaethics, vol. 8 (pp. 252–82). Oxford: Oxford University Press. Weatherson, B. (2014). Running Risks Morally. Philosophical Studies, 167 (1), 141–63. Wedgwood, R. (2007). The Nature of Normativity. Oxford: Oxford University Press.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

7 It Ain’t Necessarily So Nomy Arpaly

There is a view according to which virtue is the royal road to flourishing. Call it eudaemonism. Since I share with Anscombe (1958) the suspicion that it is an anachronism to say that Aristotle is concerned with specifically moral virtue, I should rather say that there are several different versions of the view; two important ones can be called Aristotelian eudaemonism and neoAristotelian eudaemonism. While one can doubt that Aristotle, whose ideal person has such excellences as wit and the ability to give good parties, is talking exactly about morality, the neo-Aristotelian eudaemonist (the “NAE” in what follows) writes in the twentieth or the twenty-first century, and is indeed concerned with moral virtue, whether or not she refers to it as such. The NAE sees herself as offering a competing theory to utilitarianism and Kantianism, and so a theory of the same subject matter: morality. Her virtuous person is disposed to do the right thing (though the right thing may also be the fine or noble thing) and usually, if she comes into some money, she spends it on helping the poor or fighting for human rights rather than on building a grand but tasteful house. Still, the NAE holds that the virtues— her kind of virtues, moral virtues—are the character traits jointly needed to live well. Moral virtue is the secret to flourishing. It is either the case that living virtuously constitutes flourishing or that the flourishing person is someone who lives virtuously and who also has a certain minimum of luck on top of that (is not, for example, the victim of a terrible disaster).¹ Neo-Aristotelian eudaemonism is both a normative ethical view and a metaethical view, as it gives at least a conditional answer to the question of why one should be moral: because being morally virtuous will give you the good life.

¹ Some current works in what I call neo-Aristotelianism include: Annas (2011), Badhwar (2014), Bloomfield (2014), and LeBar (2013). Not all virtue ethicists who are inspired in some way by Aristotle hold the view I am talking about (for example, Swanton (2003) does not).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

146

Nomy Arpaly

Doubts about the connection between virtue and flourishing are, of course, as ancient as virtue theory itself. Are there not some flourishing bad people and good people who are not flourishing? In response to the more obvious arguments in this vein, Hursthouse (2000) offers an analogy between virtue and flourishing, on the one hand, and health and healthful habits, on the other. Thus, to the observation that some virtuous people do not flourish Hursthouse has a compelling reply: Suppose my doctor said, “you would benefit from a regimen in which you gave up smoking, took regular exercise and moderated your drinking.” Her grounds are that this is the way to flourish physically, to have a long, healthy life. If, despite following her advice, I develop lung cancer or heart disease or my liver fails [ . . . ] this does not impugn the correction of what she said; I can’t go back to her and say “you were wrong to tell me I should quit smoking etc.—look, it hasn’t worked!”. She and I both know that doing as she says does not guarantee perfect health; nevertheless, if perfect health is what I want, the only thing to do is follow her advice and hope that I shall not be unlucky. Similarly, the claim is not that the possession of the virtues guarantees that one will flourish. The claim is that they are the only reliable bet—even though it is agreed, I might be unlucky and, precisely because of my virtue, wind up dying early or with my life marred or ruined.²

Hursthouse likewise acknowledges that some vicious people flourish, but deploys her analogy again: Does my doctor’s right answer to my question about how I should live claim that following the regimen she outlines is necessary for a long healthy life? No, because if it did, it would be readily falsified: the newspapers regularly describe the lives of people who achieved remarkable longevity [ . . . ] despite flouting at least one of the requirements that she had laid down (as I write, a splendid old lady in France had clocked up 120 years—and gave up smoking at 115). To claim that the virtues, for the most part, benefit their possessor, enabling her to flourish, is not to claim that virtue is necessary for happiness. It is to claim that no “regimen” will serve one better—no other candidate “regimen” is remotely plausible.³

Hursthouse’s analogy to health has a lot to recommend it. Alongside Hursthouse, I would argue that such a theory of the relation of virtue to flourishing is superior to the main alternative theory open to the NAE, a theory that identifies the virtuous life with the flourishing life. This view is subject to a number of serious, well-known objections that are worth reviewing, if only to make salient just how difficult it is to link the virtuous life to the good (flourishing, well-lived) life in a credible manner that differs significantly from the manner Hursthouse favors. ² Hursthouse (2000: 172).

³ Hursthouse (2000: 173).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

It Ain’t Necessarily So

147

Most strikingly, consider the person who acts virtuously, and genuinely is virtuous, but who also suffers from persecution, torture, illness, abject poverty, tragic losses, and any number of other horrible things. If her luck is bad enough, she is not flourishing, even if she is strong enough to keep living virtuously; to insist otherwise is both radically revisionist and insulting to the suffering virtuous individual. Such a person is not a problem for Hursthouse’s view, since her mere existence does not undermine the thesis that there is an appropriately reliable connection between virtue and flourishing, but the existence of even one person who lives virtuously but is not flourishing refutes the view that living virtuously is flourishing. Hursthouse sharply warns the virtue ethicist to avoid denying the harsh reality of non-flourishing virtuous lives by insisting that even the virtuous person who, say, lost all her children is really flourishing, though in a special way that only another virtuous person can discern.⁴ A concept of flourishing that has no affinity whatsoever with what people seek under such descriptions as “a happy life,” “a good life,” or “well-being” loses its philosophical usefulness. Hursthouse mentions a cartoon she has seen in which a woman having cheerful, drunken fun is looked at by some dour, gloomy-looking, pious character who pronounces her an “unhappy woman.” No particular fan of drunken debauchery herself, Hursthouse nonetheless takes the cartoon to be an effective criticism of any account of flourishing that has no place for “the smile factor,” that is, the normal signs and symptoms we associate with a person who is fundamentally happy with her life.⁵ A human who has lost her beloved children is deprived of this syndrome of happiness, and so cannot be living the good life. One can suggest otherwise by claiming, like McDowell (1979), that to a sufficiently virtuous person, even losing one’s children does not have the psychological effects that disasters have on ordinary people. This, however, is a thesis Hursthouse rejects in the same commonsensical way that Aristotle rejects the idea that a courageous man will altogether fail to find (heroic) death objectionable.⁶ This would not be the inner life of a reasonable human being, much less a virtuous one. And one might add that, in regard to moral virtue, in particular, the impressiveness of virtue in action involves the virtuous person’s willingness to surrender or risk her well-being so as to do the right thing.⁷ This is compatible with Hursthouse’s optimism about virtue being the best “regimen.” It is, after all, credible that she who would guard her well-being so jealously as to never accept any loss would live a life of lesser well-being. However, the commonsensical claim that the virtuous

⁴ Hursthouse (2000: 182–4). ⁶ Hursthouse (2000: 183).

⁵ Hursthouse (2000: 185). ⁷ See Seidman (2005) among others.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

148

Nomy Arpaly

person is a person who sometimes would relinquish her well-being is not compatible with the view that one is immune to any loss of well-being so long as one lives virtuously. Other reasons to reject the identification of flourishing and the virtuous life concern the appropriateness of emotions. As Brad Hooker (1998) has pointed out, compassion is an appropriate response to someone whose life is going badly, but compassion is hardly the appropriate response to hearing that someone is deeply dishonest, unkind, or otherwise vicious. It would sound strange if I were to respond to the news that a stranger has been cheating or stealing with “oh, the poor thing!” The news that a stranger has cancer often leads to expressions of sympathy, but the news that a stranger is an asshole (if I may be permitted to add that term to the philosophical lexicon) does not. Yet, if the flourishing life is the virtuous life, then the vicious life is a life going badly, and surely compassion would be apt. And then, there is the complementary truth to that discovered by Hooker. Just as compassion is not an apt response to the news that an acquaintance is a thief, moral disapprobation is not, in general, an apt response to the news that an acquaintance is utterly miserable. If told about someone’s good character and good deeds and then informed that she is utterly miserable, people say that it is very sad that a good person has to suffer. They do not stare confusedly and say, “Just a moment, I thought you said she was a good person!” They implicitly recognize the space that exists between virtue and flourishing, and theory should follow suit. For these sorts of reasons, it seems that the identification of the virtuous life and the flourishing life will not do for the contemporary NAE. Better to work with Hursthouse, and her view that the virtuous life is the best “regimen” for a person who desires to flourish. Of course, Hursthouse’s theory of the relation of virtue to flourishing can be the best available to an NAE and still be untenable. Showing that it is indeed untenable is the work of the rest of this chapter. The conclusion I aim to defend is that Hursthouse, too, fails to offer a credible defense of the claim that the good, happy, well, flourishing life either is or is most effectively achieved through the life of virtue. It should catch the reader’s attention that, when Hursthouse describes the way in which the virtuous life conduces to flourishing, she contrasts the virtuous life’s effects on flourishing with the effects of the life of wickedness.⁸ She tells the reader of Nazis who ran away to Argentina, for example.⁹ Similarly, throughout the relevant chapter she brings up the failure to flourish that we (but not the “immoralist”) expect from people who are ⁸ Hursthouse (2000: 163–91).

⁹ Hursthouse (2000: 173).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

It Ain’t Necessarily So

149

utterly dishonest and selfish, living lives of crime, interested in nothing but money and power, and treating other people as mere objects to be controlled and exploited. To have a viable argument against her view, she says, one needs to find a pattern “that we can all see in life” according to which “evil” regularly triumphs over virtue.¹⁰ However, as Copp and Sobel (2004) point out, the virtuous and the wicked are not the only people about. I would not presume to add up just how many genuinely vicious people can be said to flourish, and, for the sake of the argument, I am willing to assume that flourishing villains are as rare as smoking centenarians. I will also happily grant what seems plausible in general: that serial killers have emotionally wretched lives and that the lives of most common criminals are nothing to aspire to if one seeks happiness. I will also grant that the lives of the purely self-centered, even if they are not criminals, are at least likely to be barren. The word from empirical psychologists of happiness appears to be that people who are satisfied with their lives are very likely to have interests beyond money, power, the physical pleasures, and happiness itself, and this is consistent with our ordinary experiences. But the person of interest for this chapter is neither the criminal nor the purely selfish agent. It is the morally mediocre person, the person who has some measure of virtue but not very much. The morally mediocre person is most of us, for most of us are people who lie somewhere between being a decent person and being a “jerk”: not many of us are saints or heroes, but not many of us are unrepentant assholes either. The flourishing of the morally mediocre person is the topic that Hursthouse and other NAEs should consider more closely. Does the morally mediocre person have strong reason to become more virtuous, insofar as she is pursuing a life of abundant well-being? The person who takes mediocre care of her health has strong reason to cut out those cigarettes she only smokes on the weekends while out drinking, to take the stairs more often and the elevator less often, and so on, insofar as she pursues good health. Is the morally mediocre person who wishes to flourish in the same position? In the spirit of Hursthouse’s challenge to those who think vice is a better route to flourishing than virtue, I would like to pose the following challenge to the NAE: there appears to be a straightforward and reliable path to the good life for many people of mediocre morality. Can the NAE theorist show that this is not so, and that virtue is a generally better bet than the mediocre life, for those who would live well? Talk of mediocrity immediately brings into view questions about the line between being ordinarily virtuous and being extraordinarily virtuous: a saint

¹⁰ Hursthouse (2000: 174).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

150

Nomy Arpaly

or a hero of some sort. Am I asking whether the life of the ordinarily virtuous person is more likely to be a flourishing life than the life of the moral hero? Or am I asking whether the life of the ordinarily virtuous person is no more likely to be a flourishing life than the life of the person who falls a bit too far short of virtue to be counted as “ordinarily virtuous,” without yet being a life of unqualified vice? To keep the discussion interesting, and avoid certain unnecessary controversies, I will soon focus this chapter on the second of these two questions. But which question is asked, and so which challenge is raised, does not matter as much as it might seem at first. If the relation of moral virtue to flourishing is like the relation of healthful habits to health then one cannot, as far as flourishing goes, be too morally virtuous. While one can exercise more than is required for one’s health, in fact so much that it is harmful to one’s health, one does not reliably become less healthy because of habits that are too perfectly healthy. Any exercise in excess of what is generally good for one’s health is not too perfectly healthy, or even healthy at all: it is unhealthy, at that point. One can overvalue health, but insofar as what one seeks is health, there is no such thing as excessively acting on healthful habits. Likewise, insofar as one seeks to flourish, there is no such thing as being excessively virtuous, on the Hursthouse-style approach to the relation between virtue and flourishing. Thus, one can challenge the Hursthouse-style NAE by arguing that the life of heroic virtue is not reliably happier than the life of ordinary virtue just as much as one can challenge her by arguing that the life of ordinary virtue is not reliably happier than the life of moral mediocrity. This line of thought can be sharpened a little further by considering the analogy to health in a little more detail. While it might not change our view of smoking substantially if it turned out that smoking a cigarette once a week is not very harmful, it would change it quite a bit to discover that smoking a cigarette once a week is actually beneficial for one’s health, that is, to discover that the most reliably healthy course of action is to smoke cigarettes occasionally. For most NAE theorists, conceding that the moral equivalent of one cigarette a week can contribute to one’s flourishing is conceding too much: it is conceding something like the thesis that, as far as flourishing is concerned, there is such a thing as being too close to perfect moral virtue. And actually it is worse than that: it is conceding that what looks like moral heroism or moral sainthood, that is, any substantial disposition toward impressive moral supererogation, is in fact a moral vice. After all, the NAE thinks that being virtuous is the best route to flourishing, not that being virtuous-but-not-too-virtuous is. Thus, the NAE does have a problem if moral sainthood and heroism do not have something more to recommend, in terms of flourishing, than ordinary virtue. However, as people of genuinely heroic or saintly virtue

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

It Ain’t Necessarily So

151

are so rare, and since the precise demands of such virtue are so disputed, it is perhaps unwise to rely too heavily on our suspicions about how happy their lives are likely to be when compared to people of more ordinary levels of kindness, equitableness, and so on. So for the remainder of the chapter I will set aside the question of the statistically expected flourishing of the person of heroic or saintly virtue, and focus instead on the putative gap in happy living between the person of virtue and the person of ordinary moral mediocrity. While I will not try to draw precise lines between moral mediocrity and merely imperfect virtue, I take moral mediocrity to involve having a character conducive to reliably taking morally wrong courses of action on some ordinary occasions, but not conducive to committing truly evil actions on any ordinary occasion. The morally mediocre person is more flexible with the truth than the person of ordinary virtue, but would not conspire to send a business rival to prison; she does not qualify for the virtue of generosity, but she is willing enough to buy a box (just one) of fundraiser cookies now and then and is not particularly likely to leave someone to die after a terrible car accident. And so on. It is clear that in order to make my point, I should do more than indicate the cases where a morally mediocre person’s less-than-fully-virtuous disposition happens to do her good. As per Hursthouse, that one centenarian survived a life of smoking cigarettes does not mean that smoking is good for your health. Similarly, if one’s tendency to be late for work causes one to meet a person with whom one will later fall in love and have a relationship that will contribute to one’s flourishing, this does not suggest that lack of punctuality leads to flourishing as a matter of course. It is also not enough to point out cases where a generally virtuous person benefits from a bad action that is out of character, as the question here is whether a mediocre character can be as or more conducive to flourishing than an ordinarily virtuous one. Care should be taken, though, to remember that not every rare or surprising action is out of character for the agent. A normally loyal and faithful woman who cheats on her romantic partner after she gets drunk with an old flame to whom she thought she was no longer attracted (and we can add: her getting drunk is an understandable reaction to enduring a frightening and sleep-deprivation-inducing fire in the hotel both were staying in for a work conference, and so on) may well be acting out of character on that occasion. On the other hand, imagine a sober, well-rested agent who had never cheated on her spouse before but who always maintained that if she finds a more “exciting” match she is entitled to cheat, as “you only live once.” Such an agent does not act out of character when, after many years of finding her spouse “exciting” enough, she meets a more thrillingly romantic person and follows her old conviction by having a fling with her.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

152

Nomy Arpaly

So a case that is meant to trouble the Hursthouse-style NAE can make use of rare or occasional, but not out-of-character, morally wrong actions performed by the person of mediocre morality, so long as it is convincing that the rare or occasional actions contribute in a statistically reliable manner (and not through lucky accident) to the flourishing, well-being, and general happiness of that person. Before considering the happily mediocre person, though, look first at three things other than moral virtue that seem tightly connected to the ability to live a good life: concern for one’s own well-being, caring about and loving other individuals and having good relationships with them, and the passionate pursuit of things (other than relationships with other people) that one values intrinsically but not in a moral way. First, consider concern for one’s own well-being. At first blush, it seems reasonable that concern for one’s own well-being would be non-accidentally connected to well-being—and having well-being, unless viewed simply as feeling pleasure, is tightly connected to living a good life or flourishing, if not identical to it. On second thought, though, reasonable doubts might arise. It is often suspected that the pursuit of well-being is paradoxical, as well-being tends to be a side effect of caring about other things, such as one’s fellow human beings and “rewarding” projects or passions. People whose sole aim is well-being, it is said, or who would never sacrifice well-being for anything, or who spend a great deal of their time thinking about their wellbeing, tend not to be satisfied with their lives. There is surely some truth in all of this. Still, being almost single-mindedly concerned with one’s wellbeing is not the same thing as simply making one’s well-being a significant priority. That this is conducive to well-being can be hard to see at first, because most people are significantly concerned with their well-being, and some are all too concerned, but the advantages of concern for one’s wellbeing become evident when one encounters people who do not seem committed to their well-being, or whose concern for their own well-being is weakened by depression or by their culture. Such people stay in situations that make them utterly miserable even if a little bit of effort could help them leave these situations. They let themselves be exploited by family and strangers alike for poor reasons. If lucky enough to be able to choose a career, they might still gravitate to the careers their parents tell them to choose even though they know it means emotional death to them. If you find yourself the friend of such a person and wish her to flourish, you wish she would care about herself, that she would give more weight to her own well-being in her choices. Second, the observation that non-selfish relationships with specific people—friendships, romances, parenthood, and the like—are part of what makes a person flourish is an old truism that is too wise and too obvious to be bolstered by any philosophical argument I can conjure.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

It Ain’t Necessarily So

153

Third, it is only a little less of a truism that a good life involves having some other pursuits, pursuits that might be artistic or athletic, culinary or mechanical, intellectual or sensuous, but which are both valued by their pursuers for themselves (not for how they make one feel, for instance) and that expand one’s world beyond morality, the self, and one’s loving relationships. One tries to do one’s work as well as possible, or is loyal to one’s beloved sports team, or one works to save the cheetah: all of these things and thousands like them can contribute to one’s well-being if one cares about them intrinsically first. Some philosophers, such as Susan Wolf (2010), would add that the things valued intrinsically should be in fact valuable, but even among these it is not generally held that they need to be morally valuable. Many of the pursuits one can think of fulfilling this role in a person’s life are related to things that “objective list theories” of well-being put on their lists of things required for well-being, though I suspect that some are not.¹¹ As Wolf would say, even if no such pursuit is a “necessary element of a life welllived,” a life with none of them can be “strangely barren” (Wolf 1982). It is easy to see that concern for any of these three things can potentially conflict with concern for the moral, right, and good. Conflicts between morality and self-interest abound, at least some levels of partiality toward one’s loved ones are morally impermissible (and occasions for such immoral partiality are not exceedingly rare), and Williams’s case of Gaugin deserting his dependents in order to realize himself as a painter (Williams 1982) provides a dramatic illustration of the conflict that can exist between morality and a person’s morally neutral pursuits (of course the conflict can come up in much less dramatic cases than that of Gaugin). How demanding one thinks morality is will shape one’s view as to how common or significant these three kinds of conflicts are, but no one, not even an NAE, should deny that these conflicts are possible. A Hursthouse-style NAE should and will, however, deny that these conflicts are systemic and deep. Since she holds that the (morally) virtuous life is the most reliable path toward happiness, she holds that conflicts between these three non-moral sources of well-being and morality are exceptions to a broad pattern (occasionally, but not reliably, it will turn out that immorally favoring a sibling will better promote one’s flourishing; and so on) or that the best bet for flourishing is to be the person who, in any such conflict, chooses morality over the non-moral source of well-being. In the case of loving personal relationships the NAE has something stronger to say—namely, that having such relationships requires moral virtue

¹¹ For example, see Parfit’s list (1984), which includes rational activity, awareness of beauty, and the development of abilities.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

154

Nomy Arpaly

(Hursthouse 2000: 168, 185–91). Friendships and rewarding romantic commitments are especially thought to require honesty, as they involve a great deal of trust, and involve such pleasures as self-disclosure and “being yourself ” with another. They also, of course, require the ability to be altruistic, which a person who is not virtuous obviously does not have. Unfortunately for the argument of the NAE, what is clear is that rewarding personal relationships are incompatible with wholesale dishonesty and pure selfishness. One does not need to be a radical skeptic on the subject of character in order to admit that it is possible for a person with a decisively mediocre record in matters of honesty and altruism to make exceptions for specific individuals and groups. It might be strange, in the same way that it is strange that a person who eats beef and pork can be selflessly devoted to her dog or cat (even dog and cat welfare generally), and yet the latter sort of strangeness is such a part of normal life that until recently very few people have noticed it at all. Even downright dubious individuals (who are not actual psychopaths) often have one or two people in their lives for whom they will act unselfishly and to whom they are loyal and substantially honest. This is both obvious to anyone who knows such people personally and is obvious through oblique measures, such as the fact that it is possible to blackmail and coerce such individuals by threatening these special relationships. The reader of this chapter might well have a friend who has such dishonest and selfish habits as stealing music (downloading it for free off the Internet), twisting and stretching the truth on tax returns, or being late submitting documents by firmly promised deadlines, but the fact that some people have all of these three habits and little or no remorse about them has never made a dent in my trusting them to be good friends to me. Likewise, many people who think eating meat is wrong do not for a moment doubt the loyalty, selflessness, and interpersonal honesty of their meat-eating friends (even if these friends also steal music, cheat in a small way on their taxes, and so on). Some particular vices might be incompatible with some relationships (as when sexism ruins a man’s ability to have rewarding relationships with women), but all in all I suspect that, even though moral perfection need not be as off-putting as Wolf (1982) takes it to be, some of us would find moral perfection (or, for that matter, bad aesthetic taste) a greater barrier to friendship than many types of moral imperfection. It is possible, then, to have non-selfish, honest, trusting, and otherwise good personal relationships without being even ordinarily virtuous. And, through parallel sorts of observations, it is just as obviously possible to be a morally mediocre person who has a modicum of concern for her own wellbeing and who intrinsically values a range of other pursuits. This profile seems to fit many of the flourishing people I know, and to be validated by

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

It Ain’t Necessarily So

155

many writers of fiction who concur that such people can and often do live well (sometimes to our delight, and sometimes to our dismay). For the sake of concreteness, imagine a specific character (with the admission that stories of people who seem happy with their lives are usually boring to read and often a touch annoying as well). I will tell his story at length. Frank has a great job at a high-tech company. He relishes it, partially because he loves the creative work and because he delights in having a much higher income than the very modest one he grew up on. His job involves some drudgery as well, but Frank is great at weaseling out of the drudgery, delegating and avoiding obligations, and making sure he has an interesting and enjoyable time in the office at the expense of his subordinates and other co-workers. He gets away with it because of his creative gifts, which make him hard to replace, and a touch of personal charm that makes people forgive him more often, and also through a bit of sneakiness. It was typical of him that, the last time it seemed he would have to go to a boring meeting, he called in sick, telling a direct lie, because his young daughter seemed unbearably cute that day, the sun was shining and he realized he hadn’t managed to spend enough time with her recently. He proceeded to spend the day playing with her instead. Some people would feel very guilty acting as he does. He wishes his good friend Rajiv wasn’t spending so much of his time worrying about fairness and obligations, for instance, but Frank, though he knows not to confess it, just does not feel guilty about this kind of thing. “What the heck, we only live once” is something he thinks fairly regularly, along with “what are they going to do, fire me?” He also knows better than to confess that he subtly helped himself get promoted by implicitly taking some of the credit belonging to his subordinates for his successes. Frank loves his partner, Felicia, his daughter, and his two close friends. He is as loyal to them as anyone can be loyal to friends, lovers, and children, and is willing to sacrifice for them. They know they can count on him. As he is richer than his friends, he often helps them a little in times of trouble and spoils them in better times, and has no qualms about using his influence to get them better jobs or carefully giving them small but morally questionable investment or taxavoidance tips (it is evident to him that many people in his position do “this kind of thing”). He relishes having the ability to make them happy—he didn’t have that sort of ability when young. His personal life has not always been as good as it has been in the last fifteen years. His first marriage, which he regards as the outcome of a mistake made when he was young, was deeply unsatisfying. His wife was a good person and accommodated him in many ways, but they did not have a deep communion of the minds, the physical spark did not last, and he came to the conclusion that she was not smart enough and too

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

156

Nomy Arpaly conventional for him. For a while he stayed, because he was comfortable and because it was clear that his wife, who seemed happy enough, would react very badly if he was to leave. He was not entirely oblivious to her potential pain, as he is only morally mediocre, not evil. He also did not want to waste time and energy on disorder, noisy fights, and legal struggles. During this time he took a lot of comfort in exciting extramarital affairs and became an accomplished liar for that purpose. Then he met Felicia, with whom he fell in love. Even when he realized Felicia was the partner of his dreams he was in no rush to get a divorce and create a mess, but she told him she could no longer stay with him if they needed to live a lie, and he made his choice without hesitation. In addition to work and to his loved ones, Frank is passionate about jazz music, which he plays and creates with his friends. That, too, is a priority for him, and that is a bit of a challenge given the kind of schedule he has. Because of the scarcity of time, Frank would never dream of stopping to help a person by the side of the road change a tire, for example, or staying late to help a young co-worker. Who has the time for such things? He gives a bit of money to a local society that promotes jazz music—he nearly cried with joy when one of his idols invited him to play onstage with him—but otherwise, he does not give to charities or causes except when there is a tax deduction to be had. He feels entitled to the money he has earned and almost required to catch up on the fun he did not have as a child and to provide it to his own child. Whatever he doesn’t spend on himself and loved ones he saves up: after all, he has some (not too bad, but some) anxieties to quiet about the possibility of being poor again. Frank refuses to worry about whether any ingredient in his food is morally objectionable. He loves food! It is so dreary to have a long list of dietary restrictions to complicate fun social events with. It’s almost as dreary as being the one who preaches to others in favor of such restrictions; he never listens. He refuses to worry about whether his clothes come from sweatshops, and if told that some luxury good he has comes from slave labor he inwardly shrugs. Lately, one of Frank’s close friends and some people in his wider circle have been worried about current events and have been suggesting to him that he might join them in fighting the rampant injustice they hear about in the news. Frank is not completely immune to being angry about some types of injustice, but this is exactly why he watches the news as little as possible. His friends tell him one must be evervigilant against evil, but they themselves admit that their psychiatrists tell them that constant vigilance is bad for you. One is told that the righteous sleep better, but that is patently false these days: those of his friends who care about causes such as combating racism or climate change seem bleary-eyed, pale, tense. One is told that they feel satisfaction, but to him they look frustrated. He wants to sleep well.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

It Ain’t Necessarily So

157

He doesn’t want the “negative energies,” as he puts it, that being politically conscious would give him. And since his friends have never known him to be a political person, this preference of his puts very little strain on their relationships. Not afraid to sound sentimental with his partner, he tells her honestly that he would not trade his life for anyone else’s.

My challenge question to the NAE is: does Frank have a clear, compelling reason, in so far as he wants to flourish, to change over to living virtuously? Of course, he might have reasons to be more moral that have nothing to do with his flourishing, perhaps reasons to be more moral at the expense of his flourishing, but would it be better for him if he compromised his “we only live once” policy of prioritizing three things—his well-being, the well-being of the people he loves, and perfection in software and music—and instead made more room in his heart for the suffering and rights of people he does not love? It is not clear to me that it would benefit him to press a button to become the sort of person who does not deceive anyone (bosses, his first wife, the government) without excellent cause, who is decent and generous with co-workers, underlings, and strangers, and who feels motivating righteous anger and worry in response to national and global injustice. It is even less clear that it would be worth it for him to put on the gargantuan effort typically required to change one’s character. It is true that Frank’s flourishing depends on some amount of luck. If, for example, his wife were to die or the computer industry were to collapse he would no longer flourish, and if he were born without talent or charm things might have gone differently. However, Hursthouse acknowledges that luck affects us all. Virtuous people need it too, and it is not at all clear that they need it less than Frank does. We see, for example, that Frank’s more virtuous friends are more susceptible to dips in mental health as a result of political upheaval than he is, and he has grasped economic security just ruthlessly enough to protect himself and those he loves from many changes of fortune. Given this world, in which a person’s luck can change every moment, it is still not clear that Frank should press that button for the sake of his own well-being. It is, of course, possible to bring testimonies of people who risked their lives to save Jewish people from the Nazis, or who have dedicated their lives to adopting and taking care of as many children with difficult-to-meet special needs as they could afford, and who have experienced a level of joyous satisfaction with their lives to the extent that they rarely, if ever, envied anyone, except maybe someone who did even more. However, to turn a Hursthousian question on its head, could these people not be the

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

158

Nomy Arpaly

equivalents of the “splendid old lady” who was over a hundred years old before she quit smoking? If Hursthouse’s medical analogy turns the question of the relationship between virtue and flourishing into a quasi-empirical one, the quasiempirical answer seems to be that we do not have clear evidence that the life of a morally mediocre agent who has loving personal relationships, intrinsically valued projects, and an average level of concern for her own interests is less “healthy” than that of a morally better agent, who has a devotion to the right and the good even to the extent that such a devotion leads to a different direction than devotion to self, loved ones, and pursuits. There might be an argument that the NAE can use, but it needs to go beyond tales of the worse-than-mediocre rotting in their personal hells, and beyond rare cases of happy moral saints. Here is one such argument. We tell our children to be honest, charitable, and so on in the same way that we tell them to eat their vegetables, Hursthouse points out, and when we do, we tell them just that: to be charitable, not just mostly charitable, to be honest, not just generally honest, and so on.¹² Hursthouse relies on the assumption that a parent tends to teach her child with an eye not (mainly) to the parent’s own benefit nor (mainly) to the benefit of society at large but rather to the child’s benefit. If parents tell their children to be virtuous the way they tell them to eat vegetables, that is a sign that virtue is regarded by them as good for the child, as eating her vegetables is. This is a powerful argument, but three things make me doubt that it ultimately works. One is the fact, mentioned by Copp and Sobel (2004), that when children grow up to be teens the advice they receive is often more mixed. A college student who wants to devote her life, or even just the next few years, to the greater good will often hear objections from her parents, who would prefer that she do something more financially prudent. Just recently, I have overheard the parents of a high-school student telling him pompously that he should be good but not “too good,” because “sometimes you have to be a little selfish. Sad but true.” Ask people who say things like “nice guys finish last” where they picked up their sayings and they will most likely attribute them to their parents. Furthermore, even if we only consider children who are not yet teens, it is easy to see that many parents are quite inconsistent in what they say to their children. Parents tell their children, often through stories and poems, to always be faithful and always be true. However, not a long time has to pass before they also tell the (sometimes confused) child such things as “tell them ¹² Hursthouse (2000: 175).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

It Ain’t Necessarily So

159

your dad isn’t home,” or “just tell them you weren’t feeling well.” Parents tell their children to be charitable and self-sacrificing, and yet they will also castigate a child who returns from pre-school having given away his coat to a less fortunate peer. In this context I always remember a story from my own childhood. The story, which was popular among Israeli kindergarten teachers, concerned a little girl walking about in her brand new, immaculate white dress. On the way home, the girl meets a poor, elderly neighbor struggling with a heavy bag of coal. She helps him, which results in dark stains on the beautiful dress. When she sees the stains, the girl is terrified that her parents will be furious with her for having made her dress dirty. At that point, an angel from heaven turns the stains of coal into shining gold. It was quite obvious to the children listening to the story that the gold was not simply a reward for the child—chocolate would have served that purpose better—but a way to protect her from her parents, who of course would have otherwise screamed their heads off at her for having stained the dress. There is no question that more parents would have penalized the girl for staining the dress than would have praised her for helping the poor man. I suspect that if the child of natural virtue were to say something along the lines of “but you said it’s good to help the poor,” most decent parents would soften up, but would still say something like, “It’s complicated. You’ll understand it when you grow up. Don’t do it again, OK?” Furthermore, the fact that parents tell their children to always be good (or true, or kind) does not imply that, given a button to press that would make the child a morally perfect (or even clearly above average) person, they would confidently press it. Ordinary parental lecturing takes place within a context, a context in which everyone knows that children will not in fact act as perfectly as their parents tell them to act but, at the very best, will act better following the lecturing than they would have acted if it weren’t for it. Children imitate a lot more than they obey, and the average mother or father is a morally average person and thus, of course, does not model perfect virtue. In short, the preaching parent is not really at any risk of creating a saint, and preaching perfect virtue is done not with the hope that it will make the child perfectly virtuous but rather merely with the hope that it will make her more virtuous than otherwise, or the fear that without it she will become a villain. What parents would do if they were to be given a magic button that they could press to make their child morally perfect is one of these questions too removed from ordinary life to answer with any confidence, but I doubt that it would be as popular as a button that would grant the child lifelong physical health, a stable and happy marriage, and skills that guarantee a good living. Ethicists and metaethicists have always been fascinated by the figure of the amoralist. I agree the amoralist and the somewhat less frightening

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

160

Nomy Arpaly

“wicked” person are fascinating figures, and it is worthwhile to think about them in the various ways philosophers do. However, when people who are not metaethicists, in real life and in fiction, ask why one must be moral, it is only rarely because they are considering becoming serial killers or even conscienceless sharks on Wall Street. More often they wonder why they shouldn’t be doing those less-than-virtuous things that one is sometimes told everyone does.¹³ Whether one is considering the view that moral virtue brings flourishing or another view according to which rationality and morality are tightly related, the morally mediocre can be as interesting and challenging as the amoralist and the wicked person can be.

References Annas, J. 2011. Intelligent Virtue. Oxford University Press. Anscombe, G. E. M. 1958. Modern Moral Philosophy. Philosophy 33(124): 1–19. Badhwar, N. K. 2014. Well-Being: Happiness in a Worthwhile Life. Oxford University Press. Bloomfield, P. 2014. The Virtues of Happiness: A Theory of the Good Life. Oxford University Press. Copp, D. and D. Sobel. 2004. Morality and Virtue: An Assessment of Some Recent Work in Virtue Ethics. Ethics 114(3): 514–54. Hooker, B. 1998. Does Moral Virtue Constitute a Benefit to the Agent? In R. Crisp, ed., How Should One Live? Essays on the Virtues. Oxford University Press. Hursthouse, R. 2000. On Virtue Ethics. Oxford University Press. LeBar, M. 2013. The Value of Living Well. Oxford University Press. McDowell, J. 1979. Virtue and Reason. The Monist 62(3): 331–50. Parfit, D. 1984. Reasons and Persons. Oxford University Press. Seidman, J. 2005. Two Sides of “Silencing.” Philosophical Quarterly 55(218): 68–77. Slote, M. 1992. From Morality to Virtue. Oxford University Press. Svavarsdóttir, S. 1999. Moral Cognitivism and Motivation. Philosophical Review 108(2): 161–219. Swanton, C. 2003. Virtue Ethics: A Pluralistic View. Oxford University Press. Vogler, C. 2002. Reasonably Vicious. Harvard University Press. Williams, B. 1982. Moral Luck. Cambridge University Press. Wolf, S. 1982. Moral Saints. The Journal of Philosophy 79(8): 419–39. Wolf, S. 2010. Meaning in Life and Why It Matters. Princeton University Press.

¹³ The non-virtuous non-amoralist is rarely discussed by ethicists—Svavarsdóttir (1999) is an exception. In the context of virtue, Slote (1992) and Vogler (2002) come close to the topic.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

8 Moral Uncertainty and Value Comparison Amelia Hicks

8.1. INTRODUCTION Imagine that you describe to someone Singer’s argument for the conclusion that many of us are morally obligated to donate to charity.¹ The person to whom you present that argument can respond in several ways. First, they can accept the conclusion. Second, they can reject the conclusion. But a third possibility is that they are thrown into doubt about that conclusion. If they are thrown into doubt (and can’t resolve that doubt quickly), then they can practically respond to their doubt in different ways. On the one hand, they can maintain that they’re not going to donate until they’re convinced of the truth of Singer’s conclusion. But, on the other hand, they can reason in the following way: Either Singer’s conclusion is true or false. If it’s true, then I need to give more money away. But even if it’s false, no harm will be done by giving money away that I would have spent only on trivial things. So I’ll give more of my money away, even though I’m not convinced that I’m morally required to.

This type of reasoning raises two questions. First, because this type of reasoning might seem morally admirable, it raises the question: What are the moral norms governing deliberation under moral uncertainty? And second, because this type of reasoning might seem like a reasonable way of making decisions when one is morally uncertain, it raises the question: What are the rational norms governing deliberation under moral uncertainty? Although both questions are important, in this chapter I will only address the second. However, I won’t provide a complete answer to the second question. I have two aims. First, I aim to vindicate some instances of ¹ Singer (1972).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

162

Amelia Hicks

the type of reasoning described above. Second, I aim to motivate the idea that the second question is a meaningful, potentially fruitful question for philosophers to address. I will work toward these aims by arguing that moral uncertainty can, in fact, affect what it’s rational for one to do. In Section 8.2, I begin by clarifying the concept of moral uncertainty. Then, in Section 8.3, I begin my defense of: Moral Sensitivity: rational obligations are “sensitive” to (morally based) moral uncertainty. (That is, moral uncertainty can affect what it’s rational for one to do.)

I argue that when one is morally uncertain but (a) cares about acting in accordance with moral norms and (b) is presented with an action that “dominates” all other available actions, then one is rationally obligated to choose in accordance with the principle of dominance. In Section 8.4, I consider what one’s rational obligations are when one is morally uncertain but no available action dominates (“no-dominance” cases). I argue that, given the rational requirement to reason in accordance with dominance, it’s implausible to think that moral uncertainty never affects one’s rational obligations in no-dominance cases. I then describe the problem of value comparison, which allegedly undermines Moral Sensitivity. In Section 8.5, I argue that the problem of value comparison, as raised in the moral uncertainty literature, is an instance of a general puzzle about rationality. Other instances of the puzzle do not show that non-moral uncertainty doesn’t affect our rational obligations. Thus, this instance of the puzzle shouldn’t convince us that moral uncertainty doesn’t affect our rational obligations. In Section 8.6, I address the objection that my position relies on an overly optimistic view of the capacities of everyday decision makers. I conclude by surveying several questions raised by my position.

8 . 2. M O R A L U N C E R T AI N T Y Uncertainty is an epistemic state in which one’s credences are “split” between mutually exclusive propositions. In this chapter, I will assume that the type of uncertainty one experiences is determined by the type of propositions that one’s credences are split between. Thus, moral uncertainty is a state in which one’s credences are split between mutually exclusive moral propositions.²

² I will be sloppy in my use of terms like “belief ” and “credence,” because my argument can be put in terms of either.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Moral Uncertainty and Value Comparison

163

(The reader might object to this way of delineating types of uncertainty, on the grounds that there is no clear distinction between types of propositions. I’m sympathetic to this concern, but I will nevertheless assume a distinction between moral and non-moral propositions. This is because, if one dispenses with that assumption, then one can arrive at my conclusion— that Moral Sensitivity is true—easily. Thus, I will burden myself with that assumption, and show that even with that assumption one can still defend Moral Sensitivity.) There can be different reasons why one’s credences are split between mutually exclusive moral propositions. On the one hand, one might be morally uncertain because of prior uncertainty about a non-moral claim. For example, I might be uncertain about whether or not I should lie to someone, but only because I’m uncertain about whether or not I could lie convincingly—in that case, my moral uncertainty reduces to uncertainty about a descriptive claim. On the other hand, one might be morally uncertain because of prior uncertainty about another moral claim; I might be uncertain about whether or not to lie to someone because I’m uncertain about whether lying is intrinsically morally wrong. So, we must distinguish between descriptively based and morally based moral uncertainty. Because there are different types of moral propositions—propositions that express normative theories, axiological claims, and claims about particular moral obligations and permissions—we can further distinguish between different types of morally based moral uncertainty. In this chapter, when I use the phrase “moral uncertainty,” it will usually refer to morally based moral uncertainty about the true normative theory. However, my argument also applies to the other types of morally based moral uncertainty. The reason my argument is applicable to all forms of morally based moral uncertainty is this: I will argue that we should treat morally based moral uncertainty in the same way that we treat descriptively based moral uncertainty, because the source of moral uncertainty is irrelevant. Given that the source of moral uncertainty is irrelevant, it follows that the source of morally based moral uncertainty is also irrelevant. It’s not controversial that one should take one’s descriptively based moral uncertainty into account when deliberating—that’s an issue on which my opponents and I agree. For example, if one is uncertain about whether it’s permissible to shoot at a target, but only because one is uncertain about whether there is a toddler playing near the target, then most agree that one’s descriptively based moral uncertainty (combined with the fact that one doesn’t want to shoot a toddler) entails that one is rationally obligated to refrain from shooting, all else being equal. The focus of this chapter is whether morally based moral uncertainty affects what one rationally ought to do in the same way that descriptively based moral uncertainty affects what one rationally ought to do.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

164

Amelia Hicks 8.3. MORAL UNCERTAINTY AND DOMINANCE

To begin, I will provide some positive motivation for Moral Sensitivity, the view that morally based moral uncertainty can affect what it’s rational for one to do. I will argue for moral weak dominance, the view that when one is morally uncertain but one available action weakly dominates the others, then one is rationally required to choose the dominating action. (I will argue for this under the assumption that one cares about acting in accordance with moral norms. If one doesn’t care about moral norms, then one’s moral uncertainty doesn’t affect the rationality of one’s action. To see this, consider a person who doesn’t care about whether or not they get rained on; all else being equal, there will be no rational norms governing that person’s decision about whether to take an umbrella with them on their walk.) An action A weakly dominates another action B if and only if (1) under every state of nature, A yields an outcome at least as good as the outcome B yields and (2) under at least one state of nature, A yields a better outcome than B. Everyone agrees that one ought to comply with weak dominance when uncertain about descriptive propositions. For example, imagine that I’m trying to decide whether to take an umbrella with me to work. If it rains, then carrying the umbrella with me is better than leaving it at home. And if it doesn’t rain, the umbrella is nevertheless light enough that I won’t notice it. Thus, because it will either rain or not rain (those being the two possible states of nature), carrying the umbrella weakly dominates not bringing an umbrella with me; carrying the umbrella is sure to yield a result that is as great as or better than not bringing it with me. We can apply weak dominance to some cases of moral uncertainty (in which the propositions expressing the states of nature are moral propositions). So, for example, imagine that I’m uncertain about whether Kantianism or Utilitarianism is true, and I’m trying to decide whether or not to steal my neighbor’s television. It turns out that according to each theory, I shouldn’t steal my neighbor’s television—not stealing my neighbor’s television leads to an outcome morally as good as or better than stealing the television, no matter which theory is true. In that case, not stealing weakly dominates stealing. Thus, we can apply dominance-style reasoning when deliberating under moral uncertainty. But should we? Here is an argument that we should: 1. We should reason in accordance with weak dominance when uncertain about descriptive propositions (such as propositions about whether or not it will rain).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Moral Uncertainty and Value Comparison

165

2. Weak dominance is insensitive (for the most part)³ to the content of the propositions that express the states of nature. 3. So, if we should reason in accordance with weak dominance when uncertain about one type of proposition (e.g., descriptive), then we should reason in accordance with weak dominance when uncertain about propositions of other types. (2) 4. So, we should reason in accordance with weak dominance when we’re uncertain about moral propositions. (1, 3) I think that the conclusion expressed by (4) is correct. Some philosophers disagree with me about (4), but that’s because they are working with nonstandard—and, I think, unmotivated—definitions of dominance.⁴ So, we have our first piece of motivation for Moral Sensitivity: moral uncertainty affects one’s rational obligations when one available action dominates the others. The problem, however, is that this conclusion isn’t practically interesting. The interesting cases of moral uncertainty—the cases that we usually face, and also the cases that are worth worrying about—are those in which no available action even weakly dominates the others. Thus, although one ought to perform whichever action weakly dominates, moral weak dominance will rarely apply to actual circumstances.⁵ I will refer to cases of moral uncertainty in which no action dominates as “no-dominance” cases.

8 . 4. M O R A L U N C E R T AI N T Y WITHOUT DOMINANCE I will argue that we have good reason to think that moral uncertainty affects one’s rational obligations in some no-dominance cases; however, it turns out ³ There are only three constraints on the content of those propositions: they must be (1) mutually exclusive, (2) jointly exhaustive, and (3) “independent” of the actions one is choosing between. A set of moral propositions can satisfy all of these constraints. For example: It is sometimes permissible to kill a human and It is never permissible to kill a human. ⁴ The foregoing argument highlights one source of confusion in Nissan-Rozen’s paper “Against Moral Hedging,” in which (in one section) he argues that dominance reasoning under moral uncertainty conflicts with the actual rule of dominance. His argument goes astray when he introduces an additional constraint on the propositions that express the states of nature: he asserts that they must be descriptive, by which he means non-moral. All that Nissan-Rozen shows is that dominance reasoning could recommend one action with respect to one partition, but not with respect to another partition. That conclusion only demonstrates that we would need to work with a finer-grained partition that allows us to take into account all of the outcomes that we care about. See Nissan-Rozen (2015). ⁵ Jacob Ross makes a similar point (2006, 753).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

166

Amelia Hicks

that the way in which this happens is dependent on one’s specific preferences. Moreover, in some no-dominance cases one lacks rational obligations. Nevertheless, we should not infer from this complexity (or from the lack of rational obligations in some instances of moral uncertainty) that moral uncertainty never affects one’s rational obligations in no-dominance cases. I’ll develop my argument by considering two possibilities. First, it could be that one’s moral uncertainty never affects what one rationally ought to do in no-dominance cases. Second, it could be that one’s moral uncertainty sometimes affects what one rationally ought to do in no-dominance cases. I’ll argue against the first possibility, and then defend the second possibility from the problem of value comparison.

8.4.1. The First Possibility: We Have No Rational Obligations in No-Dominance Cases One natural thought is that the choice a morally motivated person rationally ought to make in no-dominance cases is determined by (a) their credencelevels in the competing moral propositions and (b) how morally valuable, for each of those propositions, their available actions would be if that proposition were true.⁶ This natural thought suggests that we extend utility theory to cover cases of “moral risk” in which no available action dominates the others.⁷ As we’ll see in Section 8.4.2, some philosophers reject this natural thought because of the problem of value comparison. They then infer the falsehood of Moral Sensitivity from the failure of utility theory to adequately address decision making under moral uncertainty; they think that one’s rational obligations are determined by one’s descriptive beliefs, but not by one’s moral beliefs, and they thereby endorse the first possibility: Possibility 1: in no-dominance cases, one’s moral uncertainty never affects what one rationally ought to do (even if one cares about satisfying moral norms).

My argument against the first possibility proceeds from the claim (argued for in Section 8.3) that moral uncertainty affects one’s rational obligations in cases of dominance. We shouldn’t accept this first possibility—at least, not without a very persuasive argument—because we don’t typically believe that there’s a drastic difference between dominance and no-dominance cases ⁶ Ross also discusses this position on no-dominance cases (2006, 753–4). ⁷ Notice that one can endorse this natural thought without endorsing the view that utility theory describes the decision-making process that one should consciously engage in.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Moral Uncertainty and Value Comparison

167

when dealing with other types of uncertainty. Consider the case in which I’m trying to decide whether to carry my umbrella with me, and it would be very mildly annoying to carry it in the event that it doesn’t rain. No one would say that there are no longer any rational norms governing my choice. After all, the probability of rain might be high, and I might greatly prefer to not get rained on. In light of those considerations, it would be worth risking a mild annoyance—I would be irrational if I were to risk a very high probability of something I hate only for the sake of avoiding an improbable slight annoyance. So, because we don’t believe that rational norms vanish in nodominance cases involving non-moral uncertainty, we shouldn’t believe that rational norms vanish in no-dominance cases involving moral uncertainty, at least not without a good argument. I infer that the burden of proof falls on those who think that (1) we have rational obligations to reason in accordance with dominance when morally uncertain, but that (2) moral uncertainty is irrelevant (with respect to rationality) in no-dominance cases. In the remainder of this chapter I’ll examine and respond to an attempt to take up that burden.

8.4.2. The Problem of Value Comparison I endorse the second possibility: Possibility 2: in some no-dominance cases, one’s moral uncertainty affects what one rationally ought to do (if one cares about satisfying moral norms).

This second possibility is supported by the failure of the first possibility. However, the second possibility faces the problem of value comparison. The following description of the problem of value comparison relies heavily on Brian Hedden’s presentation of the objection.⁸ However, my response to the problem applies equally to other presentations of the objection.⁹ Let’s begin with a brief description of the problem of value comparison. In order for a morally concerned person to take into account (in the course of deliberation) that they are uncertain between several moral propositions, they would need to determine the “expected moral value” of each of their available actions. To do this, they would need to determine how morally valuable each available action would be under each state of nature (where ⁸ Hedden (2016, 102–28). ⁹ Many have presented this objection and have offered responses to it, all of which Hedden rejects. See: Hudson (1989); Lockhart (2000); Ross (2006); Sepielli (2009); Riedener (2015). Nissan-Rozen (2015) also appeals to the problem of value comparison to argue against the view that rational norms are sensitive to one’s moral beliefs.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

168

Amelia Hicks

each state of nature is expressed by one of those moral propositions), and then weight those values according to their credence-level in each proposition. However, there’s no guarantee that an action’s moral value under the assumption of one moral proposition is comparable to that action’s moral value under the assumption of another moral proposition.¹⁰ However, this brief way of putting the problem is not adequate; it’s not clear what’s meant by “comparable,” and even if we fix the type of comparability that’s at stake it’s still not clear why we get a failure of comparability. At this point, it will be helpful to work with a distinction between comparability and commensurability that’s similar in spirit to Chang’s distinction. Things are comparable just in case it is possible to ordinally rank them, that is, just in case it’s possible to determine that one is greater than, equal to, or less than the other.¹¹ Things are commensurable just in case it’s possible to cardinally rank them, that is, just in case it’s possible to “precisely” rank them “by some unit of value.”¹² In this chapter, I will say that things are commensurable when they can be ranked on the same interval scale. There are two things to notice about these definitions. First, on this understanding of the comparability/commensurability distinction, comparability does not entail commensurability (and incommensurability does not entail incomparability). For example, you might think that the value of someone’s privacy being respected is more morally valuable than the small thrill you would get from prying into that person’s affairs, even if you think that the value of privacy can’t be represented using the same units used to represent happiness (or even if you think that the values of privacy and happiness can’t be represented by units of measurement at all). Second, a set of alternatives are only comparable/commensurable relative to some value (which Chang calls a “covering value”). So, for example, we can compare going bowling and going mountain climbing with respect to many different values: with respect to the fun the activity will produce for me, with respect to the money the activity will cost, with respect to how safe the activity is, and so on. With these clarifications in place, we can attempt to state more clearly the problem of value comparison: the moral value of action A under the assumption that moral proposition p is true is not commensurable to the moral value of A under the assumption that moral proposition q ¹⁰ I assume that the moral propositions are mutually exclusive, jointly exhaustive, and act-independent. ¹¹ Chang (1997) would reject this paraphrase of the definition of comparability, since she endorses the existence of a fourth relation that can hold between sets of alternatives. ¹² Chang (1997, 2).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Moral Uncertainty and Value Comparison

169

is true (where p and q are mutually exclusive, jointly exhaustive, and act-independent). As a result, it’s not clear how we could determine the “expected moral value” of performing A. We face the same problem when attempting to determine the expected moral value of any of our available actions, and thus we are unable to determine which available action has the highest expected moral value. Some philosophers infer from this that there is no action that one is rationally required to choose under moral uncertainty. This gives us a clearer statement of the problem—the problem is, in fact, a problem of commensurability. But how can we motivate it? Hedden motivates the problem in two ways. First, he argues that the value functions we associate with different moral theories are the results of different sets of “preferences,” and thus the theories’ evaluations of actions are not commensurable. Second, he argues that not all theories can be assigned a value function. (Hedden uses the phrase “value function” instead of “utility function,” presumably because it sounds strange to say that an outcome can have utility for a theory. However, Hedden’s “value functions” just are utility functions, since we’re treating “utility” as simply the value given by a function that represents a certain type of ordinal ranking of lotteries. Thus, I will use the phrase “utility function,” even when referring to those functions that represent the preferences of moral theories.) To understand Hedden’s first argument, we first need to understand how we can assign a utility function to a theory. The idea is that a moral theory expresses a set of “preferences”—an ordinal ranking of lotteries (including actions)¹³—to which we can apply a representation theorem, assuming that those preferences satisfy certain requirements.¹⁴ That is, assuming that a theory ordinally ranks lotteries in the right sort of way, we can represent that theory as recommending that one act so as to maximize some value, where that value can be numerically represented on an interval scale. One might wonder: how could we numerically represent how valuable lotteries are, according to a moral theory? The idea is this: we can “set the interval scale” for that theory by looking at the lotteries that the theory ordinally ranks the highest and the lowest. The highest-ranking lotteries mark the top of the scale, while the lowest-ranking lotteries mark the bottom of the scale. We can set those top and bottom values however we like. Then, again assuming that the theory has the right sorts of preferences over lotteries, we can determine where all other lotteries fall on that scale, according to the ¹³ A lottery is a set of outcomes in which (a) each outcome is paired with a probability and (b) the probabilities in the set equal 1. ¹⁴ For a description of axioms close to the von Neumann–Morgenstern axioms and of the von Neumann–Morgenstern Representation Theorem, see Luce and Raiffa (1957, especially 23–31).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

170

Amelia Hicks

theory. In this way, we can numerically represent the utility of any lottery (and thus any action) according to the theory.¹⁵ Let’s imagine that we represent two competing moral theories using two different utility functions, and that we thereby determine the (numerically represented) moral value of available actions according to each of those two theories. One might think that we can then use those numerical representations of the moral values of actions to determine the expected moral value of each available action; that is, one might think that we can look at how valuable each action would be if each moral theory were true, and then weight those values using our credence-levels in each theory. However: how we set each of the theory’s interval scales will affect our calculation of the expected moral value of our actions, and it isn’t reasonable to assume that we ought to set the scales in the same way (using the same numerical values to represent the top and bottom of each theory’s scale).¹⁶ Thus, the evaluations of actions provided by competing moral theories aren’t commensurable, and so any calculation of expected moral value will be meaningless. Hedden considers several solutions to the problem of value comparison. Lockhart, for instance, attempts to solve the problem by introducing the principle of equity among moral theories (PEMT), according to which The maximum degrees of moral rightness of all possible actions in a situation according to competing moral theories should be considered equal. The minimum degrees of moral rightness of possible actions in a situation according to competing theories should be considered equal unless all possible actions are equally right according to one of the theories (in which case all of the actions should be considered to be maximally right according to that theory).¹⁷

However, PEMT is implausible.¹⁸ Imagine that I’m uncertain between Kantian deontology, on the one hand, and a theory according to which some actions are only a little bit better or worse than other actions, on the other hand. We could call the latter theory the “Meh” theory. Clearly, there will be some situations in which Kantianism and the Meh theory do not hold that there are equal amounts of moral value at stake.

¹⁵ Ross (2006, 755) has a description of how a moral theory can be represented by a utility function. ¹⁶ Another way to put the worry: given that a utility function is unique up to positive affine transformation, which transformations of the utility functions (associated with competing moral theories) should we use? ¹⁷ Lockhart (2000, 84). ¹⁸ This objection to PEMT is not original to Hedden, and has been discussed by Sepielli. See Sepielli (2013). See also Hedden (2016, 110).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Moral Uncertainty and Value Comparison

171

Sepielli shows¹⁹ that one can compare the utility functions of competing theories if one also has a sufficiently large set of background beliefs (beliefs that establish points of comparison between the theories). However, Hedden points out that one will often lack a sufficiently large set of such beliefs, and that certain sets of background beliefs could actually yield contradictory comparisons between theories.²⁰ Ross and Riedener show that one can compare the utility functions of theories if one has some pre-existing beliefs about how to choose rationally when morally uncertain. That is, both Ross and Riedener argue that having intuitions about how to compare the moral values of some actions according to several moral theories can provide one with enough information to compare those theories’ utility functions, and thus to compare the evaluations delivered by the theories in less obvious cases.²¹ Hedden, however, claims to lack any intuitions of that sort, and thus rejects this proposal on the grounds that it fails to offer prescriptions for how to act rationally under moral uncertainty without presupposing some facts about how we rationally ought to act under moral uncertainty.²² Hedden’s second source of motivation for the problem of value comparison proceeds from the observation that it isn’t possible to apply a representation theorem to every moral theory, because some moral theories have “preferences” that make the application of such a theorem impossible.²³ For example, moral nihilism denies the existence of moral values, and thus does not ordinally rank outcomes with respect to moral value; and, of course, if a theory does not ordinally rank lotteries, then we cannot represent it using a utility function.²⁴ And if we can’t represent a theory using a utility function, then we can’t numerically represent how valuable outcomes are according to that theory. Thus, in these cases, we again can’t calculate the expected moral value of an action.²⁵

8 .5 . R E S P O N D I N G T O T HE P R O B L E M OF VALUE COMPARISON I will defend Moral Sensitivity from the problem of value comparison by arguing that there does not need to be an action with the “highest expected moral value” in order for there to be rational norms governing choice under ¹⁹ Sepielli (2009). ²⁰ Hedden(2016, 111–12). ²¹ Ross (2006); Riedener (2015). ²² Hedden (2016, 113–14). ²³ Hedden (2016, 114–20). ²⁴ There are other theories to which we cannot apply representation theorems. For example, a very demanding deontological theory could violate the axiom of continuity. ²⁵ This problem is discussed at length in MacAskill (2013).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

172

Amelia Hicks

moral uncertainty. I will first describe why it’s incorrect to infer from the problem of value comparison that Moral Sensitivity is false. I will then illustrate why this is a bad inference by examining a case of descriptive uncertainty that’s analogous to a “no-dominance” case of moral uncertainty. I’ll end this section by locating the source of confusion about the problem of value comparison.

8.5.1. The Two Roles of Representation Theorems Representation theorems play two roles in helping us understand rational choice under moral uncertainty. First, we can (in some cases) apply a representation theorem to the preferences of a theory, and thereby represent that theory using a utility function that tells us how that theory ranks lotteries on an interval scale. Second, we can (in some cases) apply a representation theorem to the preferences of a person (the decision maker), and thereby represent that person using a utility function that tells us how that person ranks lotteries on an interval scale. The problem of value comparison is a problem for the first role of representation theorems; the problem is that even if we can associate utility functions with competing moral theories, those functions only deliver rankings on distinct interval scales. As a result, there’s no meaningful way of determining the “expected moral value” of an action. This is an interesting result: in at least some cases, there is no expected moral value that we can assign to an action when we’re morally uncertain. However, it doesn’t follow from the problem of value comparison that there are no rational norms governing choice under moral uncertainty; this is because even if the evaluations delivered by competing theories aren’t commensurable, the decision maker can still ordinally rank lotteries (including lotteries over the possible outcomes of decisions made under moral uncertainty) and can assign utilities to the possible outcomes of their choice. It is those utilities—not the evaluations of actions provided by the competing moral theories—that determine which action has the highest expected utility for the decision maker. Thus, although the problem of value comparison is a problem if one wants to perform whichever action has the highest expected moral value, it is not a problem for a decision maker who cares about acting morally and can ordinally rank lotteries. To summarize so far: the problem of value comparison illustrates that the evaluations of actions provided by competing moral theories are not commensurable. However, that incommensurability does not entail that the decision maker cannot compare lotteries (including the actions that one must decide between under moral uncertainty) and thereby be represented by a

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Moral Uncertainty and Value Comparison

173

utility function. This conclusion is negative, since I’ve only shown the failure of an entailment relation. However, we have reason to think that there are, in fact, rational norms governing choice in no-dominance cases. There are two observations that support this stronger conclusion. First, in cases of moral uncertainty, the states of nature expressed by moral propositions must be mutually exclusive. And second, the problem of value comparison is analogous to the problem of interpersonal utility comparison. To see the first point, consider an example described by Ross:²⁶

Order veal Order veggie wrap

Singer’s theory (0.5)

Traditional morality (0.5)

Bad Not so bad

Permissible Permissible

Perhaps we can’t associate the competing moral views with utility functions, or perhaps we can but the evaluations of actions provided by the theories aren’t commensurable. My earlier, negative conclusion was that it doesn’t follow from either of those possibilities that a decision maker can’t compare (ordinally rank) lotteries in such a way that allows them to assign values to the outcomes above. However, we have good reason to think that a decision maker can ordinally rank lotteries over these outcomes, at least in many cases—this is because the evaluations provided by the theories must be comparable if the competing moral theories are theories about the same thing. If the two theories are theories of different types of values (incomparable values, no less), then the theories are not actually in competition with each other; two theories with very different structures can both be true, so long as they’re theories about different things. (For example, we can accept both a psychologist’s theory of clinical depression and a physicist’s theory of gravity in spite of their deep structural differences, because they’re theories of different things; they’re not mutually exclusive.) When one is uncertain between mutually exclusive moral propositions, one must be uncertain between moral propositions that express “different takes” on the same thing. This observation supports the idea that in genuine cases of moral uncertainty, it is possible to compare what each moral proposition has to say about an action.²⁷ Exactly what it is that competing theories disagree ²⁶ Ross (2006, 763). ²⁷ To put this point in terms of Chang’s (1997) notion of a covering value: so long as there is a covering value with respect to which one can compare options, those options are comparable in that respect.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Amelia Hicks

174

about will depend on the theories in question;²⁸ nevertheless, genuinely competing theories must provide comparable evaluations of actions, even if those evaluations aren’t commensurable. Thus, a morally uncertain person can reasonably compare lotteries over the outcomes of actions, even when the moral theories’ evaluations of actions aren’t commensurable.²⁹ The second reason for thinking that there are rational norms governing choice under moral uncertainty is the fact that the problem of value comparison is analogous to the problem of interpersonal utility comparison.³⁰ Consider a case in which you’re going on a blind date, and you have to choose what type of outing to have. In this case, you don’t know what your date’s preferences will be (since you don’t know who your date will be); however, you do know that you’ll go out with either Frankie or Frances, and you know a bit about their preferences:

Museum Kayaking

Frankie (0.5)

Frances (0.5)

Less happy More happy

More happy Less happy

Assuming that each of your possible dates has the right type of preference set, we can represent each possible date using a utility function. Those utility functions allow us to represent numerically (on an interval scale) how happy each date will be if you choose certain activities. Perhaps that would result in a decision matrix like this one:

Museum Kayaking

Frankie (0.5)

Frances (0.5)

3 7

10 1

Note, however, that the numbers in the decision matrix above do not represent the utility of each outcome for you, the decision maker. They represent how your possible dates would feel about the outings. Moreover, we can’t assume that both of your dates are using the same interval scale to evaluate the outings. As a result, we cannot determine which action has the ²⁸ They might disagree about the deontic status (permissible, impermissible, obligatory, supererogatory, etc.) of actions, or about some scalar moral value. ²⁹ I want to thank Joshua Schecter for helping me clarify this point. ³⁰ For a basic description of that problem, see Luce and Raiffa (1957, 33–4). It’s already been noted that the problem of value comparison is similar to the problem of interpersonal utility comparison. See Hedden (2016, 110) and Sepielli (2009, 12, 27).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Moral Uncertainty and Value Comparison

175

highest expectation of making your date happy. But it doesn’t follow that there is no way for you to rationally choose what to do—this is because what’s rational for you to do, according to traditional decision theory, depends on your preferences over lotteries, not just the preferences (or utility functions) of your possible dates. And that’s true even if you’re motivated to find an activity that your date will enjoy. To see how this works in the blind date scenario, imagine that you know Frances thinks that kayaking is the worst, whereas Frankie prefers kayaking but would tolerate the museum, too. You could have a perfectly rational preference—by the lights of decision theory—for going to the museum. That is, your preference for the museum-going lottery could be perfectly consistent with all of your other preferences over lotteries. The point here is this: so long as you can rank lotteries (including lotteries over these outcomes) in the right sort of way, then there can still be an available action that has the highest expected utility for you even though there’s no action that, in a formal sense, has the highest expectation of making your date happy. Similarly, what’s rational for you to do in cases of moral uncertainty is determined by your preferences over lotteries, not by the preferences of the moral theories you’re uncertain between. The incommensurability of outcomes with respect to some value—such as moral value—doesn’t entail that you, the decision maker, can’t ordinally rank lotteries over those outcomes. Notice: this doesn’t mean that decision theory “tells you what to do” when you’re morally uncertain. That’s because in order for there to be an action that has the highest expected utility for you, you already have to have a certain type of set of preferences over lotteries, and that set will include the lotteries you’re deciding between in cases of moral uncertainty. But it’s true of all forms of uncertainty that decision theory doesn’t tell you what to do when you don’t already know what to do—this isn’t anything unusual about moral uncertainty. Thus, the problem of value comparison is analogous to the problem of interpersonal utility comparison (which involves a type of descriptive uncertainty). But we don’t infer from the problem of interpersonal utility comparison that rational norms aren’t “sensitive” to descriptive uncertainty. Similarly, we shouldn’t infer from the problem of value comparison that rational norms aren’t sensitive to moral uncertainty. That is, we shouldn’t infer that Moral Sensitivity is false.

8.5.2. Locating the Source of the Confusion about Value Comparison Philosophers take the problem of value comparison to entail the nonexistence of rational norms governing choice under moral uncertainty

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Amelia Hicks

176

because they slide between the (a) incommensurability of the actionevaluations delivered by competing moral theories and (b) the incomparability of lotteries over outcomes in instances of moral uncertainty. This slide takes place because of a failure to distinguish between the utility functions used to represent the theories and the utility function used to represent the decision maker. In “The Infectiousness of Nihilism,” William MacAskill argues that Jacob Ross fails to show that a decision maker can rationally ignore “uniform” moral theories—theories according to which all available actions are equally morally valuable or disvaluable—when deliberating about what to do. MacAskill uses the example of nihilism, construed as the view that no actions have moral value or disvalue, to attack Ross’ argument. He writes, According to nihilism, no positive value relation obtains between any two options [i.e., actions]. That is, the value of every option is undefined. However, if the value of every option is undefined, according to nihilism, then, for any decision maker with nonzero credence in nihilism, there is a big problem for her if she attempts to incorporate moral uncertainty into her reasoning about expected value. The problem is as follows. If we take an expectation over possible states of nature, taking the sum, for each state of nature, of the value of that state of nature multiplied by its probability, and the value of one state of nature is undefined, then the expectation as a whole is undefined. Because, according to nihilism, the value of every option is undefined, for a decision maker with nonzero credence in nihilism, the expected value of every option is undefined, too. Nonzero credence in nihilism is therefore sufficient to infect practical reason, resulting in there being no subjective reason for preferring any option over any other.³¹

Everything in this passage is true—except for the final sentence. It’s true that if one of the possible states of nature is moral nihilism, then there is no “expected moral value” that one can maximize. However, as we’ve seen, it doesn’t follow from this that there are no norms that could govern one’s choice. Consider the following, only slightly modified example from Ross:

Order veal Order veggie wrap

Singer’s theory (0.5)

Nihilism (0.5)

Bad Not so bad

Doesn’t matter Doesn’t matter

Even though the outcomes in this situation cannot be assigned units of moral value (and even though, as a result, there is no action that has the highest expected moral value), there could still be an action that has ³¹ MacAskill (2013, 510).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Moral Uncertainty and Value Comparison

177

the highest expected utility for the decision maker—that just requires that the decision maker ordinally ranks lotteries (including lotteries over the outcomes in this choice situation) in the right sort of way. One doesn’t need to be able to describe outcomes numerically in order to have preferences over lotteries involving those outcomes. Thus, the fact that nihilism doesn’t assign moral values to actions doesn’t entail that there is “no subjective reason for preferring any option over any other.” Hedden, too, confuses the non-existence of an action that has the highest expected “moral value” with the non-existence of an action that one rationally ought to perform when morally uncertain. Hedden begins his paper with a view he calls the MITE (Maximize InterTheoretic Expectation) view of the super-subjective ought:³² Given the attractiveness of the expected value maximization framework for theorizing about the subjective ought, it is tempting to try to extend it to the super-subjective ought. If it is possible to represent all moral theories in expected value terms . . . then there is an apparently straightforward way in which to extend the expected value framework to deal with moral uncertainty as well. Expected moral value (EMV ) is an intratheoretical notion. When we take the expected moral value of an action on each moral theory and sum them up, weighted by the probability of each theory, we get an intertheoretical notion, which we can call the ‘intertheoretic expectation.’ . . . Now, the proposal is that what you super-subjectively ought to do is to make-true the act-proposition with the highest intertheoretic expectation.³³

Hedden then attacks MITE in the ways we’ve already seen; in cases of moral uncertainty, there is no action with the highest intertheoretic expectation. And, Hedden is correct to conclude that MITE fails, at least in many cases—for reasons that are now familiar, we should not say that one always rationally ought to maximize expected moral value when one is morally uncertain. But Hedden goes on to assert that, [W]hen it comes to trying to devise a formal theory of what you super-subjectively ought to do, MITE (or some slight variant thereof ) is the only game in town. This is important, since if MITE ultimately fails, as I will argue it does, then this casts serious doubt on the prospects for coming up with any formal theory of what you super-subjectively ought to do.³⁴

We can see that Hedden infers from the failure of MITE that there can be no theory of what one super-subjectively ought to do, that is, no theory of how one rationally ought to respond to moral uncertainty. This is because, as Hedden writes, “MITE, and probably any plausible theory of the ³² Where the super-subjective ought is a rational ought that’s sensitive to one’s descriptive and moral beliefs. ³³ Hedden (2016, 105). ³⁴ Hedden (2016, 106).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

178

Amelia Hicks

super-subjective ought, requires that the different moral theories in which an agent has some credence be translated into a common currency so as to allow them to be weighed against each other,”³⁵ and the problem of value comparison makes such a common currency between moral theories impossible. However, as we’ve already seen, there need not be any “common currency” between moral theories in order for there to be facts about what one rationally ought to do when morally uncertain; all that’s required for the existence of such facts is that the decision maker has the proper sort of ordinal ranking of lotteries. Thus, we should not conclude, as Hedden does, that “[t]he default position should be that there are no rules for how to act in light of moral uncertainty; beliefs about descriptive matters make a difference to how you ought to act, while beliefs about moral matters do not.”³⁶ I should highlight that my response to Hedden is distinct from the solution Ross offers in “Rejecting Ethical Deflationism.”³⁷ Hedden describes Ross’ strategy as “start[ing] with facts about what agents supersubjectively ought to do in certain cases and us[ing] those facts to reverseengineer the desired intertheoretic value comparison,”³⁸ and Hedden rejects this strategy because he has “few if any brute intuitions about what agents super-subjectively ought to do in various cases.”³⁹ (Note that Ross’ solution to the problem of value comparison is analogous to Harsanyi’s (1955) solution to the problem of interpersonal utility comparison, and thus is subject to the same sorts of objections.)⁴⁰ My response to Hedden is different; I’m not describing a way of determining which of one’s actions has the highest expected moral value. Instead, my response to the problem of value comparison consists in the observation that in order for some action to have the highest expected utility for a decision maker, that decision maker only has to ordinally rank lotteries in the right sort of way; such a ranking does not require that the theories one is uncertain between deliver commensurable evaluations of actions.

8 .6 . OB J E C T I O N : A N A C T U A L D E C I S I O N M A K E R C A N ’ T R EA S O N A B LY C O M P A RE T HE S E O U T C O M ES My response to the problem of value comparison relies on the assumption that we can represent the decision maker using a utility function and that, as a result, there are rational norms governing what the decision maker decides

³⁵ Hedden (2016, 119–20). ³⁷ See Ross (2006, 763). ³⁹ Hedden (2016, 114).

³⁶ Hedden (2016, 126). ³⁸ Hedden (2016, 113). ⁴⁰ Hedden (2016, 113–14).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Moral Uncertainty and Value Comparison

179

(even when the states of nature are expressed by moral propositions). But the reader might worry that a decision maker can’t compare lotteries over possible outcomes of choices made under moral uncertainty, and thus can’t be represented by such a function. One version of this objection comes from the familiar concern that most people cannot ordinally rank lotteries in the way required to be represented by a utility function. More specifically, a set of preferences must be complete in order to be represented by a single utility function, and actual decision makers do not have complete preferences. (One’s preferences are “complete” when one can ordinally rank all lotteries, that is, when, for any two lotteries A and B, one prefers A to B, prefers B to A, or is indifferent between A and B.) The reader might worry that, according to decision theory, one must be representable by a unique⁴¹ utility function in order to count as rational. And since an agent’s preferences must be complete in order to be representable by a unique utility function, the reader might worry that an agent with incomplete preferences is not (at least by the lights of decision theory) a rational agent to whom rational norms apply. However, we need not assume that an agent must be representable by a unique utility function in order to count as rational, and thus we need not assume that one must have complete preferences in order for rational norms to apply to them. When an agent has incomplete preferences, they cannot be represented by a unique utility function; however, they can be represented by a set of utility functions. Moreover, in many cases, the utility functions in that set will significantly overlap—that is, an agent with incomplete preferences will still nevertheless have utility functions that “agree” about what to do in some cases. In those cases in which one’s utility functions agree, one is subject to rational norms. Thus: even if there are some cases in which one is not rationally obligated to act in any way in particular because of incomplete preferences, it doesn’t follow that such a person is never subject to any rational norms. The fact that a morally uncertain person has incomplete preferences does not entail that there are no rational norms that are sensitive to their moral uncertainty. A second version of this objection comes from acceptance of a different conception of rationality. According to this version of the objection, decision theory gets rationality wrong; the rational requirements decision theory describes are not in fact rational requirements, and the real rational requirements don’t apply to cases of moral uncertainty. Note that this version of the objection shifts the target of the problem of value comparison. The problem of value comparison in the moral uncertainty literature has been a ⁴¹ Up to positive affine transformation.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Amelia Hicks

180

problem concerning the impossibility of providing a formal framework for rational decision making under moral uncertainty. But this version of the objection says that even if that problem has been solved, there nevertheless remains a related problem. I’m not optimistic about the prospects for this sort of objection. Note that this objection can’t simply assert that the conception of rationality assumed by decision theory is “too thin” (that is, it can’t simply rely on adding to the requirements of rationality described by decision theory). Instead, the objection has to show that (a) decision theory is demanding things of agents that rationality doesn’t actually demand, and that (b) there are other, incompatible rational requirements that aren’t operative in cases of moral uncertainty. The appropriate response to this second version of the objection will depend on which constraints on rationality the objector wants to introduce. For the purposes of this chapter, I’m happy to argue only that one can be subject to decision theoretic rational requirements (which amount to a consistency requirement) when deciding under moral uncertainty. For those readers who remain convinced that there’s no reasonable way of comparing the outcomes of a choice made under moral uncertainty, I’d like to make two points. First, we regularly make such comparisons in ways that seem reasonable. Second, even if there are some cases in which we can’t make these comparisons, there can still be rational norms governing choice in other cases of moral uncertainty. For an example of a case of moral uncertainty in which reasonable comparisons of lotteries over outcomes are clearly possible, consider again the example introduced earlier, in which you must choose what to do when you have some credence in moral nihilism:

Order veal Order veggie wrap

Singer’s theory (0.5)

Nihilism (0.5)

Bad Not so bad

Doesn’t matter Doesn’t matter

It may be that no available action has the highest expected moral value in this case, because of the problem of value comparison. But that fact doesn’t prevent the decision maker from reasonably preferring the veggie wrap option to the veal option, given that they want to avoid the outcome in which they order the veal and Singer is correct. To elaborate further on my first point, we can reasonably make such comparisons when dealing with interpersonal utility. (You’ll recall that cases of interpersonal utility comparison are analogous to cases of value comparison.) Let’s say that you have two friends, Eva and Evan. They’re your

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Moral Uncertainty and Value Comparison

181

friends, and so you’d like to benefit them. For some reason, you have to choose between giving Eva an ice cream cone and giving Evan an ice cream cone. Eva says that she would really enjoy the ice cream cone, and Evan says that he doesn’t like ice cream. Given these details, it’s clear what the rational course of action is: you should give Eva the ice cream cone. This isn’t because giving Eva the ice cream cone maximizes expected (interpersonal) utility, since we can assume for the purposes of this example that there is no such notion; the utility functions representing Eva and Evan are incommensurable. Nevertheless, even though the levels of utility Eva will get from the ice cream cone aren’t commensurable to the levels of utility Evan would get from the ice cream cone, you can still make a reasonable judgment about which outcome is better—it’s better that Eva get the ice cream cone. (This is analogous to a case in which you’re forced to decide between respecting someone’s privacy and getting a small thrill from prying into their affairs; even if your utility function isn’t commensurable to theirs, you can still reasonably judge that it’s better to respect their privacy than to get the thrill.) These sorts of examples are artificial and simple; one might worry about tougher cases of moral uncertainty in which it really isn’t clear how the decision maker can compare lotteries over the outcomes. I’m sure there are such cases; but all that those cases show is that the decision maker has incomplete preferences. And, as we’ve seen, the fact that a decision maker has incomplete preferences does not entail that the decision maker isn’t subject to rational norms in some cases (namely, those cases in which the set of utility functions used to represent the decision maker “agree” about what the decision maker should do). Consider the example in which you’re planning for a blind date. If you suffer from an incompleteness of preferences that makes it impossible to decide which outing to plan, then utility theory remains silent on what you should do in that case. But note: we should not infer from this that there are never rational norms governing how we should plan for blind dates. To use Hedden’s language, we should not infer that there are no “oughts” that are “sensitive” to our beliefs about who our dates will be. Similarly, the fact that utility theory remains silent on what one rationally ought to do in some cases of moral uncertainty does not entail that one never has rational obligations that are sensitive to moral beliefs.

8.7. CONCLUSION: FOUR QUESTIONS To conclude, I’d like to briefly survey several questions that my position raises. First: How should we make decisions under moral uncertainty when we don’t already have the right sorts of preferences over lotteries? All that I’ve

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

182

Amelia Hicks

shown in this chapter is that a formal problem in decision theory should not lead us to reject Moral Sensitivity; but, unfortunately, nothing in my argument describes the right sort of procedure for forming preferences. This is an issue on which traditional decision theory is (mostly) silent. However, I hope to have motivated this question; I hope to have shown that this is a question that philosophers should, in fact, be working to answer. Second: Are there other, better reasons for rejecting Moral Sensitivity? In this chapter, I’ve only argued that results in decision theory shouldn’t convince us to reject Moral Sensitivity; but, as I mentioned in Section 8.6, the conception of rationality assumed by decision theory is a thin, “instrumental” conception. It may turn out that other conceptions of rationality entail the non-existence of rational norms governing choice under moral uncertainty, or even entail that there are rational norms requiring that one ignore their moral uncertainty.⁴² I have not discussed any of those alternative conceptions of rationality in this chapter, in part because it’s difficult to formulate an alternative conception of rationality that permits a clear formulation of (some version of) the problem of value comparison.⁴³ Third: Are the rational norms governing choice under moral uncertainty different from the rational norms governing choice under uncertainty about rationality? If the arguments in this chapter are sound, then it may turn out that we should treat moral norms and rational norms very differently. This is because what I’ve argued for in this chapter cannot obviously be extended to cases of uncertainty about rational norms; if we were to extend my conclusion in that way, then it would turn out to be possible for someone to rationally choose irrationally.⁴⁴ The conclusion that we should treat moral uncertainty differently from uncertainty about rationality would be surprising, since philosophers often assume that all forms of “normative uncertainty” should be treated in the same way. And fourth: What is the proper account of rational moral motivation under moral uncertainty? I’ve suggested that a morally motivated and morally uncertain person can choose rationally without aiming to maximize expected moral value. But this means that those of us interested in the type of rationality described by decision theory need to develop a compatible account of rational moral motivation.⁴⁵ ⁴² Perhaps a view such as Philippa Foot’s (1995) or Warren Quinn’s (1993a, 1993b) could provide alternatives to the thin conception of rationality assumed by decision theorists. ⁴³ Perhaps Tarsney responds to a different formulation of the problem of value comparison (a formulation developed in terms of a different conception of rationality). See Tarsney (forthcoming). ⁴⁴ Michael Titelbaum (2015) discusses this sort of problem. ⁴⁵ I’m grateful to many people for their feedback on multiple drafts of this chapter, especially Jason Konek, Amy Lara, Elliott Wagner, David Faraci, Graham Leach-Krouse,

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Moral Uncertainty and Value Comparison

183

References Chang, Ruth, “Introduction,” Incommensurability, Incomparability, and Practical Reasoning, ed. Ruth Chang (Cambridge, MA: Harvard University Press, 1997). Foot, Philippa, “Does Moral Subjectivism Rest on a Mistake?,” Oxford Journal of Legal Studies, Vol. 15, No. 1 (Spring 1995), 1–14. Harsanyi, John, “Cardinal Welfare, Individualistic Ethics, and Interpersonal Comparisons of Utility,” Journal of Political Economy, Vol. 63, No. 4 (1955), 309–21. Hedden, Brian, “Does MITE Make Right? On Decision-Making Under Normative Uncertainty,” Oxford Studies in Metaethics, Volume 11, ed. Russ Shafer-Landau (New York: Oxford University Press, 2016). Hudson, James, “Subjectivization in Ethics,” American Philosophical Quarterly, Vol. 28 (1989), 221–9. Lockhart, Ted, Moral Uncertainty and its Consequences (New York: Oxford University Press, 2000). Luce, R. Duncan and Howard Raiffa, Games and Decisions: Introduction and Critical Survey (New York: Dover, 1957). MacAskill, William, “The Infectiousness of Nihilism,” Ethics, Vol. 123 (April 2013), 508–20. Nissan-Rozen, Ittay, “Against Moral Hedging,” Economics and Philosophy, Vol. 3 (2015), 1–21. Quinn, Warren, “Rationality and Human Good,” Morality and Action (Cambridge: Cambridge University Press, 1993a). Quinn, Warren, “Putting Rationality in its Place,” Morality and Action (Cambridge: Cambridge University Press, 1993b). Riedener, Stefan, Maximizing Expected Value under Axiological Uncertainty, dissertation, University of Oxford (2015). Ross, Jacob, “Rejecting Ethical Deflationism,” Ethics, Vol. 116 (2006), 742–68. Sepielli, Andrew, “What to Do When You Don’t Know What to Do,” Oxford Studies in Metaethics, Volume 4, ed. Russ Shafer-Landau (New York: Oxford University Press, 2009), 5–28. Sepielli, Andrew, “Moral Uncertainty and the Principle of Equity Among Moral Theories,” Philosophy and Phenomenological Research, Vol. 86, No. 3 (2013), 580–9. Singer, Peter, “Famine, Affluence and Morality,” Philosophy and Public Affairs, Vol. 1, No. 3 (Spring 1972), 229–43. Tarsney, Christian, “Intertheoretic Value Comparisons: A Modest Proposal,” Journal of Moral Philosophy, forthcoming. Titelbaum, Michael, “Rationality’s Fixed Point (or: In Defense of Right Reason),” Oxford Studies in Epistemology, Vol. 5 (2015), 253–94.

Jon Herington, John Brunero, Terence Cuneo, and Rosa Terlazzo. Audiences at the 2nd Annual Chapel Hill Metaethics Workshop, the 9th Annual Rocky Mountain Ethics Congress, the Humboldt Conference on Normativity (at Humboldt University), the Conference on Metaethics and Practical Reason (at the University of Nebraska— Lincoln), and the philosophy department at the University of Kansas have all provided me with valuable guidance.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

9 What is (In)coherence? Alex Worsnip

Recent work on rationality has been increasingly attentive to “coherence requirements.”¹ Coherence requirements are the requirements of rationality that have to do with structural coherence between one’s mental attitudes; about how they (in a broad sense) fit together; about which combinations of attitudes (and absences of attitudes) it is rational or irrational to hold jointly. Familiar examples of possible coherence requirements include the enkratic requirement (which requires one to intend to do what one believes one ought to do), the instrumental requirement (which requires one to intend the means to one’s ends), and the noncontradiction requirement (which requires one not to believe contradictory propositions). Those are some of the most familiar examples, but in principle there could be coherence requirements on combinations of any kind of attitudinal mental states: for instance, perhaps certain combinations of beliefs and hopes are incoherent, or certain combinations of beliefs and fears. Many formal epistemologists propose that there are requirements not to have certain incoherent combinations of graded credal states; some propose that there are requirements not to combine certain graded credal states with attitudes of full belief. All of these are at least candidate coherence requirements in the sense I am interested in. As should be obvious from this list of candidate coherence requirements, coherence in my sense is not restricted to logical consistency or to probabilistic coherence. Indeed, I take it to be a substantive question whether putative requirements of logical consistency or of probabilistic coherence are actually genuine (as opposed to merely putative) coherence requirements. More generally, “coherence” is not here being used in a stipulative fashion whereby certain combinations of states count as incoherent by

¹ See, e.g., Broome (2013); Kolodny (2005); Scanlon (2007); Worsnip (2018); Fogal (ms).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

What is (In)coherence?

185

stipulation; it is always open to philosophical debate whether some particular combination of states is really incoherent or not. The foregoing loose characterization gives us some idea of the form of coherence requirements. Such requirements, we’ve just said, pertain to the rational permissibility or impermissibility of combinations of mental states. This may get us at least some way to being able to tell, given some putative requirement, whether it has the form of a putative coherence requirement (as opposed to some other kind of putative requirement). But what it does not give us is an account of which coherence requirements are genuine as opposed to merely putative: of the conditions under which a combination of mental states really does count as incoherent. For example, the putative “requirement” that, for any proposition p, one is rationally required not to (hope that p is true and have a credence of 0.6 in p), has the right form to be a coherence requirement by the foregoing account, but it is presumably not plausible that this is a genuine requirement.² As such, a purely formal account still leaves us with the substantive question: what is (in)coherence, really? There are both metaphysical and epistemological questions here. Metaphysically: what is it for two or more mental states to be jointly incoherent, such that they are banned by a coherence requirement? In virtue of what are some putative requirements genuine and others not? Epistemologically: how are we to know which of the requirements are genuine and which aren’t? Typically, theorists proceed by just listing candidate requirements and considering potential counterexamples.³ But in the absence of a general account of what coherence is, and thus of what we are looking for or having intuitions about, this procedure seems unguided. These questions are made more pointed in two ways. First, the list of candidate coherence requirements is diverse, governing both doxastic and practical mental states and including everything from bans on weakness of will, to norms supplied by deductive logic, to axioms of decision theory. They include requirements on combinations of doxastic states, requirements on combinations of practical states, and requirements on combinations of doxastic and practical states. One might reasonably wonder what, if anything, all of these requirements have in common. We should hope that

² As well as this obviously non-genuine requirement, there will also be controversial cases, such as the already-mentioned example of requirements of logical consistency and probabilistic coherence. ³ So, e.g., Broome (2013: 150): “How can we identify requirements of rationality? I wish I could describe a general method of doing so, but I am sorry to say I cannot. I shall defend a number of requirements one by one, on particular grounds that seem appropriate [ . . . ] I find myself forced to appeal largely to our intuitions.”

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

186

Alex Worsnip

an account of coherence requirements will explain why these requirements all belong to a single category, by giving some general account of what it is for mental states to be jointly incoherent. Second, one of the most lively debates in the literature on coherence requirements addresses the question of whether such requirements are normative, in the sense whereby a requirement is normative iff one necessarily has reason to comply with it.⁴ But this debate often takes place against a shared assumption that there are coherence requirements. The assumption here is not merely that there “are” requirements in the sense that there are putative or “candidate” requirements. Rather, the assumption is that there are coherence requirements in the sense that some putative coherence requirements are genuine and others aren’t—that one can make mistakes about whether some putative coherence requirement is a genuine requirement or not.⁵ Thus, participants in the literature seem willing to countenance the idea that there are coherence requirements, but yet that these requirements are not normative. But this only intensifies our question about what coherence is, and what coherence requirements are. If an attribution of incoherence does not necessarily amount to a charge of a normative failing, what exactly does it come to? And if we cannot necessarily determine whether a coherence requirement is genuine by thinking about whether one really ought, normatively speaking, to satisfy it, we seem to have even less to go on in figuring out which requirements are genuine. Indeed, one might worry that perhaps the notion of a non-normative requirement doesn’t even really make sense. As it stands, this worry is overstated. There are clear examples of genuine but non-normative requirements: for example, the requirements of British Victorian etiquette, of Mafia morality, of the grammar of the French language c.1931, and so on. The reasons to comply with such requirements are at best both derivative and contingent: in themselves, they lack normative force. Yet there are genuine requirements of this sort, in the same way that some things are actually requirements of Victorian etiquette, and some things aren’t. In enumerating

⁴ See e.g. Kolodny (2005); Raz (2005); Broome (2005, 2013: ch. 11); Southwood (2008). ⁵ For example, Kolodny (2005), the leading opponent of the view that coherence requirements are normative, nevertheless holds that various variations on the enkratic requirement comprise the “core” rational (coherence) requirements, to which all other rational requirements might be reduced (Kolodny 2005: 557), and holds that these requirements are narrow-scope rather than wide-scope (Kolodny 2005: 518–39). These views only make sense if we understand Kolodny as holding that these enkratic requirements are genuine (yet non-normative!) requirements. Similarly, Broome (2013: ch. 11) takes seriously the possibility that coherence requirements are not normative, but never questions the idea that they are nevertheless genuine requirements.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

What is (In)coherence?

187

the requirements of Victorian etiquette, one can get them right or get them wrong. What is a myth is the normativity of these requirements, not their existence. However, all of these requirements are plausibly conventional in nature. They are requirements in virtue of conventional facts about the practices of Victorian Brits, Mafiosi, and French speakers in 1931. Yet it is less compelling that the requirements of coherence can be understood as fixed by any convention. So what we really are struggling to make sense of are not non-normative yet genuine requirements per se, but genuine requirements that are neither normative nor purely conventional. Again, a general account of what coherence is would be of great help here. This chapter is an attempt to give such an account, and it will be guided by three aims. First, it will aim to unify different, diverse coherence requirements, and to show what they have in common. Second, it will aim to provide us with principled criteria for determining whether coherence requirements are genuine or not (in a sense of “genuine” that encodes more than a putative requirement’s merely having the right form to be a coherence requirement, but less than an assumption that the requirement is robustly normative). That is not to say that it will on its own clearly settle every controversial case, but it will at least show how the debate is to proceed. Third, it will aim to assign coherence requirements an important philosophical role (again, irrespective of their normative status). If we can find an account of coherence and of coherence requirements that satisfies these three aims, then coherence requirements will, I believe, earn their ontological keep. I do not aim to be giving a conceptual analysis of whatever is being picked out by all uses of the word “coherent” in English. Nor will I treat our pre-theoretical, intuitive list of genuine coherence requirements (if there be such a list) as unrevisable (though no doubt the account shouldn’t make extensional predictions that deviate too wildly from this list). Rather, I will be looking for an interesting, well-regimented, philosophically important notion in the neighborhood of what we are talking about when we talk about coherence. In asking questions like “why care about coherence?” it is easy to slip from the point of view of the agent asking whether coherence requirements matter normatively, to the point of view of the theorist asking whether coherence requirements matter philosophically. But not every philosophically important phenomenon is important normatively. In particular, the account I will give assigns coherence requirements an important and highly distinctive role in our (descriptive) philosophical psychology. So I hope that my account will make coherence requirements interesting even for those who are skeptical about their normativity. But equally, I think that my account could be accepted by someone who thinks that coherence requirements are normative. I myself will take no firm stand on the debate about the normativity of coherence requirements here.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

188

Alex Worsnip

The account of coherence (and of coherence requirements) that I am offering here might be thought of as a kind of naturalistic, reductive realism about this property, though the proper application of these labels is a notoriously fraught matter. In any case, I do not take the account developed here for coherence requirements to generalize to other, more “substantive” reasons and requirements. If anything, the particular version of the view I develop—especially its claims about the role of coherence requirements in our philosophical psychology—reinforces the metanormative disunity of these coherence requirements on the one hand, and substantive (moral, prudential, epistemic, etc.) norms on the other.

9 .1 . T HE V I E W Here is the view. A set of attitudinal mental states is jointly incoherent iff it is (partially) constitutive of the mental states in question that, for any agent that holds these attitudes, the agent is disposed, when conditions of full transparency are met, to give up at least one of the attitudes. That is, human agents are disposed such that they are (at least normally) not able to (or at least find it difficult to) sustain such combinations of attitudes under conditions of full transparency. A putative coherence requirement is genuine iff every combination of states that it forbids is jointly incoherent.⁶ “Attitudinal mental states” as I use the term includes both mental attitudes and absences of mental attitudes, such as the absence of a particular belief or intention. Some coherence requirements (for instance, the instrumental requirement) effectively ban one from having some attitude while lacking some other attitude, so this broadness is required for full generality. By “conditions of full transparency,” I mean conditions under which the agent knows, and explicitly and consciously believes, that she has the states in question, without self-deception, mental fragmentation, or any failure of self-knowledge (pertaining to those attitudes). Notice that it is not required for these conditions to be met that the agent acknowledge that her mental states violate a requirement as such. It is merely required that she acknowledge that she has the states that (perhaps unbeknownst to her) violate the requirement.

⁶ This allows, correctly, that the set of attitudinal mental states that is incoherent might have only one member. Most obviously, a single belief in a contradictory conjunction, (p and not-p), is incoherent (cf. Broome 2013: 153). By contrast with a purely formal account of coherence, my substantive account can explain why this requirement belongs with the others.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

What is (In)coherence?

189

The present account makes coherence a matter of whether (or how easily) agents can psychologically sustain the states in question under conditions of full transparency. However, notice that for a combination of states to be incoherent, it has to be true that any agent would be unable (or at least find it difficult) to sustain the states under full transparency. So it won’t suffice for incoherence that some individual agent has a psychological quirk such that they are disposed not to hold two attitudes jointly: the disposition has to be present in all agents. Moreover, this fact has to be constitutive of the mental states in question. There may be some states that all agents will find it hard to simultaneously sustain, but where this is not constitutive of the states in question; the account does not count such combinations of states as incoherent. To clear one potential objection out of the way: the present account does not make violations of coherence requirements impossible. The view I am defending does not say that combinations of mental states are incoherent only when they are held under conditions of full transparency. Rather, it says that some combination of attitudes is incoherent in cases where the agent is disposed such that, were conditions of full transparency to be met, she would at least find it difficult to sustain the attitudes together.⁷ These attitudes are still incoherent when these conditions of full transparency are not in fact met (and when, consequently, they may not in fact be difficult to sustain). Moreover, conditions of full transparency are often not met.⁸ Consequently, my view allows that coherence requirements can often be violated. It does say that such violations will tend to involve some kind of failure of transparency. But I do not think that this is a bad consequence: in Section 9.2, I will argue that, for paradigmatic coherence requirements, our making sense of violations of them relies on a tacit assumption that conditions of full transparency fail to obtain. Though this naïve objection fails as it stands, some philosophers may worry that nothing could be a requirement in virtue of descriptive facts about which states agents cannot psychologically sustain, even under conditions of full transparency. Others may worry that no combination of mental states is such that all agents would be disposed not to sustain it under conditions of full transparency, and that the present approach smacks of a priori armchair psychology. I will return to these objections later. ⁷ In this respect the view is like reductive subjectivist analyses of value that identify something’s being valuable with its being the case that an agent would value it under certain ideal conditions (cf., e.g., Lewis 1989). Such analyses obviously don’t say that the thing in question ceases to be valuable when the ideal conditions aren’t met. ⁸ For arguments that such conditions are often not met, see, e.g., Williamson (2000: ch. 4); Schwitzgebel (2008); Srinivasan (2015).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

190

Alex Worsnip 9 . 2 . H E L P F U L IL L U S T R A T I V E C A S E S

We begin with some helpful cases that (I hope) make my view more plausible. We’ll come to a harder case later.

9.2.1. Instrumental Irrationality Consider first the instrumental requirement. This requirement says, roughly, that the following combination of states is incoherent: intending an end, believing some means is necessary for that end, but not intending the means. Suppose you know that your friend’s partner is cheating on her, and that she will discover this soon. You believe that it would be better if she heard it from you, both for her and for you (since she will also find out that you knew). So you intend to be the one who tells her about the infidelity. You also believe that today is the last day on which you have the opportunity to tell her, and that the only way to do so is to call her. So, you would violate the instrumental requirement if you lacked the intention to call her today. Might you lack that intention? Yes.⁹ But let’s contrast two ways in which we might try to tell the story about how you do so. The easiest way to make that possibility clear and intelligible is to reach for some story on which your mental states are not fully transparent to you. So, perhaps your mental states are fragmented.¹⁰ “Somewhere” in your mind you intend to tell your friend, and “somewhere” in your mind you know that to do this you must call her. But you never put these two mental states together or reflect on what they jointly commit you to. By putting at least one of the two states out of your occurrent reach—perhaps subconsciously motivated by the awkwardness of calling your friend—you never come to intend to call her. This is a familiar kind of failure. It wouldn’t be correct for you, if pressed after the fact, to deny that you intended to tell your friend about the infidelity, nor for you to deny that you knew that you had to call her. You just avoided simultaneous, conscious consideration of the fact that you had both of those states. And so you managed never to form the intention to call her. Suppose now, however, that we try to tell the story so that your mental states are fully transparent to you. So here, you explicitly say: “Absolutely: I intend to tell my friend about the infidelity. And the only way to do that is ⁹ Pace Finlay (2009), who argues that violation of the requirement of instrumental rationality is impossible simpliciter. I think that much of what motivates Finlay’s argument is right, but that it over-reaches. Violations of instrumental rationality require nontransparency, but are not impossible simpliciter. ¹⁰ See, e.g., Stalnaker (1984); Egan (2008).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

What is (In)coherence?

191

to call her today. But I have no intention whatsoever to call her today.” The most natural way to hear your speech here is as a joke. Why is that? Because if you were to utter these words sincerely, you would be confused about what it is to have an intention. Why is that? Because someone who is in this position doesn’t really count as genuinely intending to tell her friend about the infidelity. It is part of what it is to intend an end that one also be disposed, under conditions of full transparency, to form corresponding intentions to intend the means that one believes to be necessary to that end (or to give up the intended end). That isn’t to say that one cannot be in various, weaker states with respect to the end. One can certainly desire or wish that one tells one’s friend about the infidelity, or think that it would be good if one did so, while not forming the intention to call one’s friend. On any account, we need some way of distinguishing these weaker states from intention proper. On the present account, it is part of what distinguishes these weaker states from the stronger state of intention that they can persist even in the face of conscious, reflective recognition that one is not following through on the (believed) means. And this, in turn, is why the instrumental requirement is genuine, but analogues of it that substitute desire or wishing for intention are bogus.

9.2.2. Transitivity We want to demonstrate that the present account can unify apparently disparate coherence requirements, showing how violations of them can each be incoherent in a single, core sense. So let’s next turn to a rather different sort of coherence requirement, often found in decision theory and economics: that of transitivity of preference. This requirement bans one from simultaneously preferring A to B, preferring B to C, and preferring C to A. Here is a case of violation of transitivity that is easy to imagine. Consider the three following things that a philosopher might do with his Saturday: working on his new article, volunteering at the homeless shelter, or rewatching series four of Friday Night Lights. • Attending to the options of working on his article and volunteering at the homeless shelter, working on the article seems like an important project that he can justifiably pick over volunteering, and which allows him to stay in his pajamas and not have to interact with anyone. So he prefers working on his article to volunteering. • Attending to the options of volunteering at the homeless shelter and rewatching series four of Friday Night Lights, choosing to do something

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

192

Alex Worsnip

so trivial as watch TV rather than volunteering seems callous. So he prefers volunteering to watching TV. • But, attending to the options of watching Friday Night Lights and working on his new article, the writing of the article seems difficult and energyconsuming after his long week, and it’s not like it’s morally required. So he prefers watching TV to working on his article. These preferences are intransitive. Yet, again, what is crucial in the telling of the story here is that the philosopher only thinks about the pairwise comparisons one at a time. This isn’t to say that the philosopher doesn’t have the preferences simultaneously; non-occurrent, dispositional preferences are possible. But again, consider what it would be for such a violation to be fully transparent to the agent. Imagine the philosopher having all three options vividly before his mind, and sincerely declaring, “I prefer working on my article to volunteering, I prefer volunteering to watching TV, and I prefer watching TV to working on my article.” Again, this sounds like a joke. Once the philosopher vividly attends to the intransitivity, he will feel a pressure to resolve it. If he does not, these are not all-things-considered preferences, but only pro tanto desires. Yet he can get away with never vividly attending to it. Here’s one way that it’s depressingly likely to go: by focusing first on the choice between the article and the volunteering, he rules out the volunteering and puts that out of his mind. Then he compares the article and the TV, and picks the TV. So, he ends up watching TV, never attending to the comparison between volunteering and watching TV.

9.2.3. Inter-Level Coherence For our third example, let us turn to a requirement on doxastic states only, that I will call “inter-level coherence.”¹¹ Inter-level coherence bans incoherent combinations of first-order and higher-order doxastic attitudes, where the latter are judgments about which first-order attitudes one’s evidence supports. For instance, it forbids believing p while also believing that one lacks adequate evidence for p. Again, we can make sense of violations of inter-level coherence.¹² For example, suppose that Fabian considers himself to be extremely attractive to most members of the opposite sex. Suppose also that Fabian is aware of a ¹¹ See Worsnip (2018) for an explication and defense. ¹² Interestingly, some have been tempted to suggest that violations of what I’m calling inter-level coherence are impossible simpliciter: see esp. Hurley (1989: 130–5, 159–70) and Adler (2002). I think this over-reaches, for reasons I’m about to explain.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

What is (In)coherence?

193

body of psychological research that shows that people like him tend to systematically overestimate their attractiveness to the opposite sex, and that the women he tries to seduce often ask him to leave them alone. When Fabian reflects on all of this, he is inclined to admit that his evidence that he is extremely attractive to most members of the opposite sex is pretty lousy. But he doesn’t like to dwell on that. When he starts to think like that, he just jumps into his sports car, rolls down the windows, turns the volume on his stereo up to eleven, and goes for a spin, and very soon he stops thinking about it. His belief that he is extremely attractive to most members of the opposite sex survives. Again, this isn’t to say that there is never a single moment where Fabian both believes that he is extremely attractive to most members of the opposite sex and believes that this belief of his not well supported by the evidence. One does not count as having suspended one’s beliefs merely because they are not occurrent. But—again—what is hard to make sense of in Fabian is a persistent, stable state whereby he consciously and transparently violates inter-level coherence. There is something incredibly odd about an utterance like “all my evidence suggests that I’m not very attractive to members of the opposite sex. Nevertheless, in fact I am very attractive to members of the opposite sex.” Again, it sounds like a kind of joke. There is a strong pressure to interpret the agent either as not really believing that his total set of evidence suggests that he is not attractive, or as not really believing that he is attractive. One of the cognitive states may be weaker: it may be a fantasy, or a wish, or a hope, or an assumption, or faith, but not a belief. Part of what it is for something to be a belief, in contrast to these weaker states, is for it not to be reflectively sustainable in the face of an acknowledged judgment that it is not supported by the evidence. I think that something like the story that I am sketching here is implicit in many explanations as to why it is so hard for us to “believe at will.” In at least the most paradigmatic cases of (trying to) believe at will, one tries to believe something for pragmatic reasons, despite taking oneself to lack evidential grounds for this belief. Numerous prominent philosophers have claimed that this is because our beliefs are in some sense “controlled” by our evidence.¹³ One finds writers saying things like “believing in opposition to one’s evidence is motivationally unintelligible,”¹⁴ or “one particular belief-forming process, reasoning, is regulated solely by evidential considerations.”¹⁵ On their most obvious interpretations, however, these claims are false, for a simple reason: we fail to believe in line with what our evidence supports

¹³ Cf., e.g., Williams (1973); Foley (1993: 16); Adler (2002); Hieronymi (2006). ¹⁴ Adler (2002: 8). ¹⁵ Shah (2003: 462).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

194

Alex Worsnip

absolutely constantly. Moreover, such failures are not always results of nontransparency: they can simply be the result of mistaken assessments of what the evidence supports. However, if the picture I am suggesting is right, there may be a truth in the neighborhood here. What these writers should have said (or, if we’re feeling charitable, meant to say) is that our beliefs, when they are formed reflectively in ways that are transparent to us, are controlled by our judgments about the evidence. In other words, we cannot reflectively sustain transparent inter-level coherence. That may explain why it is so hard to believe at will, where this involves transparently defying one’s own judgment about the evidence.¹⁶

9 . 3 . T A K I N G S T O C K : T HE EM E R G I N G PICTURE OF COHERENCE Here is the picture that is emerging. Suppose we have an apparent violation of a coherence requirement. On my account, (at least) one of two things must be the case. First, it could be that the agent’s mental states are not fully transparent to her on this occasion. In that case, the violation could be perfectly genuine. Second, though, it could be that we ought not really to attribute to the agent the mental states that violate the coherence requirement. The idea that I have tried to make plausible is that an apparent intention, belief, or other attitude that, given the agent’s other mental states, will put her in sustained, transparent violation of a coherence requirement does not really count as an instance of that attitude: it is not intention or belief proper, but something less—for example, instead of an intention, a wish or desire; instead of a belief, a pretense or a supposition. That may seem like too much of a just so story. But on any account of attitudinal mental states, we need something that will distinguish particular attitudinal mental states from others: that will explain just when something is not merely a supposition but a belief; not merely a desire but an intention. My proposal is that we do this by appeal to a particular kind of disposition that agents have to revise these mental states. For example, it is part of what it is for one to have an intention (rather than, say, a desire) that one be disposed such that if one finds oneself transparently holding that intention and a belief that some means is required for the carrying out of that intention, but not intending the means, one revises one’s attitudes such that one either comes to intend the means, gives up the means–end belief, or

¹⁶ I tried to sketch such an explanation in (Worsnip 2015: ch. 4). See also Winters (1979) and Setiya (2008).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

What is (In)coherence?

195

gives up the original intention. If this is not the case, one falls short of genuinely intending.¹⁷ This puts us in a position to answer several possible objections to the present account. First, the charge of a priori psychology. There are, of course, many questions about our attitudes that cannot be answered without detailed empirical investigation. But there are also prior philosophical questions: what is it for something to be a belief, or an intention? How do we know what to look for when we do our empirical investigation of our belief and intention-forming practices? On the account I have developed, the claims about how it is hard (or even impossible) to hold two attitudes jointly under conditions of full transparency turn out to fall out of our best answers to these prior philosophical questions. They are, thus, not empirical predictions made from the armchair.¹⁸ To the extent that a distinction between observation and philosophical theory is possible, they do not rule out any particular observational data, but are rather claims about how to philosophically interpret such observational data. It is thus not outrageous to say that these claims apply to all agents. Indeed, any account of a mental state in terms of meeting a condition or set of conditions C rules out, in advance, the possibility of agents who meet C and lack the state (or have the state and fail to meet C). So any objection to my account on these grounds would generalize to any such theory. Relatedly, here is an objection that can be cooked up to any claim that some particular coherence requirement is hard (or impossible) to transparently violate: “consider person X. Person X reports herself as violating this coherence requirement. Surely person X is possible. But it’s an ad hoc, theory-driven move to say that person X must be mistaken when she reports her mental states. So it is possible to transparently violate this coherence requirement.” Such an objection can be sharpened if we make person X a sophisticated person with a philosophical theory that rationalizes the combination of mental states from their point of view. She might be someone who (rightly or wrongly) denies that the states in question are, in fact, incoherent. For example, for the noncontradiction requirement, we can imagine a dialetheist who thinks there can be true contradictions. Surely, it is said, a sophisticated person might transparently violate this requirement by believing p and believing not-p. The dialetheist’s mental states, it will be said, may still be irrational or incoherent, but surely they are still possible.¹⁹ How can I say that a sophisticated dialetheist doesn’t even know her own beliefs?! ¹⁷ Must there be some deeper (e.g., functionalist) explanation of why such dispositions are constitutive of the mental states in question? I leave this open. It is consistent with my account if there is. ¹⁸ See also Blackburn (1998: 54–9). ¹⁹ Bruno Whittle pressed this objection particularly forcefully.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

196

Alex Worsnip

I won’t disagree that we can always imagine someone who says that she (transparently) violates some coherence requirement. The question is whether we are beholden to interpret her attitudes in the way that she herself reports them. Again, the problem with the assumption that we are is that it can be used to generalize the above objection to all accounts of individual attitudinal mental states. Again, suppose you have a theory of belief of the generic form: for an agent to believe some proposition p just is for that agent to satisfy condition C with respect to p. Now, I can object: well, I can imagine someone who reports herself as believing p, but as not satisfying condition C with respect to p. And, I might add, this person has a sophisticated theory of belief, which involves rejecting the claim that believing p is a matter of satisfying condition C. Surely, I may now say, it’s an ad hoc, theory-driven move to say that she must be mistaken when she reports her mental states. Therefore, I conclude, believing p is not a matter of satisfying condition C. Thus, we have a recipe for objecting to any theory of belief (or indeed any other attitude) in terms of some condition(s). Thus, any non-vacuous theory of belief will issue the verdict that particular subjects, even sophisticated subjects, misclassify themselves as having or not having particular beliefs. If this is a problem for my view, it is a problem for every view. Once the generality of the objection is laid bare, it is apparent that it is too quick. If we have good theoretical grounds to say that subjects misclassify their own mental states, we should not be held hostage to their self-attributions. Indeed, in the case of the dialetheist specifically, I do find it hard to make sense of what it means to attribute a transparent state of believing p and believing not-p to someone, no matter how much a person professes those beliefs.²⁰ And my view is not alone in this: it’s not clear how a dispositional theory of belief, for example, or a possible worlds theory of belief, makes sense of such a person. Of course, you may disagree with me here, and think that it is easy to make sense of a dialetheist’s beliefs, even under transparent conditions. But, I submit that, to the extent that you think that, you think that the dialetheist isn’t really incoherent, and thus you reject the noncontradiction requirement, at least in its universal and exceptionless form. In that case, the example is no threat to the theory of incoherence that I have offered here. Next, the present account makes explicit why coherence requirements, independently of their normative status, have an important philosophical role to play. In particular, coherence requirements—or, more specifically, the

²⁰ Note that this isn’t to say that the dialetheist can’t (transparently) hold the belief that there are true contradictions.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

What is (In)coherence?

197

assumption that agents are disposed to obey such requirements under conditions of full transparency—play a constitutive role in our attributions of mental states. We can attribute mental states to individuals by backgrounding other mental states of theirs and seeing what is needed to make coherent sense of their intentions as manifested in their behavior—at least when we take conditions of transparency to be generally met. A similar process allows us to explain and predict this behavior. If this sort of story sounds familiar, it should: it is highly reminiscent of the “interpretationist” theory of belief and other mental states associated especially with Donald Davidson.²¹ Davidson memorably claimed that in order to be able to interpret other agents, we have to assume that these agents are rational. However, the view developed here represents a particular version and precisification of this view. When Davidson says that we must assume that agents are rational, many have interpreted him as claiming that to attribute mental states to agents, we must see them as conforming to substantive norms: as doing and believing what they really have reason to do and believe; as believing the truth, and pursuing the good.²² This assumption seems to many to be crazy. Human beings are very bad at doing what they have most reason to do. Indeed, on many plausible enough normative theories—views that combine a demanding view of morality with the view that morality gives us weighty and categorical reasons—we almost never do what we have most practical reason to do. It would be very odd if our philosophical psychology ruled these substantive normative theories out in advance, and utterly misguided to predict behavior on the assumption that we will comply with them. My claim is that things are quite different when it is coherence requirements that are at issue. On my view, it is not a precondition of interpretation that we assume that agents do or believe what they have most reason to do. But it is a precondition of interpretation that we assume that agents are disposed to be coherent under conditions of mental transparency.²³ Take a variant of a classic, well-worn example: I know that Tim intends to drink a beer, and I see him heading for the fridge (manifesting his intention to open the fridge). On that basis, I attribute the belief that there is beer in the fridge to Tim. I am assuming that Tim’s intentions, desires, and beliefs fit together ²¹ See, e.g., Davidson (1980: essay 12, 2004: essay 6). See also, inter alia, Dennett (1971). ²² I doubt that this common interpretation of Davidson is correct. See especially Davidson (2004: essays 11–12, e.g., p. 170), where he restricts his interpretationist doctrine at least primarily to rationality as coherence. ²³ The “under conditions of mental transparency” qualification also (alongside the restriction of the doctrine to coherence requirements) represents a qualification of interpretationism as standardly understood.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

198

Alex Worsnip

coherently here. If I didn’t think that, I would have no reason to favor attributing the belief that there is beer in the fridge over the belief that the fridge is empty and that the only available beer is in the garage. The latter interpretation literally doesn’t make sense of Tim’s behavior, namely his heading for the fridge rather than the garage. An assumption of coherence is thus needed to attribute mental states to Tim, and to explain and predict his actions. However, I need not assume that Tim is really responding to his substantive reasons. Perhaps he ought not to be drinking beer; perhaps he ought to be attending his child support hearing. Perhaps he ought not even believe that there is beer in the fridge: in fact, he is basing his belief on a vague memory of having put the beer there, and he could well have drunk it last night and forgotten, or it could well have been taken by his brother Billy in the intervening time. Neither of these possibilities interfere with me reading the belief that there is beer in the fridge off of his behavior. They do not make that behavior unintelligible in the way that his actually believing that there is no beer in the fridge would.

9.4. A HARD CASE: ENKRASIA We want our account to deliver an extensionally adequate account of incoherence that covers the clearest examples of coherence requirements. I cannot consider every putative coherence requirement. But let’s focus on one that threatens to make particular trouble for the account: the “enkratic” requirement. According to numerous philosophers, there is a coherence requirement forbidding akrasia: that is, forbidding one from simultaneously believing one ought to Φ but not intending to Φ.²⁴ But there is widespread consensus that clear-eyed akrasia is possible. One can think that one ought to do something, and have this thought quite clearly at the front of one’s mind, but realize that one’s intentions fail to match up to what one ought to do. This appears to be a violation of a coherence requirement without any kind of failure of transparency. Yet we still want to describe akrasia as irrational, in the sense of rationality that deals with coherence. We thus seem to have a counterexample to my claim that two or more states are incoherent only if all agents must be disposed not to sustain them under conditions of full transparency.

²⁴ Cf. esp. Broome (2013); see also (e.g.) Kolodny (2005), Scanlon (2007), and Setiya (2007).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

What is (In)coherence?

199

One could try to straightforwardly resist this counterexample to my account in one of two ways. One way would be to hold a very hard-line version of motivational internalism about normative judgment, on which if a putative normative judgment does not produce an intention to comply with that judgment (under conditions of transparency), it isn’t a normative judgment after all.²⁵ Putative normative judgments that do not produce intentions should actually be classified as other cognitive states (perhaps as purely descriptive beliefs or as what Hare called “inverted commas” judgments).²⁶ This would precisely mirror the treatment I gave of putative transparent violations of our paradigm requirements in Section 9.2. While I have more sympathy for this line of thought than many philosophers do, I still believe that as it stands it is unreasonably strong. In the cases of instrumental irrationality and inter-level coherence, it was (I think) not too much of a stretch of the ordinary notion of belief and intention to say that the putative transparent violations of the requirements in fact involved something less than full-blown belief and intention. But it is a real stretch to say that the ordinary notion of normative judgment does not allow for an intelligible notion of (clear-eyed, or transparent) akrasia. The second line of resistance would be simply to deny that akrasia really is incoherent, at least in the sense that we have identified and in which violations of the other requirements we have discussed are incoherent. Again, I have some sympathy with this. There is, I think, a good sense in which someone who says “there’s conclusive evidence that giving to charity saves lives, but giving to charity doesn’t save lives” is incoherent in a deeper way than someone who says “I ought to give most of my earnings to charity, but I’m not going to do so.” It is precisely the fact that the former is harder to make sense of than the latter that makes it appropriate to brand the former as a more radical kind of incoherence. This reveals an asymmetry between practical akrasia and inter-level incoherence that questions whether they should really be thought of as pure analogues of one another (as the label “epistemic akrasia” suggests).²⁷ Nevertheless, as it stands the second proposal also feels too strong. Rejecting the enkratic requirement wholesale, and saying that there is nothing irrational about believing one ought to do something but not intending to do it, is a drastic move. So what I propose is a kind of compromise between the two lines of resistance that moderates each. We should allow that incoherence is something that comes in degrees, and that violations of some requirements are ²⁵ Cf. Hare (1952: 19–20, 169–70). ²⁶ Hare (1952: 164–5). ²⁷ In Worsnip (2018), I argue that there is a rationale for inter-level coherence that finds no analogue in the practical case.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

200

Alex Worsnip

more incoherent than others.²⁸ For example, violations of inter-level coherence (or indeed of the instrumental requirement) are more incoherent than violations of the enkratic requirement; but the latter are still somewhat incoherent. This can be accommodated by our account of incoherence by saying that, correspondingly, the strength of the disposition not to sustain attitudes jointly (under conditions of full transparency) can also come in degrees. The most incoherent sets of mental states are ones whereby the disposition is so strong that it cannot be blocked; these sets of states will be impossible to sustain jointly under conditions of full transparency. But in less incoherent cases, such as akrasia, the disposition is weak enough to sometimes be blocked. We can then agree with the motivational internalist that it is partially constitutive of normative judgment that the agent have some disposition not to be in a sustained and transparent state of holding that normative judgment while having no intention to comply with it.²⁹ One of the ways that we get a grip on what it is to make a normative judgment is by focusing on this motivating role. Accordingly, there is some pressure on us not to interpret agents as consistently defying their own normative judgments. But this pressure is not always insurmountable: if enough other markers of normative judgment are there, we can attribute clear-eyed akrasia to agents nevertheless. Some may wish to say that this is the right way to handle the other examples of coherence requirements we considered—allowing that the disposition not to hold such states jointly under conditions of full transparency can be blocked in certain cases. This would allow for the metaphysical possibility of sustained and transparent violations of those requirements also. It would be enough that agents must, to count as having such states, have some disposition not to engage in such sustained and transparent violations—so that the assumption that agents fulfill these requirements plays some constitutive role in our mental state attribution. Although I am inclined to take a harder line on at least some coherence requirements, as shown by my treatment of cases in Section 9.2, I would still count this stance as a version of my view.

²⁸ Fogal (ms) argues that the degreed nature of incoherence makes talk of coherence requirements inappropriate. I am not fully persuaded of this: it may be that the strength or “force” of a requirement can itself come in degrees, or alternatively that something’s being a requirement is a matter of its forbidding states that are sufficiently incoherent, where there is some minimum threshold for this sufficiency. ²⁹ For congenial views, see Jackson and Pettit (1995: esp. 35–8) and Blackburn (1998: 61).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

What is (In)coherence?

201

9.5. NATURALISM AND NORMATIVITY This completes my defense of my account of (in)coherence and of coherence requirements. For any individual requirement, there is of course room for dispute. But I think that, in general, to the extent that someone disputes that agents must be disposed not to transparently sustain violations of some putative requirement, they will also be inclined to dispute that it really is incoherent to violate this putative requirement: that is, to dispute that the requirement is genuine. If this is so, it is actually confirming evidence for my account. As I said at the start, the account I have offered might be understood as a form of naturalistic, reductive realism about coherence. I have identified the property of coherence, as it attaches to sets of mental states, with a psychological property of these sets of mental states: that agent’s being disposed not to sustain them jointly under conditions of full transparency.³⁰ Now, one might in principle agree with me that these two properties are co-extensive—that is, that incoherent states are states that are such that we are disposed not to sustain them under conditions of transparency, and vice versa—without agreeing that they are identical. After all, this move is often made by non-naturalists in metaethical debates about moral properties. I don’t have a knockdown objection to this view, and I think the view I’ve offered would be interesting even understood as a claim about (mere) necessary coextension rather than property identity. However, I do think there is one way in which the move from necessary coextension to property

³⁰ One might hesitate to call the view “naturalistic,” on the grounds that the attitudinal mental states are themselves not “naturalistic” on the interpretationist view that I have endorsed. Dennett and Davidson themselves encourage this, but I doubt that there is really any tension between interpretationism and naturalism. I cannot resolve this here, but in any case, my view is as naturalistic about coherence requirements as it is about mental states. A distinct worry is that the account isn’t reductive because it explicates coherence in terms of dispositions to sustain or give up mental states, but those mental states themselves have to be understood partly in terms of coherence requirements. Thus, it is at most a virtuously circular explication, rather than a reductive analysis. There is a subtle sleight of hand in this objection. All that needs to enter into the attribution of mental states is the assumption that we are disposed to combine or not combine mental states in various ways. Such dispositions can be described without any reference to coherence or coherence requirements. My account does say that we should ultimately identify incoherence with the sets of attitudes that we are disposed not to combine. So, it will be true that the dispositions not to combine states are effectively dispositions not to be incoherent. Nevertheless, coherence is still reduced to psychological dispositions that can be described without reference to coherence, which is enough to avoid circularity.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

202

Alex Worsnip

identity is on stronger ground when it comes to coherence as compared to morality, as I’ll now explain. Reductive naturalists about morality often say that property identity is the best explanation of the co-extensiveness and thus of the supervenience of the moral on the natural.³¹ But non-naturalists have a reply to this: that the co-extensiveness of moral rightness with certain natural properties is explained by one or more very general, irreducibly normative, principles that specify that an act is right iff it has some particular natural features. If these principles are necessary, then we get an explanation of the co-extensiveness of moral and natural properties, and of the supervenience of the moral on the natural.³² This is not the place to adjudicate the adequacy of this reply. What I want to point out is that such a non-naturalist reply is a non-starter in the case of coherence requirements. There is not, on any view I know of, some master normative principle of coherence that enjoins one not to be in states that one is disposed not to sustain under conditions of transparency. So there is no fundamental, irreducibly normative principle that can explain the co-extensiveness of the property of coherence and the property of being sustainable under conditions of transparency. Thus, the claim that those properties are in fact the same property seems the most plausible way to go. This suggests that the naturalist is better off (or even better off!) in the case of coherence requirements than in the case of substantive moral requirements. This connects to the more general point I foreshadowed at the start, that I see no reason to think that the account I have offered offers any particularly strong support for reductive realism about other requirements or normative claims, for example substantive moral requirements. I have already explained why I do not think that substantive reasons and requirements play the same role in the interpretation of mental states, or the explanation and prediction of behavior, that coherence requirements do. I am much less sympathetic to naturalistic, reductive realism about moral requirements than to naturalistic, reductive realism about coherence requirements.³³ The possibility that we might give very different metanormative accounts of coherence requirements and of other, substantive (e.g. moral) requirements reinforces the distinctness of coherence requirements on the one hand, and these other, substantive requirements on the other. No doubt some will think this shows that, on the account developed here, coherence requirements are not really normative, or perhaps that they are ³¹ See, e.g., McPherson (2012). ³² Cf. Enoch (2011: ch. 6); Scanlon (2014: ch. 2). ³³ Similarly, Ridge (2014) advocates reductive realism about “rationality” (that is, coherence), but an “ecumenical expressivism” about moral (and other substantive normative) judgments.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

What is (In)coherence?

203

not really requirements, in any good sense, at all. They have, these people will say, stopped functioning as normative requirements and become merely descriptive truths about psychology. As I said at the beginning of this chapter, I want to be neutral on whether coherence requirements are normative in the most robust sense of that term. My aim here was not to show that they are, but rather to earn them their ontological keep in a different sort of way. Nevertheless, let me make three preliminary points on this issue. First, to the extent that there is a phenomenon to be saved here, it is not most naturally characterized in terms of the language of “normative requirements” but rather in terms of the language of “rationality.” What we want to earn the license to say is that incoherent combinations of mental states are irrational. But the account of coherence that I have given does connect with a long and venerable tradition of thinking about rationality that is recognizable in the ways that ordinary people actually use the concept. One important idea is that that rationality consists in intelligibility: irrationality is a way of being harder to make sense of as an agent.³⁴ Another is that rationality is the norm—in the “default” or “statistical” rather than the robustly normative sense of “norm”—on the basis of which we predict human behavior and ascribe mental states (an assumption engrained both in academic social science and in ordinary talk). So, even if this notion of coherence turns out not to count as normative in the most robust sense, I do not think that it “changes the subject.” If someone concedes this but simply insists that the word “requirement” must be expunged from my account, I will feel contented overall.³⁵ Second, however, it is not obvious that in giving a reductive, naturalistic account of coherence requirements, we thereby preclude them from counting as normative even in a robust sense. At least, this should be a matter for debate. Naturalists about moral norms are often accused of being unable to account for the normativity of morality³⁶—and while such criticisms may hit their target, it is not just obvious that they do so. Just because we have identified the property of coherence with a psychologically describable property does not immediately entail, without argument, that this property cannot be normative.³⁷

³⁴ Cf., again, Davidson (2004). ³⁵ If, as I suggested, coherence is a necessary condition for rationality, there is at least one sense in which there are coherence requirements: there are conditions of coherence that one must satisfy if one is to count as being rational. Broome (2013: 109–10) calls this the “property” sense of “requirement.” ³⁶ Cf., e.g., Nagel (1986: ch. 8); Parfit (2011); Enoch (2011: ch. 5, esp. 107–8). ³⁷ See Railton (1989); Schroeder (2005).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

204

Alex Worsnip

Third, as participants in both the debate about the normativity of rationality and the debate about reductive realism in metaethics have noted,³⁸ the term “normative” can be used to stand for a range of things, and some requirement might be normative in one sense but not in another. Naturalist realists tend to argue that morality is normative in some but not all of these senses.³⁹ It may turn out that coherence requirements are as naturalist realists say that moral requirements are, in this respect. Moreover, those who are dissatisfied with anything other than the most robust kind of normativity in the moral case need not take the same view when it comes to coherence requirements. For the intuitive appearance that coherence requirements are normative is, in my view, somewhat less robust than the intuitive appearance that moral requirements are normative. Ultimately, there may be less normativity to account for in the former case than the latter. I cannot resolve these questions here. But I hope that the account of coherence that I have offered lends adequate determinacy to that notion for the debate about the normativity of coherence requirements to be conducted in a reasonably orderly manner. And however we eventually resolve this debate, I hope that the account I’ve given evidences an underlying unity behind talk of coherence in many of its superficially disjunct guises, and that this unified notion is philosophically interesting independently of its normative status.⁴⁰

References Adler, J.E. (2002). “Akratic Believing?” Philosophical Studies, 110: 1–27. Blackburn, S. (1998). Ruling Passions: A Theory of Practical Reasoning. Oxford: Oxford University Press. Broome, J. (2005). “Does Rationality Give Us Reasons?” Philosophical Issues, 15: 321–37.

³⁸ For the former, see, e.g., Southwood (2008) and Ridge (2014: ch. 8). For the latter, see, e.g., Copp (2007). ³⁹ See, again, Copp (2007). Some naturalist realists, such as the early Railton (1986), argued that morality is normative (in the most robust sense) only derivatively on instrumental rationality. ⁴⁰ For helpful comments/discussion, I’m grateful to Selim Berker, Nic Bommarito, Brian Cutter, Jane Friedman, Max Hayward, Chris Howard, Alex King, Boris Kment, Wooram Lee, Ram Neta, Kate Nolfi, John Phillips, Geoff Sayre-McCord, Josh Schechter, Miriam Schoenfield, Ralph Wedgwood, Bruno Whittle, and two anonymous referees for Oxford University Press. Special thanks to Daniel Fogal, Harvey Lederman, and Conor McHugh for highly detailed written comments. I’m also extremely grateful to participants at the 2016 Chapel Hill Metaethics Workshop and the Fordham Ethics and Epistemology Workshop for many excellent questions, comments, and suggestions.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

What is (In)coherence?

205

Broome, J. (2013). Rationality Through Reasoning. Chichester: Wiley-Blackwell. Copp, D. (2007). “Moral Naturalism and Three Grades of Normativity,” in Morality in a Natural World. Cambridge: Cambridge University Press. Davidson, D. (1980). Essays on Actions and Events. Oxford: Oxford University Press. Davidson, D. (2004). Problems of Rationality. Oxford: Oxford University Press. Dennett, D. (1971). “Intentional Systems,” Journal of Philosophy, 68/4: 87–106. Egan, A. (2008). “Seeing and Believing: Perception, Belief Formation and the Divided Mind,” Philosophical Studies, 140/1: 47–63. Enoch, D. (2011). Taking Morality Seriously: A Defense of Robust Realism. Oxford: Oxford University Press. Finlay, S. (2009). “Against All Reason? Scepticism about the Instrumental Norm,” in C.R. Pigden (ed.), Hume on Motivation and Virtue. Basingstoke: Palgrave Macmillan. Fogal, D. (ms). “Rational Requirements and the Primacy of Pressure,” draft manuscript. Foley, R. (1993). Working Without a Net. Oxford: Oxford University Press. Hare, R.M. (1952). The Language of Morals. Oxford: Oxford University Press. Hieronymi, P. (2006). “Controlling Attitudes,” Pacific Philosophical Quarterly, 87/1: 45–74. Hurley, S.L. (1989). Natural Reasons. Oxford: Oxford University Press. Jackson, F. and Pettit, P. (1995). “Moral Functionalism and Moral Motivation,” Philosophical Quarterly, 45/178: 20–40. Kolodny, N. (2005). “Why Be Rational?” Mind, 114/455: 509–63. Lewis, D. (1989). “Dispositional Theories of Value,” Proceedings of the Aristotelian Society Supplementary Volume, 63: 113–37. McPherson, T. (2012). “Ethical Non-Naturalism and the Metaphysics of Supervenience,” in R. Shafer-Landau (ed.), Oxford Studies in Metaethics, Volume 7. Oxford: Oxford University Press. Nagel, T. (1986). The View from Nowhere. Oxford: Oxford University Press. Parfit, D. (2011). On What Matters. Oxford: Oxford University Press. Railton, P. (1986). “Moral Realism,” Philosophical Review, 95/2: 163–207. Railton, P. (1989). “Naturalism and Prescriptivity,” Social Philosophy and Policy, 7/1: 151–74. Raz, J. (2005). “The Myth of Instrumental Rationality,” Journal of Ethics and Social Philosophy, 1: 1–28. Ridge, M. (2014). Impassioned Belief. Oxford: Oxford University Press. Scanlon, T.M. (2007). “Structural Irrationality,” in G. Brennan, R. Goodin, F. Jackson, and M. Smith (eds.), Common Minds: Themes from the Philosophy of Philip Pettit. Oxford: Oxford University Press. Scanlon, T.M. (2014). Being Realistic About Reasons. Oxford: Oxford University Press. Schroeder, M. (2005). “Realism and Reduction: The Quest for Robustness,” Philosophers’ Imprint, 5/1: 1–18. Schwitzgebel, E. (2008). “The Unreliability of Naïve Introspection,” Philosophical Review, 117/2: 245–73. Setiya, K. (2007). “Cognitivism About Instrumental Reason,” Ethics, 117/4: 649–73.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

206

Alex Worsnip

Setiya, K. (2008). “Believing at Will,” Midwest Studies in Philosophy, 32: 36–52. Shah, N. (2003). “How Truth Governs Belief,” Philosophical Review, 112/4: 447–82. Southwood, N. (2008). “Vindicating the Normativity of Rationality,” Ethics, 119/1: 9–30. Srinivasan, A. (2015). “Are We Luminous?” Philosophy and Phenomenological Research, 90/2: 294–319. Stalnaker, R. (1984). Inquiry. Cambridge, MA: MIT Press. Williams, B.A.O. (1973). “Deciding to Believe,” in Problems of the Self. Cambridge: Cambridge University Press. Williamson, T. (2000). Knowledge and its Limits. Oxford: Oxford University Press. Winters, B. (1979). “Believing at Will,” Journal of Philosophy, 76/5: 243–56. Worsnip, A. (2015). Rationality’s Demands on Belief. Doctoral dissertation, Yale University. Worsnip, A. (2018). “The Conflict of Evidence and Coherence,” Philosophy and Phenomenological Research, 96/1: 3–44.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

10 The Authority of Formality Jack Woods

10 . 1. INTR O DU CT ION Etiquette gets a bad rap. Theorists sometimes claim that etiquette is merely formally normative whereas morality is substantively normative: that morality is normatively special in some way which eludes requirements of mere manners.¹ Some theorists flavor these claims with the etiquette norms of our grandparents and moral norms of pressing contemporary interest. It’s hard, in the face of this, to do much but nod along. Who could seriously think that wearing pants to dinner was in any way on a par with treating others with compassion? I do. There’s at least one important sense of normativity which is shared by both morality and etiquette. Both explain a distinctive kind of normative reason for action in virtue of their being standards we take seriously. I’ll call these reasonsδ for action where δ ranges over standards such as etiquette, morality, or even the norms of pool.² Reasonsδ come from our normative reasons to respect δ as a standard governing our behavior. They are different from the “institutional reasons” which are part of systems of norms (Joyce 2001: §2.1)—etiquette might “say” that we have decisive reason to ψ, but we might only have weak reason to ψ as we only have weak reason to respect etiquette. Reasonsδ are also independent of instrumental reasons to obey standards—we can have instrumental reason to obey any standard of correctness, regardless of our relationship with the standard. Reasonsδ instead capture how our relationship with certain standards makes an important normative difference to what we ought, all things considered, to do. Some standards, like morality, might also fully explain unsubscripted normative reasons for action. This property—which I’ll call intrinsic ¹ For recent articulations, see Ridge (2014: 93), Parfit (2011: 144), Broome (2013: 26). ² I focus on practical standards, though my considerations also apply to epistemic standards. See Maguire and Woods (manuscript).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

208

Jack Woods

reason-providingness—is what I take substantiveness to be. My aim, though, isn’t to investigate whether morality or etiquette are intrinsically reasonproviding, but rather to explain how morality and etiquette yield structurally identical reasonsδ because of our intimate relation with morality and etiquette as important standards for us. My view is that certain standards become normatively important because we treat them as guides to practical behavior. When this happens, I’ll say these standards are in force for us. I’ll argue that we then have reason to respect their norms exactly because they’re in force standards for us. These reasons to respect norms in turn explain reasonsδ for action. Both morality and etiquette are treated by us as guides to practical behavior, so in terms of reasonsδ, there’s no structural difference between morality and etiquette. As I’ll say, both are formally normative standards. I highlight this symmetry of morality and etiquette, but I don’t insist that morality isn’t substantive.³ My sneaky hope is that the picture I’ll paint in what follows suggests that the existence of substantive obligations is unnecessary.⁴ But the primary aim is developing a detailed account of the normative character of formally normative standards. In my view, formal normativity is under-explored because of the temptation to treat most formal standards as merely arbitrary standards of correctness. But they are so much more than that. I open by arguing that morality and etiquette share a type of normativity not possessed by all standards. I then give more details about standards and their relationship to other normative notions (in Section 10.2). Returning to formal normativity in Section 10.3, I claim our actual formal standards aren’t escapable even when there are reasons to act against them, use this fact to evidence standards being in force for us, and develop an account of standards being “in force” drawing on these claims. This sets up the discussion of reasonsδ as the characteristic normative oomph of formal standards (in Section 10.4). I dismiss two worries for my account: one involving the genericity of reasons, the other fetishism, and close by suggesting that my account may undermine the grounds for believing in substantive normativity at all. 1 0. 2 . N O R M A T I V E S T R U C T UR E A N D T H E “ S C H - ” C H AL L E N G E Why think that there’s a further distinction between formal normativity and mere standards of correctness? It’s tempting to just collapse these two. This ³ For compelling reasons to reject substantive obligations, see Williams (1979) and Copp (2004). ⁴ To be honest, slightly less sneaky now.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

The Authority of Formality

209

temptation is aggravated by what I’ve come to think of as the “sch-” challenge.⁵ Let schmorality be a system of norms covering the same field of actions as morality, but slightly differing on its verdicts about what to do. Intuitively, violations of the schmorality standard aren’t important, but schmorality is nevertheless a standard of correctness: we can be good and bad schmoralizers. Just as intuitively, violations of morality are important— being immoral isn’t merely incorrect. What explains this difference? What’s so special about morality?⁶ The problem with the “sch-” challenge is that it misleadingly suggests that the only important division between standards is one that puts morality and like intuitively substantive standards on one side and the rest on the other. But this is incorrect. We can turn the “sch-” challenge into a demonstration of this conflation. Let schmetiquette be a system of norms covering the same territory as etiquette, but slightly differing on its verdicts about what to do. Intuitively, violations of the schmetiquette standard aren’t important—schmetiquette is a mere standard of correctness. Just as clearly, violations of etiquette are a serious matter—being impolite isn’t merely incorrect. What explains this difference? What’s so special about etiquette? The problem is that there are potentially three types of standards here: mere standards of correctness, like schmetiquette and schmorality, which aren’t important to us and aren’t intrinsically reason-providing; merely formal standards, like pool rules, the law, and etiquette, which are important but are not intrinsically reason-providing; and substantive standards, like morality perhaps, which are important and intrinsically reason-providing. I haven’t yet said what it is for a standard to be important to us (more on this later) but the idea is clear enough—we take both morality and etiquette seriously. The original “sch-” challenge runs roughshod over this three-way difference by suggesting the sole important normative difference is between substantive standards and the rest. But schmorality, schmetiquette, and etiquette shouldn’t be classed together. Etiquette is a normative standard which we take seriously, schmorality and schmetiquette unimportant standards of correctness. Even if etiquette isn’t intrinsically reason-providing, it’s still a part of our normative outlook. We regard violations of etiquette as sufficient for criticizability; we regard etiquette as giving us reasons to be

⁵ The “sch-” challenge crops up in different guises, but it should be recognizable to the reader. ⁶ For discussion, see McPherson (2011) and works cited within. Note that McPherson analogizes chess with schmess and schmeasons in the misleading way I’ll now suggest.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

210

Jack Woods

polite. These features of both etiquette and morality aren’t possessed by schmorality or schmetiquette.⁷ So there are two potential differences between morality and schmorality: substantiveness and whether they’re important to us. Since the latter difference is also shared between etiquette and schmetiquette, the “sch-” challenge provides little reason to think that morality is special in a way that etiquette and the rules of games are not. Rather, it gives significant reason to think that both morality and etiquette are special in a way that schmorality and schmetiquette are not. The “sch-” challenge shows that we need an explanation of the difference between both merely formal and putatively substantive standards and those standards of correctness which aren’t even formal norms. We turn to this shortly after giving a more detailed picture of standards.

10.2.1. Subscripted Obligations The standards we’re interested in—morality, etiquette, correct pool playing—arise from systems of norms.⁸ For our purposes here, a system of norms is any collection of demands, permissions, and forbiddings of actions, as well as favorings and disfavorings of the same. Systems of norms underwrite our target standards of correctness—being in accordance with or variance with the norms. We typically lexicalize this correctness property when the standard is one we care about: morally upstandingness, politeness, rationality. These are gradable properties: one can be more or less morally upstanding, polite, or rational, though we sometimes use these terms to indicate being perfectly correct. Often the exact character of a system of norms isn’t always clear to those taking the corresponding property seriously. Morally upstandingness is important to us even though it’s unclear to many of us—it’s unclear to me—what being morally upstanding requires. It’s correspondingly difficult to carve systems of norms apart precisely, but in practice we can generally distinguish them—we have little trouble distinguishing what morality requires from, say, what etiquette requires. Systems of norms themselves may consist entirely of forbiddings and permissions with no non-trivial favoring and disfavorings, like legal systems ⁷ This point isn’t novel, but its importance has been downplayed. See the discussions of rule-implying normativity in Parfit (2011) and Broome (2013) (see Section 10.2.3 for why “reason-implying” isn’t the right property to explicate authoritativeness). See also Berman (forthcoming) for related discussion of the “normative isomorphism” between substantive and formal obligations. ⁸ I’ll generally default to talking about standards, instead of the system of norms that underwrite them or the obligations that standards induce. This should cause no confusion.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

The Authority of Formality

211

of norms.⁹ They may consist entirely of favorings and disfavorings where we concoct forbiddings and permissions in terms of what’s most favored, as we find in culinary norms. They may consist entirely of rough aims which underwrite favorings and disfavorings, forbiddings and permissions. Finally, they might be entirely particularistic, having no explicit rules or aims, but where we have a sense of which things are favored, disfavored, forbidden, and permitted. Any of these can be massaged into a standard of correctness. Consider the rules of tic-tac-toe. One is permitted to mark any open square on one’s turn. One is forbidden from making two moves at once. A mark is an “X” or an “O”; each player may only have one mark. Player X marks with “X,” Player O marks with “O.” A win is a configuration of the board where there are three consecutive “X”s or “O”s. A player is the winner if their mark makes a win; all other players are losers. A draw is a full configuration of the board which is not a win. The aim of play is to configure the board into a win; a move is favored insofar as it advances this goal. And so on. This is only a partial spelling out of the permissions, forbiddings, favorings, and disfavorings of a game so notoriously simple as tic-tac-toe, but it’s enough to see what correct tic-tac-toe play is. Standards of correctness play a fundamental role in the orthodox semantics for “ought.” On this account, background information, such as a contextually salient standard, ranks ways things could go in terms of how well they accord with it.¹⁰ We oughtδ to do something just in case the contextually salient δ ranking places ways in which I do it above worlds in which I don’t. This semantic story is fully compatible with every standard being substantive, none being substantive, and any mediate position. Since we can raise any standard to contextual salience with sufficient background, we won’t find any difference in normativity between our three types of standards in terms of the semantics of “ought” or “obligation.”¹¹,¹²

⁹ “Non-trivial” since we can cook up favorings and disfavorings given a system of forbiddings and permissions by claiming disfavor of anything we’re forbidden from doing, favor of anything we’re forbidden to not do, etc. ¹⁰ See (Kratzer 1977) for origins of this account and (Chrisman 2015: chs. 2–5) for an overview from a philosophical standpoint and details. We sidestep worries arising from differences in the various flavors of “ought” since the cases under consideration all more or less involve agentive “oughts” with different ordering sources. ¹¹ I use “substantive” and “formal” for both standards and the obligations they induce. I’ll likewise not carefully distinguish between “ought” and “obligation.” ¹² For the cognoscenti: I’m running roughshod over a number of important semantical differences between “ought,” “must,” “should,” and their connection to obligations. I’m also leaving unspecified many important issues in the metasemantics of “ought.” These details aren’t important for my point and the details are contentious, so let it go.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

212

Jack Woods

Standards and induced notions of “ought” are cheap; the interesting normative questions lie elsewhere.¹³ Consider again our trivial example of tic-tac-toe. We oughtttt to block any two-consecutive pattern unless we can win immediately by playing elsewhere. And the tic-tac-toe standard might be contextually salient. Yet this is insufficient for normative reason to satisfy these obligations—so what if we’re obligedttt to play a particular way? If, following Parfit (2011), we treat the issue of genuine normativity in terms of reasons, then we should look to connections to normative reasons in order to distinguish formal and substantive standards from mere standards of correctness.

10.2.2. Generic Reasons In order to do this, we need to make use of something like a generic sense of reason or ought.¹⁴ What these notions are is, of course, a matter of substantive philosophical dispute. As is what it is to act in a way which is based on them. I’ll abstract from these worries here, intending a broadly ecumenical picture of generic normative notions and how we can base our actions on them. I’ll briefly argue in favor of the existence and nonemptiness of these notions shortly. I’ll use reasons-talk, though I’m strictly agnostic on whether (generic) reasons or (generic) oughts are more fundamental. The reason for my agnosticism is that we can interdefine these notions. Starting with generic normative reasons, we generically ought to do what we have most reason to do. Starting with generic oughts, we have generic reason to do as we ought.¹⁵ We can flesh out these connections in different ways. Perhaps value is explanatorily fundamental in grounding our generic reasons (Maguire 2016). Then we generically ought to do the value-maximizing things.¹⁶ Perhaps desire explains generic reasons. If so, we generically ought to do those things which are maximally desire-satisfying. Even if there are multiple, equally fundamental, grounds for our reasons, then we can construct a partial generic ought in terms of what these reasons all agree upon. ¹³ I think it’s natural to typically treat the obligations true in a context as plenitudinous so that right now I oughtskippedy-do ψ, etc. If more restrictive accounts are wanted, I trust the reader can make the obvious adjustments. ¹⁴ See Baker (this volume) for worries about the existence of these kinds of flat notions. The position I construct in what follows concedes many of his worries while avoiding his charge of “changing the subject.” See Finlay (forthcoming) for discussion of exactly what the subject is. ¹⁵ This latter reduction requires complications involving weighing reasons. As I prefer generic reasons, I won’t worry about this. ¹⁶ Maximization is a simplification; substitute your favored weaker notion if you like.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

The Authority of Formality

213

I don’t care which underlying story we tell about generic reasons and oughts. I only care that there are recognizable and suitably related generic notions of reason and ought which, if they existed, would serve to “close deliberation” (Schroeder 2011) and structure our ordinary practical deliberations.¹⁷ My task is not to defend the existence of such notions, but a few remarks might help to see where I’m coming from.¹⁸ Our practical deliberation typically proceeds by means of subscripted normative notions (Tiffany 2007). Starting with reasons, we talk about what prudence favors, what morality favors, and the like. It seems, though, that no matter which norms we invoke, we can always ask whether we should do what they jointly favor. In other words, constructions like the following always seem reasonable, no matter which named systems of norms we stack before “but”: (OQR₁) Morality favors doing it, but should I do it? (OQR₂) Morality and prudence favor doing it, but should I do it? (OQR₃) Morality, prudence, and etiquette favor doing it, but should I do it? ...

In each case, I’m asking what I have most reason to do given the listed facts preceding the “but.” I take the generic notion of a reason to be the limit of “should” in the OQR sequence. If that’s right, then it’s extremely plausible we recognize a coherent generic normative notion, like generic reasons, in making sense of instances of the OQR schema. In further defense, note that many normative conflicts seem easily resolvable.¹⁹ If all my subscripted reasons, except for my mild penchant for elegance, direct me to do something, it seems clear that I should. This is easily explained with the existence of generic reasons or oughts; it seems nearly impossible to explain otherwise, especially as we complicate the case to get deep systematic conflicts between systems of norms. Our taking many of these questions to have determinate answers is significant evidence that there really are generic normative notions.²⁰ One could claim that ways of making sense of conflict resolution differ between contexts and between sets of subscripted norms (Baker, personal ¹⁷ Baker (this volume) claims that the notion of generic obligation is too vague and metaphorical to do its characteristic work. I disagree; there may be no such oughts (Copp 2004; Tiffany 2007), but it’s manifest that we presume there are such in our practical deliberations. ¹⁸ McPherson (this volume) takes steps towards a plausible defense. Steve Finlay (2014) offers an interpretation of all-things-considered ought judgments that fits nicely with the sort of picture I’m sketching here. ¹⁹ Dorsey (2013) makes a compelling case for this theoretical virtue of generic reasons. ²⁰ Generic reasons shouldn’t arbitrate all conflicts between systems of norms as sometimes our reasons really are just on a par (Chang 2002).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

214

Jack Woods

correspondence, this volume). This, however, itself demands an explanation; such explanations can typically be massaged into an explanation of generic reasons. For example, suppose the value of resolving the conflict a particular way breaks ties. This suggests grounding generic reasons in value. Such explanations are far more elegant and theoretically satisfying than explanations denying any systematic understanding of the “should” in each instance of OQR. Does presuming generic reasons concede that there are substantive formal obligations? Not exactly. The ought constructed out of generic reasons is distinct from a subscripted ought. To see this, consider taking what we’ve generic reason to do as a particular system of norms: call the resulting standard practicality. We could then sensibly ask questions about what we oughtpractical do that seem incoherent when we don’t “standardize” generic reasons, like “why should I do what I have most reason to do?” So to play their characteristic role generic reasons shouldn’t be treated as just one standard among others. That is, they are plausibly not a subscripted bit of normativity in the relevant sense. Anyways, if I’m mistaken and we can and should treat generic reason and oughts as the sole substantive standard, that isn’t a large cost to my view as my target is the normativity of non-generic oughts like morality and etiquette.

10.2.3. Summary: Norms—Substantive and Formal Summing up, I’ve said that systems of norms give rise to gradable standards of correctness. Each standard δ yields a notion of what we oughtδ to do—act as δ demands. What then would it be for δ to be substantive? I view the best explication of substantiveness as holding that a standard δ is substantive insofar as oughtδ facts totally explain generic reason to do as δ demands. Then, according to (my cashing out of ) myth and legend, that I oughtmorality to do something totally explains the generic reason to do it; that I oughtetiquette not so much.²¹ I believe, but won’t insist, that there are no substantive obligations in this sense—myth and legend is just that. Why “explains” as oppose to “implies”? After all, Parfit (2011) treats the question of substantiveness in terms of a distinction between a rule-implying ²¹ A referee worries that it’s more natural to say morality is substantive because our obligation to be moral is totally explained by our generic reasons. Presumably the idea is that we necessarily have non-instrumental generic reason to be moral. This view strikes me as strange—presumably we don’t cordon off morality by what we have necessary generic reason to do. Especially since we may have intuitively non-moral, yet necessary, generic reason to act toward each other in various ways. Perhaps prudence or politics furnishes such reasons. This suggests that any non-foot-stampy explanation of how morality is substantive will make use of features of the moral standard itself. But, given that explanation is transitive, this means that our generic reasons are in turn explained by morality.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

The Authority of Formality

215

sense of obligation (etiquette and rules of games) and a reason-implying sense (morality and prudence). This, however, does a serious disservice to the idea that various normative domains are special. Let Γ be a set of norms defined so that if we have non-instrumental reason to do something, we oughtΓ to do it. On the common presumption that non-instrumental reasons are necessary, that we oughtΓ to do something will imply that we have reason to do it. Yet Γ is intuitively not substantive. The problem is that entailment is cheap.²² A domain like Γ can entail that we have reasons even though Γ plays no role in explaining why we have reasons. It might merely necessarily track our generic reasons.²³ But theorists should want morality to be substantive in a stronger sense; they should want the fact that morality says “ψ!” to explain, by itself, reason to ψ. This would be a type of substantiveness worth its salt. We can have many types of generic reason to act as δ demands: instrumental reason—I’d rather not act incorrectly for various reasons—as well as independent reason to ψ regardless of what δ says. And, of course, the distinctive class of reasonsδ I’ll explain shortly. None of these are sufficient for substantiveness since (a) they aren’t explained totally by oughtδ facts and (b) we have such reasons for etiquette as well as morality. If morality is substantive, then there will be generic reasons to ψ which are totally explained by the fact that I oughtmorality ψ. But there will also be reasonsmoral to act morally; the explanation of these reasons is entirely analogous to our reasons to act politely. Morality, that is, is also formally normative. Distinguishing mere standards of correctness from formal norms giving rise to these distinctive reasons will occupy Section 10.3.

10.3. FORMAL OBLIGATIONS: WHAT’S W RO N G W I T H S C H M E T I Q U E T T E ? I’ve characterized substantive standards as those standards which are intrinsically (generic) reason-providing. But we can’t characterize formal standards nearly as neatly. I agree with philosophical consensus that obligations of etiquette don’t totally explain generic reasons to act as etiquette directs. It’s ²² No ought claim logically entails a reason claim unless “ought,” “reason,” and the bridge-principles connecting them are themselves of logical character. Inspection of the standard semantics for “ought” mentioned earlier bears out that this is implausible and unnecessary. At best, the connection between substantive obligations and reasons is conceptual or metaphysical. ²³ McPherson (2011: 236) makes a complementary point against Scanlon’s metanormative quietism. The problem is that we might have necessary coincidence between what a substantive normative standard demands and what Hades currently wants; yet intuitively the Hades standard isn’t “robustly normative.”

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

216

Jack Woods

only in tandem with reason to do as I oughtetiquette that the fact that I oughtetiquette ψ explains why I’ve generic reason to ψ. Of course, that I oughtschmetiquette would likewise explain that I’ve generic reason to be schmolite with appropriate background factors. It’s tempting to distinguish formal standards from mere standards of correctness by whether there are any reasons obey them, but we shouldn’t. There can be reasons, instrumental or intrinsic, to do as a mere correctness condition demands. Perhaps we’ve reason to obey schmetiquette because of its nose-snubbing charm. Nevertheless, schmetiquette is not a standard that’s important to us. Etiquette, in contrast, is the way that we do things; it’s an important standard for us. It’s this contrast that we need to zoom in on to distinguish formal standards from mere standards of correctness.

10.3.1. The Inescapability of Formal Obligation The key to distinguishing formally normative standards from mere standards of correctness is the fact that formal obligations aren’t escapable even when we have decisive reason to disobey them. In particular, others are in a position to criticize and upbraid us for breaking our formal obligations, no matter why we broke them; whereas, when we have no reason— instrumental, intrinsic, or whatever—to obey a mere standard of correctness, then it plays no role in our practical behavior. We’ll evidence this by looking closer at some of our formal systems of norms: promissory, legal, and etiquette norms. Plausible accounts of promises allow us to promise to do things which we pretty clearly ought not all-things-considered do. We might promise to do away with the person who cut in line, we might promise to pursue our mother’s campaign of total world domination. Are these promises obligatory in the promissory sense? Intuitively, and as I’ve argued elsewhere, yes (Woods 2016). We seem promissorily obliged to keep immoral promises— that’s how the promise game is played. If we fail to do so, some criticism on the behalf of the promisee is licensed by our acceptance of promissory norms even when, on balance, our reasons favor breaking our promise. Similarly, consider perjury. Suppose I’m in a position to perjure myself to put a horrible criminal away for a crime they obviously and clearly did. Suppose the risks of being caught are low, it would be a real service to society, etc. It seems plausible that I have all-things-considered reason to perjure myself.²⁴ Nevertheless, if I’m punished for perjury, I can’t complain ²⁴ For a good case of this, consider Omar Little’s testimony against Bird in The Wire.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

The Authority of Formality

217

that I did nothing wrong. After all, I broke the law. I can, of course, complain that the law stinks, that it should be changed, and that it shouldn’t be enforced. These, though, are quite different complaints.²⁵ Finally and famously, our etiquette obligations likewise don’t lapse even when we are indifferent to or should break them. As Foot put it, both morality and etiquette “are inescapable in that behavior does not cease to offend against either morality or etiquette because the agent is indifferent to their purposes and to the disapproval he will incur by flouting them” (Foot 1972: 311). For example, though it’s laudable and maybe even morally required that we cease using gendered norms of polite address, we are currently formally criticizable for so doing. People are within their rights to criticize someone for violating these norms. Rudeness is rudeness, even when warranted, and we can’t respond to a charge of particular rudeness except by giving reasons to violate the norm in this context; even then, the critic has legitimate grounds for complaint, even if it’s overly pedantic, obnoxious, or even immoral to actually criticize someone for doing so. That’s what it is for us to treat a standard as our standard.²⁶ Contrast the cases just described with standards not in force for us. Suppose we see people moving chess pieces around on a board and we correct a seemingly illicit play. The players can respond that they’re playing checkers with chess pieces. This would be a complete defense against our criticism. Likewise, suppose our bizarre uncle upbraids us for wearing pants to dinner. Again, pointing out that that it’s customary to wear pants to dinner is a complete defense. For standards like legality, etiquette, and promising, it’s typically odd to respond to this kind of criticism with the claim that you are not subject to these norms. We can say “I don’t care about being polite” or “I reject your bourgeois lifestyle,” of course, but there is a strong residue of strangeness. Criticism of us for violating politeness norms is licensed by our acceptance of politeness norms as our norms. Whether we have reason to be polite is

²⁵ A referee complains that this suggests Rosa Parks was obliged, in some subscripted sense, to sit in the back of the bus (and that she was criticizable for not doing so). This, again, is a feature, not a bug. Rosa Parks is liable to criticism for violating a norm presumably in force then (see discussion of liability in Section 10.3.2). She also had overwhelming reason to break this obligation and anyone actually leveling criticism at her for doing so would be doing something seriously immoral. She’s not criticism-worthy, in other words. ²⁶ Details vary, of course. In some contexts, I should sanction someone indirectly for being rude, perhaps by complaining about their behavior to someone else. Elsewhere, I should tell them to their face.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

218

Jack Woods

another matter; we might not. Still, we’ve acted rudely.²⁷ We can object when someone points out that we have violated some mere standard of correctness; “what’s schmetiquette to us” is a complete defense.²⁸ The distinction between formal norms and mere standards of correctness has to do with whether and when they are in force for us, as evidenced by their inescapability: when they are in force, we are bound by them. When they are not, we are not. Our next task will be to characterize what it is for a standard to be in force.

10.3.2. When Norms Are in Force: The Internal Point of View There are two obvious ways that standards can fail to be in force. First, they might not be part of our normative outlook. Schmetiquette, for instance, plays no part in structuring our practical behavior. Second, they might structure our practical behavior, but only somehow, somewhere, somewhen. Rules of games are like this, as are legal norms, bits of etiquette, and possibly morality (many of us think morality takes no stand on the color of my shoelaces). Chess norms apply when playing chess. We needn’t comply with such norms unless we’re engaged in the activity they govern and we need not so engage. This shouldn’t obscure the sense in which we are bound by formal norms when we are engaged in their target activity. We don’t get out of our obligation to move our bishop only diagonally because we didn’t want to play chess in the first place. We’re playing chess and when playing chess, chess norms are in force. When we break these norms, we are criticizable on this basis. Legal norms, in contrast, apply more broadly, but only in particular contexts. I am bound to obey British law given my UK residency, but not once I move back to the US. Similarly with etiquette: I am permitted, but not required, to tip in a pub. I am required, when tipping, to say something like “get one for yourself,” not just leave money on the bar. When I violate these norms, fellow members of my British community have the right to snigger at me. When in the US, I ought to tip. Not doing so legitimates significantly stronger sanctions than just sniggers. These considerations about applicability and inescapability limn what it is for a standard to be in force, but they don’t yet explain it. So we need to ²⁷ We can, of course, object to someone invoking norms we’re subject to for other reasons. Actually invoking certain rights we have can be immoral, pedantic, cruel, and politically dangerous. This is yet another reason that criticism of Rosa Parks was awful, even if licensed by in force norms. ²⁸ This doesn’t mean that we’re immune to criticism for being imschmolite. Such criticism needs, though, to be grounded in independent reason to avoid imschmoliteness.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

The Authority of Formality

219

explain how formal standards yield “real” obligations. To do so, I’ll draw on Hart’s idea of an internal point of view: [F]or it is possible to be concerned with the rules, either merely as an observer who does not himself accept them, or as a member of the group which accepts and uses them as guides to conduct. We may call these respectively the “external” and the “internal points of view.” (Hart 1961: 89)

The internal point of view involves at least taking a system of rules to guide our practical deliberations.²⁹ In his postscript, Hart explains “accepts”: [Acceptance of rules] consists in the standing disposition of individuals to take such patterns of conduct both as guides to their own future conduct and as standards of criticism which may legitimate demands and various forms of pressure for conformity. (Hart 1961: 255)

These quotations, in combination, suggest that the internal point of view involves taking a system of norms as generating “live” obligations in the sense described above. We’ll say that norms are in force for us when we—qua social group—take the internal point of view toward them.³⁰ The pluralized pronoun is important; being in force is a relationship between a social group qua social group and the norms it accepts, though exploring the details of this must wait for another occasion.³¹ Even if I’ve reason to violate legal norms, I’m open to being punished for so doing because we (a we of which I’m a part) view the pronouncements of judges as legitimate on the matter. I may disagree with us, but no matter. Even though I’m part of the legitimators, I need not agree ²⁹ An anonymous reviewer worries that it only need be legal officials who take the internal point of view for Hart. This is correct, but unworrisome as I’m only drawing inspiration from Hart. Hart’s position is that legal officials must take the internal point of view and that this is enough to undermine Austinian pictures of legal validity. Nevertheless, it’s clear that Hart allows that we non-officials can also use the law to guide our practical deliberations in the relevant sense—these are well-functioning legal systems (and ones I’d take to be clearly formally normative). Moreover, even when we don’t take the internal point of view to the law, in my view we need to treat certain folks—the “officials”—as in a position to set the law in order for it to be formally normative for us. This complication—vindication of the internal point of view from one remove—isn’t crucial for my picture, so I’ll bracket it for now. ³⁰ See Shapiro (2006; 2011: 95–8) for useful discussion. See Dorsey (2013) for critical discussion of the analogy between legality and other normative domains. My (2016) goes into detail on the legal analogy; in particular, I there discuss how to use something akin to rules of recognition and adjudication to determinately fix the extension of conventional normative standards. ³¹ I’ll stick with intentionally generic language like “us taking δ seriously” and cognates in what follows, but it would be vastly preferable to give an account in terms of collective agency and group action since it might be that a majority of society are “bad [people]” in Hart’s sense (1961: 91). I hope to revisit this issue elsewhere.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

220

Jack Woods

entirely with them in order for our view to be that the judge fixes my liability to punishment. Connecting up our earlier discussion with the internal point of view, we’ll say standards get their status as formal standards when violations are taken to license criticism and where we take this criticism seriously in the sense of being a sanction. Criticism need not involve putting someone in the stocks. Hart recognized this early on: rules are conceived and spoken of as imposing obligations when the general demand for conformity is insistent and the social pressure brought to bear upon those who deviate or threaten to deviate is great. (Hart 1961: 84)

As suggested by Hart, in many cases, explicit recognition of violations is sanction enough. Consider someone pointing out failure to vote, rudeness, or unsporting behavior. It’s embarrassing for the gross majority of us to be pulled up on such charges.³² When we are part of a social group that takes these norms seriously— which is, again, to take them as licensing sanction like blame and criticism— then they are in force for us. It’s important to be clear about what this exactly means. First, taking the internal point of view toward a standard doesn’t require most individuals taking an internal point of view toward particular norms that underwrite that standard. We might not know what they are, after all, as is often the case in the law. Second, taking the internal point of view toward legal norms isn’t treating violation of legal standards— failure to meet our subscripted legal obligations—as domain-independent justification for punishing the offender. Justification for applying punishment involves additional materials such as the usefulness, morality, and reasonableness of punishment. Not every case of sanction-liability is a case of sanction-worthiness—just consider jaywalking and minor drug offenses. It’s easy to confuse sanction-liability with sanction-worthiness, but we should avoid doing so. Blameworthiness tends to be read as “being all-thingsconsidered worthy of blame.” It’s rather implausible that every violation of social norms is sanction-worthy in this sense. It’s plausible that someone is sanction-liable for doing so. Compare the intuitive distinction between

³² A reviewer complains that telling a gangster they’re a lawbreaker isn’t criticism. I don’t think that’s right—gangsters are typically part of our community and we take lawbreaking seriously as a criticism of them, even if they don’t seem to (obviously, it’s not obvious that gangsters don’t care about the law at all; normative psychology isn’t behaviorally transparent). If they’re appropriately related to legality being in force for us (on which, see below), they’ll also have reasonslegality to obey the law. Alternatively, if there were enough socially independent gangsters such that they aren’t part of our community, then presumably they wouldn’t take our laws as important for them.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

The Authority of Formality

221

prosecution-liability and prosecution-worthiness.³³ If, say, I wander over a do-not-cross line unmaliciously, then intuitively I am liable for, but not worthy of, prosecution.³⁴ It’s thus better to say that a social group takes the internal point of view to systems of norms when they treat offenders as being liable for sanction on that basis and view liability to sanction as a serious matter.³⁵ If this is right, then norms governing a particular activity ψ are in force when we are ψ-ing if a social group we are members of are disposed to regard the subscripted obligations grounded by this standard to be serious. That is, when they are disposed to treat us as liable to sanction because of our behavior with regard to these obligations and treat sanction-liability as to be avoided. Again, this needn’t require having detailed knowledge of when and how to apply these sanctions. Even etiquette, a clearly conventional standard, requires advice columnists. Whether a system of norms licenses sanction for particular actions is often only loosely grasped at. The degree of clarity varies from standard to standard. It’s often roughly clear what’s required to be legally upstanding—since we accept the rule of law as a constraint on acceptable legal systems—but this seems a somewhat special case.³⁶ There might not be any corresponding rule of morality or rule of etiquette. Moral and etiquette norms may only be loosely accessible to us. Again, we often focus on the lexicalized unspecific properties they ground like moral upstandingness. Don’t confuse our account of in-force-ness and use of games and legality as examples with commitment to treating morality as conventional. The only conventional feature of formal obligations that I’m committing to is that our view of certain standards is what gives them formal normativity. This is entirely consistent with these standards themselves being fixed by mind-independent properties.³⁷ Likewise, regardless of ³³ Daniel Wodak suggests (personal correspondence) that legality is special in this regard, pointing out that it might be rude to point out rudeness. This strikes me as an unfortunate feature of Commonwealth etiquette; we can be licensed to do something, the doing of which licenses comparable sanction against us. ³⁴ This is complicated by cases where the hands of the judges and police are tied, such as three strike laws and the like. See (Woods 2016: §2). ³⁵ Hart (1960) distinguishes between the justification of instances of punishment (in terms of a social good being realized by the practice) and the justification from within the practice of punishing someone (in terms of retributive considerations). This means we can hang someone out to dry even though it’s known to the authorities that they’re innocent. It’s more plausible, in my view, that there’s generic reason punish them, but they’re neither punishment-liable nor punishment-worthy. ³⁶ Of course, this constraint is violated often in practice. Thanks to Daniel Wodak for discussion. ³⁷ How to fix the content of morality is a notoriously difficult matter; I favor fixing it conventionally, but I recognize that I haven’t argued for that here. Stay tuned for future work on this.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

222

Jack Woods

our treatment of reasonsδ, there will likely be non-conventional reasons as well as instrumental reasons to do as morality directs. It’s undeniable that many putatively substantive systems of norms, like morality and prudence, have sanctions built into them which we take seriously. We regard moral blame as serious and take claims about imprudence to guide our behavior. Clearly we also take etiquette, rules of games, and many other merely formal standards seriously. So, at least in my defined sense of a standard being in force, it seems that both putatively substantive and merely formal standards can be and often are in force for us.

1 0. 4 . T H E D I S T I N C T I V E R E A S O N S O F F O R M A L O B L I G A T I O N : R E S P E C T - B A S E D R E AS O N S Recapping briefly, I’ve argued that norms are in force when we take the obligations they support to be real obligations for us, in the sense of licensing sanction for violating these obligations. This might range from something like social sanction (etiquette), actual punishment (criminal behavior), or mere recognition of being a rule-breaker. I distinguished this sanctionliability from sanction-worthiness since license to sanction isn’t justification for so sanctioning. So, formal norms are in force when (a) we are engaged in the activity they govern and (b) members of a social group who are disposed to treat us as liable to sanction when we act as we shouldn’t (by its lights) and who are disposed to take the sanctions so licensed seriously. We still need one more piece before we can tie this all together into a neat picture of the distinctive reasons that come from formal obligations. Consider again that mere standards of correctness can, with appropriate background conditions, explain reasons. Schmetiquette can explain instrumentally— because it would further an end of ours to be schmolite—or indirectly— because schmetiquette tracks some other source of reasons. But these aren’t the distinctive reasons we’re after.

10.4.1. The Distinctive Reasons of Formal Obligation: Respect-Based Reasons To isolate reasonsδ, we need there to be something special about taking standards as giving us reasons in their role as our standards. The idea is that we’ve reason to obey δ because δ is in force for us. There’s something significant about taking standards to be standards. When we do so, we’ll get a particular type of reasons (which I’ll call respect-based reasons) to act out

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

The Authority of Formality

223

of the fact that these norms are in force for us. That’s what’s special about inforce standards; they are literally guides to behavior in that there are reasons to comply with them in virtue of them being our standards. Any account of reasons where that the fact that I desire something yields pro tanto reason to do it implies that we have respect-based reasons. This is because what it is for δ to be in force for us is, partially, for us qua community to regard liability to δ-sanctions as undesirable.³⁸ So we, qua community, want (pro tanto) to avoid liability to δ-sanctions. So we’ve reason to obey δ in order to avoid δ-sanction-liability (this assumes the instrumental principle adumbrated in Section 10.4.2). This particular explanation turns on desire-based reasons, but this isn’t essential—I hazard that on all plausible pictures of generic reasons, we’ll have reason to be both morally upstanding and polite because both morality and etiquette are in force for us. Strictly speaking, I’ve only so far explained why we, qua social group, have respect-based reason to obey norms. Since individuals don’t have to agree with their community about realness of δ-obligations in order for δ to be in force, these individuals might not have reason to obey δ. Nevertheless, as a community, we’ve reasons fully explained by the fact that δ is in force for us. Drawing this together, we have respect-based reason to obey δ when the fact that we have reasons to obey δ is fully explained by δ’s being in force for us. Derivatively, an individual will have respect-based reason to obey δ when they’re appropriately related to the fact that δ is in force for their community such that this latter fact, in tandem with their relationship to it, fully explains their reasons for to obey δ. Being appropriately related will differ with particular accounts of generic reasons. On a value-based picture, my membership in the community for which δ is in force might make obeying δ valuable, thereby giving me reason to obey δ.³⁹ Perhaps my observance of δ reinforces our acceptance of δ as our norms. On desire-based pictures, my respect-based reasons might come from my attitudes partially constituting δ being in force for us—that is, from my desire to avoid δ-sanction.⁴⁰

³⁸ Undesirable in a general sense. Individuals will often have more particular desires to avoid liability to particular folks for particular infractions. See (Woods 2016: 93) for discussion. ³⁹ For a useful example of how to spell out “appropriately related” on a value-based picture, see (Scheffler forthcoming). Scheffler’s account denies reason to act on immoral demands from non-moral standards, but this restriction can be removed without additional cost to the view. ⁴⁰ See my (2016) for extended discussion of desire-based pictures of this type for promissory obligations. Desire-based pictures need to avoid Schroeder’s “elusive reasons” cases (2007), but this can be finessed by a proper account of what it is to act on a reason (Sinclair 2016).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

224

Jack Woods

10.4.2. The Distinctive Reasons of Formal Obligation: Reasonsδ For our full account of reasonδ, we need a minimal instrumental constraint on generic reasons: INSTRUMENTALISM: If I’ve reason to ψ and φ is a necessary means of ψ-ing, then I’ve correspondingly strong reason to φ.⁴¹

Now, suppose δ is in force for us. We’ll then have respect-based reasons to obey δ which are totally explained by δ being in force for us. The combination of these reasons and the actual contours of δ explain why we have reason to ψ—because ψ-ing is how to obey δ and we have respect-based reason to obey δ. Individuals, in turn, will have analogous reason to ψ when they’re appropriately related to δ being in force for their community and thereby have derivative respect-based reason to obey δ. We now can define the distinctive reasons arising from formal obligations, our reasonsδ: REASONSδ: A reasonδ to ψ is a reason to ψ whose weight derives from our respect-based reason to obey δ.

For example, reasonprudence to sleep derives from my respect-based reason to have prudent behavior; its weight is a function of the strength of my respectbased reason to be prudent and how imprudent staying awake would be. Staying awake is strongly imprudent so, by instrumentalism, insofar as I’ve strong respect-based reason to be prudent, I’ve strong reasonprudence to sleep. Again, reasonsδ differ from so-called “institutional” reasons—a “reason” from “inside” of δ (Joyce 2001: §2.1).⁴² For example, etiquette might say decisively that we shouldn’t play practical jokes on friend Fulya. But, given a lack of strong respect-based reason to be polite, we might have significantly weaker reasonetiquette to refrain. Poor Fulya. Institutional reasons are not real reasons though they approach such when they are backed by appropriately strong respect-based reasons. ⁴¹ Be careful here not to confuse sanctionable—in the sense that someone might actually sanction us—with liability to sanction. The central application of this principle, in constructing reasonsδ is obtaining reasons to avoid liability to sanction, not avoiding the possibility of someone being in a position to sanction. The latter requires something like knowledge that I’ve violated δ. But even if everyone thinks I’m being polite, I might be rude, and thereby liable to criticism. Systems of norms δ where there’s a referee with fancy illocutionary powers complicate matters, but our reasonsδ are likewise complicated there. I’m open to complicating this principle for the usual reasons, but simplicity suffices here. ⁴² See Schafer (2016) for useful discussion of how institutional reasons of, say, morality can privilege prudence over itself. The topography of normative reasons becomes complicated once we take seriously the plurality of normative domains.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

The Authority of Formality

225

For formal systems of norms δ and γ, our reasonsδ and reasonsγ will be commensurable, on standard pictures of generic reasons, as they and their weight come from a single source. On the other hand, our respect-based reasons to obey δ and obey γ might be on a par. We shouldn’t expect all formal dilemmas to be resolvable, but we’ll only face the dualism of (formal) reasons worried about by Dorsey (2013) and Copp (2004) when we really do face a generically normative dilemma. Of course, if some formal systems of norms are also substantive, we may still face a dualism involving the generic reasons so explained.⁴³ We now turn to some closing objections to our view of formal normativity.

10.4.2.1. Objection: The Generality of Reasonsδ Reasonsmorality are derived from our respect-based reasons to be moral. This raises questions about individuation and extension-fixing conditions of moral correctness. After all, substantive norms like morality are purportedly universal, unlike context-sensitive properties like legality. If we fix moral correctness by local moral conventions, we’ll have to explain away this intuition.⁴⁴ Luckily, this isn’t a worry for the general picture I’ve painted. I’ve said nothing about how to fix the extension of moral correctness. If we do so via conventions, then our reasonsmorality will vary from context to context just as our reasonsetiquette clearly do. If not, then many of us make mistakes about morality, just as we might do with etiquette and legality when we’re not careful. My account is consistent by design with any number of ways of fixing the extension of moral correctness. My claim is that our behavior toward the property of being morally upstanding makes morality in force for us. If, then, we didn’t take morality seriously we wouldn’t have reasonsmorality to act. There might still be generic reasons explained by morality, but there wouldn’t be the distinctive reasons of the sort I’m investigating. I don’t worry much about this. We are moralizers, we thereby have distinctively moral reasons, and it’s unclear whether we could fail to be moralizers. The possible lack of reasonsmorality shouldn’t be terrifying since it’s difficult to understand what a non-moralizing community would even look like (Street 2009). Moreover, we’ll typically have other generic reasons to do recognizably moral actions. Being unconcerned with morality doesn’t eradicate our sympathy toward other beings, the social benefits of

⁴³ The worry isn’t that substantive reasons are of different types; the worry is that the status of morality and prudence as both substantive does nothing to explain the weight of the reasons explained by each. ⁴⁴ This can be done. See my (in press).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

226

Jack Woods

behavior coordination, or the value of treating each other fairly. It’s a good result if the distinctively moral character of our reasons is bound up with our treatment of morality as a formal normative standard; that’s how moralizing feels to many of us. I’ve addressed these issues in detail elsewhere (2016, in press), so I won’t pursue them further here. A more significant problem is how to individuate morality from etiquette and honor codes. I cannot address this difficult project here. It’s presumed by the theory I’m offering that there’s a way to do this, but, admittedly, it’s not a presumption I’ve made good on. We should, though, be optimistic. Even though the borders of etiquette and morality bleed together, we’re typically capable of distinguishing paradigmatic cases of each.⁴⁵

10.4.2.2. Norm-Fetishism Another worry is that my account fetishizes norms (Smith 1994).⁴⁶ As I used respect-based reasons to explain reasonsδ, I’ve built on bare concern for being morally upstanding or polite, not responsiveness to underlying features of actions that explain why doing such and so is moral or not. I shouldn’t want to care to save my partner from dying because it’s morally right; rather, I should be responsive to the moral-making features of saving my partner. Saving my partner because (de dicto) it’s the right thing to do is somehow objectionable. This objection misunderstands my account. I took no stand on moral motivation. Correctly acting on morality might involve responsiveness to underlying features which make it moral. All I’ve claimed is that reasonsmorality are explained by morality being in force for us and that this involves us taking morality seriously as a system of rules governing practical behavior. Moreover, the explanation of why we take morality seriously may itself be explained by the properties which fix the moral standard. For example, some of us take morality seriously as a system of rules governing practical behavior because we take immorality as indicative of a lack of empathy for others. This doesn’t imply any sort of objectionable fetishism. On the other hand, it would be costly if my account ruled out δ-fetishism entirely. Fetishistic concern with obeying a standard is possible, actual, and even sometimes desirable. Our distinctively bureaucratic reasons are a bit like this; spending time in bureaucratic anarchy long ago convinced me

⁴⁵ A natural suggestion uses types of sanction to distinguish them. If there’s a distinctive moralized form of blame, for instance, then we can treat morality as that which licenses morally blame-liability. Another attempt uses platitude-first accounts of normativity familiar from Jackson (1998) and Wright (1992) to divide up standards. ⁴⁶ See Svavarsdóttir (1999: §6) for a useful overview and trenchant criticism.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

The Authority of Formality

227

that fetishistic concern with bureaucratic procedure can be good for a community.⁴⁷

10 .5 . CONC L U S ION Our distinctive reasons to do as we oughtδ are totally explained by δ being one of our normative standards. We have such distinctive reasons to be polite, moral, prudent, and even to play pool correctly. The structure of our explanation of reasonsδ, for each standard δ, is the same, whether δ is traditionally substantive or merely formal. Importantly, my account explains how both morality and etiquette play their actual role in guiding behavior, informing our practices of kudos and sanction, and regulating normative deliberation. It shows how clear conflict between different normative systems is sometimes resolvable, avoiding some worrisome normative dualisms. So long as we’ve enough respect for morality, etiquette, and prudence—as I hope and believe we do—this account looks tidy, informative, and accurate to Foot’s point about the seriousness of formal obligation, moral or otherwise. If morality is also substantive, the major difference between it and etiquette is that I’ll have generic reason to ψ just because I oughtmorality ψ. But there will also be reasonsmorality to ψ. So regardless of what else you believe about morality and other putatively substantive standards, we should all recognize formally normative standards and their corresponding reasonsδ. But, given all this, why should we continue to believe that some standards are substantive? Why aren’t reasonsδ enough? After all, we already accept instrumental generic reasons to ψ and generic reasons to ψ. Since we take morality seriously, we also have reasonsmorality to do as morality commands. So we’ll often have reasonsmorality, on top of generic and instrumental reasons, to do as morality directs. Given the tidy package these three kinds of reasons comprise, why ask for more? My account raises a serious challenge for any theorist who accepts that some systems of norms are substantive: find some role for the generic reasons of substantive domains which can’t be subsumed under a plausible account of generic reasons to take morality seriously as a normative standard. I bet any remaining interesting distinctions between putatively substantive obligations and merely formal obligations can be reduced to a

⁴⁷ Thanks to Jamie Dreier and Geoffrey Sayre-McCord for discussion of this worry for my account.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

228

Jack Woods

distinction between our reasonsδ or our generic reasons to take δ seriously. But I’ll leave selling that wager to another occasion.⁴⁸

References Broome, J. 2013. Rationality Through Reasoning. Oxford: Wiley-Blackwell. Chang, R. 2002. “The Possibility of Parity,” Ethics 112(4): 659–88. Chrisman, M. 2015. The Meaning of “Ought”: Beyond Descriptivism and Expressivism in Metaethics. Oxford: Oxford University Press. Copp, D. 2004. “Moral Naturalism and Three Grades of Normativity,” in P. Schaber, ed., Normativity and Naturalism. Oxford: Oxford University Press, 7–45. Dorsey, D. 2013. “Two Dualisms of Practical Reason,” in R. Shafer-Landau, ed., Oxford Studies in Metaethics, vol. 8. Oxford: Oxford University Press, 114–39. Finlay, S. 2014. Confusion of Tongues: A Theory of Normative Language. Oxford: Oxford University Press. Finlay, S. forthcoming. “Defining Normativity,” in D. Plunkett, S. Shapiro, and K. Toh, eds., Legal Norms, Ethical Norms: New Essays on Metaethics and Jurisprudence. Oxford: Oxford University Press. Foot, P. 1972. “Morality as a System of Hypothetical Imperatives,” Philosophical Review 81(3): 305–16. Hart, H. L. A. 1960. “Prolegomenon to the Principles of Punishment,” Proceedings of the Aristotelian Society 60(1): 1–26. Hart, H. L. A. 1961. The Concept of Law. Oxford: Clarendon Press. Jackson, F. 1998. From Metaphysics to Ethics: A Defense of Conceptual Analysis. Oxford: Oxford University Press. Joyce, R. 2001. The Myth of Morality. Cambridge: Cambridge University Press. Kratzer, A. 1977. “What ‘Must’ and ‘Can’ Must and Can Mean,” Linguistics and Philosophy 1(3): 337–55. McPherson, T. 2011. “Against Quietist Normative Realism,” Philosophical Studies 154(2): 223–40. Maguire, B. 2016. “The Value-Based Theory of Reasons,” Ergo 3(9): 233–62. Maguire, B. and J. Woods. manuscript. “Explaining Epistemic Normativity.” Parfit, D. 2011. On What Matters: Vol. 1. Oxford: Oxford University Press. Ridge, M. 2014. Impassioned Belief. Oxford: Oxford University Press.

⁴⁸ Thanks to (at least) Derek Baker, Mitch Berman, Jamie Dreier, Ulrike Heuer, Nathan Howard, Zoë Johnson-King, Gerald Lang, Errol Lord, Barry Maguire, Tristram McPherson, Erum Naqvi, Geoffrey Sayre-McCord, Lucas Thorpe, Pekka Väyrynen, Ken Westphal, Robbie Williams, Daniel Wodak, Bill Wringe, and Gözde Yıldırım for useful discussions of this material and to audiences at the Uppsala Language and Metaphysics of Normativity conference, the Chapel Hill Metaethics Workshop, Bilkent University, and Normativity in Action II for useful feedback.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

The Authority of Formality

229

Schafer, K. 2016. “The Modesty of the Moral Point of View,” in E. Lord and B. Maguire, eds., Weighing Reasons. Oxford: Oxford University Press, 241–56. Scheffler, S. Forthcoming. “Membership and Political Obligation,” Journal of Political Philosophy. Schroeder, M. 2007. Slaves of the Passions. Oxford: Oxford University Press. Schroeder, M. 2011. “Ought, Agents, and Actions,” Philosophical Review 120(1): 1–41. Shapiro, S. 2006. “What is the Internal Point of View?” Fordham Law Review 75(3): 1157–70. Shapiro, S. 2011. Legality. Cambridge, MA: Harvard University Press. Sinclair, N. 2016. “On the Connection between Normative Reasons and the Possibility of Acting for those Reasons,” Ethical Theory and Moral Practice 19(5): 1211–23. Smith, M. 1994. The Moral Problem. Malden, MA: Blackwell Publishers. Street, S. 2009. “In Defense of Future Tuesday Indifference: Ideally Coherent Eccentrics and the Contingency of What Matters,” Philosophical Issues 19(1): 273–98. Svavarsdóttir, S. 1999. “Moral Cognitivism and Motivation,” Philosophical Review 108(2): 161–219. Tiffany, E. 2007. “Deflationary Normative Pluralism,” Canadian Journal of Philosophy 37(5): 231–62. Williams, B. 1979. “Internal and External Reasons,” in R. Harrison, ed., Rational Action. Cambridge: Cambridge University Press, 101–13. Woods, J. 2016. “The Normative Force of Promising,” Oxford Studies in Normative Ethics 6: 77–101. Woods, J. in press. Footing the Cost (of Normative Subjectivism). Wright, C. 1992. Truth and Objectivity. Cambridge, MA: Harvard University Press.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

11 Skepticism About Ought Simpliciter Derek Baker

What ought I to do? A lot of things, and frankly, too many things. Too many things, because there are too many oughts. Morally I ought to give to charity, prudentially I ought to invest. Epistemic reasons may demand that I begin to doubt my friend’s innocence, while loyalty forbids it. So in some cases it is impossible to satisfy all of these oughts.¹ A natural thought in response to this kind of situation is to ask “what ought I really to do?” or “what ought I to do simpliciter?” But this natural thought, I will argue, should be rejected. That’s because the only coherent candidate notion of an ought simpliciter, an all-things-considered ought, or what Philippa Foot called “the free and unsubscripted” sense of ought (1972/1977: 169)² comes with normative commitments that are strange and unpalatable. This is not to deny normativity altogether. There may be facts about what you morally ought to do. There are almost certainly facts about what you rationally ought to do and what you prudentially ought to do. But there is no ought that has the job of adjudicating conflicts between these other oughts.³

11.1. PRECEDENTS: COPP AND TIFFANY Arguments denying the existence of an ought simpliciter have been offered before. It’s worth looking at these both to see the wider consequences

¹ I will talk in this chapter about multiple oughts and multiple senses of ought. None of this should be interpreted as a commitment to polysemy (i.e., massive ambiguity of meaning) about the word “ought.” I am sympathetic to recent contextualist accounts (e.g., Dowell 2012 and 2013; and Finlay 2009 and 2014) which attempt to give a unified theory of different uses and senses of ought. ² The quote is from the footnote 15 added in 1977. Also see Tiffany (2007: 232 and 233). Note that Foot wrote that “I have never found anyone who could explain the use of the word in such a context” (1972/1977: 169). ³ But see McPherson (this volume) and Woods (this volume).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Skepticism About Ought Simpliciter

231

of accepting such a denial, and to mark out where the argument of this chapter differs. Evan Tiffany’s (2007) argument begins with the plurality of evaluative standards. When offering reasons to justify our actions, we cite, “desires, legal statutes, social norms, aesthetic value . . . moral value, norms of formal etiquette, future interests, to name a few” (Tiffany 2007: 239). Perhaps there is some way of reducing all of these to a single kind of consideration, but that requires considerable theoretical interpretation. On the face of it, there are distinctive kinds of reasons which don’t share a common justificatory source (Tiffany 2007: 239). But if the kinds of reasons we cite are plural, how do we combine them to reach a conclusion about what to do, especially when they compete? What would be the common scale on which considerations of morality, desire satisfaction, aesthetic value, and etiquette could be compared? Actually, there is no real mystery here. The question of how we combine potentially competing pluralistic reasons is one about our psychology, about what we in fact do (Tiffany 2007: 241ff.). This is an empirical, not a normative problem, and presumably not every agent will combine their reasons in the same way.⁴ There seems to be a further question, however, of how we should combine these reasons. But this, Tiffany argues, can only be answered from a particular normative perspective, of which there are many (2007: 241–3). Tiffany admits that this “deliberative pluralism” is not logically entailed by pluralism about reasons. But it is the most natural consequence: “[I]f some standpoint is capable of generating genuine contributory reasons, why could it not also serve as a legitimate source of deliberative evaluation?” (2007: 247). Once we acknowledge the variety of normative considerations that are out there, it looks extremely unlikely that there could be some way of adding these considerations up that isn’t guilty of arbitrariness in its weightings. Consequently, there will be a plurality of (equally arbitrary) ways of combining these considerations. We can criticize one way of combining reasons from another standpoint, but that standpoint is itself equally criticizable from others, and none deserve the status of “reason as such” (2007: 240). One can, nonetheless, hold on to the conviction in the rightness of one’s own standards: “Just as one may be a partisan supporter of the Canucks over the Maple Leafs . . . without thinking that there is some deep metaethical truth backing up one’s partisanship; so too can one be similarly partisan toward, e.g., morality, prudence, or authenticity” (2007: 244–5). What Tiffany’s pluralism rules out is establishing the rightness of one’s standards ⁴ Tiffany talks about an agent’s “standards of practical deliberation” (2007: 241), but it is clear that these standards can be realized by inarticulate psychological tendencies to choose one way or another as much as by consciously held principles.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

232

Derek Baker

through philosophical argument, or recognizing it through noetic insight into the truth. Coming to see certain standards as right is ultimately a matter of “existential choice” (2007: 243–5). This is essentially an argument from pessimism. Given the apparent plurality of kinds of reasons we cite to support our responses, we are owed grounds for the apparently widespread optimism about the possibility of a nonarbitrary standard for weighing seemingly different kinds of reasons against one another.⁵ I am in agreement with Tiffany on this. But a stronger form of skepticism is available. His argument gives grounds for skepticism that any normative standard could exist which satisfies our concept of reason as such or which issues prescriptions satisfying our concept of an ought simpliciter. I will argue instead that the only coherent concept of an ought simpliciter comes with unacceptable normative commitments. David Copp’s (1997) argument is closer to mine in its aims. If reason as such is to play the role of adjudicating between competing normative perspectives, it must have a special normative authority that they lack. Copp aims to show that any attempt to explain this authority forces us to embrace a contradiction. Copp starts with a familiar thought: self-interest and morality can conflict. Gyges would be better off as king, whatever Plato thought, but murder is morally wrong. So, Gyges ought self-interestedly to kill the king, while morally he ought to refrain. But now we face the question: which of these two standards, self-interest or morality, is “normatively more important” (Copp 1997: 93ff.)? Well presumably we ought self-interestedly to listen to the ought of self-interest, and morally we ought to heed the moral ought. Morality is morally more important, and self-interest is prudentially more important. What we need is an independent standard from which to assess these two competing demands. This standard, moreover, must be the normatively most important standard: “We want to know whether moral reasons override . . . period, not merely whether moral reasons are overriding as assessed by some standard or other” (Copp 1997: 94). But how could there be a normatively most important standard, or a standard of reason as such? After all, it seems that assessments of the importance of some standard can only be made from the perspective of some standard. Copp explains: Hence, the claim that the candidate [standard] S has the property of supremacy is the claim that it is normatively more important than any other standpoint, as assessed from a relevant authoritative standpoint. That is, if S is normatively the most ⁵ See Chang (2004) for an example of how one might defend this optimism.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Skepticism About Ought Simpliciter

233

important, then there is some authoritative standard R that yields the verdict that S is normatively the most important standpoint. (1997: 101)

Copp argues that this leads to a dilemma: “either standard R is identical to S, or it is not” (1997: 101). But if R and S are identical, then R cannot serve as the standpoint from which to establish S ’s normative supremacy. “For a standard cannot be normatively the most important in virtue of its meeting criteria that it itself specifies as criteria to be met by standards” (1997: 101–2). After all, morality tells you to listen to morality; self-interest tells you to listen to self-interest.⁶ Self-endorsement is unimpressive. On the other hand, if R is some standard other than S, we run into absurdity. It won’t do if R is just some normative standard or another—if self-endorsement is unimpressive, the fact that there is some other normative standard giving S the thumbs up is also unimpressive. We want to know whether S is most important according to the most important standard. So R would have to be the most important standard. But then S is not the most important standard. This contradicts our initial assumption (Copp 1997: 102). The argument I will offer will have obvious affinities with Copp’s. But his argument assumes that S ’s normative supremacy must be explained in terms of the supremacy assigned to it from some standpoint. S’s supremacy, however, could be normatively primitive.⁷ From which point of view would reason as such be assessed as normatively most important? From its own. R = S. Copp’s mistake is assuming that the philosopher who says this must go on to explain S ’s normative supremacy on the basis of the normative importance R assigns it, and thus must say that S is “most important in virtue of . . . meeting criteria that it itself specifies as criteria to be met by standards.” But it could be that S is most important in virtue of nothing at all (McLeod 2001: 274 and 286, italics mine). Taking on brute primitives is of course a mark against a position, but consider the theoretical landscape. If normative non-naturalism is correct, then whichever normative properties are most fundamental—the property of being a reason, say—admit of no further explanation. Once we’ve said this, it doesn’t seem like much more of a theoretical burden to go on and add that the status of those reasons as normatively most important is primitive as well. On the other hand, if some variety of naturalism about the normative is true, then the normative supremacy of S presumably does hold in virtue of something else, but that something else is not itself a

⁶ But see Schafer (2016). ⁷ For other criticisms of Copp’s argument, see McLeod (2001) and Dorsey (2013).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

234

Derek Baker

normative standard; rather it will be, for example, S ’s reduction base, or something similar.

1 1. 2 . T H E AR G UM E N T F O R S K E P T I C I S M Let’s say the following is true of Gyges: he ought prudentially to kill the king, but morally he ought to refrain. Or we can think up other conflicts. Zach Snyder ought self-interestedly to direct Batman v Superman, but from the perspective of aesthetic value he ought to decline. Perhaps in such cases the conflict is resolved by what I ought to do simpliciter, or full stop. But the initial conflict resulted from too many oughts, not too few. So how does adding another ought help? Why doesn’t it simply breed more conflicts? Let’s say Gyges ought simpliciter to kill the king. Refraining from murder remains, by stipulation, what he ought morally to do. The choice remains immoral, even if he ought full stop to do the immoral thing. At first glance an ought simpliciter simply adds an additional conflict. It’s unclear how it resolves anything. If the conflict is resolved, it is because the prescriptions of the ought simpliciter trump⁸ other prescriptions, or are overriding,⁹ or have greater normative authority¹⁰ than the demands of self-interest, or the ought simpliciter tells you what you really ought to do; or the ought simpliciter is robustly or genuinely normative;¹¹ or the ought simpliciter has normative force¹² whereas the rival ought does not (or possesses less of it). But talk of normative force is, to put it bluntly, completely metaphorical. The claim that the ought simpliciter is the ought that really tells you what you ought to do relies on the table-thumping sense of “really” (Baker 2017). In the literal sense, the prudential ought really—that is, just as genuinely and accurately—tells you what to do as any other ought. “Overriding” and “normative authority” rely on metaphors of political power, or they are simply vague ways of gesturing at some normative property which needs clearer characterization. Talk of one ought trumping another is a metaphor referring back to card games. These metaphors and table thumps need interpretation. The most natural interpretation is that one ought to act as the ought simpliciter prescribes, rather than the prudential ought. But on this interpretation the initial ⁸ McLeod (2001: 271). ⁹ Copp (1997); McLeod (2001: 271). ¹⁰ Copp (1997: 101); Tiffany (2007: 248). ¹¹ McPherson (2011: 233); Broome (2013: 11 and 26–7). Broome contrasts “true normativity” with normativity in a wider sense. ¹² Parfit (2011: 35).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Skepticism About Ought Simpliciter

235

conflict will reappear at the level of which ought we ought to heed. We ought in some sense to heed to the ought simpliciter when it conflicts with the prudential ought; but obviously one ought prudentially to heed to the prudential ought, otherwise one is a sucker or sap. This conflict forces us to ask whether this new sense of ought—the one telling us we ought to heed the ought simpliciter—is overriding or not. If it is, our attempt to characterize overridingness non-metaphorically makes tacit appeal to the original metaphor.¹³ If not, invoking such an ought could not resolve our original dilemma.¹⁴ Alternately, certain oughts might be psychologically overriding. The agent’s conclusion about what this ought prescribes determines what the agent will in fact do. If Gyges judges he ought full stop to refrain from murder, then the conflict is settled in the sense that he will act in compliance with the moral ought and not the prudential. But cashing out “overridingness” in psychological terms seems to change the subject. If all we meant by resolving the conflict was making a choice in the face of a conflict between different flavors of ought, it is unclear why a further ought would be needed at all. Gyges’ personal preferences would have worked just as well. This is my case against an ought simpliciter. This ought is supposed to have some special property in virtue of which it resolves conflicts between other flavors of ought. The characterization of that special property is typically metaphorical or otherwise hopelessly vague and ambiguous. Attempts to cash out the metaphor or otherwise eliminate the vagueness face a dilemma: either they characterize the special property in more familiar normative terms, leading to vicious circularity, or they are non-normative characterizations that seem to change the subject. There is one characterization that avoids this dilemma, but it comes with highly counterintuitive normative commitments. These commitments are distasteful enough that it is reasonable to conclude that we have failed to characterize a concept of the ought simpliciter that we currently use or could learn to use. My reason for assuming that a characterization of the ought simpliciter is needed is straightforward: this, and phrases like “authority” or “normative force,” are philosophical terms of art. The burden is thus on those who would invoke these terms in explanations to tell us what they mean. Sections 11.3 and 11.4 will call this into question, considering reasons why perhaps ¹³ “The relevant notion of [normative] authority is difficult to characterize noncircularly” (Tiffany 2007: 248). ¹⁴ There are obvious similarities to Copp’s argument described earlier. The key difference is that I am not asking for an explanation of normative authority or normative force but rather a clear account of the property in a manner free from metaphor. Also see Baker (2017).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

236

Derek Baker

we are not owed a definition or why ought simpliciter is not a term of art. I will explain why these arguments fail. But this section will deal directly with attempts to characterize ought simpliciter informatively.

11.2.1. The Ought that Settles What to Do Let’s say that the ought simpliciter is the one which settles what to do (or feel, or believe).¹⁵ Here the ambiguity between a normative and a psychological reading is obvious. We could mean that this ought settles what one ought to do, or we could mean that it settles what one will do. On the former reading, we have failed to identify anything special about the ought simpliciter. It is a clear triviality that ought in any sense settles what one ought in that sense to do. On the second reading, the ought simpliciter is special in that it motivates an action. But, as noted before, if all we needed was that Gyges be motivated to act in the face of the dilemma, it’s not clear why a special sense of ought was necessary at all. A strong enough desire would serve just as well.

11.2.2. The Categorical Ought Perhaps we can say that the distinctive feature of the ought simpliciter is that it is the categorical use of ought, as opposed to all the others which are hypothetical. Unfortunately “categorical ought” is ambiguous (Foot 1972/1977; and Joyce 2001). Kant’s original example of a categorical imperative was an ought-judgment that remains valid even when one lacks subjective desires that would be furthered by following it. Notoriously, the ought of etiquette and the legal ought demonstrate this kind of categoricity. On the other hand, if we mean that these oughts entail the existence of reasons, we are again giving a normative characterization. This might not seem obviously problematic. At least we are not characterizing the ought simpliciter in terms of a further ought, or in terms of vague bits of philosopher-speak, but in terms of reasons. The problem is that other oughts entail the existence of reasons in some sense. If you morally ought to do something, then there are moral reasons to do it. If you prudentially ought to do something, there are self-interested reasons in its favor. If etiquette requires something, there are reasons of etiquette to do it. This is just to note that for any normative standard, we can specify considerations in virtue of which, in these particular circumstances,

¹⁵ Wedgwood (2007: 25); Joyce (2001: 50).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Skepticism About Ought Simpliciter

237

this particular action (or attitude) is prescribed or favored.¹⁶ But when we say that categorical oughts entail reasons, we presumably mean reasons simpliciter.¹⁷ This doesn’t help, however, because we can simply run our initial argument again. Gyges’ prudential reasons favor killing the king. His moral reasons favor refraining. So there’s a conflict. We appeal to reasons simpliciter to settle the conflict. But the dilemma resulted from too many kinds of considerations, not too few. So how does introducing another class of reasons help? There must be something special about reasons simpliciter, but what? Do they possess normative force? Are they the reasons we ought to heed (or have most reason to heed)? We are merely pushing the bump in the rug.

11.2.3. The All-Things-Considered Ought Another thought is that the authority of the ought simpliciter consists in its status as the all-things-considered ought.¹⁸ Other oughts are based on a partial and incomplete collection of the relevant considerations. The ought simpliciter is special in that it is comprehensive—based on all of the considerations in favor and against. There is actually an ambiguity built into the notion of an all-thingsconsidered ought. We could mean that it is a prescription based on the total state of things—there is no further fact about how the world is that is left out in determining what one ought to do. If one likes, one can think of this as the ought issued as a bit of advice by an omniscient guru, who considers all the things, in some way or another. In contrast, by the all-things-considered ought one might mean the all-things-considered-as-they-ought-to-beconsidered ought. The former interpretation—on which what we mean is the all-thingsconsidered-in-some-way-or-another ought—is unacceptable. Whatever overridingness or normative authority amounts to, it cannot simply be the fact that more considerations go into the relevant prescription. Assume for ¹⁶ Compare this with Joyce’s discussion of institutional reasons (2001: 39ff.). Stephen Finlay’s end-relational account of reasons would make similar predictions (2014: 84ff.), see especially his brief discussion of perverse reasons (2014: 109). Also see Stroud (1998) and Dorsey (2013: 118). ¹⁷ Sarah Stroud (1998: 172–3) distinguishes reasons of social climbing from reasons simpliciter. Only the latter have “force” in practical reasoning. Dale Dorsey (2013: 118) likewise distinguishes domain reasons from practical reasons, writing that only the latter are normative (2013: 135ff.). ¹⁸ Wedgwood (2007: 24); this also seems to be the characterization found in Stroud (1998: 175).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

238

Derek Baker

argument that there is an all-things-considered ought. Notice that another normative system could take into account all the same considerations that serve as inputs to what one ought all-things-considered to do, but assign them radically different weights, resulting in a different prescription. “All-things-considered” must mean, then, “all-things-considered-as-theyought-to-be-considered.” But this is a normative characterization. So once again, we have a problem of circularity.¹⁹ We could mean all-thingsconsidered-as-they-ought-morally-to-be-considered, or all-things-consideredas-they-prudentially-ought-to-be-considered, or all-things-considered-as-theyought-simpliciter-to-be-considered. Presumably we mean the last—but what do we mean by the last?

11.2.3.1. All-Things-Considered, Reconsidered Perhaps this is too quick. There may be another way to characterize the allthings-considered ought, which is neither normative nor psychological, but metaphysical. Here the idea is that there is a single class of reasons, and terms such as “moral” and “prudential” designate restrictions on this class of reasons.²⁰ Moral reasons are just a subclass of the reasons, those reasons concerned with the interests of others, perhaps. Prudential reasons are those reasons bearing on one’s own interests. “Morally ought” just designates what’s favored when we restrict ourselves to those reasons that are moral. The all-things-considered ought, on the other hand, designates what all the reasons favor: it is the ought based on unrestricted consideration of the reasons. The advantage of this picture is that it seems to leave no question of how the all-things-considered ought is normatively superior: the all-thingsconsidered ought actually tells you what you ought to do, whereas the merely moral ought tells you what it would be the case that you ought to do, if only moral considerations were in play. Another way of putting this might be, all oughts besides the all-things-considered are merely prima facie; hence their inferiority. The difficulty with this characterization is how it handles ought-claims based on prescriptive social conventions, such as etiquette or positive law. Consider the ought of etiquette. As noted in Section 11.2.2, it will be connected to reasons of some sort. If you ought to reply to an invitation in the third person, there are considerations in virtue of which etiquette ¹⁹ But see Chang (2004: 11ff.) for the claim that we should leave room for the normative judgment of “What values should be at stake in an all-things-considered judgment?” ²⁰ Thanks to Daniel Wodak and Geoff Sayre-McCord for proposing an account like this one.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Skepticism About Ought Simpliciter

239

prescribes this action and not some other (that the invitation addressed you in the third person, for example). This leads to the question, does the all-thingsconsidered ought take these reasons of etiquette as inputs into what it prescribes, or not? Or to put this another way, we can follow Dale Dorsey (2013: 118, 135ff.) in naming the reasons on which the all-things-considered ought is based ATC (all-things-considered) reasons.²¹ Moral reasons are presumably a subclass of ATC reasons, as are reasons of prudence. The question is, are reasons of etiquette a subclass of ATC reasons?²² Neither answer will do. Let’s say that they aren’t. This looks like the more intuitive horn—that the reasons of etiquette don’t play a role determining the authoritative ought. But notice what this implies, given our explanation of normative authority. The strategy we are considering characterizes being less authoritative in terms of being prima facie, of being based on a subclass of the ATC reasons. But the ought of etiquette is based on all of the (non-ATC) reasons of etiquette. It is not prima facie in any sense. The all-things-considered ought is authoritative with respect to the moral ought in the sense that the moral ought is based on a restriction of the ATC reasons; this relation does not hold between the all-things-considered ought and the ought of etiquette. So we have not yet accounted for the authority of the all-thingsconsidered ought with respect to etiquette. Could one say instead that the all-things-considered ought is authoritative because it is based on the ATC reasons and the ought of etiquette is not?²³ But again, this is a normative characterization of the all-things-considered ought: we are characterizing it in terms of a relation to reasons. In fact, it is just a notational variant of the proposal considered in Section 11.2.2. The allthings-considered ought is tied to ATC reasons, but the ought of etiquette is connected to reasons of etiquette. What makes ATC reasons special? Their normative force? Again, we have merely moved the bump in the rug. One might argue that reasons of etiquette are not normative, whereas ATC reasons are (Dorsey 2013). But in what sense are reasons of etiquette not normative? They determine ought-facts that are prescriptive, not predictive. They favor and disfavor options. Claims about what you have reason of etiquette to do are not falsified by non-compliance. Non-compliance licenses criticism. They are connected to norms. The idea might be that they are not genuine normative reasons. But unless this means that claims about reasons ²¹ Note that Dorsey does not use ATC reasons in the manner in which they are being used here. ²² Thanks to Tristram McPherson and Michael Smith for independently suggesting an answer like this. ²³ Thanks to an anonymous referee for raising this question.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

240

Derek Baker

of etiquette are systematically false (which they aren’t), this is an emphatic, table-thumping sense of “genuine,” similar to the use of “really” in the phrase “what you really ought to do.” Again, we’ve pushed the bump in the rug. So perhaps we should agree that reasons of etiquette are ATC reasons. Notice, though, that we are now saying that etiquette has the same kind of normative standing that morality and prudence are typically taken to have. What’s more, this must also be true of every prescriptive social convention (or else our characterization of normative authority will remain incomplete). And, as I think everyone is aware, some prescriptive social conventions are awful. We have the Mafia’s code of Omerta, feudal norms, the rules of etiquette surrounding race under Jim Crow, the rules of etiquette surrounding gender, conventions governing blood feuds, and general norms of machismo. All of these conventions give reasons of the same normative kind as morality and prudence. That it would send a message about what happens to people who snitch on the Family would be an ATC reason for a Mafioso to torture someone to death, and play some role in determining what the Mafioso ought, all-things-considered, to do. That it would show a manly lack of concern for considerations of morality, suffering, or personal health would be an ATC reason for a man to reject vegetarianism. These ATC reasons apply simply as long as the people in question fall under the scope of the relevant convention, regardless of their personal opinions. The Mafioso can have come to hate the Mafia; the man may have long rejected the ideal of being a bro. They still have ATC reasons to torture or reject vegetarianism. But isn’t the skeptic about normative authority also committed to saying these agents have reasons to torture or eat meat? Yes, but so is everyone. This just follows from the fact that for any prescriptive social convention, we can specify the considerations in virtue of which it prescribes as it does in the particular case. The skeptic can also reject these reasons on the basis of partisanship. It’s only if we take the horn of the dilemma under consideration that we have to say the reasons are authoritative (since they are partial determinants of the authoritative ought). This horn may result in a coherent characterization of authority, but the normative implications are bizarre. That we find these commitments preposterous is good evidence that our concepts do not in fact work this way, and we have failed to characterize a notion of authority that we in fact use or could learn to use to resolve dilemmas.

11.2.4. The Ought of Rational Criticizability Perhaps an agent who fails to do what she ought simpliciter to do is irrational; or perhaps an agent who fails to do what she judges she ought

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Skepticism About Ought Simpliciter

241

simpliciter to do is irrational (e.g., Korsgaard 1986; Smith 1994; Hubin 1999; Wedgwood 2007; Kiesewetter 2011; Broome 2013; Dorsey 2013: 116; Dreier 2015). Either way, this would establish something special about the ought simpliciter: it is connected to rational criticism in a way that other oughts aren’t. But this is another normative characterization. To see why this is the problem, notice that the agent who fails to do what she ought morally to do is immoral. The agent who acts contrary to prudence is imprudent. The agent who acts contrary to etiquette is rude. It cannot simply be, then, that the ought simpliciter is overriding in the sense that it is tied to criticism. The tie to criticism is not unique to the ought simpliciter. It must be that there is something special about rational criticism. But as Donald Hubin asks in a similar context: “what, exactly, is it about the charge of irrationality that makes it carry more weight . . . than that of immorality? It is certainly not that being irrational is morally worse than being immoral” (1999: 40). Presumably we don’t want to know if irrationality is morally worse, we want to know whether it is worse simpliciter. Now we can run the argument again. It is morally worse of Gyges to do what he prudentially ought, but it is worse for Gyges to do what he morally ought. We have a conflict between values. Now let us add that it is worse simpliciter if Gyges listens to self-interest, and kills the king. But our problem resulted from too many evaluations, not too few, so how does adding a third evaluation help? The conflict still remains. It must be that there is something special about evaluation simpliciter, in virtue of which it resolves the conflict. But what could that be? Overridingness? Evaluative weight, as the corollary of normative force? Once again, we have merely moved the bump in the rug.

11.2.5. The Ought We Cannot Question Perhaps the special authority of the ought simpliciter is that “it is not available for legitimate questioning” ( Joyce 2001: 51). For any other ought-claim, a claim about what I morally ought to do or a claim about what I prudentially ought to do, it is sensible for the person to whom it applies to ask, “Why should I?” But this question is unintelligible for the ought simpliciter. I am asking for a reason to do what I have most reason to do; I am already implicitly granting the authority of reasons in my very question (Joyce 2001: 49ff.). Let’s grant that the questions, “Why should I do what I should do?” or “What reason do I have to do what I have most reason to do?” are selfundermining. As Joyce puts it: “[T]o question practical rationality is

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

242

Derek Baker

unintelligible—it is to ask for a reason while implying that no reason will be adequate” (2001: 51, italics mine). It doesn’t follow that the question “Why morally ought I to do what I ought simpliciter to do?” is self-undermining. This question does not imply that no reason will be adequate. It implies that a moral reason will be adequate. Someone could even fill her challenge in like this: “I get that I ought simpliciter to X. But I don’t care about that. I’m a moral fanatic. So you need to give me some moral reason to X, or I’m not going to do it.” Joyce addresses this possibility (2001: 50), but claims that insofar as one accepts moral reasons one accepts practical reasons. His grounds for this, unfortunately, are not entirely clear. If he means this as a psychological claim, it seems false. Intuitively, the moral fanatic or committed egoist are fully conceivable agents. The only way to rule them out is to hold that one regards a reason as authoritative simply in virtue of having some motivating commitment to act on it. But this is a psychological reduction of the notion of authority: we are now characterizing authority in terms of the agent’s motivations. On the other hand, without this implicit psychological characterization of authoritative reasons, the whole trick to showing challenges to reasons or the ought simpliciter self-undermining seems to rest on putting selfundermining words in the challenger’s mouth. “What reason do I have to do what I ought simpliciter to do?” is a self-undermining challenge. But so is “What moral reason do I have to do what I morally ought to do?” On the other hand, “What reason do I have to do what I have most moral reason to do?” is not self-undermining; but neither is “What moral reason do I have to do what I have most reason full stop to do?” All this shows is that it doesn’t make sense to challenge a kind of consideration or a sense of ought in terms of that very same kind of consideration or sense of ought. So don’t imagine that those who would challenge the ought simpliciter would phrase their challenges that way!

11.2.6. Accepting Psychological Characterization Perhaps we should accept a psychological characterization of the ought simpliciter’s special authority. One could easily imagine a Humean at this point saying, “Of course you can only characterize it in terms of psychological dispositions to be motivated: that’s what we’ve been trying to tell you!”²⁴ On this interpretation, ought simpliciter is the ought for which a

²⁴ See, for example, Finlay (2007).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Skepticism About Ought Simpliciter

243

motivational internalist constraint holds. It is the ought that makes up the content of those ought-judgments that are by their nature motivating. This is what its special authority consists in. This position could be seen as a kind of “conceptual retreat” about normative authority, motivated by skeptical worries. On one reading of Hume, for example, he can be seen as deploying skeptical arguments to motivate similar conceptual retreat about causation. Ideas are based on impressions for Hume, and terms must express some idea to be meaningful. Now, if causation were a kind of necessary connection or power or force, the term “cause” would be meaningless, as we have no impressions of necessary connections or powers or forces, and so could have no idea of them either. But the more conservative response is to realize that “cause” means something of which we do have impressions, perhaps regular succession, or perhaps one’s feeling of being compelled to imagine the effect upon encountering the cause. By analogy, if normative force were anything other than some species of psychological force, we would have no clear concept of it at all. So the conservative response is to identify normative force with some species of psychological force. My objection to this move is that first, unlike “cause,” the notions of an ought simpliciter and normative force are not terms with day-to-day familiarity, but philosophers’ terms of art. There is, therefore, no general reason of conservatism to accept the conceptual retreat. We can just jettison the terms. We could put this point another way: why doesn’t the advocate of psychological characterizations simply become an eliminativist about normative force, or authority? There is an action-determining sense of ought. It has the philosophically interesting property of being the content of those normative judgments that actually (perhaps necessarily) cause action. This allows it to resolve conflicts between other oughts, albeit in the minimal sense of causing the agent to side with one or the other. No other senses of ought have these properties.²⁵ This seems like enough. What else do we gain by hanging on to metaphors about authority or overridingness when what we really mean is causes, or tends to cause?

²⁵ Donald Hubin’s (1999 and 2001) position, while not exactly like this, is similar enough to merit note (see especially his 2001: 465ff.). Hubin accepts a Humean, instrumentalist account of practical rationality. There is a sense in which rationality, so explained, is just one more normative standard among many. Nonetheless, rationality is of special philosophical interest because it defines what an agent should do from that agent’s own evaluative point of view.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

244

Derek Baker

11.2.7. Other Options The characterizations so far obviously don’t exhaust the possibilities. What, then, justifies our confidence that the pattern—of circularity or changing the subject—will hold?²⁶ There are of course limitations on our ability to object to proposals that have yet to be proposed, but there is a general problem that any proposal, to be credible, must address. First of all, when asked what we mean by “authority” or by one ought “trumping” another, it is very natural to answer in further normative terms: the authoritative prescriptions are the ones we ought to listen to, or the ones that have importance. But this can’t serve as a characterization for reasons of circularity. Furthermore, as we have seen, there is no class of normative terms that seem unique to intuitively authoritative standards: etiquette supports ought-claims; there are reasons (internal to etiquette) why one ought to do these things; doing so realizes the value of politeness. So attempts to characterize authority in terms of reasons or values will not pull us out of the circle. We end up forced to characterize the authority of ought in terms of its tie to authoritative reasons or authoritative values. But it was authority, not ought, that we wanted characterized. On the other hand, non-normative facts seem irrelevant to what phrases like “trumping” or “authority” were meant to evoke (this is probably why the normative characterization seemed so natural). I have focused here on psychological accounts, because many of the characterizations in the literature seem ambiguous between a normative and psychological reading. But we could say the same about metaphysical characterizations. Imagine Gyges’ dilemma again. The ought simpliciter tells him not to kill the king. But, he asks, how does that help, given that the prudential ought advises murder. Ah, we respond, the ought simpliciter has a special metaphysical status: it is a non-natural, sui generis relation, whereas the ought of prudence can be given a naturalistic reduction. Gyges could be understandably perplexed at this point. He asked for the sense in which the ought simpliciter overrides conflicting oughts, and he’s told that its metaphysics is more exotic. This seems to change the subject. Naturalistic stories about the ought simpliciter don’t look better. Perhaps the ought simpliciter has a less gruesome reduction base than the ought of prudence. Fine, but how does that, in itself, capture what we were gesturing at earlier—that its prescriptions are more genuinely prescriptive? Of course, a philosopher wishing to pursue a metaphysical characterization might have more to say. But the point is she must say it. Metaphysical ²⁶ Thanks to an anonymous referee for posing this problem.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Skepticism About Ought Simpliciter

245

claims do not wear their connection to normative authority on their face. In fact, it’s difficult to imagine any non-normative characterization of the ought simpliciter illuminating what authority is supposed to be. The obvious non-normative characterizations seem simply to be about something else. But normative characterizations won’t do; hence the general pessimism that the terms philosophers have introduced here have any definite meaning.

1 1. 3. L A C K O F T H E O R E T I C AL O R P RA CT ICA L U T IL IT Y My case against an ought simpliciter is that such an ought would have to have some special property, but that property cannot be characterized except in normative terms that invoke the very same property (e.g., “the authoritative ought is the one you ought—in the authoritative sense—to heed”), or in psychological terms that seem to change the subject. This argument for skepticism may seem to trade on an open-questiontype problem facing the normative in general: namely, the apparent impossibility of giving a satisfying definition of normative terms in anything except further normative terms. It may seem I am demanding an analysis of a normative term, then complaining that that analysis is either in further normative terms, and so circular, or else in non-normative terms, and so a violation of the Moorean ban on defining the normative in terms of the natural. But such an argument would prove too much—that we should be skeptics about the coherence of normative concepts in general. This would be to misrepresent my argument, however. The normative terms which are standardly denied to be non-normatively analyzable are familiar terms of everyday use: “ought,” “reason,” “good,” “bad,” “permitted,” “forbidden,” “right,” “wrong”. We can claim an implicit understanding of what these terms mean, even if informative definitions elude us. By contrast, terms such as “normative authority,” “normative force,” “overridingness,” and “ought simpliciter” are philosophers’ terms of art. What I am asking for is some characterization of these terms of art in something besides further terms of art. There may still be two reasons why even this demand is illegitimate, however. First, one might point out that in other domains of inquiry we accept new terms of art that cannot be defined in more familiar terminology. Metaphysicians have by and large accepted a relation of metaphysical grounding, for example, a relation which according to these metaphysicians cannot be defined in more familiar terms. Second, one might claim that “normative authority” is a term to express a concept that we already deploy

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

246

Derek Baker

implicitly, but for which we lack a distinctive word. The evidence that we deploy such a concept is presumably deliberative practice. Sometimes we face conflicting oughts, and we make a further judgment about what we ought to do, or we ask for advice about the same. What sense of ought could we be deploying in these cases, besides the one that resolves the dilemma? In response to the first point, assessing the case for the use of a relation of grounding in metaphysical theorizing is beyond the scope of this chapter. That said, if the case for grounding really is the same as the case for an ought simpliciter—we can only indicate what this relation was supposed to be through metaphor, for example—then we should be skeptics about grounding as well. To the extent that we should accept grounding, it is because the cases are not analogous. First, grounding is linked to a familiar concept we obviously have—explanation. Second, advocates of the relation have put in considerable work to show the theoretical utility of this concept, mostly in giving a more perspicuous characterization of existing theoretical debates, and have further argued that this theoretical work cannot be done by some more familiar relation, such as supervenience (e.g., Fine 2001; Schaffer 2009; Rosen 2010). The theoretical utility of the relation is a reason to accept it, even if a reductive definition in more familiar terms cannot be offered.²⁷ In the case of the ought simpliciter, by contrast, we lack any sort of link to a more familiar theoretical notion. Instead, we have vague but evocative phrases (“normative force,” “what you really ought to do”) that gesture at it. One may argue that it is linked to reasons,²⁸ but as we saw (Sections 11.2.2 and 11.2.3.1) oughts in general are linked to reasons. To identify a linkage unique to the ought simpliciter we have to restrict ourselves to authoritative reasons or reasons simpliciter—that is, we must use more terms of art. More importantly, the notion lacks theoretical utility. There is, to put it bluntly, no theoretical problem that the ought simpliciter solves, or helps us to characterize more perspicuously. Consider our initial case. Gyges ought self-interestedly to kill the king, and he ought morally to refrain. This is how the world is. There is no mystery here crying out for explanation.²⁹ The problem which the ought simpliciter might seem to solve is: how is he to resolve the conflict? ³⁰ But this is not a theoretical problem; it is a practical one. There is no mystery about how the world is which this sense of ought ²⁷ But see Wilson (2014) for arguments against the theoretical utility of grounding. ²⁸ Thanks to an anonymous referee for raising this objection. ²⁹ But see McLeod (2001: 274ff.) for an enumeration of possible theoretical uses of the concept. Constraints of space prevent me from addressing all of these, but the most substantial—that the concept has deliberative utility—is addressed in the following paragraphs. ³⁰ See, for example, Stroud (1998: 175), McLeod (2001: 278–9), and Kiesewetter (2011: 2–3).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Skepticism About Ought Simpliciter

247

clears up. (Gyges is in the unfortunate position of not being able to pursue self-interest and morality at once; that is not mysterious at all.) Rather, there is a practical dilemma for Gyges about which choice to make.³¹ But there is no need to assume that the practical problem admits of any solution. Every theory of practical reason must allow that some possible problems admit of no solution: if only because there must be conceivable circumstances in which the reasons favoring option A tie with those favoring B. But even if such problems can’t be solved, they can be resolved in the minimal sense already discussed: the agent As or the agent Bs, in either case arbitrarily. But, again, everyone needs to accept that some decision problems admit of no solution and must be resolved arbitrarily. The skeptic about ought simpliciter simply thinks that dilemmas like this are more common. She is not alone in holding they exist. Unsolvable dilemmas, then, are not some extra theoretical burden unique to the skeptic alone, nor is positing a capacity to act arbitrarily in the face of such dilemmas. What if instead of theoretical or practical utility we appeal to familiarity? It is true that ought simpliciter is a philosopher’s term of art, but it doesn’t follow that we only learn the concept through philosophical theorizing. It may simply be that there was no special word for this concept, perhaps because we didn’t need one. The ought simpliciter is a philosopher’s name for the default sense of ought we deploy in advice-giving and deliberation. The metaphors and table thumps aren’t there to help gesture at some new concept; they’re to help one focus on a concept that’s already familiar. In reply, the skeptic is entitled to ask what the evidence is that we deploy such a concept. It’s not enough to say that we do come to a decision about what we ought to do in the face of conflicts from different normative domains. That could be explained just as easily if we were, as Tiffany put it, partisans of one normative domain or another. Some people care about morality more than self-interest, others about self-interest more than morality. An ought simpliciter is unnecessary, so long as we are constituted to care about some oughts more than others. A better case for such an ought is offered by Judith Jarvis Thomson (2001: 46): Suppose that Alfred is ill, and that only a dose of a certain medicine will cure him. It tastes truly awful, however. Alfred asks us “Ought I really take it?” It is a wildly ³¹ A referee wonders whether the ought simpliciter is necessary to solve the theoretical problem of how Gyges should resolve the dilemma. It is not necessary to explain how Gyges should resolve the dilemma morally, nor how he should resolve it prudentially. But how Gyges should resolve the dilemma simpliciter? This question presupposes that there is something one should do simpliciter. Deny there is a should simpliciter, and there is no mystery to solve.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

248

Derek Baker

implausible idea that we can reply only: “Well, your taking it would be very unpleasant, so in one sense of ‘ought,’ it’s not the case that you ought to take it, namely the ‘oughtenjoyable’ sense of ‘ought.’ But your taking it would be good for you, so in another sense of ‘ought,’ you ought to take it, namely the oughtgoodness-for-Alfred’ sense of ‘ought.’ ” It is likely that Alfred will repeat his question: “But ought I take it?”

Alfred’s question is intelligible. And that looks like a problem for the skeptic. There is not some obviously relativized ought that Alfred is asking for information about. Nor could Alfred plausibly be asking us what Alfred ought to do according to Alfred’s personal deliberative standards—how would we know, or at least, how would we know better than Alfred? So here we have a prima facie case of the use of a non-relativized ought shared in common by both Alfred and ourselves, an ought simpliciter. To see the skeptic’s answer, we have to first notice that the skeptic can agree that sometimes we make judgments about what we ought to do, and sometimes this causes us to act. But sometimes it doesn’t. When it doesn’t, this might be because what Tiffany calls my personal deliberative standards are silent on how to resolve this particular dilemma. It could also be instead that my deliberative standards do tell me what to do, but they fail to move me on this occasion (that can happen—I don’t always live up to my commitments). In situations like these, I might try to imagine the options and competing values in a variety of ways, from a variety of perspectives. I might read Epictetus or Emerson, and try to get fired up. I might ask a friend for advice. Maybe I can take what they tell me I ought to do as an indication of what they would do, and that might be enough to move me. Maybe my friend is good at getting me fired up. Maybe I already know what I’m going to do and just want a thumbs up first. People ask for advice for lots of reasons. Bernard Williams noted that “Practical reasoning is a heuristic process, and an imaginative one, and there are no fixed boundaries on the continuum from rational thought to inspiration and conversion” (1981: 110).³² The skeptic about ought simpliciter should accept this, since there is an awful lot of thinking leading up to decision making that cannot plausibly be seen as the working out of what’s required by one’s prior deliberative standard or desires. But if we accept this about the intrapersonal case, it should be true in the interpersonal, advice-giving case as well. So when we ask for advice we may be asking what some normative standard (morality, self-interest, authenticity, our own personal deliberative standards, our interlocutor’s standards) favors doing. But we may be asking, instead, that ³² Also see Hubin (1999: 42).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Skepticism About Ought Simpliciter

249

our interlocutor convert us to a standard, or inspire us to live up to some standard we already accept, or to help us construct a new standard—while converting and inspiring us in the process. In short, the skeptic can say that Alfred is asking us to supply an ought-claim that will move him to act. The oughts initially invoked don’t do this, so Alfred wants us to offer another one, maybe with more effective rhetoric behind it. Of course, Alfred has no special sense of ought in mind in advance: he doesn’t know in advance which ought will move him. He’s hoping we might have one in mind that he hasn’t thought of yet, or might invent an attractive (to him) way of weighing dissimilar kinds of considerations on the spot. By way of analogy, consider when you ask your friend to suggest which restaurant you both eat at that night. One reason you might ask her is none of the options you have thought of strike you as attractive, and so you are asking to see if she can think of one you can’t call to mind, but which, once it’s mentioned, you’ll endorse. Sometimes we do the same with oughts.

1 1 .4 . D E N Y I N G M U L T I P L E SE N S E S O F OU G H T Thomson offers another argument on behalf of the ought simpliciter—it is the only sense of ought that we have (2008). Thomson’s argument is simple: if there were relativized ought, no sense could be made of an ought simpliciter. But we can make sense of an ought simpliciter. Therefore, there are no relativized oughts (2008: section X.2). Thomson, then, can be seen as accepting the validity of the basic argument here,³³ but concluding that the correct response to such an argument is to apply modus tollens, not ponens. The question, then, is what it is more reasonable to give up on: an ought simpliciter or various relativized oughts. Thomson’s case is intuitive. She regards it as extremely implausible that there is no ought simpliciter: “I draw attention, first, to the intuitive implausibility of the following conclusion that the argument issues in: that there is no such thing as the proposition that A ought to Vact. (Just as there is no such thing as the proposition that A is taller.)” (2008: 168). But this argument is weak. First, the skeptic agrees that sometimes statements of the form “A ought to Vact” are true. She simply holds that they are elliptical, and so, in conjunction with context, they denote a proposition of greater complexity than the surface grammar of the sentence reveals. Thomson’s argument here depends on our possessing highly reliable intuitions not simply about the surface

³³ Thomson’s argument is simpler than mine. Roughly, if any uses of ought are relativized, then an unrelativized use of ought would be ill-formed (2008: 168).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

250

Derek Baker

grammar of acceptable claims, but also about whether the meaning of these claims possesses any hidden structure. It is very doubtful, however, that we have reliable intuitions directly about such matters. The recent wave of contextualist treatments of “ought” argue that, as a modal verb, ought-claims possess considerable hidden structure (e.g., Dowell 2012 and 2013; Finlay 2009 and 2014).³⁴ These accounts may be right or wrong—but if they are wrong, this is shown by demonstrating that they are inadequate for explaining our intuitions about various ought-claims in various contexts (or incapable of being integrated into our overall semantic picture, etc.). The brute intuition that ought-propositions are logically simpler than that is not something such accounts should be expected to answer. This greatly weakens the case for tollens-ing over ponens-ing. What’s more, insisting that there is only an ought simpliciter forces us to attribute very odd normative commitments to people who utter the following, very sensible sentences: (1) “Nixon ought to have burned the tapes.” (2) “Instead of pushing eastward into Russia, Hitler ought to have sent the Wehrmacht south to seize the Romanian oil fields.” (3) “The killer should have covered his tracks better.” These are normal things for people to say, and do not imply that the speaker is a committed amoralist, Nazi, or whatever. Someone could utter (1) and still agree that Nixon ought to have handed himself in and confessed his crimes.³⁵ According to Thomson, this person must have contradicted herself. But it’s more likely that “ought” is being used in a sense that implicitly restricts the kinds of considerations in play—to those of selfinterest, or desire-satisfaction, or good military strategy, for example. So, if we have to choose between relativized oughts and ought simpliciter, as Thomson and I both agree we must, it is the latter that must go.

11 .5 . CONC L U S ION Perhaps we can accept some very strange normative commitments. We can hold that conventional standards have independent normative importance ³⁴ Finlay’s contextualist account of an ought simpliciter is especially relevant here: “s ought to ϕ ‘simpliciter’ just in case it would be correct and felicitous in the absence of any special context to assert simply ‘s ought to ϕ’ ” (2014: 151). Note that my argument in this chapter has nothing to say against his interpretation of the ought simpliciter on which it has no special normative standing, only special conversational relevance. ³⁵ Similar examples can be found in Dowell (2012) and Finlay (2014: 50 and 137).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Skepticism About Ought Simpliciter

251

simply in virtue of being standards. If we reject this, however, as I think we should, we must reject the ought simpliciter. This would mean, admittedly, a rejection of the practicality of philosophy. There is no absolute perspective from which different kinds of considerations can be weighed. Individual agents may have their deliberative standards for measuring one kind of consideration against another, but these standards are themselves arbitrary, in the sense that each is simply one more standard among many. Conflicts between self-interest and morality, or any other two normative standards, cannot be resolved by coming to appreciate some philosophical (or everyday) truth. Rather, it is resolved through what Tiffany calls partisanship and existential choice. Despite this, most of moral philosophy will remain in place. Since most of us are partisans of morality, we naturally have an interest in what morality in fact requires. We can still ask questions about morality’s naturalistic bona fides and the relation between morality and rationality. But there is one question to which we as philosophers have no special answer, and that’s “So what?”³⁶

References Baker, D. 2017. “The Varieties of Normativity.” In Routledge Handbook of Metaethics, edited by T. McPherson and D. Plunkett, 567–81. Routledge. Broome, J. 2013. Rationality through Reasoning. Wiley-Blackwell. Chang, R. 2004. “All Things Considered.” Philosophical Perspectives 18 (1): 1–22. Copp, D. 1997. “The Ring of Gyges: Overridingness and the Unity of Reason.” Social Philosophy and Policy 14 (1): 86–101. Dorsey, D. 2013. “Two Dualisms of Practical Reason.” In Oxford Studies in Metaethics, edited by R. Shafer-Landau, 8: 114–39. Oxford University Press. Dowell, J.L. 2012. “Contextualist Solutions to Three Puzzles about Practical Conditionals.” In Oxford Studies in Metaethics, edited by R. Shafer-Landau, 7: 271–303. Oxford University Press. Dowell, J.L. 2013. “Flexible Contextualism about Deontic Modals: A Puzzle about Information-Sensitivity.” Inquiry 56 (2–3): 149–78. Dreier, J. 2015. “Can Reasons Fundamentalism Answer the Normative Problem?” In Motivational Internalism, edited by G. Björnsson, C. Strandberg, R. Francén Olinder, and J. Eriksson, 167–81. Oxford University Press. Fine, K. 2001. “The Question of Realism.” Philosophers’ Imprint 1 (1): 1–30.

³⁶ Thanks to Brad Cokelet, David Copp, Tristram McPherson, Daniel Wodak, Jack Woods, anonymous referees, and the audience at the 2nd Annual Chapel Hill Metaethics Workshop for comments and criticisms. I apologize that considerations of space prevent me from addressing many of the excellent objections raised at the conference.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

252

Derek Baker

Finlay, S. 2007. “Responding to Normativity.” In Oxford Studies in Metaethics, edited by R. Shafer-Landau, 2: 220–39. Oxford University Press. Finlay, S. 2009. “Oughts and Ends.” Philosophical Studies 143 (3): 315–40. Finlay, S. 2014. Confusion of Tongues: A Theory of Normative Language. Oxford University Press. Foot, P. 1972/1977. “Morality as a System of Hypothetical Imperatives.” The Philosophical Review 81 (3): 305–16. Reprinted with additional footnote in her Virtues and Vices, University of California Press, 1977: 157–73. Hubin, D. 1999. “What’s Special about Humeanism?” Noûs 33 (1): 30–45. Hubin, D. 2001. “The Groundless Normativity of Instrumental Reason.” The Journal of Philosophy 98: 445–68. Joyce, R. 2001. The Myth of Morality. Cambridge University Press. Kiesewetter, B. 2011. “ ‘Ought’ and the Perspective of an Agent.” The Journal of Ethics and Social Philosophy 5 (3): 1–24. Korsgaard, C. 1986. “Skepticism About Practical Reason.” The Journal of Philosophy 83 (1): 5–25. McLeod, O. 2001. “Just Plain ‘Ought.’ ” The Journal of Ethics 5: 269–91. McPherson, T. 2011. “Against Quietist Normative Realism.” Philosophical Studies 154 (2): 223–40. Parfit, D. 2011. On What Matters, Volume 1. Oxford University Press. Rosen, G. 2010. “Metaphysical Dependence: Grounding and Reduction.” In Modality: Metaphysics, Logic, and Epistemology, edited by B. Hale and A. Hoffman, 109–36. Oxford University Press. Schafer, K. 2016. “The Modesty of the Moral Point of View.” In Weighing Reasons, edited by E. Lord and B. Maguire, 241–56. Oxford University Press. Schaffer, J. 2009. “On What Grounds What.” In Metametaphysics: New Essays in the Foundations of Ontology, edited by D. Manley, D.J. Chalmers, and R. Wasserman, 347–83. Oxford University Press. Smith, M. 1994. The Moral Problem. Wiley-Blackwell. Stroud, S. 1998. “Moral Overridingness and Moral Theory.” Pacific Philosophical Quarterly 79 (2): 170–89. Thomson, J.J. 2001. Goodness and Advice. Princeton University Press. Thomson, J.J. 2008. Normativity. Open Court. Tiffany, E. 2007. “Deflationary Normative Pluralism.” Canadian Journal of Philosophy 37 (5): 231–62. Wedgwood, R. 2007. The Nature of Normativity. Oxford University Press. Williams, B. 1981. “Internal and External Reasons.” In Moral Luck, 101–13. Cambridge University Press. Wilson, J. 2014. “No Work for a Theory of Grounding.” Inquiry 57 (5–6): 535–79.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

12 Authoritatively Normative Concepts Tristram McPherson

12 . 1. INTR O DU CT ION There are many species of norms. An act can be an illegal chess move, and impolite to boot, but still morally required. And there are famous philosophical puzzles about the generic normativity shared by all of these species.¹ For example, Saul Kripke’s Wittgenstein (1982) famously challenged our ability to explain how someone could count as following one norm rather than another. However, metaethicists have not typically focused on understanding normativity in this generic sense, because puzzles about generic normativity arguably do not cut to the heart of the metaethical dialectic.² One illustration of this is that many of the most influential and wellmotivated classes of views in metaethics—non-naturalism, expressivism, error theory, and fictionalism—are less attractive views about merely generic norms like the norms of chess.³ Work in metaethics has instead focused on moral normativity, or— increasingly—on practical normativity. What is this latter notion? Quick glosses usually focus on two ideas. The first is a contrast with morality: sometimes one has a choice where morality appears silent, but it seems clear that one option is nonetheless better than the other. For example, suppose that I have no morally significant options, but can choose whether to spend the afternoon in pleasant conversation or counting blades of grass. In this ¹ The “generic” label is from Copp 2005a. I previously called this sort of normativity “formal” in my 2011. ² In the text, I use “metaethical” and related terms in a familiar, very loose sense. For a more careful taxonomy, see McPherson and Plunkett 2017. ³ For a discussion that usefully and sharply distinguishes the task of explaining practical normativity from that of explaining generic normativity, see FitzPatrick 2008. Gibbard 2012 takes the core issues at stake in metaethics to carry over at least to semantic norms. In this respect, the project of this chapter sides with FitzPatrick against Gibbard. For an important challenge to Gibbard’s attempt to extend his master argument for expressivism to semantic norms, see Baker 2016.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

254

Tristram McPherson

case I arguably practically ought to choose conversation. The second is the idea that practical normativity is distinctively authoritative. The practical norms are the ones that settle what to do, in a way that somehow contrasts with the deliverances of etiquette (for example). These glosses orient us to the idea of authoritative normativity. This chapter aims to help us to better understand this idea by developing an account of one authoritatively normative concept: PRACTICAL OUGHT. (I use small caps to denote concepts.) My account analyzes this concept in terms of the constitutive norms for the activity of non-arbitrarily selecting an option (Section 12.5). I argue that this analysis permits an attractive and substantive explanation of what the distinctive normative authority of this concept amounts to. I briefly show how my account can answer “schmagency”-style objections to constitutivist explanations of normativity (Section 12.6). Finally, I briefly sketch some of the central metaethical implications of my account, showing that it can be used to help realists, error theorists, and fictionalists address central challenges to their views (Section 12.7). I begin by more carefully introducing the concept that I analyze, the style of account I offer, and the constitutivist resources that this account deploys (Sections 12.2–12.4).

1 2 .2 . E L US I VE P R A C T I C A L N O R M A T I V I T Y Because the notion of authoritative normativity is central to my project here, I will augment the brief orientation in the introduction with a more careful characterization. To begin, consider the following deliberative scenario: Sticky Situation You find yourself in a sticky situation. You conclude that morality requires you to stay and help, while prudence dictates that you take the money and run. Torn, you ask yourself: given all of this, what ought I to do? ⁴

Sticky Situation concludes with an interesting question, about which you might agonize. Because of this, it is implausible to read “ought” in this question as expressing⁵ either the concept MORALLY OUGHT, or the concept ⁴ My initial characterization and discussion of Sticky Situation is indebted to Wedgwood 2004, 406, who in turn credits Cullity and Gaut 1997. ⁵ I use expression to pick out the relation between linguistic tokens and the mental states they are associated with in virtue of the meanings of those linguistic tokens. In doing so, I commit myself to the (plausible but controversial) idea that the meanings of (some) linguistic tokens entail such a connection to contentful speaker mental states. I do not intend the more controversial thesis that the meanings of linguistic entities are, or are grounded in, their conventional relation to the contentful mental states they express.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Authoritatively Normative Concepts

255

PRUDENTIALLY OUGHT.

For you already take yourself to know the answer to the question, understood in those ways. (Note that the answer you take yourself to know might be incorrect: perhaps morality and prudence cannot conflict. For my purposes, it is important only that someone can intelligibly think that they can conflict, and pose this question in light of the perceived conflict.) We can step back from the specific question highlighted in Sticky Situation, to query the normative significance of conflict between morality and prudence more generally. Here, one familiar view is: Moral Rationalism If moral requirements and prudence conflict, one ought to do the morally required thing.⁶

Whether or not it is true, Moral Rationalism appears to be an interesting and substantive thesis. The upshot is the same as in Sticky Situation: if we interpret “ought” here as expressing MORALLY OUGHT, Moral Rationalism is trivially true; if we interpret it as expressing PRUDENTIALLY OUGHT, it is trivially false. In both Moral Rationalism and Sticky Situation, it seems most plausible to read “ought” as expressing a concept that purports to wear a distinctive normative authority on its sleeve, in a way that even moral and prudential “ought”s do not. This explains why the truth of Moral Rationalism is seen by many as crucial to vindicating the normative significance of morality. Similarly, when faced with the apparent conflict imagined in Sticky Situation, one might ask what is the honorable thing to do? or what would best promote my life’s ambition? etc. It is natural to see answers to these questions as identifying new considerations that complicate the choice one faces. By contrast, in asking what ought I to do? one seeks an answer that normatively settles, rather than complicates, one’s choice.⁷ It will be useful to have a label for the distinctively authoritatively normative concept that plays these roles: PRACTICAL OUGHT. ⁶ For discussion of related principles, see Smith 1994, Darwall 1997, van Roojen 2010, Lord and Plunkett 2017, and several of the papers in Jones and Schroeter forthcoming. ⁷ My aim here is to help the reader to latch on to a concept, not to claim that a certain form of words always expresses that concept. Because “ought” is a context-sensitive term, it could easily be used to express another concept in some refinement of Sticky Situation. Consider an idea proposed to me by David Copp: to read the question in Sticky Situation as asking what my values instruct me to do, given the conflict. It is a familiar point that we can understand talk of one’s “values” in several ways. If we think of my values as my beliefs about what I ought to do, then this would be a variant of my gloss. Suppose instead that we think of values in terms of the felt importance of certain features. In this case, however, one can ask what one ought to do, in the settling sense, given a perceived conflict between morality or prudence and one’s values.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

256

Tristram McPherson

PRACTICAL OUGHT is part of a family of authoritative concepts, which range across various dimensions of normative structure. Just as we can talk of a range of narrowly moral concepts—MORAL REQUIREMENT, MORALLY BETTER, MORAL REASON, etc.—we can talk about their explicitly authoritative structural correlates: PRACTICAL REQUIREMENT, PRACTICALLY BETTER, PRACTICAL REASON, etc. The last of these is arguably the most infectious contemporary locution for gesturing at authoritative normativity.⁸ This chapter focuses almost exclusively on PRACTICAL OUGHT, postponing the important question of how my account could be extended to other authoritatively normative concepts. Authoritatively normative concepts are an especially natural locus for metaethical inquiry. Suppose first that Moral Rationalism (or something like it) is true. Then it would be very natural to expect that an account of the authoritatively normative concepts or properties will be crucial to understanding moral thought, talk, and reality. This will also hold true if Moral Rationalism is false, but a weaker distinctive connection between morality and practical normativity holds. Suppose instead that moral requirements turn out to be like the requirements of chess or etiquette, in being only contingently connected to what we practically ought to do. Then there would be less reason to think that morality raised distinctive metaethical puzzles (as opposed to the sorts of puzzles also raised by etiquette norms, for example). On this supposition, it would be natural for metaethicists to turn their attention to the authoritatively normative concepts. For these reasons it is not surprising that over the past generation, the focus of metaethical work has shifted significantly from morality to what I am calling authoritative normativity.⁹ The nature of the authoritatively normative concepts can seem elusive, however. This can be illustrated by considering four unfruitful strategies for illuminating them. First, one cannot illuminate authoritatively normative concepts generally simply by analyzing one member of the family of such concepts in terms of another. For example, even if it were possible to do so, it would not suffice to analyze all other authoritatively normative concepts in terms of PRACTICAL REASON. This is because the core question here is about what is distinctive of the whole family of authoritatively normative concepts, ⁸ Contemporary philosophers very often deploy locutions that are plausibly intended to convey the distinctive authority characteristic of this family of concepts: compare Scanlon’s talk of reasons in the “standard normative sense” (1998, 17–19), Schroeder’s “normativity of the normative” (2007, 79), and Hampton’s talk of “normative authority” (1998, 85ff.) which my talk of “authoritative” norms echoes. In conversation—if less often in print—philosophers will sometimes speak of normative “oomph.” ⁹ Representative examples include Bedke 2010, Gibbard 2003, Schroeder 2007, Street 2008, and Wedgwood 2007.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Authoritatively Normative Concepts

257

or—if inter-normative conceptual analyses are possible—whichever of these concepts are conceptually basic. Second, it would be a mistake to gloss authoritativeness as categoricality, where a norm is categorical if you cannot escape its application simply by changing your desires or intentions. Categoricality appears neither necessary nor sufficient for our understanding of authoritativeness. Its insufficiency was well established by Philippa Foot (1997). The norms of etiquette are categorical in the sense just mentioned: indifference to these norms does not make their violation any less impolite. Its necessity is rendered doubtful by the Humean research program, according to which facts about authoritative normativity are grounded in facts about agents’ contingent desires. If categoricality were central to our understanding of authoritativeness, then familiar forms of Humeanism would be guilty of a transparent category mistake, which is hardly plausible. Third, in order to mark a contrast with norms like etiquette, Derek Parfit contrasts normativity in the “rule-implying” sense, with normativity in the “reasons-implying” sense (2011, §88), where the latter is his way of adverting to what I am calling authoritative normativity. But this is potentially misleading: there are lots of ways of using “reason,” many of which fail to be transparently authoritative. One can talk about moral or aesthetic reasons, or reasons of etiquette, all of which are manifestly generically normative, but none of which is obviously authoritatively normative. In these cases, “reason” is best understood as adverting to a certain kind of normative structural kind, which can have instances across both authoritative and merely generic normative systems.¹⁰ Fourth, Ralph Wedgwood (2004) glosses our target concept as the allthings-considered ought. The locus classicus for “all-things-considered” talk is Donald Davidson’s discussion of weakness of will, which contrasted the judgments that X is better than Y simpliciter, with the judgment that it is better prima facie, and that it is better all-things-considered (2001 [1969]). Davidson’s contrast is unhelpful in this context, because it is also structural: the simpliciter/prima facie/all-things-considered contrast will show up within moral and prudential judgments, and even within chess judgments. For example, one might say (pedantically), “Weaknesses around white’s king prima facie support mounting an attack there, but in light of the concentration of my pieces, and the open c-file, it is better all-thingsconsidered to secure strategic advantages on the queenside instead.” Here ¹⁰ There is a reading of “aesthetic reason” (for example) as meaning something like a practical reason simpliciter that has an aesthetic basis. However, “reason” is also used in careful philosophical contexts with the structural meaning (for one example, see the discussion of moral rationalism in Lord and Plunkett 2017).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

258

Tristram McPherson

“better all-things-considered” is most plausibly read as a claim internal to the norms of chess. Similarly, a gloss of “all things considered” as all reasons considered is unhelpful. If this means all authoritative reasons considered, then we have simply moved the bump in the rug. If it means all reasons, including the nonauthoritative ones, considered, then it is hard to see how this distinguishes the authoritative ought from other oughts: just as the function from reasons to the authoritative ought will presumably assign zero weight to some reasons (the non-authoritative ones), the function from reasons to the prudential ought will presumably assign zero weight to the non-prudential reasons. As these examples show, common attempts to provide an informative gloss on the notion of an authoritatively normative concept appear to fail. At this point, an objector might claim that she has no idea what is being gestured at with talk of “distinctively authoritative” normativity. She might point out that the term “authoritative” is so far simply a label, and should in no way convince us that we have a grip on the alleged concept being deployed. She might continue: morality is distinctively morally authoritative, prudence is distinctively prudentially authoritative (etc.), and there is no other coherent notion of authority which can be used to give us purchase on PRACTICAL OUGHT. This might lead her to suggest that philosophers’ attempts to discuss authoritative normativity simply fail to latch on to a genuine concept. Call this view deflationary pluralism about normative concepts.¹¹ The cost of deflationary pluralism is high, however. It appears to deny that Sticky Situation as I have described it raises an interesting question, and that Moral Rationalism is an interesting thesis. And it suggests that the range of central metaethical views mentioned in Section 12.1—non-naturalism, expressivism, error theory, and hermeneutic fictionalism—are confused at a fairly fundamental level, if (as I suggested) they are often tacitly motivated in part by the thought that metaethics has authoritative normativity as its explanatory target, either directly or indirectly. In light of this, one might claim instead that the elusiveness of authoritative normativity is explained by concept primitivism. On this view, perhaps some authoritatively normative concepts can be analyzed in terms of others, but the fundamental authoritative concept(s) are unanalyzable: nothing non-circular can be said to illuminate their distinctive nature (compare Scanlon 1998, 17 on reasons). The strategy of positing a primitive concept here arguably pairs best with a non-naturalist metaphysics: on this view, what is distinctive of authoritatively

¹¹ Compare Copp 2005a, 2005b, Tiffany 2007, and especially Baker this volume for relevant discussion.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Authoritatively Normative Concepts

259

normative thought is that it alone is about a sui generis, distinctively normative part of reality. But arguably, our only way of understanding the idea that this bit of reality is distinctively normative is that we talk about it using these concepts.¹² If so, this metaphysical explanation seems unhelpful, and primitivism remains unsatisfying. Against these views, I argue that it is possible to provide an informative account of the concept PRACTICAL OUGHT that vindicates its distinctive normative significance. The nature of the account will also suggest a plausible explanation of why the nature of authoritative normativity appears so elusive.

1 2 . 3 . CO N C E P T U A L A N A L Y S I S A N D T H E VINDICATION OF AUTHORITY My account will consist of an analysis of the concept PRACTICAL OUGHT paired with an explanation of how this analysis vindicates the distinctive normative authority that concept purports to have. Because the very possibility of providing informative analyses of philosophical concepts is controversial, this section sketches how I understand this project. I have sought to locate the concept I am discussing—the PRACTICAL OUGHT—in part by adverting to a kind of thought that most of us recognize (for example, in Sticky Situation) and to familiar moves in ethical theorizing (for example debates about Moral Rationalism). There may be a word in non-philosophical English that uniquely picks out this concept, but for the purposes of this project, I set that question aside. My aim is to understand the concept—whatever it is—that plays these distinctive roles in deliberative and theoretical contexts. This means that the central desideratum on the analysis of this concept is that it permits an illuminating explanation of the concept’s distinctive normative purport. I take this desideratum to require in turn that the analysis be informative and non-circular. For otherwise the explanation it is paired with risks presupposing the very thing it seeks to explain. However, I do not take it to require analyzing the concept PRACTICAL OUGHT in wholly non-normative terms. I will make use of generic normativity in my analysis without precisely explicating the idea of a generically normative concept. This is because I aim to explain the contrast between authoritative and merely ¹² For important discussion of the underexplored question of whether the normativity of our concepts is parasitic on the normativity of the reality they represent—or vice versa—see Eklund 2017.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Tristram McPherson

260

generic normativity, not to solve the puzzles that are shared by authoritative and merely generically normative concepts. I take the project of this chapter, as just glossed, to be compatible with a range of views about the methodology for investigating our concepts. For example, on the expansive conception to conceptual analysis championed by Frank Jackson (1998), my project could potentially be understood as an instance of conceptual analysis.¹³ Alternatively my project could be understood as an instance of the sort of “reforming definition” approach defended by Peter Railton (1989). Here the idea is, roughly, that the correct account of PRACTICAL OUGHT is the one that best articulates and refines the theoretically crucial elements of our conceptual practice. Finally, it might be that the best way to understand the project is as introducing novel technical concepts that answer to specific theoretical aims, rather than analyzing or refining an existing folk concept. (For more discussion of relevant methodological complications, see McPherson and Plunkett, forthcoming.)

1 2 .4 . C O N S T I T U T I V I S T R E S O U R C E S The account that I will offer appeals to the contrast between the norms constitutive of certain activities, and merely generic norms. To see this contrast, consider the following norm, which I hereby introduce: Touch Nose Move move.

You must touch your nose while playing any chess

To be clear, this is neither an interpretation of the rules of chess, nor a proposed amendment to them. Nor is it an (absurd) claim about your moral, prudential, or authoritative obligations. It is introduced as an independent norm. It is generically normative: actions can clearly satisfy or violate it. If you play chess moves without touching your nose, you are violating it. What then is the contrast between merely generic norms and the constitutive norms for an activity? In my view, the constitutive norms for an ¹³ Given that my project may be well understood as offering a sort of conceptual analysis, I should explain why I am unworried by the open question argument. In anything like its canonical form (à la Moore 1993 [1903]), I take the argument to simply be hopeless. Consider: suppose that I offer you a novel and surprising analysis A of some philosophical concept C. As a result of reading my argument, your credence in the analysis goes from negligible to 0.7. As philosophy goes: extraordinary success! But now you ask yourself: I know that X is A; but is it C? This is likely an open question given that your credence in the analysis is only 0.7. As far as I can tell, the openness of this question gives you no information you did not have before, and is no reason to lower your credence in the analysis.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Authoritatively Normative Concepts

261

activity fix whether one performs that activity correctly or successfully, not merely relative to some norm or another, but relative to the very activity one is engaging in.¹⁴ Consider an example: if, while playing chess, you move a knight diagonally, you violate the rules for the movement of the pieces, and hence play incorrectly. If you play legal moves, but get checkmated, you have thereby played unsuccessfully. By contrast, because the Touch Nose Move norm is a merely generic norm, and not constitutive of the activity of playing chess, there is a natural sense in which one does not play chess incorrectly or unsuccessfully in virtue of violating it. To bring out the significance of this contrast, suppose that I clearheadedly intend to play chess. Suppose next that while doing so, I routinely make moves without touching my nose, and you point out that by doing so, I violate the Touch Nose Move norm. It seems that without any confusion I might simply note that I don’t care about that. Suppose by contrast that I move a knight diagonally, and you point out that this violates the rules of chess. I might reply in all sorts of intelligible ways: I might decide that I am not playing chess after all; I might evince confusion about the rules of chess; I might accept the correction to my play, etc. But it would be puzzling for me to say that I am playing chess, but I simply don’t care about playing according to its rules. It would become tempting to impute some rational failing to me if I were to say this: perhaps a failure to understand what it is to play chess. The precise nature of the criticism that is warranted here is controversial. My aim is only to establish that, insofar as one is engaged in an activity, the constitutive norms for that activity appear to have a kind of grip on one that the merely generic norms lack. And this grip is naturally understood as marking an asymmetry in the normative significance of constitutive norms, compared to merely generic norms.¹⁵ The apparent promise of this asymmetry has launched a thousand constitutivisms in ethics. But by itself this asymmetry is not enough to explain the distinctive normativity of the concept PRACTICAL OUGHT. After all, the asymmetry does not entail that if one engages in an activity like chess, one practically ought to follow its constitutive norms. To make this vivid,

¹⁴ For a different analysis of what is distinctive of constitutive normativity, which focuses on constitutive aims, see Katsafanas 2013, 39. ¹⁵ A referee suggests that this sort of grip only seems to be a feature of some constitutive norms for an activity, but not others. For example, amateurs playing a friendly game of chess might simply ignore the touch-move rule. This is true, but I think it is best understood as a case where the players are playing a variant of chess with different constitutive rules (there are uncontroversially many such variants in circulation). If one did recognize the touch-move rule as constitutive of the activity one is engaged in, it would strike me as puzzling to react to criticism for violating that rule with indifference, in just the same way as in the example in the text.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

262

Tristram McPherson

suppose that someone invented a game in which players compete to torture a puppy in the most awful way possible. Finding oneself playing this game surely does not entail that one ought to torture a puppy. At least intuitively, it does not even entail that one has a practical reason to torture a puppy.¹⁶ Thus, the constitutivist must identify the added ingredient that, when combined with the constitutive element, will yield authoritative normativity. The most familiar proposal here is that the authoritative norms are constitutive norms for an activity that is in some sense inescapable.¹⁷ To see the appeal of this proposal, notice that one striking feature of chess or the puppy-torturing game mentioned above is that one can stop playing them. When one does stop playing them, any sense that one is under normative pressure to abide by their constitutive norms evaporates. Identifying an inescapable activity seemingly promises to prevent such evaporation. Despite this intuitive appeal, I am pessimistic about the inescapability approach, for reasons suggested by Matthew Silverstein (2015) and especially David Enoch (2006, 2011). On the one hand, it is unclear in what sense deliberation or agency (or whatever else the constitutivist points to as the relevant activity) is inescapable. On the other hand, it is unclear why the inescapability of an activity makes the norms of that activity authoritatively normative. In what follows, I propose a very different way of developing the constitutivist idea.

12.5. AN ACCOUNT OF THE CONCEPT PRACTICAL OUGHT This section develops an analysis of the concept PRACTICAL OUGHT that appeals to the constitutive success norms for a distinctive activity: nonarbitrary selection. After sketching a partial analysis, I explain how it permits an attractive informative account of what it is for a concept to be authoritatively normative. I then complete the analysis in light of challenges related to the scope of applicability of the concept, including the conditional fallacy. I begin by introducing the activity at the heart of my account.

¹⁶ Compare Enoch 2006, 185–6. Note, however, that Schroeder’s case against the reliability of “negative existential intuitions” about reasons (2007, 92–7) could be used to challenge this intuition. ¹⁷ As Korsgaard memorably puts the inescapability idea, “Human beings are condemned to choice and action”(2009, 1, emphasis hers). See also e.g. Ferrero 2009, 304, Velleman 2009, 137, and Katsafanas 2013, 47.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Authoritatively Normative Concepts

263

12.5.1. Non-Arbitrary Selection Return to Sticky Situation, at the moment when you have concluded that morality and prudence conflict in your current circumstances. There are many ways you could react to this perceived conflict. For example, you might flip a coin: heads for morality, tails for prudence! Or you might plump for following morality. Or you might shrug your shoulders, ignore both prescriptions, and look for a beer. In each of these cases, you select an option in the face of the conflict. On the most natural reading of Sticky Situation, you do none of these things. Instead, you ask yourself what you ought to do in order to then select an option on the basis of your answer. Focus on this natural reading of the case, and consider the distinctive nature of the activity you are engaged in. A straightforward gloss on this activity is that you are seeking to select the option that you practically ought to select. However, this gloss is also relevantly uninformative, if we assume that we do not yet have a grip on the concept PRACTICAL OUGHT. We do better by focusing on the contrast with coin-flipping and plumping. What are you trying to achieve in Sticky Situation, that you would not be guaranteed to achieve if you selected an option in one of these ways? It is striking that these look like entirely arbitrary ways of selecting an option. Arbitrary selection is an ordinary part of human life, but in Sticky Situation you take there to be a striking conflict between significant norms that bear on your action. This is exactly the sort of context where we might seek to avoid arbitrary selection. This in turn suggests a natural and informative characterization of the activity you are engaged in, in Sticky Situation: you are seeking to select an option in a nonarbitrary way. Plumping and coin flipping are obvious cases of arbitrary selection. Consider a series of cases which illustrate less obvious arbitrariness. Suppose first that in Sticky Situation Zoe takes there to be a conflict between morality and prudence, and then selects an option on the basis that it is the most prudent. By selecting on the basis of a norm, Zoe avoids the most obvious sort of arbitrariness. However, suppose that Zoe had initially selected the prudence norm as the norm to guide her deliberation by flipping a coin. This would constitute arbitrariness in the etiology of reasoning that led to the resolution of the conflict. Suppose that instead of flipping the coin, she had reasoned as follows: always following the prudence norm is the prudent thing to do, so I shall do it. Here there is no overt picking or coin-flipping in the background. But this reasoning ignores a troubling symmetry: perhaps always following the moral norm is the moral thing

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

264

Tristram McPherson

to do.¹⁸ Upon noticing this symmetry, a reasoner would either need to plump for resolving the conflict by deploying the prudential norm (which is arbitrary), or she would need to find some further basis for choice between the two norms. The same point applies to a reasoner who arbitrarily picks a third norm to adjudicate the conflict. Suppose, for example, that Yan plumps for a policy of appealing to etiquette to guide his choice in cases like Sticky Situation. These examples show that selection can count as arbitrary in virtue of the etiology of that selection involving either (a) relevant arbitrary picking or (b) a failure even to consider a relevant normative conflict.¹⁹ This gloss on ARBITRARY SELECTION may be incomplete, but together with the examples, I take it to provide readers with a substantial grasp of the concept. For example, I think it should allow a careful reader to confidently categorize novel cases as involving arbitrary selection, or not. I aim to analyze the concept PRACTICAL OUGHT in terms of ARBITRARY SELECTION. Because of this, it is important that my account of ARBITRARY SELECTION not covertly deploy the concept PRACTICAL OUGHT, on pain of circularity. One might worry that the account just sketched fails this condition: that our only grip on the concept NON-ARBITRARINESS is parasitic on our grasp on PRACTICAL OUGHT. In reply, consider how I introduced the concept ARBITRARY SELECTION. I pointed to recognizable paradigms, explained how to apply the concept to more complex cases, and proposed an informative tentative gloss on it. None of these elements advert to the concept PRACTICAL OUGHT, and I claim that they jointly put the reader in a position to understand and apply the concept ARBITRARY SELECTION. It is thus unclear how the concept PRACTICAL OUGHT could covertly enter this account. A second natural worry is that the activity of non-arbitrary selection that I aim to elucidate doesn’t make sense. Distinguish two versions of this worry. First, one might worry that the activity makes no sense because we cannot make sense of the concept of ARBITRARINESS. This version of the worry can again be answered by pointing to my constructive gloss on the concept. The second version of the worry is that the concept NON-ARBITRARY SELECTION is coherent but necessarily empty. On this version, seeking to ¹⁸ The symmetry suggested here is intended only as an illustration. If we think of following a norm as an intentional activity guided by a representation of the norm, it is unlikely that either morality or prudence always endorses following itself, due to familiar “rational irrationality”-style phenomena. ¹⁹ The mention of relevance here gestures at a pattern familiar from other contexts. For example, anti-luck epistemologies do not object to knowledge acquired via luckily acquired evidence. Similarly, if I flip a coin, and on that basis decide to non-arbitrarily select an option, which I then go on to do, the coin flipping is outside of the relevant scope of the deliberation.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Authoritatively Normative Concepts

265

make a non-arbitrary selection is like seeking to identify the largest prime: you can engage in this activity, but it is wrongheaded to do so, because you will necessarily be unsuccessful. While I am optimistic that this worry can be answered, I do not aim to do so in this chapter. My aim here is to analyze the concept PRACTICAL OUGHT, and such analysis does not require that successful non-arbitrary selection is possible.

12.5.2. A Preliminary Account of the Concept PRACTICAL OUGHT Recall that my aim here is not necessarily to provide an analysis of a folk concept. Rather it is to provide an account of the concept PRACTICAL OUGHT that vindicates its apparent philosophical significance in debates about Moral Rationalism, and especially in cases like Sticky Situation. I have just argued that in Sticky Situation one is engaged in the activity of non-arbitrary selection. My account is motivated by a pair of hypotheses. First, what it is for the concept PRACTICAL OUGHT to be authoritatively normative is (in part) for it to be the concept of a norm that is distinctively appropriate to appeal to in cases like Sticky Situation. Second, the constitutive norms for the activity of non-arbitrary selection satisfy this description. The first step in developing an account from this intuitive idea is to return to the contrast between constitutive correctness conditions and constitutive success conditions. Consider addition as an example. If I seek to add 17 and 34, I add them correctly if, without error, I use an appropriate mathematical procedure to arrive at the sum. On the other hand, suppose that in adding the two numbers I make a pair of errors, on the basis of which I am lucky to conclude that the sum is 51. I clearly did not add the numbers correctly here, in light of these errors. However, I did add them successfully: my goal was to identify the sum, and I did that. In some activities, correctness ensures success. Simple addition is like this: I cannot correctly add two numbers and get the wrong answer. There are other activities whose constitutive correctness norms do not ensure success. For example, on a toy constitutivist epistemic theory, to believe a proposition correctly is to believe it on the basis of sufficient evidence. And for this belief to be successful is just for the proposition to be true. Fallibilism about sufficient evidence allows that in cases of misleading evidence, one can believe correctly but unsuccessfully. We can distinguish constitutive correctness and success for the activity of non-arbitrary selection. Consider a motivating example: suppose that in Sticky Situation you sought to make a non-arbitrary selection, and thereby engaged in some reasoning that led you to stay and help. Later, you might

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

266

Tristram McPherson

revisit your choice, and conclude (a) that your reasoning was fallacious, but (b) that the correct reasoning in Sticky Situation would have led you to the same conclusion. It is natural here to take yourself to have succeeded in your constitutive aims in Sticky Situation, despite having engaged in the activity incorrectly.²⁰ The question of whether correctness entails success for practical norms is interesting and substantive, and I will not settle it here.²¹ I propose to analyze the concept PRACTICAL OUGHT in terms of the constitutive norms for the activity of non-arbitrary selection. To be plausible, this analysis must appeal to constitutive success as opposed to correctness conditions. For example, in the fallacious reasoning case above, it would be perverse to insist that because you selected the right option on the wrong basis, you failed to do what you ought to do. This puts me in a position to propose the following conceptual truth: Conditional When an instance of the activity of non-arbitrary selection concludes in selection of an option A, for the agent’s judgment I practically ought to do A to be true is for the selection to have satisfied the constitutive success norms of that activity.

Because Conditional provides only a sufficient condition, it falls short of being a conceptual analysis. However, it allows me to sketch how the analysis that builds on Conditional can vindicate the distinctive normative authority that the concept PRACTICAL OUGHT purports to have. This vindication begins with the fact that cases like Sticky Situation are especially powerful tools for orienting us to authoritative concepts like PRACTICAL OUGHT. A compelling diagnosis of this fact is that Sticky Situation is a case of perceived conflict between intuitively important norms. And conflict between such norms makes the need for a non-arbitrary basis for selection especially vivid. I claim that this sort of case is crucial to orienting us to PRACTICAL OUGHT exactly because the distinctive normative authority of this concept is authority in the context of non-arbitrary selection. More precisely, the activity of non-arbitrary selection is the fundamental context in which the question of the relative authority of different normative standards arises. Our grip on relative normative authority in other contexts

²⁰ One implication of this clarification is that the name I have given this activity—nonarbitrary selection—is slightly misleading, because it references the correctness conditions of the activity, and not its success conditions. ²¹ One model for entailment failure would be if correctness (but not success) is information-relative. Note as well that in rational-irrationality style cases, success may be directly incompatible with correctness. For example, a modest assumption about the relevant success conditions for practical deliberation entails that if an evil demon will torture everyone forever if I do whatever is required for correct non-arbitrary selection in C, then successful selection in C will require incorrect selection.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Authoritatively Normative Concepts

267

is parasitic on the relationship that normative standards bear to non-arbitrary selection. This suggests a job description for the concept PRACTICAL OUGHT: it is the concept of a norm that is the norm to appeal to in the context of nonarbitrary selection. The fact that the concept PRACTICAL OUGHT satisfies this job description is what constitutes its distinctive authoritativeness. The deflationary pluralist introduced in Section 12.2 has a seemingly powerful retort to the initial gloss on the vindication just sketched: what is the relevant priority suggested by the talk of the “norm to appeal to” here? As in any context, morality arguably has moral priority in cases of nonarbitrary selection, prudence arguably has prudential priority, etc. So what could talk of a norm being the “norm to appeal to” be here, except a cheat? In response, consider the constitutive success norm for the activity of non-arbitrary selection. If you are engaged in this activity, and your selection satisfies this norm, you will thereby have concluded your activity successfully. And this is a very natural gloss on what it is for a norm to be the one to appeal to in the context of this activity. Returning to Sticky Situation makes this vivid. I have argued that in this scenario you are engaged in the activity of non-arbitrary selection. Suppose for concreteness that you were convinced that the constitutive success norms for this activity require you to always privilege morality over prudence. And suppose that you knew that you are engaged in this activity. There would then be little to intelligibly deliberate about: you already take yourself to know both what morality requires, and that if you complete your deliberations successfully, you will do what morality requires. And this in turn shows why Conditional is so plausible as a conceptual truth. In Sticky Situation, you ask the question what ought I to do? to identify the selection that would constitute success in your activity. Given the role of that question in that context, it is hard to see how you could conclude that activity successfully without doing what you practically ought to do. This concludes my provisional account of what it is for a concept to be authoritatively normative. One reason the account is provisional is that Conditional—the conceptual truth I proposed earlier—only identified a sufficient condition for the truth of the judgment schema I practically ought to do A. Section 12.5.3 takes up two central challenges that arise en route to completing the conceptual analysis.

12.5.3. Scope and the Conditional Fallacy The aim of this section is to complete the analysis of the concept PRACTICAL OUGHT, explaining how it applies to agents not engaged in the activity of non-

arbitrary selection. I begin by suggesting a natural counterfactual extension

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

268

Tristram McPherson

of my account. I then address two important objections to this extension. The first concerns the scope of application of constitutive norms, and the second is an instance of the “conditional fallacy” style of objection to counterfactual accounts. We do not always engage in the activity of non-arbitrary selection. For example, we plump or make unreflective decisions all of the time. (When I argued in Section 12.5.1 that a natural way to interpret Sticky Situation involved non-arbitrary selection, I did so in part by contrasting that activity with other things we might do.) It is plausible that in many cases of unreflective decision, there is something that one practically ought to do. In any case, no account of authoritatively normative concepts should rule this out. In order to satisfy this constraint, I must extend Conditional to provide an analysis that can apply to such decisions. I do so by offering a natural counterfactual extension of my account. An initial gloss on the extension is this: in cases where I act unreflectively, for it to be true that I practically ought to have done A, is for it to be that were I to have successfully non-arbitrarily selected, I would have done A. An immediate objection to this proposal is that my counterfactual participation in an activity governed by constitutive norms does not make those norms actually apply to me. For example, were I playing chess right now, and I touched my queen, the rules of chess would require me to move that piece (provided it was legal to do so). But because I am not playing chess right now, touching a chess piece entails no such requirement. So, the objection continues, why should facts about counterfactual non-arbitrary selection have any normative significance for me in a situation when I do not engage in this activity? I reply to this objection by emphasizing the contrast between the account of normative authority that I offer, and the more standard form of constitutivist explanation. To simplify brutally, this standard explanation is that normative authority just is the grip that constitutive norms have on the participants of the activities they govern, for an activity we cannot escape. My account of normative authority is different. Put abstractly, the standard form of constitutivist explanation locates normative authority in the relation of a norm to an agent (inescapability), while my account locates normative authority in the nature of the activity itself, independently of its relationship to any actual agent. On my account, as sketched in Section 12.5.2, what makes a norm distinctively authoritative is that it is distinctively the norm to appeal to in the context of non-arbitrary selection. Crucially, the fact that a norm has this status is independent of whether anyone is actually engaged in the activity of non-arbitrary selection. Thus, on my account, authoritative norms can lack any “grip” on actual agents: they can apply to agents who are wholly and coherently indifferent to its prescriptions.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Authoritatively Normative Concepts

269

A second natural worry about my counterfactual proposal is that it is vulnerable to the conditional fallacy (e.g. Shope 1978, Johnson 1997). The basic worry is this. Suppose (à la Conditional) that any time someone actually successfully non-arbitrarily selects, she selects the option that she ought to perform. Now consider a case in which I do not seek to nonarbitrarily select. It seemingly might be that the nearest possible world in which I seek to do so is relevantly different from my actual circumstances, such that while I ought to perform a certain action in that world, I ought not to perform it in my actual circumstances. If this were possible, then the counterfactual extension of my account would deliver the wrong results. A theoretically contentious example may make the worry more vivid. My view about the concept PRACTICAL OUGHT should be compatible with familiar Humean theories about the substance of what we practically ought to do. According to these theories, facts about what one ought to do in a context are a (non-constant) function of one’s desires in that context. Consider the thesis that I ought now to deliberate, supposing that I currently have no desire to do so. The closest worlds in which I seek to non-arbitrarily select concerning this question are worlds in which I deliberate, and hence, in which I have at least some desire to deliberate. My counterfactual account thus seemingly might entail that I ought now to deliberate, in light of these counterfactual desires, despite my actually lacking any relevant desire. This is inconsistent with the familiar Humean thesis. Crucially, my account is only vulnerable to the conditional fallacy if the constitutive success norms for the activity of non-arbitrary selection have a certain structural feature: that success conditions can vary with features of the context of assessment. This feature is absent from some familiar constitutive norms. To see this, consider chess: holding fixed a certain chess position, every correct assessment of the legal moves available to black in that position will be the same, no matter the context of assessment. It is plausible that we are conceptually committed to the constitutive success norms for non-arbitrary selection exhibiting a similar invariance. To see this, focus on a case where an agent engages in non-arbitrary selection prospectively. For example, suppose that yesterday I attempted to non-arbitrarily select an option to perform now, and that I now recheck that reasoning. I would be confused if I believed now that I had successfully non-arbitrarily selected yesterday, but that I might get a different answer if I successfully non-arbitrarily select now. The same thing holds for possible cases of retrospective or hypothetical non-arbitrary selection. A retrospective example: sitting in jail after staying to help, you might ask yourself: should I have taken the money and run? Or, reading this chapter you might engage in non-arbitrary selection hypothetically, considering what to do were you in Sticky Situation. When we imagine such cases, we notice that a form

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

270

Tristram McPherson

of invariance is built into the concept NON-ARBITRARY SELECTION: at any two contexts of evaluation where I successfully non-arbitrarily select an option for a single context of action, I must come to the same conclusion. Call this property of the concept NON-ARBITRARY SELECTION intrapersonal invariance.²² This property is plausible when combined with my background account of the nature of normative authority. Suppose that you were convinced by the argument of Section 12.5.2 that it is a conceptual truth that successful non-arbitrary selection in the context of action entails that one does what one ought to do in that context. And now consider the sort of case just envisioned, where I had previously engaged in successful non-arbitrary selection prospectively, for my current circumstances. It would be perverse to think that the mere fact that I engaged in the activity prospectively prevents me from drawing from this reasoning a conclusion about what I practically ought to do now. The same goes for beliefs about successful retrospective and hypothetical non-arbitrary selection: these are well understood as conceptually entailing conclusions about what I hypothetically or retrospectively ought to have done. This fits in with my account of normative authority: in short, the question of distinctive normative authority arises in the context of the activity of non-arbitrary selection, but we can answer such questions about decisions reached outside of that context. The conditional fallacy worry arises because of the apparent possibility that success conditions for the activity of non-arbitrary selection concerning a certain context of selection might vary with features of the counterfactual context of deliberation. But intrapersonal invariance blocks exactly this possibility, and so blocks the conditional fallacy from arising for my account. This puts me in a position to refine my counterfactual extension into a conceptual analysis of PRACTICAL OUGHT: Constitutive S practically ought to do A in context of action C =def the constitutive success conditions for S’s activity of non-arbitrary selection concerning C require doing A.

In Section 12.2 I suggested that the thought that a concept is authoritatively normative is elusive. My account neatly explains this elusiveness. The connection between PRACTICAL OUGHT and NON-ARBITRARY SELECTION is not obvious; I have argued that it becomes vivid only when we think carefully about scenarios like Sticky Situation. Further, only with conditions like intrapersonal invariance on the table is it possible to see how an account ²² Note that intrapersonal invariance is compatible with the constitutive success conditions for non-arbitrary selection varying across evaluators. I am attracted to interpersonal invariance here as well, but nothing in this chapter depends on that.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Authoritatively Normative Concepts

271

of PRACTICAL OUGHT in terms of NON-ARBITRARY SELECTION can have adequate scope. If Constitutive is the correct analysis of PRACTICAL OUGHT, it is very natural to think that the concept would appear highly elusive, seemingly supporting primitivism or deflationary pluralism.

1 2 .6 . SC H M A G E N C Y My account offers a kind of constitutivist explanation of the distinctive normative authority of the concept PRACTICAL OUGHT. The most powerful contemporary objection to this broad explanatory strategy is David Enoch’s “schmagency” challenge. I thus now consider two ways of attempting to adapt Enoch’s challenge to apply to my account. As I read it, the core of Enoch’s challenge begins with the constitutivist’s insight that our various ordinary motivations are too arbitrary a basis on which to ground normativity. By appealing to features that are constitutive of action (for example), the constitutivist aims to eliminate this objectionable arbitrariness (Enoch 2006, 178). Against this, Enoch argues that the problematic arbitrariness cannot be eliminated by the inescapability of the relevant activity (Enoch 2006, §6), or by the related idea that we (can’t help but) care about agency or its constitutive norms (Enoch 2011, 212–13), or by insisting that one cannot even raise doubts about normativity except in a context where one is already committed to certain constitutive norms (Enoch 2011, §6). Because I reject the mechanisms for arbitrariness reduction that Enoch considers on behalf of the constitutivist, I can simply embrace the arguments just mentioned. However, it is worth addressing two important ways of adapting Enochian ideas to challenge my account. Both challenges focus on the concept NON-ARBITRARINESS that is central to my analysis. First, it may seem that I face a dilemma. The concept NON-ARBITRARINESS will either: (a) be too thin to generate facts about what we ought to do, or (b) be thick enough to prompt reasonable normative resistance: why care about non-arbitrariness? (This dilemma is inspired by Enoch 2011, 213.) The first horn fails against my account, simply because my account is not committed to there being such facts. As I noted in Section 12.5.1, my aim is to analyze a concept, not to show that we can derive substantive norms from that analysis. The second horn is a variant of the central “schmagency” idea that Enoch prosecutes: he imagines someone who is “normatively indifferent”

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

272

Tristram McPherson

to the alleged constitutive features of agency. The possibility of such indifference is no challenge to my account, for, as I noted in Section 12.5, nothing in my account entails that anyone cares about what they ought to do (or about non-arbitrariness). It is also true that a reasonable agent could resist the explanation of normative authority that I sketched in Section 12.5.2. But again, this is compatible with my account: I have argued that this account is true, not that it must be accepted by any reasonable agent. A second schmagency-style challenge starts by supposing that there is a set of slightly different arbitrariness concepts: ARBITRARINESSA, ARBITRARINESSB, etc., which agree on central cases but diverge at the penumbra. These could seemingly be referenced by the constitutive norms of a set of activities: non-arbitrary selectionA, etc. A schmagent might thus ask: why go in for nonarbitrary selectionA, rather than another variant activity? The reply to this challenge begins by observing that the aim to make a non-arbitrary selection is incompatible with the existence of such higherorder arbitrariness. If the choice between the activities non-arbitrary selectionA and non-arbitrary selectionB were itself arbitrary, then neither of these activities satisfies our concept of non-arbitrariness. This shows that it is a conceptual constraint on the concept of NON-ARBITRARINESS that this concept is privileged in some way over nearby notions. I favor a picture where the presupposed privileging is metaphysical: where we take the arbitrary/nonarbitrary distinction to be part of the genuine structure of the world. In adverting to non-arbitrariness, we aim to be latching on to that structure. If it turns out that there is no such structure—or that there are multiple candidate structures that we might be talking about—then the uniqueness assumption of the activity is undercut, and the result will again likely be that the concept PRACTICAL OUGHT is necessarily empty. This would be an unappealing result, but it does not threaten the conceptual analysis itself.

1 2 . 7 . CO N S E Q U E N C E S O F T H E A N AL Y S I S This chapter has aimed to illuminate the nature of authoritatively normative concepts. The centerpiece of my account is an analysis of the concept PRACTICAL OUGHT, which draws in a novel way on some resources from the constitutivist tradition: Constitutive S practically ought to do A in context of action C =def the constitutive success conditions for S’s activity of non-arbitrary selection concerning C require doing A.

I have argued that this analysis permits an elegant explanation of what the distinctive authoritativeness of the normativity of this concept amounts to.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Authoritatively Normative Concepts

273

Roughly: what makes this concept distinctively authoritatively normative is that it is (constitutively) the norm to appeal to when engaged in the activity of non-arbitrary selection. I also suggested that this account permits an attractive explanation for why the nature of this authoritativeness has seemed so elusive. Besides these intrinsic merits, my account illustrates the flexibility of constitutivist metanormative accounts. For example, I have just argued that my account escapes Enoch’s schmagency challenge to constitutivist accounts of normative authority. One might reasonably regret that this flexibility is purchased at the cost of reduced ambition: for example, there is no hint in my account of an attempt to derive any substantive normative conclusions—let alone the categorical imperative—from constitutivist materials. However, I take it to be a mistake to think that constitutivism must accomplish everything its proponents have dreamed of, in order to deserve our attention or assent. I conclude by briefly sketching some implications of my account for broader metaethical theorizing. Consider first the dialectic between the metaethical cognitivist and the non-cognitivist (where, roughly, the cognitivist claims that, at the fundamental explanatory level, the thought that I ought to do A is a belief ). One way to motivate the non-cognitivist view is to insist that primitivism about authoritatively normative concepts is unacceptable. Once this is done, the non-cognitivist conjectures that what marks off authoritatively normative thoughts as distinctive is that they are constituted by certain non-cognitive states. One dialectical strength of the non-cognitivist view, if we grant the objection to primitivism, is that plausible and non-trivial cognitivist accounts of fundamental authoritatively normative concepts are thin on the ground. However, with my account in hand the dialectic shifts. The cognitivist can point out that her view accommodates the plausible thought that authoritatively and merely generic concepts are species of a unified conceptual genus: the normative concepts. The non-cognitivist, by contrast, faces pressure to posit a deep and not obviously plausible discontinuity in the fundamental nature of authoritative and merely generic normative concepts, on pain of offering an implausible non-cognitivist gloss on chess norms, for example. My account also helps the cognitivist with respect to the dialectic concerning normative disagreement. R. M. Hare (1952) and Terence Horgan and Mark Timmons (e.g. 1992) have argued persuasively that genuine moral disagreement seems possible between members of linguistic communities whose moral thought seems to track different properties. Such disagreement phenomena seem to carry over with even more force to the case of authoritatively normative thoughts. My account can potentially help to explain how this fact is compatible with cognitivism: my account analyzes

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Tristram McPherson

274

the authoritatively normative concepts in terms of an extremely thin functional role. As long as a linguistic community’s word behaves in accordance with this role, it will express the same concept. My account can also help to defend particular cognitivist metaethical projects against some of the characteristic challenges that they face. Consider first the way it can be useful to the metaethical error theorist and the fictionalist. Schematically, the most straightforward way of arguing for error theory about the authoritatively normative is as follows: Conceptual claim:

All authoritatively normative claims commit us to the existence of X.

Metaphysical claim: Conclusion:

X does not, or could not, exist. Authoritatively normative claims are systematically erroneous.

One central difficulty for the error theorist is to provide a plausible version of the conceptual claim (Finlay 2008). Illustrative here is the derision rightly heaped on John Mackie’s claim that it is part of the ordinary concept of objective value that such values somehow magically make us pursue them (1977, 40). My account can help the error theorist here. For if she accepts my conceptual claim, she can argue as follows: Conceptual claim:

The concept of an authoritative norm is the concept of the constitutive success norm for the activity of non-arbitrary selection.

Metaphysical claim:

There are no constitutive success norms for this activity. Authoritatively normative claims are systematically erroneous.

Conclusion:

Metaethical fictionalists face an analogue of the error theorist’s conceptual challenge. In this case, the task is to tell us what, at least roughly, makes a fiction a fiction about authoritative normativity (compare Hussain 2004). My account can again help. The fictionalist could treat the claim that there are success norms for the activity of non-arbitrary selection either as a characterization of the content of the fiction that we already implicitly accept (for the hermeneutic fictionalist), or as an account of the fiction that we should adopt (for the revolutionary fictionalist). As I have been emphasizing, my account is neutral concerning whether there are any authoritatively normative facts. So it does not by itself promise to vindicate normative realism. And while the account is compatible with the strategy of arguing for a constitutivist derivation of normative content, I am pessimistic about the prospects for that strategy. Readers who join me

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Authoritatively Normative Concepts

275

in this assessment may think that the account of the concept PRACTICAL offered here makes it hard to see how that concept could have a nonempty intension. I will close with the barest sketch of the picture that makes me optimistic on this front. I am compelled by a thesis David Lewis (1983, 1984) proposed about reference determination: that a factor in determining reference is the eligibility of candidate referents. Adapted to content-determination, the idea is that the content of a concept is a function of two factors: how well the candidate content satisfies the functional role of the term, and how well the candidate content carves nature at its joints. A natural way of characterizing the metaphysical commitment of normative realism is this: that the authoritative norms are distinctive joints of nature among the space of possible normative systems. The functional role suggested by Constitutive is extremely thin. Being the distinctive joint of nature among the space of normative systems is (a) a non-arbitrary basis for selection, and (b) an eligibility maker. On this view, the determinate and stable reference of the concept PRACTICAL OUGHT is a function of the thin conceptual role suggested by Constitutive, the realist’s distinctive metaphysical claim that there is a distinctive normative joint of nature, and (an appropriately developed version of) the sketched theory of content-determination. This is, of course, the barest of sketches of an ambitious view (for more detailed discussion of related metasemantic ideas in a metaethical context, see Dunaway and McPherson 2016). However, I think it is highly promising in broad outline, and it at least suffices to show why accepting Constitutive does not force the metaethical realist to abandon all hope.²³

OUGHT

References Baker, Derek. 2016. “Intuitions About Disagreement Do Not Support the Normativity of Meaning.” Dialectica 70(1): 65–84. Bedke, Matt. 2010. “Might All Normativity Be Queer?” Australasian Journal of Philosophy 88(1): 41–58.

²³ I am grateful for illuminating discussion of this chapter at the Melbourne Moral Rationalism Workshop, the Australasian Annual Workshop in Moral Philosophy, the Hong Kong Metaethics Workshop, University of Kentucky, the Virginia Tech philosophy faculty reading group, Ohio State, the CRNAP Varieties of Normativity Workshop at Princeton, the Varieties of Normativity Workshop at Uppsala, and the Chapel Hill Metaethics Workshop. Special thanks to Derek Baker, David Copp, Garrett Cullity, David Faraci, Anandi Hattiangadi, Sarah McGrath, David Plunkett, and Michael Smith, and two anonymous referees for Oxford University Press, whose comments and conversations dramatically improved my understanding of the issues in this chapter.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

276

Tristram McPherson

Copp, David. 2005a. “Moral Naturalism and Three Grades of Normativity.” Normativity and Naturalism. Ed. Peter Schaber. Frankfurt: Ontos-Verlag, 7–45. Copp, David. 2005b. “The Normativity of Self-Grounded Reason. Social Philosophy and Policy 22: 165–203. Cullity, Garrett, and Berys Gaut (Eds.) 1997. Ethics and Practical Reason. Oxford: Clarendon. Darwall, Stephen. 1997. “Reasons, Motives, and the Demands of Morality.” Moral Discourse and Practice. Eds. Stephen Darwall, Allan Gibbard, and Peter Railton. New York: Oxford University Press, 305–12. Davidson, Donald. 2001 [1969]. “How is Weakness of the Will Possible?” Essays on Actions and Events, 2nd edition. Oxford: Oxford University Press, 21–42. Dunaway, Billy, and Tristram McPherson. 2016. “Reference Magnetism as a Solution to the Moral Twin Earth Problem.” Ergo 3(25): 639–79. Eklund, Matti. 2017. Choosing Normative Concepts. Oxford: Oxford University Press. Enoch, David. 2006. “Agency, Schmagency.” Philosophical Review 115(2): 169–98. Enoch, David. 2011. “Schmagency Revisited.” New Waves in Metaethics. Ed. Michael Brady. Basingstoke: Palgrave Macmillan, 208–33. Ferrero, Luca. 2009. “Constitutivism and the Inescapability of Agency.” Oxford Studies in Metaethics, volume 4. Ed. Russ Shafer-Landau. Oxford: Oxford University Press, 303–33. Finlay, Steven. 2008. “The Error on the Error Theory.” Australasian Journal of Philosophy 86(3): 347–69. FitzPatrick, William. 2008. “Robust Ethical Realism, Non-Naturalism and Normativity.” Oxford Studies in Metaethics, volume 3. Ed. Russ Shafer-Landau. Oxford: Oxford University Press, 159–205. Foot, Philippa. 1997. “Morality as a System of Hypothetical Imperatives (including her 1994 ‘Recantation’).” Moral Discourse and Practice. Eds. Stephen Darwall, Allan Gibbard, and Peter Railton. New York: Oxford University Press, 313–22. Gibbard, Allan. 2003. Thinking How to Live. Cambridge, MA: Harvard University Press. Gibbard, Allan. 2012. Meaning and Normativity. Oxford: Oxford University Press. Hampton, Jean. 1998. The Authority of Reason. Cambridge: Cambridge University Press. Hare, R. M. 1952. Language of Morals. Oxford: Oxford University Press. Horgan, Terence, and Mark Timmons. 1992. “Troubles for New Wave Moral Semantics: The Open Question Argument Revisited.” Philosophical Papers 21(3): 153–75. Hussain, Nadeem. 2004. “The Return of Moral Fictionalism.” Philosophical Perspectives 18: 149–87. Jackson, Frank. 1998. From Metaphysics to Ethics. Oxford: Oxford University Press. Johnson, Robert. 1997. “Reasons and Advice for the Practically Rational.” Philosophy and Phenomenological Research 57(3): 619–25. Jones, Karen, and François Schroeter (eds.) Forthcoming. The Many Moral Rationalisms. Oxford: Oxford University Press.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Authoritatively Normative Concepts

277

Katsafanas, Paul. 2013. Agency and the Foundations of Ethics. Oxford: Oxford University Press. Korsgaard, Christine. 2009. Self-Constitution: Agency, Identity, and Integrity. Oxford: Oxford University Press. Kripke, Saul. 1982. Wittgenstein on Rules and Private Language. Cambridge, MA: Harvard University Press. Lewis, David. 1983. “New Work for a Theory of Universals.” Australasian Journal of Philosophy 61(4): 343–77. Lewis, David. 1984. “Putnam’s Paradox.” Australasian Journal of Philosophy 62(3): 221–36. Lord, Errol, and David Plunkett. 2017. “Reasons Internalism.” Routledge Handbook of Metaethics. Eds. Tristram McPherson and David Plunkett. London: Routledge, 324–39. Mackie, J. L. 1977. Ethics: Inventing Right and Wrong. Harmondsworth: Penguin. McPherson, Tristram, and David Plunkett. 2017. “The Nature and Explanatory Ambitions of Metaethics.” Routledge Handbook of Metaethics. Eds. Tristram McPherson and David Plunkett. London: Routledge, 1–25. McPherson, Tristram, and David Plunkett. Forthcoming. “Conceptual Ethics and the Methodology of Normative Inquiry.” Conceptual Ethics and Conceptual Engineering. Eds. Alexis Burgess, Herman Cappelen, and David Plunkett. Moore, G. E. 1993 [1903]. Principia Ethica, revised edition. Ed. Thomas Baldwin. Cambridge: Cambridge University Press. Parfit, Derek. 2011. On What Matters. Oxford: Oxford University Press. Railton, Peter. 1989. “Naturalism and Prescriptivity.” Social Philosophy and Policy 7: 151–74. Scanlon, Thomas. 1998. What We Owe to Each Other. Cambridge, MA: Belknap. Schroeder, Mark. 2007. Slaves of the Passions. Oxford: Clarendon. Shope, Robert. 1978. “The Conditional Fallacy in Contemporary Philosophy.” Journal of Philosophy 75(8): 397–413. Silverstein, Matthew. 2015. “The Schmagency Question.” Philosophical Studies 172: 1127–42. Smith, Michael. 1994. The Moral Problem. Oxford: Blackwell. Street, Sharon. 2008. “Constructivism About Reasons.” Oxford Studies in Metaethics, volume 3. Ed. Russ Shafer-Landau. Oxford: Oxford University Press, 207–46. Tiffany, Evan. 2007. “Deflationary Normative Pluralism.” Canadian Journal of Philosophy, supp. vol. 33: 231–62. Van Roojen, Mark. 2010. “Moral Rationalism and Rational Amoralism.” Ethics 120(3): 495–525. Velleman, J. David. 2009. How We Get Along. Cambridge: Cambridge University Press. Wedgwood, Ralph. 2004. “The Metaethicist’s Mistake.” Philosophical Perspectives 18(1): 405–26. Wedgwood, Ralph. 2007. The Nature of Normativity. Oxford: Oxford University Press.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

13 The Rationality of Ends Sigrún Svavarsdóttir

How do we determine whether agents can be more or less rational on account of their ends? We could start with well-known examples of putative irrationality in the selection of ends. Consider the agent who adopts the end¹ of proving that π is an algebraic number. Or the agent who adopts the end of counting blades of grass in park squares. Or the agent who adopts the end of keeping Tuesdays free of pain even at the cost of greater pain on other days of the week.² If agents are not subject to evaluation in terms of rationality on account of their ends, none of the ends enumerated should raise questions about the rationality of the respective agent. But don’t they? Well, how are we to determine the answer to that question? For sure, there is nothing ungrammatical or semantically dubious about sentences like “It is irrational to adopt the final end of counting blades of grass in park squares.” Moreover, it would be questionable to accuse those who assert that sentence of misapplying the ordinary concept of rationality. However, the issue that interests me is neither about the syntactic and semantic behavior of the term “rational” (or “irrational”) nor about the application conditions of our ordinary concept of rationality. Instead, it is an issue within the theory of practical rationality, conceived of as a branch of normative theory, and the methodology of normative inquiry is, alas, in dispute. Here I will try to make some headway on questions regarding the rationality of ends by starting with some very general thoughts about the kind of assessment that is the province of theories of practical rationality. This is the task of Section 13.1, while Section 13.2 relies on these general ¹ I should be understood as talking about final ends whenever I talk about ends unless I specify otherwise. Also, I use “end” and “goal” interchangeably. ² These are well-known examples from the discussion of the rationality of desires. For the first two examples, see Rawls (1971: 419 and 432, respectively). For the third example, see Parfit (1984: 124). Both accept the view of ends as the objects of desire, so they would hardly object to rewording their examples so as to raise the question of the rationality of final ends.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

The Rationality of Ends

279

thoughts to defend the thesis that an agent can display more or less rationality in selecting ends. Section 13.3 argues that this conclusion does not decide the issue between Humeans and anti-Humeans about practical rationality in spite of the traditional association of Humeanism with the claim that an agent’s final ends are not subject to evaluations of rationality.

1 3 .1 . SO M E G E N E R A L T H O UG H T S O N P R A C T I C A L R AT I O N A LI T Y I take a theory of practical rationality to be a normative theory pertaining primarily to the assessment of fairly cognitively sophisticated agents. Such agents are practically rational to the degree that they are disposed to display excellence in the exercise of cognitive capacities in their practical endeavors. A theory of practical rationality is a theory of that excellence: a theory of what it essentially takes to display excellence in the exercise of cognitive capacities in one’s practical endeavors.³ In developing such a theory, some attention needs to be paid to the cognitive capacities of the kind of agents who are subject to evaluation as more or less practically rational. In particular, it must be examined how these capacities come into play in the practical domain. This requires forays into the philosophy of mind and action, if not cognitive science, but a normative theory of practical rationality should not be confounded with a philosophical or a scientific theory of cognition and its role in the practical domain. It lays out standards for evaluating how an agent exercises cognitive capacities in his practical endeavors rather than gives a descriptive cum explanatory account of cognition in the practical domain. Let’s call this “the cognitive excellence conception of practical rationality.” I find this general conception of practical rationality a compelling starting point for thinking about the subject matter of theories of practical rationality. Yet, this is by no means a common characterization of their subject matter. In the contemporary philosophical literature, it is more common to characterize practical rationality as consisting in the responsiveness to (normative) reasons for actions and practical attitudes⁴ or in the satisfaction of coherence requirements on practical attitudes and actions.⁵ Below I explain why I find the cognitive excellence conception to be a better starting point than the two more familiar ones. ³ This conception of practical rationality is set forth in Svavarsdóttir (2006) and (2008). ⁴ See, for example, Parfit (2011: 111) and Raz (1999: 68). ⁵ The so-called “wide-scope” requirements of rationality are coherence requirements on attitudes and actions. See especially Broome (2013).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

280

Sigrún Svavarsdóttir

13.1.1. Practical Rationality and Responsiveness to Reasons I am sitting next to my dear friend Anna, when I spontaneously put my arm around her shoulders and ask how she is doing, flashing a sympathetic smile. I am myself a bit taken back by my supportive gesture and worry that it will make for an awkward moment. I acted on an urge before having the time to get any second thoughts. I do not quite know what possessed me. Anna had not indicated any problems or solicited my support, and I see absolutely no reason to think that there is anything wrong with her. Hopefully, she takes my gesture only as an innocent expression of friendly feelings and does not find this intrusive, off-putting, or downright weird behavior. All these thoughts occur to me in a flash during the short moment before Anna, thankful for the opening, spills the beans. I had somehow picked up on her state of extreme anxiety and responded with a supportive gesture. Now, let’s assume that although I had no conscious thought of Anna being in a state of extreme anxiety and had no conscious awareness of anything in her behavior that indicated as much, my supportive gesture was in fact the result of a physiological mechanism that made her anxiety register with me and yield the urge to make the supportive gesture. Let’s say that odorants emitted by her body due to extreme stress were detected by my olfactory system that yielded neural impulses that led to brain activity manifested in my urge to extend a supportive gesture to Anna. Let’s furthermore assume that, in the circumstances at hand, the fact that a dear friend is in a state of anxiety is a decisive reason to make a supportive gesture towards her. Given all of this, it seems perfectly reasonable to say that my behavior manifested some kind of responsiveness to reasons for extending a supportive gesture towards Anna. But does my behavior manifest practical rationality? I cannot see that it does. Perhaps, it manifests my great sensitivity to other people’s state of mind, or Anna’s in particular; perhaps, it manifests my sympathetic nature; but it does not manifest my rationality any more than a dog’s sensitivity to the moods of its master does. My ground for denying that the behavior in question manifests practical rationality is not that a physiological process was involved in its production. I am of the persuasion that physiological processes are involved in the production of any human behavior, rational or not. However, in order to bear on the rationality of an agent, the behavior in question has to be the product of the kind of physiological process that yields fairly sophisticated cognitive activity in light of which the conduct makes sense to the agent herself. That cognitive activity does not have to amount to arduous self-conscious deliberation, but it cannot be as rudimentary as in my story. As I have told the story, my action was born of an

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

The Rationality of Ends

281

urge, to that extent it made sense to me, but I did not understand the urge and had nothing to say in favor of it, even if the impulse was triggered by my literally smelling Anna’s anxiety. I have disavowed reliance on linguistic or conceptual intuitions in developing and defending my position on practical rationality. I also disavow reliance on a priori intuitive insights into the nature of the evaluative property at stake. However, I come to this subject matter as a philosopher, trained within the tradition of Western thought, a tradition that has been preoccupied through the ages with the rationality of humankind. I am interested in that subject matter: a normative appraisal related to a trait that has long been claimed (correctly or incorrectly) to separate us from the animals and cast us in the image of a deity conceived as the intelligent designer of the universe. For sure, the assumption has been that the trait in question crucially involves our cognitive capacities: our capacities for representation and reasoning. If we give up on that presumption, we have changed the topic of conversation rather than advanced a radical thesis about a subject matter that we share with traditional rationalists. That is, at least, my sense of what has been at stake in philosophical conversations, spanning the ages, about rationality displayed in the practical domain. The foregoing does not amount to an argument against the conception of practical rationality as a kind of responsiveness to normative reasons for actions and attitudes. For it is possible to develop that conception of practical rationality so as to require that the relevant kind of responsiveness involve cognitive activity, presumably of high quality. However, this would create a link with the conception of practical rationality as an excellence in the exercise of cognitive capacities in one’s practical endeavors. The right kind of responsiveness to reasons and excellence in the exercise of cognitive capacities in one’s practical endeavors would, presumably, come to one and the same thing. Thus, there need not be any conflict between these two very abstract ways of conceiving of practical rationality. However, I prefer as my starting point the latter rather than the former conception of practical rationality for two reasons. First, it focuses attention directly on the subject matter of traditional disputes about the nature of rationality: the nature and assessment of cognitive activity in the practical domain. Second, it steers us towards an inquiry that seems tractable: an inquiry into the cognitive capacities of beings like us and their role in our practical endeavors,⁶ even if the

⁶ As noted earlier, this requires forays into the philosophy of mind and action, if not cognitive science. Of interest is not only the kind of linguistically well-articulated conscious reasoning that is often emphasized by traditional rationalists, but also the kind of cognitive activity that underlies intellectually skillful activities like those of chess

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

282

Sigrún Svavarsdóttir

question concerns ultimately the appropriate standards for evaluating an agent’s exercise of these cognitive capacities in her practical endeavors. This seems more tractable than an inquiry into the nature of normative reasons and of the appropriate kind of responsiveness to reasons. Unlike some contemporary reason-responders, I am not satisfied with relying on one or both of these elusive notions unexplained.

13.1.2. Practical Rationality and Coherence Practical rationality is sometimes characterized as consisting in, or at least requiring, coherence of practical attitudes and coherence of actions with those attitudes. This does not tell us much until the relevant coherence relations are specified. However, it should be uncontroversial that these are non-causal relations that hold between attitudes in virtue of their contents and between attitudes and actions in virtue of the contents of the former and intentional descriptions of the latter. Any further specification of these relations is controversial, especially against the assumption that practical rationality is to be measured in terms of the degree of coherence between attitudes and between attitudes and actions.⁷, ⁸ It seems plausible to assume that coherence relations are the kinds of relations that would hold between an agent’s attitudes in virtue of their masters, quick-witted conversationalists, or strategic tennis players who are “in the flow” rather than deliberatively working through their next move. ⁷ I do not accept a coherence view of rationality but, in a different context, I have made suggestions about how to understand the coherence of attitudes. See sections 3.1 and 3.3 of Svavarsdóttir (2015). ⁸ One prominent example of a coherentist theory of rationality is the fairly standard neo-Humean account of practical rationality as the maximization of expected utility, when the expected utility of an action is the sum of the products of the subjective utility and the subjective probability of each of the action’s possible outcomes. Since the subjective utility is a measure of the agent’s preference for the outcome and the subjective probability is his estimate of the likelihood of that outcome, an action which maximizes expected utility is an action, under a certain description, that stands in a certain noncausal relation to the agent’s preferences and beliefs in virtue of the content of these states and the action-description in question. In addition, such neo-Humean accounts lay down coherence conditions on preferences in the form of the axioms of utility theory. (However, in fairness, it should be acknowledged that the focus of neo-Humean accounts is often on the rationality of options rather than on the rationality of agents.) In a more Kantian spirit, attempts could be made to specify coherence conditions on practical attitudes such as to ensure that insofar as an agent is fully rational, the ways in which she intends to treat any rational being are ways in which she condones that any other rational being be treated in exactly the same circumstances. Yet another familiar approach spells the relevant coherence relations out in terms of relations between, on the one hand, actions or attitudes and, on the other hand, judgments about reasons for such actions or attitudes. See, for example, Scanlon (2007).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

The Rationality of Ends

283

contents and between her attitudes and actions in virtue of the contents of the former and intentional descriptions of the latter, if she had exercised, in optimal conditions, cognitive capacities maximally well in her endeavors. In other words, specifying the coherence relations goes hand in hand with specifying the standards for evaluating an agent’s exercise of cognitive capacities in her endeavors. Nevertheless, there is a crucial difference between the coherence conception and the cognitive excellence conception of practical rationality. On the coherence conception, it does not matter why the relevant coherence relations hold. It need not be because the agent has exercised cognitive capacities well in forming her attitudes or deciding on her course of action. The coherence may be a pure cosmic coincidence rather than the upshot of intellectual resourcefulness on the agent’s part.⁹ By contrast, on the cognitive excellence conception, it crucially matters for assessments of the agent’s rationality why the relevant coherence relations hold. Coherence is at best a symptom of rationality. Practical rationality consists in being disposed to exercise cognitive capacities well in one’s practical endeavors rather than in having attitudes and actions configured in certain ways, although how well one has exercised cognitive capacities will, at least to some extent, be reflected in the configuration of one’s attitudes and actions. I say “at least to some extent” because it should not be assumed without argument that for agents who exercise cognitive capacities in suboptimal conditions, the degree of rationality (on this conception of rationality) and the degree of coherence is invariably the same. I do not have a knock-down argument for the superiority of the cognitive excellence conception over the coherentist conception of practical rationality. Again, I appeal to an interest in sticking to the subject matter of traditional debates about the nature of rationality in the practical domain. As noted in Section 13.1.1, this debate concerns a trait that was traditionally supposed to separate us from the animals and cast us in the image of a deity conceived as the intelligent designer of the universe. Presumably, such a supreme intelligence would have perfectly coherent attitudes and its divine actions would perfectly cohere with these attitudes, given the assumption that coherence relations are the kinds of relations that would hold between an agent’s attitudes in virtue of their contents and between his attitudes and actions in virtue of the contents of the former and intentional descriptions of the latter, if he had exercised, in optimal conditions, cognitive capacities ⁹ This is a version of the cosmic coincidence objection that I raise to both coherentist and counterfactual accounts of practical rationality in Svavarsdóttir (2008: 26–8). Notice that the objection targets coherentist accounts of the rationality of agents rather than of options. It is in terms of the former kind of evaluation that I seek understanding of the evaluation of ends as more or less rational.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

284

Sigrún Svavarsdóttir

maximally well in his endeavors. However, it seems odd to think that our practical rationality is primarily a matter of the configuration of our attitudes and actions approximating this divine coherence in attitudes and actions, regardless of how our overall state of mind was produced or maintained, rather than a matter of approximating, or being disposed to approximate, the divine intellectual activity that yields perfect coherence of attitudes and actions in optimal conditions. On the cognitive excellence conception of practical rationality, the degree to which an agent is (overall) practically rational is the degree to which she is disposed to display excellence in the exercise of cognitive capacities in her practical endeavors over time rather than the extent to which she has actually done so, and the strength of that disposition may, to some extent, be a matter of luck. However, for embodied beings like us, the disposition is honed in the context of an actual exercise of cognitive capacities in one’s practical endeavors. Moreover, an agent’s attitudes do not manifest practical rationality unless they are the upshot of cognitive activity, and the rationality an agent displays on any given occasion is a function of how she forms an attitude or decides on an action. Thus, on the cognitive excellence conception, rationality in attitude and conduct marks a cognitive accomplishment in conducting one’s practical affairs. For embodied and limited agents like us that accomplishment is very much a matter of using well our cognitive capacities to cope with various constraints: e.g., epistemic constraints, physical or psychological obstacles, vulnerability to external events, health constraints, and constraints on an ability to cope with adversity as well as other emotional and motivational constraints. This is not merely a matter of reasoning well by standards of deductive and inductive reasoning but, also, of tailoring reasoning and other cognitive activity intelligently to what we are up against and to what we are up to. It is also a matter of making adjustments in attitude or conduct that make good sense in light of what our cognitive activity has brought into purview. Sometimes this will yield a wait and see attitude, waiting to make up one’s mind—on what to believe, on what practical stance to take, or on what to do—until more evidence is in or one has gained more strength to cope with adverse consequences, even if the wait prolongs a state characterized by a lack of coherence in attitude and action. The cognitive excellence conception of practical rationality makes room for the possibility that this is the rational thing to do on occasions. This seems to be a promising approach to developing a theory of rationality that yields evaluative standards that are reasonably applied to limited agents like us: more promising than an approach that focuses inquiry immediately on the relations that would hold among our attitudes and between our attitudes and actions, were we ideal cognitive agents in optimal conditions—conditions of omniscience and omnipotence.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

The Rationality of Ends

285

1 3. 2. T H E R A T I O N A L I T Y O F EN D S Turning to the main topic of this chapter—namely, the rationality of ends—let’s start with the question of whether an agent can display more or less rationality in selecting ends, even final ends. The general conception of practical rationality introduced earlier gives us parameters for approaching this question. We need to determine whether cognitive capacities can be exercised better or worse in the selection of ends. The reader should not expect an exhaustive treatment of this issue here, but a case will be made for the claim that an agent can exercise cognitive capacities better or worse in selecting ends, even final ends, and thus display more or less rationality on account of her ends. I start with brief remarks on the very nature of an end before giving examples of how cognitive capacities may come into play in the selection of final ends. Subsequently, I argue that, given an adequate understanding of what it is for an end to be a final end, an agent may display rationality in selecting final ends. This section is brought to a close with a discussion of the question whether the ends themselves can be more or less rational. This takes us to the issue between neo-Humeans and antiHumeans about practical rationality, which is the topic of Section 13.3.

13.2.1. An Agent’s Ends An agent’s ends are whatever he is set to pursue: whatever is on his agenda. Presumably, this is a matter of the agent having some kind of mental attitude towards the items correctly listed as his ends. The attitude in question is often identified as desire, when “desire” names the broadest category of intentional states of an individual’s motivational system: states that are assumed to be fundamentally different in kind from beliefs, when “belief” names the broadest category of intentional states of an individual’s doxastic system. I accept as a working hypothesis that intentional states can be thus divided into two non-overlapping categories depending on whether the states are part of the motivational or the doxastic system,¹⁰ although I prefer to avoid the ordinary connotations of the terms “belief” and “desire” by adopting instead the jargon “conative attitude” and “doxastic attitude” to name these very general categories of intentional states.¹¹ I also accept that the kind of attitude that fixes an agent’s ends is to be found within the ¹⁰ There is the further question about the nature of the emotions, which I will ignore here. ¹¹ This background assumption has to be ultimately defended in light of its fruitfulness in providing adequate models of mental and agential structures.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

286

Sigrún Svavarsdóttir

category of conative attitudes. However, the ends of an agent should not be identified with the intentional objects of his conative attitudes in general. We all know from experience that it is one thing to want some ice cream and quite another to have the goal of getting some ice cream, one thing to desire to go on vacation and quite another to have the goal of going on vacation, one thing to wish to become a musician and quite another to have it as an end. A man with sugar intolerance may have adopted the end of avoiding ice cream, yet continue to crave it. A woman of meager means may be reconciled to the fact that she cannot afford overseas vacations and she may be too prudent to adopt futile ends, yet she may very much covet overseas travels and spin great tales of visits to foreign shores. A person of many interests may have the sense of not adopting the end of pursuing each and every vocation that appeals to him. We are creatures of voracious, and often conflicting, appetites but limited resources. Fortunately, we have the sense and ability to settle on a relatively narrow range of attainable goals, much more narrow than the range of things that appeal to us. One of the remarkable features of human agency is that, instead of pursuing willy-nilly any object of desire, the typical human is capable of establishing more limited and, often, more stable goals. We need the theoretical resources to highlight this feature when modeling the kind of agency that is characteristic of humans. This cannot be done within a framework that identifies an agent’s ends with objects of his desires or, in my preferred terminology, with the intentional objects of his conative attitudes in general. For this purpose, we need a framework that distinguishes between various types of conative attitudes and identifies ends with the objects of a special type of conative attitude. Let’s call the relevant type of attitude, “the E-attitude.” It will be hard to state precisely what distinguishes the E-attitude from other types of conative attitude, but the difference at least shows up in the kinds of deliberative dispositions that such attitudes ground. There is a difference in the deliberative dispositions of the person who merely wants, say, to go overseas and the person who has set herself the goal of doing so. An agent does not have that as a goal unless she has a relatively strong disposition to deliberate regarding how to go overseas: where exactly to go, when to time one or more overseas trip, how to get to the chosen place, etc. Moreover, these dispositions—let’s call them, “executive deliberative dispositions”—have to be such that they are activated in circumstances that are quite similar to the circumstances in which the agent actually believes herself to be. (This condition would not be met by executive dispositions that are only activated in circumstances in which the agent is at least a hundredfold richer than she actually believes herself to be and is under no illusion about her finances.) By contrast, an agent can

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

The Rationality of Ends

287

very much want to go overseas without being disposed, in circumstances similar to the ones in which she actually believes herself to be, to deliberate about where, when, and how to go overseas. It may be an idle wish that more strongly disposes her to daydreams about foreign shores than to deliberation about how and when to get to foreign shores. This is not to deny that a mere wish grounds some deliberative dispositions. Undoubtedly, a mere wish to go on vacation disposes an agent to notice an opportunity to take a vacation (e.g., she can add a day to a work trip) and to deliberate about whether to take advantage of that opportunity. If she decides so, it has become her end and she will be disposed to deliberate about how to execute that end. This will have to suffice as an initial defense of the thesis that the ends of an agent are not to be identified with the intentional objects of his conative attitudes in general but, rather, with the intentional objects of his conative attitudes of a more narrowly delineated type.¹² Now, notice that executive deliberative dispositions are cognitive dispositions: dispositions to engage in reasoning geared towards deciding how, when, and where to execute a goal. It should not be controversial that cognitive capacities come into play, when deliberatively deciding how to execute a goal, not only in selecting the means and timing the execution of the goal but, also, in refining the goal and setting sub-goals. But do they come into play in the selection of final ends?

13.2.2. The Exercise of Cognitive Capacities in the Selection of Final Ends Consider an agent, interested in seeing the morning star, who checks the weather forecast before settling on the goal of watching the morning star tomorrow. This may seem to be an example of an agent who exercises cognitive capacities to time the execution of an end. This would be true, were a goal but the object of desire, but I have argued against that view. Given the view of ends accepted here, it is not an obvious truth that this is an example of an agent who has the longstanding goal of seeing the morning star: i.e., an agent who has an E-attitude rather than a mere desire towards seeing the morning star. Indeed, the example could be developed such as to be of an agent who has always been interested in seeing the morning star but has never set himself the goal of doing so. It is just one of the many things that he fancies. Now, he has become tired of fretting away his time aimlessly and decides to make something of tomorrow. It occurs to him that one

¹² An E-attitude is somewhat like a Bratmanian intention, but it is not clear that this is one and the same kind of conative attitude. See Bratman (1987).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

288

Sigrún Svavarsdóttir

possibility would be to take the day early and see the morning star, which appeals to him, but he has the sense of checking the weather forecast before settling on this as one of his goals. This is a case of an agent’s interests influencing what options he considers when establishing a goal rather than of an agent timing the execution of a prior goal. It is a case of an agent who exercises cognitive capacities to assess whether it makes sense to adopt an option that interests him as a goal. Another slightly different example of this phenomenon concerns an agent who struggles with an overabundance of interests but tries to find goals that allow her to stay focused and pursue at least some of her interests. She undertakes an inquiry about the various things that interest her to help sort out her priorities, and she engages in strategic thinking to find goals the pursuit of which answer to a number of her interests. Some of the goals she selects are already objects of desire for her, but others are not. For example, having no prior interest in photography, she nevertheless adopts the goal of making photography her hobby because it is a good match with her various interests: such as to engage visually with her environment, to channel her passion for gadgets and gizmos, to get a total mental break from her work as a writer, and to develop new interests to share with her partner. Her interests in the various things mentioned influence her selection of the goal of making photography her hobby without these interests manifesting prior goals to which the new goal is subservient, and she relies on diagnostic and problemsolving skills as well as various other cognitive abilities in selecting her goal. It may be questioned whether these are examples of cognitive capacities coming into play in the selection of final ends. The first example was developed so as to introduce a background goal. If our hero sets seeing the morning star on his agenda for tomorrow morning, that goal will serve his extant goal to stop fretting away his time aimlessly. Similarly, in the second example, the agent is easily seen as driven by the goals of finding a way of staying focused in spite of diverse interests and of accomplishing at least some of the many things she fancies. It may, thus, seem that both examples are of an agent who exercises cognitive capacities to find means to achieve an extant goal, though the means in these cases is to adopt another goal: a subgoal the pursuit of which serves a prior goal. What is true is that in both examples the agent adopts a goal that she believes to be conducive to attaining a prior goal. However, this does not make the adopted goal (more precisely, the possession of the goal) on a par with mere means to attain the prior goal. We need to distinguish between the following two relations between an agent’s goals: 1. A’s goal g₁ serves A’s goal g₂ (when g₁ 6¼ g₂) iff A’s having or pursuing g₁ is conducive to A’s achieving g₂.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

The Rationality of Ends

289

2. A’s goal g₁ is subservient to A’s goal g₂ (when g₁ 6¼ g₂) iff A’s having g₁ is counterfactually dependent on A’s having or aspiring to have g₂ and the belief that g₁ serves g₂. A goal g₁ is an ultimate goal or final end of A’s iff it is not subservient to any other of A’s goals, but it may still serve another of A’s goals and A may have originally adopted it because she saw that it would serve that goal. The etiology of the goal does not determine whether it is a subservient or an ultimate goal. That is determined by its persistence conditions. Thus, the first example could be developed such that the agent’s goal of seeing the morning star ends up being an ultimate goal of the agent, even if it was adopted, at least partly, because the agent saw that goal as serving a prior goal. We need only tell the story such that his longstanding desire to see the morning star suffices for sustaining the goal of seeing it tomorrow, even if that goal was set in working towards the goal of spending tomorrow purposefully. That is, it would persist even if, perchance, the agent lost his preoccupation with spending tomorrow purposefully. The second example can be developed in a similar way to be an example of an ultimate goal though adopted because the agent saw that it would serve the goal of staying focused and saw that it would connect nicely with various interests of hers. Thus, these two examples serve to demonstrate that cognitive capacities may come into play in the selection of final ends.

13.2.3. Rationality in the Selection of Final Ends As noted in Section 13.2.2, that a goal is a final rather than a subservient end does not depend on the etiology of the goal but rather on its persistence condition. It depends on whether the goal, once set, is counterfactually dependent on the agent having (or aspiring to have) some other end and a belief to the effect that the goal serves that other end. By contrast, given the general conception of rationality introduced in Section 13.1, the degree of rationality that an agent has displayed in selecting his ends depends on their etiology. It has to do with how well he has exercised cognitive capacities in selecting his ends. Now, we have seen a couple of credible examples of how cognitive capacities may come into play in the selection of final ends and, presumably, the agent can exercise cognitive capacities better or worse in that context. So, presumably, an agent can be evaluated as more or less rational on account of how he selects his final ends. Of course, it is possible that subservient goals are more likely than final goals to be set in ways that bear on the rationality of the agent, but that is far from obvious. Certainly, there is no a priori reason to expect that goals that meet the persistence

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

290

Sigrún Svavarsdóttir

condition of a final end do not have an etiology that involves cognitive activity in such a way as bears on the agent’s rationality. The examples in Section 13.2.2 may raise suspicion. After all, I had to argue that these were not cases of selecting means or sub-goals. It may seem that my case better rests on more clear-cut examples of agents exercising cognitive capacities in the selection of final ends: examples in which the selection of a goal does not serve any background goal or meet any prior interest that the agent may have. However, this demand is unreasonable because it is only in the context of purposive or interested cognitive activity that an agent exercises cognitive capacities better or worse in his practical endeavors in such a way as to demonstrate more or less rationality. The mere exercise of cognitive capacities—however exquisite the reasoning and representation is—does not enhance an agent’s rationality. An agent who compulsively and pointlessly teases out all the implications of his beliefs is not displaying rationality in so doing, however masterful his reasoning is.¹³ Indeed, he is displaying irrationality in case he is going off on a tangent rather than keeping the thread in a project that he has undertaken. It does not reflect well on an agent’s rationality to explore at length issues that— deemed against the background of his beliefs, interests, and goals—have no bearing on what he is doing. These observations accord well with the general conception of practical rationality assumed here. Even if the representation and reasoning is flawless, an agent is not exercising cognitive capacities well in his practical endeavor if he fails to tailor his cognitive activity to something he is trying to accomplish or takes interest in. This does not mean that his efforts need to be successful in order for him to be practically rational. Things can go wrong in many ways without it being in any way due to a poor exercise of cognitive capacities in his practical endeavors. He may meet with failure in his practical endeavors due to a false belief, formed on the basis of a misleading input from his environment, that he is not in a position to correct however well he exercises cognitive capacities. It is also possible that his efforts are thwarted by unforeseeable events. Or there may be an overload of information that he has no way of sorting through given the limits on his finite cognitive powers. Practical rationality is not about being free from error or being successful in one’s endeavors. Instead, it is about not making mistakes that could be avoided by exercising cognitive capacities well in one’s practical endeavors. Practical rationality is an excellence in the exercise of cognitive capacities in one’s practical endeavors. That excellence does not merely consist in

¹³ See Harman (1999: 18–19). For a discussion of the main point made here, see Svavarsdóttir (2006: 64–5).

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

The Rationality of Ends

291

high-quality representation and reasoning but also in tailoring that cognitive activity intelligently to what one is up to and adopting the ends or means that, given one’s background attitudes, make good sense in light of what the cognitive activity has brought into view. This does not require that the selected ends be subservient to some ulterior ends set by one’s background attitudes, but it does require that adopting these ends make sense to an agent with these attitudes. The new ends may make sense to the agent because they seem to serve or support his prior ends, or because they seem very interesting or valuable, or because they seem required given his circumstances, or because pursuing them seems to involve important activities, or because such pursuits seem to bring rewarding experiences, and so on. Their seeming so is a function of the agent’s various background attitudes, both doxastic and conative. In the absence of attitudes that establish such an interested perspective within which the cognitive activity takes place, there is no place for evaluations of the subject as more or less rational. Thus, a computerized machine that gathers and analyzes data from its physical environment does not display more or less rationality in doing so, whereas a fieldworker who does the same is subject to such evaluations.¹⁴ The crucial difference between them is that, unlike the machine, the fieldworker’s data collection and analysis is embedded in a web of propositional attitudes. Her activity is guided by background beliefs that partly determine to what she pays attention at the various steps in her research. Some of these beliefs concern what serves the aims of the research project. There are several possibilities with respect to her relation to these aims. She may be the principal investigator who has set these research goals; she may be an enthusiastic associate who has, full of curiosity, taken on these goals; or she may be a more or less conscientious hired hand whose only interest in these goals is to understand them well enough to do her job adequately. In other words, she has E-attitudes to the research goals or to goals served by working towards the research goals. However, not all of the conative attitudes that frame the fieldworker’s research need to amount to such E-attitudes. For example, at a crucial juncture the research may take one direction rather than another because of curiosity that is suddenly piqued in the course of inquiry, resulting in a revision of the original goals of research. That revision need not serve some prior end in order to make sense to the agent. It may make full sense in light of the curiosity aroused. That curiosity may have its roots in mental ¹⁴ The example of the machine and the fieldworker also appeared in Svavarsdóttir (2006). However, there I used it for a slightly different purpose: namely, to argue that even assessments of theoretical rationality need to take into account goals, albeit exclusively cognitive goals, of the agent.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

292

Sigrún Svavarsdóttir

attitudes that make the agent in general interested in certain kinds of issues and make her disposed to adopt the goal of investigating them, yet do not amount to a prior commitment to any such goal. However, the revision makes sense from the point of view of the fieldworker, thanks to these mental attitudes, and that is crucial for making her rational in revising her research goals, when her curiosity is piqued in a new direction in the course of inquiry. How rational she is in adopting this new goal depends on other factors as well. Her practical rationality is a matter of how well she exercises cognitive capacities in conducting her practical endeavors, which in this case consist in a research project. If the agent thoughtlessly heads in whichever direction her curiosity leads, she does not display much rationality in revising her research goals. If she does not bother to exercise cognitive capacities to examine the ramifications of changing the direction of her research, or if she exercises them poorly for that purpose, she does not display much rationality whatever sense the change in research goals makes to her, due to how her curiosity has been piqued in the course of inquiry. By contrast, if after examining well the ramifications of changing the course of inquiry, she indeed does so because (at least partly due to her newfound curiosity) it makes more sense to her than to stick to her prior agenda, then it reflects well on her rationality to modify her research goals. Indeed, it would reflect poorly on her rationality were she to stick to her prior agenda when a change in the direction of research makes more sense to her after exercising cognitive capacities as well as is feasible in thinking through the matter.

13.2.4. The Rationality of Ends We have seen that, given the cognitive excellence conception of practical rationality, an agent can display more or less rationality in selecting ends, but what about his ends? Can they be more or less rational, given this general conception of rationality? It certainly makes sense to evaluate the agent’s actual ends as more or less rationally adopted. However, could they be rationally held even if not rationally selected? Surely, there is room for that on the cognitive excellence conception of practical rationality. An agent may not have exercised cognitive capacities much or exercised them poorly when originally adopting an end but, later, rethought the case and reaffirmed his commitment to the end in a way that manifests an excellence in the exercise of cognitive capacities in his practical endeavors. In that case, it is reasonable, given the cognitive excellence conception, to consider him as rationally having the end, even if it was not originally rationally selected. There is also the possibility that an agent did not display rationality when adopting an end and has never revisited the case for having this end, yet were he to exercise cognitive capacities well in doing so, he would reaffirm this

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

The Rationality of Ends

293

end. In that case, does he rationally have the end, given the cognitive excellence conception of rationality? This question is of crucial importance because if the answer is in the affirmative, then the agent’s rationality in having an end can turn on the end’s persistence condition rather than its etiology, which would undercut my argument at the beginning of Section 13.2.3. Fortunately, the cognitive excellence conception calls for a negative answer. On this conception, agents are practically rational to the degree that they are disposed to display excellence in the exercise of cognitive capacities in their practical endeavors. This implies that an agent could be highly rational even if she had just sprung fully grown from the head of Zeus without having had any opportunity to display her cognitive excellence in her practical endeavors. However, her rationality is not yet manifested in any of her E-attitudes, since an agent’s attitude or conduct manifests practical rationality only insofar as it is the upshot of cognitive excellence displayed in the practical arena. If Athena had E-attitudes from the beginning, they would have been formed in ways that reflect neither well nor poorly on her rationality, even if it is true that were she to exercise cognitive capacities well in her practical endeavors, as she is disposed to do, she would retain some of these E-attitudes. For an agent to have an end rationally, it does not suffice that her end satisfies the counterfactual condition of being an end that she would retain, were she to exercise cognitive capacities well in her practical endeavors. If the end satisfies this condition because of some cosmic coincidence or a good design on the part of Fortuna, the agent does not rationally have the end. As noted in Section 13.1.2, rationality displayed in having ends is a cognitive accomplishment in the practical domain on the part of the agent rather than a lucky coincidence. Is there room for asking whether the end itself is rational? Well, what exactly is the entity whose rationality is so queried? Is it an actual end of a particular agent? If so, the question is whether his having an E-attitude towards the item listed as his end is rational, and we have already seen that there is room for that question. Another possibility is that the question is about a type of end, individuated with reference to the proposition expressed by the that-clause in “S has an E-attitude towards the proposition that p” when that attitude-ascription is given a de dicto reading. Let me refer to such an end-type as “a substantive characterized end.” To illustrate: the substantively characterized end of becoming a movie star is the type of end that anyone has of whom the attitude-ascription “S has an E-attitude towards the (de se) proposition that I become a movie star” is true on a de dicto reading of that E-attitude ascription. Notice that to be of this type, an end need not be the actual end of anyone—it may merely be a possible end—and two agents can have the same substantively characterized end, since they may both have ends of this type.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

294

Sigrún Svavarsdóttir

Given the cognitive excellence conception of practical rationality, it is reasonable to talk about a substantively characterized end as rational for a particular agent to adopt or reaffirm if this is a type of end that he could, given his current conditions, rationally select.¹⁵ But can we say of a substantively characterized end—i.e., of that type of end and, hence, of any instance of that type of end—that it itself is a more or less rational end? Put differently, are there any substantively characterized ends that it is rational for any agent to have? Shifting attention to the irrationality of ends, what about the substantively characterized ends whose rationality was queried in the opening paragraph of this chapter: the end of proving that π is an algebraic number, the end of counting blades of grass in park squares, and the end of keeping Tuesdays free of pain even at the cost of greater pain on other days of the week? On the cognitive excellence conception of practical rationality, if there are substantively characterized ends that are irrational for any agent to have, it is because excellence in the exercise of cognitive capacities in one’s practical endeavors invariably yields a configuration of attitudes which precludes that these substantively characterized ends make good sense to the agent, so that having these ends is a sure sign of irrationality. Whether this is true hangs on how the cognitive excellence conception is developed into a full-blown theory of practical rationality. It depends on the specification of the standards for evaluating how well an agent exercises his cognitive capacities in his practical endeavors: not least the standards by which to assess how intelligently he has adjusted his attitudes, and conduct, to the upshot of that cognitive activity. This is where the fundamental disagreement between Humeans and antiHumeans about practical rationality lies. The anti-Humeans need not (and should not) deny that an agent’s rationality is assessed with reference to where he stands cognitively and motivationally, but they are convinced that the correct specification of the standards of rationality will have the upshot that if an agent exercised cognitive capacities sufficiently well in his practical endeavors, certain substantively characterized ends would be ruled out for him as making poor sense (and perhaps others will be ruled in as making good sense), no matter what his initial starting point had been, and this is not due to some contingent facts about the human motivational system. By contrast, the neo-Humeans are convinced that the correct specification of the relevant evaluative standards will have the upshot that if there are substantively characterized ends that do not make good sense to all mature

¹⁵ Notice that if he had such a substantively characterized end, this would not be the same as saying that he rationally has that end or has rationally selected it.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

The Rationality of Ends

295

human agents who have exercised cognitive capacities well enough to count as fully rational, this is partly due to contingent facts about the human motivational system rather than solely the nature or demands of practical rationality. The issue between Humeans and anti-Humeans about practical rationality will not be settled here, but I will bring this chapter to a close by taking a closer look at how the conclusions of this section sit with Humeanism about practical rationality. 13 . 3. A N T I - H UM E A N C O N C L U S I O N S ? As noted at the beginning of this chapter, Humeanism about practical rationality is traditionally associated with the claim that an agent’s final ends are not subject to evaluations in terms of rationality. Hume himself held that strictly speaking an agent is not evaluable as more or less rational on account of her passions—or, in my terminology, on account of her conative attitudes. That is certainly an acceptable reformulation of some famous (or infamous) claims from Book II of the Treatise. Relying on the conception of practical rationality as an excellence in exercising cognitive capacities in one’s practical endeavors, I have contested this Humean thesis by making the case that an agent can exercise cognitive capacities better or worse in forming E-attitudes so as to render the agent more or less rational on account of her ends. Now, Hume himself would not have had any quarrel with the general conception of practical rationality that I have introduced, but he would have developed that conception very differently from the direction in which I have taken it. Hume accepts that human cognitive capacities are bound to a special faculty of mind, Reason, whose exercise is manifested in cognitive activities geared towards the acquisition of knowledge and understanding. He thinks that this faculty is handy in conducting our practical affairs just to the extent that it brings knowledge of what there is in the world that may pique our interest and of how it is best to accomplish what already interests us. On his view, we reliably take the most efficient means to satisfy our desires if guided by Reason in our practical affairs. This is because Reason reliably figures out what will give us pleasure or pain, as well as how to attain or avoid such things, and (according to Hume’s causal hedonistic theory of motivation) we will effectively pursue whatever we expect to maximize the balance of pleasure over pain, which is also what maximizes the satisfaction of desire since human motivation closely tracks expectations of pleasure and pain. However, Hume holds that, strictly speaking, it is only our expectations regarding pleasure and pain and the means to attain or avoid them that is in accordance with or contrary to Reason, i.e., rational or irrational. Strictly speaking, the passions

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

296

Sigrún Svavarsdóttir

and actions influenced by these expectations are not evaluable as more or less rational. Seen from a slightly different angle, his view is that we are subject to evaluation in terms of rationality solely insofar as we form beliefs. Practical rationality is but theoretical rationality displayed in the context of conducting one’s practical affairs. This view of rationality in the practical domain stems from a theory of mind that is overly simplistic. In particular, Hume has a simplistic view of mental content (the theory of ideas), of mental inference (the associationist theory), and of mental attitude (e.g., the attitude of believing he understands in terms of the liveliness of the idea that gives the content of belief). He overlooks that capacities to use and manipulate symbolic structures infuse the entire human mind, including the motivational system: just as it is one thing to believe that the evening star is bright and quite another to believe that the morning star is bright, it is one thing to want to see the evening star and quite another to want to see the morning star (given a de dicto reading of these attitude-ascriptions). It is under various modes of presentation that a human agent takes doxastic and conative attitudes towards actual or possible states of affairs. In this way, capacities to use symbolic structures to represent actual and possible states of affairs infuse not only the doxastic states but also the conative states ascribed to the typical human agent.¹⁶ This observation suggests that, pace Hume, practical rationality is not merely displayed in how the agent forms the beliefs that inform her preferences and actions. To take a simple example, consider a woman who exercises cognitive capacities well or badly when scheduling her next day so as to accomplish the various things that she has set herself to complete by the evening. If she has set herself the goal of seeing the morning star, she would fail to exercise cognitive capacities well in her practical endeavors, if she did not make it part of her plan to take the day early, given that she believes that the morning star is visible only before sunrise. In this case, her irrationality is not manifested in how she forms beliefs that inform her conduct but rather in how her beliefs inform her plans. In Section 13.2, we saw other ways in which cognitive capacities can come into play in the selection of ends, even final ends, in ways that bear on the agent’s practical rationality, conceived as excellence in the exercise of cognitive capacities in one’s practical endeavors. More generally, we should expect the manipulation of symbolic structures in reasoning to have a pervasive role not only in fixing belief but also in fixing means and ends, given that agents have conative attitudes towards objects conceptualized in one way rather than another.

¹⁶ This is why I chose the jargon “doxastic state” rather than “cognitive state” to denote the broadest category of the intentional states of an individual’s doxastic system.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

The Rationality of Ends

297

As noted in Section 13.1, a theory of practical rationality should not be confounded with a theory of cognition and its role in the practical domain. It is a normative theory that lays out standards for evaluating how well an agent exercises cognitive capacities in her practical endeavors rather than gives a descriptive cum explanatory account of cognition in the practical domain. Nevertheless, these standards must suit the kind of cognition in practical pursuits that is characteristic of agents like us in order to be of any relevance in assessing our practical rationality. A theory of cognition and its role in the practical domain is, therefore, a reasonable starting point for a theory of practical rationality. We should, however, be aware of the danger that an inadequate theory of the role of cognition in motivation may yield a faulty theory of practical rationality. That has, indeed, happened in the case of Hume: an inadequate theory of cognition led him to the mistaken thesis that agents are not subject to evaluations of rationality on account of their ends. That is the upshot of Section 13.2. Nevertheless, that conclusion does not completely undermine Humeanism about practical rationality. It would, certainly, be odd to place within the anti-Humean rather than the Humean tradition, a view that accepts that (full) practical rationality does not guarantee the absence of any substantively characterized ends, even if it embraced the conclusion from Section 13.2 regarding the possibility of rationality in the selection of ends. This view preserves the spirit of Hume’s provocative remark that “’Tis not contrary to reason to prefer the destruction of the whole world to the scratching of my finger. ’Tis not contrary to reason for me to chuse my total ruin, to prevent the least uneasiness of an Indian or person wholly unknown to me”¹⁷ and should, therefore, be treated as a neo-Humean theory of practical rationality. After all, Hume’s primary target is the rationalist view that Reason alone provides guidance along the path of virtue.¹⁸ It is most reasonable to treat a denial of the thesis that an agent can be more or less rational in selecting ends as rooted in an obsolete theory of mind that can be discarded while retaining the Humean conception of practical rationality as not normatively ruling out any substantively characterized end, however silly, imprudent, or evil it may be. The deep disagreement between neo-Humeans and anti-Humeans about practical rationality centers on the question of whether being fully practically rational guarantees the absence of some substantively characterized ends.¹⁹ More work is ¹⁷ Hume (1739 [1978]: 416). ¹⁸ Cf. the opening of the famous section iii of Book II of the Treatise. Hume (1739 [1978]: 413). ¹⁹ There is a caveat in order here. Humeans should concede that irrationality rules out any type of end that is substantively characterized by a proposition that is easily knowable

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

298

Sigrún Svavarsdóttir

needed on the standards for the cognitive excellence that has been identified as practical rationality before settling this disagreement, not least the standards for assessing how intelligently the agent has adjusted her attitudes and conduct to the upshot of her cognitive activity.²⁰

References Bratman, Michael. 1987. Intention, Plans, and Practical Reason. Cambridge, MA: Harvard University Press. Broome, John. 2013. Rationality Through Reasoning. Oxford: Wiley-Blackwell. Harman, Gilbert. 1999. Reasoning, Meaning and Mind. Oxford: Oxford University Press. Hume, David. 1739 [1978]. A Treatise of Human Nature, ed. P.H. Nidditch. New York: Oxford University Press. Parfit, Derek. 1984. Reasons and Persons. Oxford: Oxford University Press. Parfit, Derek. 2011. On What Matters, Vol. I. Oxford: Oxford University Press.

a priori to be necessarily false. A case in point would be the end of proving that 3 + 3 = 3  3. Surely, an agent with the mental capacities needed to have an attitude with the intentional object that I prove that 3 + 3 = 3  3 (on a de dicto reading of that attitude ascription) has the elementary arithmetic skills for figuring out rather easily that the equation is necessarily false and the mental capacity to understand that the equation is, therefore, impossible to prove. Moreover, on a plausible account of E-attitudes, these are attitudes taken only towards something that the agent has not ruled out as impossible to bring about more or less directly through his agency. Thus, if the agent has an E-attitude with the intentional object that I prove that 3 + 3 = 3  3 (on a de dicto reading of that attitude ascription), he has either failed to exercise his arithmetic skills as well as feasible in the context of selecting this end or he has failed to bring the conclusion of his arithmetic reasoning to bear on the possibility of attaining the end he is adopting. In either case, he has not exercised cognitive capacities well in his practical endeavor and has, thus, displayed irrationality in selecting the end. This is, indeed, in line with Hume’s central idea that cognition serves as an instrument for navigating the world, while the passions set the course. It is because of the nature of the E-attitudes that anyone who has exercised cognitive capacities well in his practical endeavors does not have ends that are substantively characterized by a proposition that is easy, for a person in his epistemic position, to know to be necessarily false. Moreover, this concession will in no way undermine Hume’s anti-rationalist thesis that Reason alone does not provide guidance along the path of virtue unless Samuel Clarke, Hume’s main rationalist foe, is right that ignoring what Reason has discerned to be fit or obligatory amounts to “the greatest absurdity imaginable” of attempting “to make things be what they are not, and cannot be” (Raphael 1969: 201). ²⁰ In the autumn of 2015, a precursor of this chapter was presented at the Chapel Hill Metaethics Workshop and at Brandeis University. During the subsequent year, drafts of the chapter were presented at Northwestern University and at Oslo University as well as at the Chambers Conference on Practical Reason and Metaethics, the University of Nebraska, and at the Philosophy Workshop at the College of Holy Cross. I thank all six audiences for helpful questions and comments. I also thank two anonymous reviewers of this volume for their feedback.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

The Rationality of Ends

299

Raphael, D.D. 1969. British Moralists 1650–1800. Oxford: Clarendon Press. Rawls, John. 1971. A Theory of Justice. Cambridge, MA: Harvard University Press. Raz, Joseph. 1999. Engaging Reason: On the Theory of Value and Action. Oxford: Oxford University Press. Scanlon, T.M. 2007. “Structural Irrationality,” in G. Brennan, R. Goodin, F. Jackson, and M. Smith (eds.), Common Minds: Themes from the Philosophy of Philip Pettit. Oxford: Oxford University Press. Svavarsdóttir, Sigrún. 2006. “Evaluations of Rationality,” in T. Horgan and M. Timmons (eds.), Metaethics After Moore. Oxford: Oxford University Press. Svavarsdóttir, Sigrún. 2008. “The Virtue of Practical Rationality,” Philosophy and Phenomenological Research 77: 1–33. Svavarsdóttir, Sigrún. 2015. “Coherence of Attitudes, Integration of the Self, and Personal Integrity,” Oxford Studies in Agency and Responsibility 3: 62–84.

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Index aboutness, semantics and 59 access to aesthetic facts 87 access to moral facts 87, 91 accurate description, evaluation for 57 acts, badness and painfulness in relation 7 aesthetics access to aesthetic facts 87 acquaintance and 71, 75, 81, 87 aesthetic supervenience (AS) 3 deference and 71, 72, 78, 83 epistemology and 76, 91 values 71, 79, 231, 234 agent’s ends, rationality of 285, 292 akrasia, epistemic 199 alethic modal thought 38 all-things-considered ought 237 analytic naturalism 140 Anscombe, G. E. M. 145 anti-realism normative supervenience and 6 theories of modality 36 applied ethics, moral philosophy and 92 Aristotelian eudaemonism 145 Aristotle 145 authoritative normativity conceptual analysis 259 conditional fallacy 267 constitutivist resources 260 introduction to 253 non-arbitrary selection 263 practical normativity and 254 practical ought 255, 259, 262 ‘schmagency’ challenge 271 authoritative values 244 badness and painfulness in relation 7 base property 5 belief epistemology and 265 moral judgments as 128 normative supervenience in 1 Blackburn, Simon 30, 43, 110 Bonaparte see Napoleon Boyd, R. N. 134, 135

categorical imperatives 236, 273 categorical ought 236 Chang, Ruth 168 classical utilitarianism 41, 42, 43, 44 cognitivism authoritatively normative concepts and 273 cognitive capacities, exercise of 287, 293, 295 cognitive dispositions 287 cognitive penetration of perception 89 cognitive states 193, 199 coherence and 283 conceivability argument and 19 moral judgments and 99, 117, 128 non-descriptive relativism and 52, 60, 63 non-naturalism and 26, 136 practical rationality and 279, 280, 284, 292, 297 psychology and 61, 62, 63 semantics and 52 truth and 60 see also non-cognitivism coherence caring about 187 cognitivism and 283 inter-level 192 picture of 194 practical rationality and 282 scope of 184 see also incoherence coherence requirements enkrasia 198 illustrative cases 190 instrumental requirement 190 introduction to 184 naturalism and 201 normativity and 201 ontology and 187, 203 transitivity requirement 191 view of 188 coherentism and moral epistemology 107 composition of language content 66 conceivability argument for NS 17, 18 conditional fallacy 267

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

302

Index

conservativism and moral epistemology 107 consistency argument for NS 10 content of language, non-descriptive relativism and 52 Copp, D. 149, 158, 232, 233 correctness correct belief in propositions 265 standards of correctness, formal normativity and 208 cruel(ty) and non-descriptive relativism 55 Dancy, Jonathan 11 Davidson, Donald 197, 257 deflationary pluralism 258, 267, 271 deliberative pluralism 231 deontology, Kantian 170 dependence and normative supervenience 7 description non-descriptive relativism and 52 semantics and 54 descriptive accuracy, evaluation for 57 desire see rational final desire disagreement moral epistemology and 116 non-descriptive relativism and 64 semantics and 64 Divers, John 35, 36, 39, 43, 44 dominance moral uncertainty with 164 moral uncertainty without 165 Dorsey, Dale 239 Dreier, Jamie 17, 26, 27, 31, 37, 38, 65 embedded evaluation 9 Emerson, Ralph Waldo 248 enkrasia 198 Enoch, David 139, 262, 271 Epictetus 248 epistemology aesthetic 76, 91 belief and 265 deference and 71, 73, 79, 81 ‘epistemic akrasia’ 199 epistemic constraint on rational final desire 131, 135, 139 epistemic pessimism 74, 79 epistemic supervenience (ES) 1, 3, 4, 12 incoherence and 184, 185, 188, 199 irreducible thickness and 8

moral see moral epistemology moral uncertainty and 162 non-epistemic ‘other’ 1 non-epistemic pessimism 75, 79, 81, 84 normative 89 ‘ought’ and 230 rationality and 284 reason and 77 error theory morality and 142 normativity and 253, 258, 274 etiquette as formally normative 207 eudaemonism 145 evaluative standards, pluralizsed 231 evolutionary explanations of moral judgments 100 explanation objection to NS 20 expressivism non-descriptive relativism and 48, 50, 53, 54, 63, 68 normativity and 253, 258 relational 43 semantics and 48, 50 fallacy, conditional 267 fetishizsed norms 226 final ends, selection of 287, 289 Fine, Kit 16 Finlay, Steven 274 Foot, Philippa 217, 230, 257 formal normativity formal obligations 215, 216, 222, 224 generic reasons 212 internal point of view 218 introduction to 207 norm-fetishism 226 respect-based reasons 222, 225 standards of correctness and 208 subscripted obligations 210 substantive normativity and 214 formal obligations 215, 216, 222, 224 foundationalism and moral epistemology 108 Frege, G. 48, 49, 50, 63, 65 Frege–Geach concerns 48, 49, 50, 63, 65, 66 Geach, P. 44, 49, 50, 63, 65 generic reasons formal normativity and 212 values and 212, 214, 223, 226 Gettier, E. 133

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Index

303

Gibbard, Alan 57 Gonzáles-Varela, J. E. 36 grounding argument for NS 12

Lewisian realism, non-naturalism and 33 Lockhart, Ted 170

Hare, R. M. 273 Harsanyi, John 178 Hart, H. L. A. 220 heavyweight modal realism, nonnaturalism and 33 Hedden, Brian 167, 169, 170, 171, 177, 178, 181 hedonistic utilitarianism 29, 43 Hills, Alison 75, 84, 85, 86 Hooker, Brad 148 Hopkins, Robert 79, 80, 81 Horgan, Terence 273 Hubin, Donald 241 Hume, David 243, 257, 269, 279, 285, 294, 295, 296, 297 Hursthouse, R. 146, 147, 148, 149, 150, 151, 152, 153, 158

MacAskill, William 176 MacFarlane, J. 57 Mackie, John 274 McDowell, J. 147 McLeod, O. 233 McPherson, Tristram 5, 20, 21, 28 meta-semantics see semantics mind-dependence moral epistemology and 111, 113, 114 moral truth 111 mind-independent conceptions of moral truth 113 modal judgment 37 modal realism, non-naturalism and 32 modal status of learned truths 92 modal strength 2, 3 modal supervenience 2, 7 modal thought 38, 40 modal verb, ought-claims as 250 modalities, sense 91 monism see eudaemonism Moore, G. E. 135 moral acquaintance 81 moral deference 72, 83 moral epistemology access to moral facts 91 coherentism 107 conservativism 107 disagreement and 116 evolutionary explanations of moral judgments 100 field of 107 formal view of 109 foundationalism 108 generic debunking argument 101 introduction to 98 mind-dependence and 111, 113, 114 mind-dependent conceptions 111 moral judgment and 106, 108, 117 moral philosophy and 92 possibility of 98, 109, 111, 118 ‘pure’ 107 quasi-realism and 118 redundancy 103 sensitivity argument 104 substantive view of 109 utilitarianism and 114, 117

imperatives categorical 236, 273 quasi-realism and 34 incoherence, epistemology and 184, 185, 188, 199 instrumental coherence requirement 190 inter-level coherence 192 intrinsic values 152, 153, 154, 158 irreducible thickness (IT) 8 Jackson, Frank 19, 136, 137, 140, 141, 260 Johnston, Mark 137 Joyce, Richard 98, 99, 100, 101, 102, 106, 112, 113, 114, 117, 118, 241, 242 Kant, Immanuel 74, 145, 164, 170, 236 Kantian deontology 170 Kaplan, D. 49, 50, 52 Kripke, Saul 135, 253 Lack of Armchair Access to facts of aesthetic relevance 87, 91 Lack of Perceptual Moral Access 87 language content composition and validity of 66 non-descriptive relativism and 52 Lehrer, Keith 1 Lewis, David 33, 111, 275

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

304

Index

moral facts access to 87, 91 moral judgments in relation 117 moral judgments as beliefs 128 cognitivism and 99, 117, 128 evolutionary explanations of 100 formation of 106, 108 moral epistemology and 117 moral facts in relation 117 moral motivation in relation 123 non-cognitivism and 100 source of 99, 117 supervenience and 4, 11 moral motivation and moral judgments in relation 123 moral nihilism, denial of moral values 171, 176, 180 moral ontology see ontology moral philosophy applied ethics and 92 moral epistemology and 92 moral properties, direct access to 88 moral rationalism authoritatively normative concepts and 256 deflationary pluralism and 258 oughtness and 255, 259, 265 moral realism, arguments against 98, 100 moral semantics see semantics moral sensitivity 171 moral supervenience (MS) 3 moral truth mind-dependent conceptions 111 mind-independent conceptions 113 ontology and 107 moral uncertainty dominance, with 164 dominance, without 165 epistemology and 162 introduction to 161 obligation and 166 representation theorems 172 state of 162 utilitarianism and 164 value comparison and 167, 171, 175, 178 moral values 167, 231 morality, error theory and 142 motivation see moral motivation Napoleon 102, 109 natural and normative in relation 25

naturalism analytic 140 coherence requirements and 201 synthetic 134 neo-Aristotelian eudaemonism as normative ethics 145 utilitarianism and 145 nihilism see moral nihilism non-arbitrariness, thick and thin 271 non-cognitivism authoritatively normative concepts and 273 moral judgments and 100 non-naturalism and 26, 30, 31, 32, 42, 43 pure moral motivation and 128 non-descriptive relativism basic idea 49 cognition and 52 composition of language content 66 content of language and 52, 56 cruel(ty) and 55 description and 52 disagreement and 64 evaluation for descriptive accuracy 57 expressivism and 48, 50, 53, 54, 63, 68 Frege–Geach concerns 48, 49, 50, 63, 65, 66 introduction to 48 normative thought and 61 ontology and 6 psychology and 61 semantics and 57 speaker subjectivism and 49, 52, 54, 61 truth in values 60 validity of language content 66 world-moral-opposition-stance pairs 56, 57 wrong(ness) and 52, 56, 58, 63 non-epistemic ‘other’ 1 non-epistemic pessimism 75, 79, 81, 84 non-natural property 5 non-naturalism cognitivism and 26, 136 heavyweight modal realism 33 Lewisian realism 33 modal realism 32 non-cognitivism and 26 non-representationalism 33, 35 normative supervenience (NS) and 18 notion of 26 possible worlds realism 33 quasi- 25

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Index quasi-realism 26, 31, 32, 40 realism 26 SUPER 26, 27, 30, 31 supervenience and 27, 37 non-reductionism, normative supervenience (NS) and 18 non-representationalism non-naturalism and 33, 35 propositions and 33, 36 normative and natural in relation 25 normative epistemology 89 normative ethics, Neo-Aristotelian eudaemonism as 145 normative supervenience (NS) aesthetic supervenience and 3 base of 5 basic idea of 2 belief in 1 conceivability argument 17 consistency argument 10 dependence and 7 epistemic supervenience and 3 explanation objection 20 grounding argument 12 irreducible thickness and 8 moral supervenience and 3 non-naturalism and 18 non-reductionism and 18 radical holism and 8 reduction objection 19 supervenience of the normative 2 understanding of 6 normative thought, non-descriptive relativism and 61 normativity authoritative see authoritative normativity coherence requirements and 201 error theory and 253, 258, 274 etiquette as formally normative 207 expressivism and 253, 258 formal see formal normativity norm-fetishism 226 reason and 207 substantive 214 thick normative terms 12, 55, 73, 74 thin normative terms 48, 50, 53, 54, 56, 73, 74, 274, 275 NS see normative supervenience objective values 274 obligation formal obligations 215, 216, 222, 224

305

moral uncertainty and 166 semantics of 211, 250, 275 subscripted obligations 210 Ockham’s razor 118 ontology coherence requirements and 187, 203 Kantian deontology 170 moral truth and 107 non-descriptive relativism and 6 Orwell, George 82 otherness non-epistemic ‘other’ 1 quality of 1 ought, oughtness epistemology and 230 moral rationalism and 255, 259, 265 ought-claims as modal verb 250 ought-propositions 249 practical ought see practical ought semantics of 211, 250, 275 simpliciter see ought simpliciter ought simpliciter all-things-considered ought 237 categorical ought 236 denial of multiple senses of 249 introduction to 230 lack of utility 245 options for characterizsation other than psychological 244 ought of rational criticizsability 240 ought that settles what to do 236 ought we cannot question 241 precedents 230 psychological characterization 242 skepticism argument 234 painfulness and badness in relation 7 Parfit, Derek 137, 212, 214, 257 perception cognitive penetration of 89 Lack of Perceptual Moral Access 87 pessimism, epistemic 74, 79 Pettit, Philip 140 pluralism, plurality deflationary 258, 267, 271 deliberative 231 evaluative standards of 231 pluralized pronouns 219 reason and 231 PMM see pure moral motivation possible worlds realism, non-naturalism and 33

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

306

Index

practical ought authoritative normativity 255 conceptual analysis 259, 262 conditional fallacy 267 consequences of analysis 272 constitutivist resources 260 definition 256 non-arbitrary selection 263 preliminary account 265 ‘schmagency’ challenge 272 practical rationality cognitivism and 279, 280, 284 coherence and 282 responsiveness to reasons 280 precedents, ought simpliciter 230 pronouns, pluralized 219 propositions coherence and 184, 185, 196 contradictory 184 correct belief in 265 descriptive 164 moral uncertainty and 162, 164, 166, 167, 173, 177, 179 non-representationalist views 33, 36 normative 26 ought-propositions 249 propositional attitudes 291, 293 representationalist views 33 structured 50 psychology non-descriptive relativism and 61 ought simpliciter 242 semantics and 63 pure moral motivation (PMM) analytic naturalism 140 epistemic constraint 131 introduction to 123 non-cognitivism and 128 non-naturalism 136 pure moral inquiry with conceptual limitations 126 pure moral inquiry with full information 125 rationality of 124 synthetic naturalism 134 Putnam, Hilary 135 quasi-non-naturalism 25 quasi-realism imperatives and 34 modal realism and 32 moral epistemology and 118 non-naturalism and 26, 31, 32, 37, 40

radical holism (RH) 8 Railton, Peter 260 rational criticizsability ought of 240 rational final desire, epistemic constraint on 131, 135, 139 rationalism see moral rationalism rationality cognitivism and practical rationality 279, 280, 284, 292, 297 ends of see rationality of ends epistemology and 284 thin 180, 182 rationality of ends agent’s ends 285, 292 Humeanism 295 introduction to 278 practical rationality 279, 282 question of 285 selection of final ends 287, 289 realism non-naturalism 26 see also anti-realism reason acquaintance and 76 epistemology and 77 generic reasons 212, 214, 223, 226 normativity and 207 respect-based reasons 222, 225 responsiveness to 280 reduction objection to NS 19 relational expressivism 43 relativism see non-descriptive relativism representation theorems 172 representationalism propositions and 33 see also non-representationalism respect-based reasons 222, 225 RH see radical holism Ridge, Mike 17 Riedener, Stefan 171 Rosen, Gideon 1, 16 Ross, Jacob 171, 173, 176, 178 Russell, Bertrand 137 ‘schmagency’ challenge 272 semantics aboutness and 59 cognition and 52 description and 54 disagreement and 64 expressivism and 48, 50 meta-semantic distinctions 26 non-descriptive relativism and 57

OUP CORRECTED PROOF – FINAL, 25/4/2018, SPi

Index of ‘ought’ or ‘obligation’ 211, 250, 275 psychology and 63 rationality and 278 semantic rules 34, 39, 42 synthetic naturalism and 135 truth and 60 sense modalities 91 Sepielli, Andrew 171 Silverstein, Matthew 262 Singer, Peter 161, 180 skepticism and ought simpliciter 234 Smith, Michael 12, 13, 15 Sobel, D. 149, 158 Socrates 36 speaker subjectivism, non-descriptive relativism and 49, 52, 54, 61 St Francis 17 standards of correctness, formal normativity and 208 Street, S. 100, 109 Sturgeon, Nicholas 5 subjectivism see speaker subjectivism subscripted obligations 210 substantive normativity, formal normativity and 214 SUPER 26, 27, 30, 31 supervenience epistemic (ES) 1, 3, 4, 12 modal 2, 7 moral judgments and 4, 11 non-naturalism and 27, 37 normative see normative supervenience synthetic naturalism pure moral motivation (PMM) and 134 semantics and 135 theorems see representation theorems thick concepts 8 thick non-arbitrariness 271 thick normative terms 12, 55, 73, 74 thick properties 12, 13, 17, 18 thickness, irreducible (IT) 8 thin non-arbitrariness 271 thin normative terms 48, 50, 53, 54, 56, 73, 74, 274, 275 thin rationality 180, 182

307

Thomasson, Amie 33, 34, 38, 42, 43 Thomson, Judith Jarvis 247, 249, 250 Tiffany, Evan 231, 232, 247, 248, 251 Timmons, Mark 273 transitivity coherence requirement 191 truth modal status of learned truths 92 ontology and moral truth 107 semantics and 60 values 54, 60 uncertainty see moral uncertainty utilitarianism classical 41, 42, 43, 44 hedonistic 29, 43 moral epistemology and 114, 117 moral uncertainty and 164 neo-Aristotelian eudaemonism and 145 ought simpliciter 245 validity of language content 66 values aesthetic 71, 79, 231, 234 authoritative 244 comparison 161, 166, 167, 171, 175, 178 competing 248 conflicting 241 functions 169 generic reasons and 212, 214, 223, 226 intrinsic 152, 153, 154, 158 moral 167, 231 objective 274 realization of 244 truth 54, 60 Van Cleve, James 12, 15 Wedgwood, Ralph 139, 257 William of Ockham 118 Williams, Bernard 153, 248 Wolf, Susan 153, 154 world-moral-opposition-stance pairs 56, 57 wrong(ness) and non-descriptive relativism 52, 56, 58, 63