Oxford Studies in Metaethics Volume 15 2019954826, 9780198859512, 9780198859529

Citation preview

OXFORD STUDIES IN METAETHICS

Oxford Studies in Metaethics Volume 15

Edited by RUSS SHAFER-LANDAU

1

3

Great Clarendon Street, Oxford, OX2 6DP, United Kingdom Oxford University Press is a department of the University of Oxford. It furthers the University’s objective of excellence in research, scholarship, and education by publishing worldwide. Oxford is a registered trade mark of Oxford University Press in the UK and in certain other countries © the several contributors 2020 The moral rights of the authors have been asserted First Edition published in 2020 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: 2019954826 ISBN 978–0–19–885951–2 (hbk.) ISBN 978–0–19–885952–9 (pbk.) Printed and bound in Great Britain by Clays Ltd, Elcograf S.p.A. 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.

List of Contributors Selim Berker is Alford Professor of Natural Religion, Moral Philosophy, and Civil Polity, Harvard University Joshua Blanchard is Visiting Assistant Professor, Oakland University M. Coetsee is Assistant Professor of Philosophy, Jepson School of Leadership Studies, University of Richmond Michelle M. Dyke is Assistant Professor and Faculty Fellow, NYU Center for Bioethics Daniel Fogal is Assistant Professor, NYU Center for Bioethics Daniel Z. Korman is Professor of Philosophy, University of California, Santa Barbara N. G. Laskowski is Assistant Professor of Philosophy, California State University, Long Beach Dustin Locke is Associate Professor, Claremont McKenna College Matt Lutz is Associate Professor of Philosophy, Wuhan University Berislav Marušić is Senior Lecturer, University of Edinburgh Olle Risberg is a graduate student, Uppsala University Connie S. Rosati is Professor of Philosophy, University of Arizona Paulina Sliwa is Senior Lecturer and Fellow of Sidney Sussex College, University of Cambridge Kenneth Walden is Associate Professor of Philosophy, Dartmouth College Preston Werner is Assistant Professor of Philosophy, Hebrew University Daniel Wodak is Assistant Professor of Philosophy, University of Pennsylvania

OUP UNCORRECTED AUTOPAGE PROOFS – FIRST PROOF, 4/7/2020, SPi

Introduction Russ Shafer-Landau This volume marks a decade and a half of Oxford Studies in Metaethics. It is, appropriately, a very rich collection that shines a light on a broad range of topics in our field. We begin with a set of papers focused on normativity and reasons. Kenneth Walden starts us off with an article that develops and defends an account of what it is to reason: to reason is to scrutinize one’s attitudes by consulting the perspectives of other persons. Walden identifies the principal attraction of this account as its ability to vindicate the unique of authority of reason. He argues that this conception entails that reasoning is a social endeavor—that it is, in the first instance, something done with other people. He then argues that such social endeavors presuppose mutual respect on the part of those participating in them. The account is designed to yield a form of Kantian constructivism: we have an unconditional duty of respect for persons because such a duty is implicit in the very nature of reasoning. M. Coetsee’s entry, whose subtitle (“How Not to Be a Psychopath”) offers guidance we’ll all want to adhere to, is focused on what she calls the phenomenal appreciation of reasons. To appreciate the notion, consider an agent such as Huck Finn, who believes that he is doing wrong by helping his enslaved friend Jim escape to freedom. Coetsee thinks that if Huck is to earn our moral esteem, then there must be some implicit way of appreciating and responding to considerations as moral reasons that does not involve explicitly believing that those considerations are moral reasons. She argues that agents can implicitly appreciate a consideration as a moral reason to φ by presenting that consideration via the light of a felt directive force that “points” towards φ-ing—lending weight to it, or soliciting it—in a particular authoritative way. In the course of motivating her account, Coetsee examines and rejects four alternative proposals for how to account for implicit reasons-appreciation: first, a de re account of appreciation and then three additional accounts of appreciation derived from major theories of mental representation (inferentialist, causal tracking, and functionalist theories). Daniel Wodak next offers his thoughts on the hotly contested issue of which normative category (if any) rightly counts as the fundamental one.

Russ Shafer-Landau, Introduction In: Oxford Studies in Metaethics Volume 15. Edited by: Russ Shafer-Landau, Oxford University Press (2020). © Russ Shafer-Landau. DOI: 10.1093/oso/9780198859512.001.0001

OUP UNCORRECTED AUTOPAGE PROOFS – FIRST PROOF, 4/7/2020, SPi

x

Introduction

What he calls “X-Firsters” hold that there is some normative feature that is fundamental vis-à-vis all others (and, often, that there’s some normative feature that is the “mark of the normative”: all other normative properties have it, and are normative in virtue of having it). This view is taken as a starting point in the debate about which X is “on first.” Rather than engage in this much-contested intramural debate, Wodak takes a step back and asks whether we should be X-Firsters in the first place, and what we should think about normativity if we aren’t X-Firsters. He returns a negative answer to the first question, offering an argument by analogy that points to the implausibility of assuming the existence of a fundamental category in other domains. He then offers an alternative view—taking normativity to be a determinable that is explained in terms of its determinates—that provides his preferred path to thinking about the structure and unity of normativity. Paulina Sliwa’s article concerns the relationship between moral ignorance, excuse, and exculpation. She argues that the philosophical debates about the relationship have been based on a mistaken assumption: namely, that excuses are all-or-nothing affairs. She rejects the idea that to have an excuse entails that one is blameless. On her view, excuses are not binary but gradable: they can be weaker or stronger, mitigating blame to greater or lesser extent. Along the way, Sliwa explores the notions of strength of excuses, blame mitigation, and the relationship between excuses and moral responsibility. Her take-away: Moral ignorance may well excuse, but it does not exculpate. Next up: a trio of papers on the merits of moral realism. N. G. Laskowski opens this stretch with a chapter that seeks to diagnose and then allay concerns that many philosophers have expressed about the tenability of reductive metaethical views. The keystone of the enterprise is a focus on the distinctiveness of our use of normative concepts. Laskowski seeks to make a cumulative case for the view that what it is to use a normative concept is to use an unanalyzable natural-cognitive concept that is related to noncognitive elements of our psychology. Along the way, he also explains why ethicists’ resistance to reductive views is not well captured by recent attempts to account for seemingly parallel resistance to reductive views in other subdomains of philosophy, such as in the philosophy of mind. Joshua Blanchard next offers a novel paper that defends pro-realism: the view that it is better if moral realism is true rather than any of its rivals. He situates the discussion within a framework that focuses on the notion of philosophical angst, offering three general arguments. The first targets nihilism. Blanchard’s view is that in securing the possibility of moral justification and vindication in objecting to certain harms, moral realism secures something that is non-morally valuable and even essential to the

OUP UNCORRECTED AUTOPAGE PROOFS – FIRST PROOF, 4/7/2020, SPi

Introduction

xi

meaning and intelligibility of our lives. His second argument targets antirealism: as he sees it, moral realism secures a desirable independence for moral justification that is qualitatively different from the anti-realistic construal of independence that is explicable only in terms of degrees of distance from our subjective responses and attitudes. The third argument focuses on quasi-realism, which, Blanchard believes, provides no comfort to the prorealist who is already angsty about anti-realism. Preston Werner’s selection focuses on metasemantic worries for moral realists who also endorse non-naturalism—as Werner conceives of it, the view that normative properties are response-independent, irreducible to natural properties, and causally inefficacious. As he notes, non-naturalist realists have had relatively little to say about their metasemantic views, likely hoping that they could remain ecumenical on such matters and take on board the best view that emerges from research in the philosophy of language. Werner argues that this hope is bound to be frustrated, and that non-naturalists must commit themselves to an epistemic account of reference determination, one that represents a modification of a view recently defended by Imogen Dickie. Before setting out his positive view, he reviews the traditional metasemantic theories and explains why they each cause trouble for the non-naturalist. If Werner is correct, a fully fleshed out moral epistemology of the sort he sketches will not only answer longstanding worries about non-naturalism, but will at the same time rebut metasemantic objections to non-naturalism. Daniel Fogal and Olle Risberg next take up a set of issues that has recently attracted a fair bit of attention from metaethicists—namely, the role (if any) that general moral principles play in the explanation of particular moral facts. One view is that such principles play no role at all—the explanations of particular moral facts are given entirely by means of “natural” or “descriptive” facts. Fogal and Risberg reject that view, and argue that general moral principles play an ineliminable role in such explanations. They argue that their view best makes sense of some intuitive data points, including the supervenience of the moral upon the natural. In making their case, they present two alternative accounts of the nature and structure of moral principles: (i) “the nomic view,” on which moral principles are laws of metaphysics of the same broad kind as the laws that (plausibly) figure in metaphysical explanations more generally; and (ii) “moral Platonism,” on which moral principles are facts about kind-applying (as opposed to particular-applying) moral properties. Along the way, they consider a number of related issues, such as the distinction between metaphysical grounding and metaphysical analysis, and conclude by discussing the sense in which moral principles obtain of necessity.

OUP UNCORRECTED AUTOPAGE PROOFS – FIRST PROOF, 4/7/2020, SPi

xii

Introduction

Selim Berker’s article shifts the focus to expressivism, and its quasi-realist variant in particular. Quasi-realists aim to account for many of the trappings of metanormative realism within an expressivist framework. Chief among these is the realist way of responding to the Euthyphro dilemma: quasirealists want to join realists in being able to say, “It’s not the case that kicking dogs is wrong because we disapprove of it. Rather, we disapprove of kicking dogs because it’s wrong.” Berker argues that the standard quasirealist way of explaining what we are up to when we assert the first of these two sentences rests on a mistaken identification of metaphysical dependence (or grounding) with counterfactual covariation. He proposes a different way for expressivists to understand such sentences, on which they serve to express complex states of mind in which an attitude bears a relation of psychological dependence (or basing) to another state of mind. In slogan form: talk of normative grounding is the expression of attitudinal basing. He argues that this proposal is a natural, versatile, and fruitful approach for expressivists to take that helps them secure the first half of the Euthyphro contrast—but at the cost of making it difficult to see how expressivists can make sense of that contrast’s second half. Michelle M. Dyke then offers a diagnosis of the shortcomings of familiar versions of moral relativism, as a way to motivate the development of her preferred version of normative relativism. She argues that well-known forms of relativism are unable to accommodate, at once, three highly intuitive theses about the distinctive character of moral reasons, especially as they compare to reasons of self-interested practical rationality. By contrast, the new version of relativism she develops is, she argues, capable of accommodating all three claims. Her view combines the idea that the normative facts are attitude-dependent with the claim that there are other agents—societies, in particular—to which it makes sense to attribute the kinds of attitudes that give rise to normative reasons. Thus not only people, but societies as well, can possess reasons to pursue their aims. On Dyke’s account, what distinguishes moral reasons from reasons of practical rationality is that the former apply directly to societies in virtue of aims held by each society as a group, while the latter apply directly to persons in light of their own individual interests. Next up: a pair of chapters that straddle the porous borders between metaethics and normative ethics. The first of these is by Connie S. Rosati, whose contribution focuses on the welfarist good-for relation and undertakes to defend a particular view about the structure and nature of this relation. The paper critically assesses alternative views, including Moore’s eliminativist view, the “private ownership” view Wayne Sumner attributes to Moore, and both Donald Regan’s and Guy Fletcher’s more recent “locative analyses.” According to Rosati’s preferred view—what she calls

OUP UNCORRECTED AUTOPAGE PROOFS – FIRST PROOF, 4/7/2020, SPi

Introduction

xiii

the rational fit theory of welfare—the welfarist good-for relation is a reasongiving relation of fit between a welfare object and a welfare subject, where a welfare subject is a valuable being. Berislav Marušić then invites us to reflect on how we do and how we ought to respond to suffering or witnessing injustice. Such responses often include a combination of anger, grief, resentment, indignation, or horror. And it seems that this is how it should be: the injustice is a justifying reason for our emotional response. However, it is a striking fact that our anger, grief or horror will diminish over time, often fairly quickly, even if the injustice persists. We accommodate ourselves to the injustice. As Marušić notes, this is good for us, and may even seem appropriate, as it is often wrong to dwell on a wrong. The puzzle is how accommodation can be appropriate if the injustice remains unchanged. And how can we make sense of accommodation when we anticipate it? Marušić argues that accommodation to injustice poses an insurmountable problem for our understanding our emotional response to injustice; if he is right, such accommodation reveals something incomprehensible at the heart of our moral outlook. We conclude with a pair of chapter on moral epistemology. The first of these, by Matt Lutz, is devoted to the so-called reliability challenge to moral non-naturalism. The challenge can be traced back at least to a skeptical worry in the philosophy of mathematics raised by Paul Benacerraf. The current reliability challenge is widely regarded as the most sophisticated way to develop this skeptical line of thinking. But Lutz demurs, arguing that much influential work undertaken since Benacerraf ’s formulation of the challenge has been misconceived and confused. As Lutz sees it, the reliability challenge is not the most potent epistemic challenge to moral nonnaturalism. That title rather belongs to a different problem deriving from the fact that there is, as Lutz puts it, a becausal condition on knowledge that he believes non-natural moral facts cannot satisfy. The last of our selections won the Marc Sanders Prize in Metaethics in 2018. The paper, co-authored by Daniel Z. Korman and Dustin Locke, is an important contribution to the burgeoning literature on moral debunking arguments. Many such arguments are formulated as challenges specifically to moral realism, and are meant to show that, by realist lights, moral beliefs are not explained by moral facts, which in turn is meant to show that they lack some significant counterfactual connection to the moral facts (e.g. safety, sensitivity, reliability). The dominant, “minimalist” response to the arguments—sometimes defended under the heading of “third-factors” or “pre-established harmonies”—involves affirming that moral beliefs enjoy the relevant counterfactual connection while granting that these beliefs are not explained by the moral facts. Korman and Locke seek to show that the minimalist gambit rests on a controversial thesis about epistemic priority:

OUP UNCORRECTED AUTOPAGE PROOFS – FIRST PROOF, 4/7/2020, SPi

xiv

Introduction

that explanatory concessions derive their epistemic import from what they reveal about counterfactual connections. They then challenge this epistemic priority thesis, with an eye to undermining the minimalist response to debunking arguments (both in ethics and elsewhere). While a couple of the chapter here had their origins in talks given at the 12th annual Madison Metaethics Workshop, the dozen others that form the bulk of this volume date from the 13th iteration of MadMeta, in September 2018. Many thanks to the program committee—Ralf Bader, Matt Bedke, Guy Fletcher, Errol Lord, and Eric Wiland—for their assistance in reviewing submissions for that event. Thanks, too, to the Marc Sanders prize panel, consisting of Tristram McPherson, Jonas Olson, and Hille Paakkunainen. Two anonymous reviewers gave copious, insightful comments on the papers here that led to numerous improvements in the works you have before you. Finally, I’m pleased to express my gratitude to my editor Peter Momtchiloff, who enthusiastically took a chance on this series a decade and a half ago and has been a wonderful steward of it ever since.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

1 Reason and Respect Kenneth Walden

Kantian constructivists believe that the norms of morality possess universal, unconditional, and categorical normative force. And this is true, they believe, not because those norms are grounded in an objective normative reality that is independent of agents’ wills, but because morality can be “constructed,” so to speak, from the point of view of practical reason as such—because any creature correctly employing practical reason will inevitably come to endorse morality. By “morality,” Kantian constructivists usually mean a cluster of closely related claims from Kant’s own moral theory, for example, that every person possesses a dignity by which they are entitled to exact respect from others. There are two principal genres of argument for this family of views. The first consists of variations on and elaborations of Kant’s “regress on the conditions of value” argument in the Groundwork. There Kant asks about the conditions of the value of humdrum goods. This question puts a regress on the conditions of value into motion. The best candidate for halting this regress, for the unconditioned condition of value, is the value of “humanity.” And not just the value of whoever is doing the valuing either, but of the rational capacity common to every person. All the lofty Kantian doctrines about dignity and respect follow from this claim, and so are implicit in the conditions of valuing as such.¹ The second kind of argument suggests that there are constraints on practical reason imposed by its connection to agency. According to those pressing this kind of argument, agency has certain constitutive requirements, which include principles that entail features of Kant’s moral theory. Insofar as agents must presuppose their own agency while engaged in practical reasoning, these philosophers argue, we can be sure that those

¹ Korsgaard (1996) and Markovits (2013).

Kenneth Walden, Reason and Respect In: Oxford Studies in Metaethics Volume 15. Edited by: Russ Shafer-Landau, Oxford University Press (2020). © Kenneth Walden. DOI: 10.1093/oso/9780198859512.003.0001

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

2

Kenneth Walden

constitutive requirements will inevitably be constructed from every practical point of view.² Both of these arguments have come in for heavy weather. On the regress argument: we might wonder why we are required to ground relational value in non-relational value, why this value must be “conferred” by those who value, and why it is shared by all persons.³ On the agency argument: the connection between normativity and the requirements of agency can be questioned, as can the idea that Kantian morality is a real condition of agency.⁴ I don’t plan to defend these arguments. I instead want to note that it is slightly odd that even though Kantian constructivism is naturally understood as a claim about what we are committed to just insofar as we are reasoners, neither of these arguments focus on the nature of reason, at least not in the first instance. Instead, they base their conclusions on an analysis of kindred notions like valuing and agency. This isn’t an objection, but it does raise the possibility of a more direct ascent to Kantian constructivism. My purpose here is to lay out such an argument. Stripped to its essentials, the argument goes like this. The first premise is that reasoning is an essentially collective activity, and thus something that one cannot undertake in perfect solitude.⁵ Indeed, I will argue that every rational agent is a party to every other agent’s reasoning. The second premise concerns the general conditions of participation in a collective activity. These activities constitutively require mutual respect between their participants within the bounds of that activity. That is, they require attitudes of respect specific to the activity undertaken. In most cases, this respect is conditional and highly circumscribed. If we are painting a house together, I must respect you as my co-housepainter, but not necessarily beyond that narrow station. And I can avoid even this requirement by opting out of our collective project. The third premise is that reason is different from other activities. Its authority over us is unconditional and the demands it makes are uncircumscribed. Putting these premises together yields the conclusion that we have an unconditional and uncircumscribed duty of respect for all persons. 1. Reasoning is a collective activity we undertake with all other persons. 2. Participation in a collective activity imposes an activity-specific duty of respect for one’s partners. 3. The authority of reason is unconditional and uncircumscribed.

² Korsgaard (2009), Velleman (2009), Schafer (2015). ³ Regan (2002), Ridge (2005), Langton (2007), and Theunissen (2018). ⁴ Enoch (2006) and Setiya (2007). ⁵ The argument will make no distinction between the theoretical and practical employment of reason.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Reason and Respect

3

Therefore, we have an unconditional and uncircumscribed duty of respect for persons. The rest of the chapter is organized around a defense of these three premises, with the preponderance of discussion addressing the first. 1.1. THE SOCIALITY OF REASON The phrase “the sociality of reason” might aptly denote a number of doctrines. In the interest of economy, I will dispense with a systematic comparison of them and leap into my conception.⁶ The version of the thesis I am interested in is inspired by a famous passage from the Critique of Pure Reason. In the final sections of the book Kant turns to the issue of reason’s own authority: Reason must subject itself to critique in all its undertakings, and cannot restrict the freedom of critique through any prohibition without damaging itself and drawing upon itself a disadvantageous suspicion. Now there is nothing so important because of its utility, nothing so holy, that it may be exempted from this searching review and inspection, which knows no respect [Ansehen] for persons. The very existence of reason depends upon this freedom, which has no dictatorial authority, but whose claim is never anything more than the agreement of free citizens, each of whom must be able to express his reservations, indeed even his veto, without holding back. (A739/B767)⁷

There are several important claims in this short but provocative passage. Teasing them out and defending them can furnish us with an argument for the sociality of reason in the sense that supports my argument. That is what I attempt. (1) Reason is a critical faculty that enables reflection and scrutiny on our attitudes, and reasoning is an activity of applying such scrutiny. This idea is more implied than stated here, but it finds a fuller expression elsewhere, especially in Kant’s Conjectural Beginnings of Human History.⁸ There Kant says that rational animals are distinguished by a special freedom from instinct. Whereas non-rational creatures are automatically moved by instinct, rational creatures are not. They can “step back” from their instincts and reflect on them—not as something that will move them, but as ⁶ E.g. just in the last half-century: Cavell (1969), Habermas (1984), Pinkard (1996), Brandom (2009), Laden (2014), and Manne (2016). I have more specific debts to Christine Korsgaard and Onora O’Neill that will become clear later. ⁷ References to Kant are to the Berlin Akademie edition of 1900– except for the Critique of Pure Reason, which uses the usual A/B pagination. ⁸ Especially 8: 111–13.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

4

Kenneth Walden

something that they may endorse and act on, or else reject and resist. This reflection is an opportunity to scrutinize an inclination, belief, or appearance: to ask what significance that inclination, belief, or appearance has, apart from its natural tendency to move us. Whereas a non-rational creature is simply moved by its hunger, the rational creature’s hunger poses a problem: what does this mean for me, what reasons do I have in virtue of this hunger, what should I do about it? According to the History, these activities—detachment from and scrutiny of one’s instincts—are the fledgling steps of reason. Fully developed, reason encompasses the many ways, simple and sophisticated, that we can apply this reflective scrutiny to our judgments.⁹ This is just a thumbnail sketch, but we can already see how this conception of reason differs from some points of conventional wisdom. For the sake of comparison, consider two familiar alternatives. There is a Clarkean conception on which reason is a perceptual or quasi-perceptual faculty for detecting some special class of contents, e.g. for grasping intellectual appearances. And there is a Lockean conception according to which reason is a capacity for transforming mental states according to particular rules, e.g. for performing logical inferences.¹⁰ My argument going forward will be premised on Kant’s critical conception, so the first step in that argument must be to explain why we ought to adopt it rather than these others. One way to argue for a conception of reasoning is by showing that it better captures our considered judgments about a range of paradigms. But this isn’t my approach. My claim is not that Clarkeans and Lockeans identify bogus examples of reasoning or fail to capture genuine exemplars. Indeed, I will come around later to try to show how the critical conception can capture much of what the they think is characteristic of reasoning. Instead, I contend that the critical conception is preferable because it better explains a vital property of reason—its pre-eminent authority. As Thomas Nagel explains: Whoever appeals to reason purports to discover a source of authority within himself that is not merely personal, or societal, but universal—and that should also persuade others who are willing to listen to it.¹¹ ⁹ Compare Korsgaard (1996: 12–16) and O’Neill (2000). ¹⁰ See Clarke, The Boyle Lectures, esp. §227 in Raphael (1991) and Locke, An Essay Concerning Human Understanding, IV.18.2. I have no special reason for choosing these philosophers as representatives other than their proximity to Kant. There are of course modern descendants of these views. Bonjour (1998) is a clear descendant of Clarkeanism. Those that view reason as a receptivity to self-subsistent substantive reasons could also be included in this group, notably Parfit (2011) and Scanlon (2013). Those who emphasize reasoning as a procedure are plausibly classed as Lockeans, viz. Setiya (2014) and McHugh and Way (2018). ¹¹ Nagel (1997: 3).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Reason and Respect

5

I take this special authority to be something like a fixed point in our understanding of reason. It is, Nagel goes on to say, established in a very special way. In order to have the authority it claims, reason must be a form or category of thought from which there is no appeal beyond itself—whose validity is unconditional because it is necessarily employed in every purported challenge to itself.¹²

The suggestion here is that reason is not just authoritative. It is the “last word” on (nearly) every subject and the “final court of appeal” for (nearly) every question. For reason to be authoritative in this ultimate, foundational sense, its authority cannot be conferred by something else because then it wouldn’t be the last word. This means that reason must be, in some sense, self-validating. But obviously not any self-validation will do. If a Magic 8-Ball tells us “always trust the Magic 8-Ball!,” we are not thereby justified in trusting the Magic 8Ball.¹³ Nagel hints at a more promising kind of self-validation. We could justify the use of reason by showing that any attempt to examine its bona fides from some extra-rational point of view will end up presupposing that authority. This idea can be cast as an argument from pragmatic paradox. If we can show that any criticism of reason will presuppose the use of reason, then a demonstration that reason has no authority will be self-defeating. That is, it will be self-defeating in the way that “A says, ‘I cannot speak’,” “A fears that she has no fears” or “X thinks that she doesn’t exist,” are selfdefeating. The content of our demonstration—what it is shown—would be incompatible with that demonstration.¹⁴ (Such an argument would not establish reason’s possession of a special property, but only that we must take it to have this property. This is of course a common concern with transcendental arguments. But here the difference doesn’t really matter: saying that I must recognize something’s authority is, from my point of view, tantamount to establishing that authority.) If we can make an argument like this succeed, then reason would have authority not because that authority is bestowed by something else or established according to some further standard. It would have authority because reason is presupposed by any process that would arbitrate such authority.¹⁵

¹² Nagel (1997: 7). ¹³ Compare Nagel’s example of tea leaves on p. 24 and Markovits (2013: 160). ¹⁴ This way of understanding the force of pragmatic paradox arguments comes from Haslanger (1992). ¹⁵ What I am claiming for reason here is sometimes claimed for the principles constitutive of agency. See Ferrero (2009) and Silverstein (2015) for attempts to flesh out the logic of this argument.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

6

Kenneth Walden

Capturing this feature of our understanding of reason seems much more important than fitting our intuitions about what counts and doesn’t count as an episode of reasoning. The critical conception is well-suited to this task. Normative questions about what has authority, what reasons we have, and what we ought to do are critical questions. They are questions we ask ourselves when we find that we have an intention, instinct, or an urge and want to know what to do about it, whether it is something we ought to endorse or renounce. If the critical conception of reason is correct, then we are engaged in reasoning simply in virtue of asking these questions and undertaking the scrutiny they express. It follows that any conclusion that reason has no authority here—over e.g. the question of whether to endorse this urge and set out to satisfy it—would constitute a pragmatic paradox. It would assert that reason has no authority while presupposing that very authority. This means that such a conclusion is something we could never reach without contradicting ourselves. On this picture, then, reason secures the authority Nagel identifies by being already enlisted in the very activity of questioning a thing’s authority.¹⁶ It is rather harder to see how the Lockean and Clarkean conceptions could have this kind of unconditional authority. If we understand reason as a productive faculty for producing new judgments—either by intuiting them or transforming extant judgments—then it seems we need a substantive demonstration of its bona fides before we rely on it. This demonstration could issue from something other than reason, like track record data for our intuitions or a proof that a certain rule is truth-preserving given a certain semantics. But this supposes that reason isn’t the “last word,” so it won’t work. On the other hand, this demonstration could come from reason itself. But how would this go? It would have to be something like reason putting forward the judgment (e.g. as an intellectual intuition) that reason has authority to review and revise some wide range of judgments. But then this proposal starts to look like the highly suspect kind of circular justification offered by the Magic 8-Ball: reasoning is self-validating, but not in a way that will appease any genuine skeptic. And that’s no good either. This is a reason to prefer the critical conception to its rivals. It doesn’t mean that there won’t be a place for Clarkean and Lockean activities in reasoning. It means that our most basic conception of reason won’t be Clarkean or Lockean. We can’t say that reason just is this quasi-perceptual capacity or the ability to apply certain rules. Insofar as these capacities are used in reasoning, it will be because they can be subsumed under the critical

¹⁶ Compare Korsgaard (1996: 46–7).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Reason and Respect

7

conception—because these rules or perceptual judgments pass the scrutiny of reason. I will say more about how this can happen later in the chapter.¹⁷ (2) The scrutiny of reason comes from the consultation of other perspectives. The most provocative part of Kant’s claim is that the “claim of reason” rests on the “agreement of free citizens.” To many readers this is a scandalous thing to say, not least because it suggests that reasoning, like “agreement,” is a social undertaking. Most of us are probably inclined to think that reasoning is sometimes, if not primarily, a solitary activity—a set of procedures that I cycle through in my own mind. I can invite others to join me in this enterprise, but I needn’t do so. Why suppose that other people will necessarily be involved in the business of reasoning at all? To defend this idea, I want to begin by suggesting a connection between the scrutiny of reasoning and other perspectives.¹⁸ According to the critical conception, reasoning consists in the scrutiny of judgments and other attitudes. But how does this scrutiny get applied? Suppose it appears to me that there is a cup of tea before me or that I find myself inclined to take a swig. I clearly don’t reason about these things simply by beholding them and asking, “so what?” There needs to be some friction or resistance applied, some gauntlet to be run, if this scrutiny is to be anything but perfunctory. What does this friction consist in? The answer of course is that there are many things that we do to scrutinize attitudes such as these. We see if they are consistent, broadly speaking, with other attitudes. We consider their consequences, logical and causal. We devise empirical tests and ponder hypothetical counterexamples. My question is whether there is anything synoptic we can say about these procedures.

¹⁷ There is a sense in which this dispute about reason is verbal. As will become clear I am interested in the nature of reason because of the very authority Nagel identifies—the sort of authority that makes demonstrating that something is a condition of reasoning much more exciting than showing it is a condition of chess. Whether we call this faculty reason or the Kritische Kraft doesn’t really matter. That said, the association between this kind of authority and reason is strong and venerable enough that I think it’s worth proceeding as I have so far and making reason the focus of our examination instead of some nameless Ramsey sentence. Notably, this is an issue that arises within Kant’s own work. Early in the first Critique he flirts with a position close to Lockeanism before offering the more critical remarks I quoted above. (Reason is a “faculty of principles” (A299/B356).) And in the Critique of Judgment he reassigns many of the tasks of reason to the power of judgment. ¹⁸ Kant doesn’t draw this connection here, but intimates it elsewhere, especially in the second of his maxims of common understanding—that one ought to think into the place of the other. See Anthropology from a Pragmatic Point of View, 7: 200 and Critique of Judgment, 5: 293–5. Analogies with spatial perspective are also important for Hume’s discussion of the common point of view. See the Treatise on Human Nature, Book III, Part III, Section III.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

8

Kenneth Walden

We might be inclined to think that we ought to scrutinize our judgments by comparing them to whatever they are about. If it appears to me that there is tea in the cup before me, I ought to critique this appearance by comparing it to the cup itself. But this hardly makes sense. You needn’t be an idealist to doubt that the cup itself—the cup as it is independent of my thoughts about it—is the sort of thing I can use as an instrument in my reasoning. If I could, I wouldn’t need to reason at all. What we can do, however, is take a closer look at the cup: we can examine its underside, weigh its contents, or give it a sniff. This suggests a rather different way of thinking about scrutiny. In doing these things, we are not employing the cup in itself but taking up different perspectives on the cup— from up close, from underneath—and considering the significance of our attitude in light of those perspectives. The same conception of scrutiny is suggested by the metaphor of reason allowing us to “step back” from our judgments. When we step back from something, we view it from a different vantage point, one from which its force or significance may appear differently. We scrutinize something by taking up a new perspective. Consider some simple examples. Suppose that it appears to me that there is a trapezoidal surface before me. The natural way to scrutinize this appearance is to ask what this surface would look like if I were standing a few feet to the left, and then a few more, or if I occupied all the different points on the circle circumscribing the surface. Depending on how the surface appears from these vantage points, I may come to believe that the surface is rectangular, trapezoidal, or rhomboid. Now suppose I have an inclination to eat bacon. There are some natural ways to scrutinize this inclination. I can consider what I would feel like—leaden or sated—after eating the bacon, and in doing this speculate about how that self would regard the bacon-eating. I can also consider the inclination in light of various commitments—to vegetarianism, God, or my own health. I may then endorse the inclination or resist it. It is natural to say in both of these cases that our scrutiny of a judgment (or some other attitude) will involve viewing it from different perspectives, be they spatial perspectives on a surface or evaluative perspectives on an intention. My suggestion is that the gauntlet we expect an attitude to run when scrutinizing it—what it is we subject it to—is further perspectives. To reason about a judgment is to scrutinize it from different perspectives. Familiar examples of reasoning can be made to fit this formula. When I reason about whether copper rusts, I am searching for perspectives where copper’s rusting, or the lack thereof, is in evidence. When I reason about whether courage is a virtue, I am imagining myself into perspectives where courage may have favorable or unfavorable effects. When I reason about whether Jim can be

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Reason and Respect

9

trusted, I am imagining perspectives that might expose his treachery or unreliability. This is, as I said before, a highly synoptic claim about scrutiny. It isn’t to say anything about the specific methods we follow in reaching a verdict about copper, courage, or Jim. Obviously those will be variable. I will say more about how familiar methods can be made to fit this picture later. The claim is one about our conception of reasoning at its most general: reflective scrutiny on a judgment can be cashed out in terms of the subjection of that judgment to additional perspectives.¹⁹ (3) The consultation of other perspectives in reasoning is second-personal. By itself the claim that reason’s scrutiny involves the consultation of other perspectives is quite vague. What, we should wonder, does this “consultation” consist in? And what is it that we are consulting? I want to distinguish two basic forms the consultation could take. First, it could be third-personal in the following sense.²⁰ There is a certain content to be found in your perspective—how things seem from where you stand on whatever subject matter we’re discussing. Consulting your perspective means integrating this how-things-seem content into some process by which we come to ratify, reject, or revise a judgment. This conception of consultation fits quite well with the geometrical model of perspective. Suppose that I have an appearance as of a trapezoidal surface, with the longer of the two parallel sides nearer me. You are oriented on the opposite side of the surface and so occupy a different perspective. How do I consult your perspective to scrutinize my judgment about the surface? A natural thought is that I take my appearance and yours and plug them into some procedure that yields a suggestion about what shape would produce these appearances from these perspectives. The procedure will reflect basic facts about particular geometrical transformations. If we both see trapezoidal forms and are oriented in a particular way, for instance, geometry would tell us to believe that the surface is in fact rectangular. What’s important about this conception of consultation—and why I call it “third-personal”—is that the occupant of the other perspective is dispensable to the activity. What matters is a fact about how things seem from a certain perspective, not what a person occupying that perspective would say, ¹⁹ This proposal is not necessarily incompatible with all other conceptions of reason. It is not implausible, for example, to describe a constellation of attitudes that has survived scrutiny from all perspectives as one that exhibits systematic unity or understanding. In this case, my proposal would coincide with a view of reason on which it aims at one of these ideals. Compare, e.g. Schafer (forthcoming) and Kant’s view in the “Appendix” to the Transcendental Dialectic. ²⁰ Compare the central distinction in Darwall (2006).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

10

Kenneth Walden

think, or do. I am not so much asking for the judgment, reaction, or criticism of a full-fledged person as for an appearance as glimpsed from a particular vantage point, one that a particular person happens to occupy, but which could, in principle, be captured by something impersonal, like a camera. In this sense, my consultation of your perspective does not essentially involve you. Your testimony is a way to acquire the perspectival content I need, but you are no more than an instrument for obtaining the information. An alternative way of thinking about this scrutiny is second-personal. On this conception, the consultation of another perspective is, fundamentally, the consultation of another person, and the significance of what we discover from this consultation for the process of reasoning depends on its being proffered by a person. So in consulting another perspective, we are not discovering content, but engaging with another person. On this view, the normative structure of consultation is less like looking at a photograph and more like having a conversation. Suppose that I think this surface is a trapezoid or that it would be prudent to kill my wife for the insurance money. Consulting your perspective means asking you, “What do you think?” about these ideas and having the conversation that ensues from that question. This conversation is not a means to something of significance for my scrutiny; it is the scrutiny.²¹ I have two arguments against the third-personal model of consultation, both of which point us toward its second-personal rival. First argument. Suppose I believe that there is a cup of tea before me. I scrutinize this belief by consulting a perspective from across the table. Maybe there’s someone sitting there, or maybe I get up to look for myself. In any event, I find that from there it appears as though the cup is empty. This appearance, we are supposing, has a certain significance for the activity of scrutinizing my belief. Presumably, it favors withholding or revising that belief. The third-personal model of consultation understands this process in a particular way, as the retrieval of an impersonal perspectival content—an appearance-from-yonder. On this model how does this content come to have the significance it has? How does it come to militate in favor of belief revision (assuming that it does)? This question is an instance of the more general one: how, to use Sellars’s well-wrought image, can impersonal perspectival content be brought into the “space of reasons”? In some cases, we seem to have a good idea of how this works. You and I are looking at a common surface from different points of view, and you have some perspectival content p. It’s innocent enough to think of this as something like sense data: colors arranged shape-wise amidst my visual field. ²¹ That reasoning is conversational is a principal theme of Laden (2014).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Reason and Respect

11

This content plus your position, the direction of your gaze, and basic facts about geometry can produce a verdict about the consistency of my belief about the surface and p. For example, trapezoidal configurations in p are consistent with my belief that the surface is a rectangle, but octagonal configurations are not. (From no perspective does a rectangular table appear octagonal.) Obviously, if p is inconsistent with my belief, then it favors revising that belief. Thus in this simple case, p comes to have significance for the scrutiny of my belief through the interaction of two things: (i) a set of claims about the consistency of different perspectival contents and propositions about the shapes of those items they are perspectives on, and (ii) the axiom that inconsistency is grounds for revision. But there’s a hitch. The claims in (i) reflect substantive assumptions about the world, namely that you, I, and the table are oriented in a certain kind of geometrical space. (Not all geometrical spaces will support this kind of reasoning.) This poses a problem. For if we want to offer the method just described as the basis for an answer to our question, then it must be a method characteristic of reasoning. But that would mean that this substantive assumption about space would have to be a constitutive feature reasoning. And that seems highly dubious, especially on the critical conception of reason. We adopted this conception because it was able to self-validate in a way that vindicated with the unique authority of reason. But the geometrical propositions on which the claims in (i) depend will certainly not be capable of that kind of self-validation. There is no paradox, pragmatic or otherwise, in supposing that the world does not have this kind of structure. The more plausible interpretation is that the procedure described above is not a feature of reason, but something that introduces substantive, extra-rational claims about the nature of space. And if this is true, then that procedure cannot be the answer to our question about how perspectival contents acquire normative significance for reason. Of course, things only get more difficult when we move beyond the spatial cases and start talking about different perspectives on the justice of capital punishment or the efficacy of psychoanalysis. The general problem is that we need some method or principles to bring impersonal perspectival contents into the “space of reasons,” to make them normatively significant, but any particular method for doing this—like the one exemplified by my consultation of your spatial perspective on the surface—is likely to rely on substantive assumptions that we should be very reluctant to call constitutive of reasoning as such. Maybe this challenge can be met, and a procedure can be discovered that does not make substantive, extra-rational assumptions, but the burden of proof is very much on the advocate of the third-personal model to show it can be. Importantly, however, the second-personal model doesn’t face it all.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

12

Kenneth Walden

For it understands the consultation of other perspectives as a conversation, and the “moves” in a conversation—asserting, assuring, challenging, questioning, demurring—come with normative significance built into them. Reasoners will of course face the question of what, exactly, to do with someone’s assurance or demurral—e.g. how much weight to assign to it— but this is something that can be settled as a conversation evolves. What they don’t face is the more fundamental question of how the contents of other perspectives gain normative significance in the first place. Second argument. The third-personal model of consultation presupposes what I will call the separability of perspective and occupant. It supposes that there are facts about how things seem from a perspective independent of what a person occupying that perspective would say, judge, or think in certain circumstances. Once more, this supposition seems justified in the spatial case. Objective facts about how things appear from various spatial perspectives is one of the things Renaissance polymaths are famous for figuring out. But the separability of occupant and perspective seems more doubtful when we move to complex cases. Think of the perspective of an actual person, or a well-developed literary character. I think I have a decent grasp of Philip Marlowe’s perspective on various questions and subjects. But I have trouble imagining there being facts about how things seem from Philip Marlowe’s perspective above and beyond what Philip Marlowe would say or think about those subjects. It’s hard to even describe the perspective in impersonal terms. We could list biographical details in characterizing the perspective, but that seems to miss something. There’s a difference between Marlowe’s perspective and facts about how things would seem under distinctively Marlovian conditions. So we cannot, as is tempting in the spatial case, understand Marlowe’s perspective as fixed by conditionals of the form, “if a had Marlovian quality m₁, then things would seem thus to them.” Things only get harder when we scrutinize more complicated judgments. Suppose we have a hunch about a murderer’s motive, an idea for getting revenge on someone who swindled us, or a moral dilemma about whether to turn a client over to the police. Can we extend our model of spatial perspective to these cases? Doing so would require us to think there is some perspectival content that is not essentially Marlowe’s but which he is simply well-placed to inform us about—a content of the form “how things seem with respect to your revenge plot from here.” I have trouble conceiving of what such a thing would even be. This is not just an epistemic problem of gaining access to this impersonal perspectival content. The best explanation for our difficulties in conceiving of the impersonal perspectival content of a complex person on a delicate question is that the separability of person and perspective is incoherent.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Reason and Respect

13

It is incoherent because a perspective is constitutively something that belongs to a person. The separability of perspective and occupant runs into problems of conceivability because it tries to separate a thing from its conditions, and this yields an antinomy on the order of “leaving our olfactory powers aside, what does peppermint smell like?” If I am right, then the understanding of spatial perspectives we have been working with is misleading. The perspective on this table from the other side of the room is not an impersonal fact about how the table seems from that point. It is representation of what a hypothetical person would judge if they were standing there. This hypothetical person may simply be a version of myself who is standing over there, or it may be a suitably idealized observer. The geometrical theory of perspective that tells us which relations are preserved under which kind of transformations is useful not because it is constitutive of reasoning but because it makes reliable predictions about what a wide range of these hypothetical observers will see. Both of the objections I have put to the third-personal conception of consultation favor its second-personal rival. Marlowe is an essential party to my consultation of Marlowe’s perspective, I have argued, and not simply an instrument for discovering perspectival content. The natural way to understand my consultation, then, is as a request for Marlowe to do something: to agree with my hunch about the motive, to object to my revenge plan, to suggest a resolution to my dilemma. In most cases, this would only be the beginning of a longer process. I may reply to his reservations or ask for clarification. And he may reply in turn. What I am asking for is a conversation about my hunch, my plan, or my dilemma. For our purposes we don’t need to say much more about the structure or aims of this conversation. Some Kantian fellow travelers have been tempted to insist that reason requires me to justify myself to others in particular ways: to make my principles and judgments intelligible to others, to make them seem reasonable, or even to secure their agreement.²² And they may want to specify the conditions under which this process takes place: with equal information, behind a veil of ignorance, freed of various distortions. These claims may be good advice about how to consult other perspectives in answering particular questions, but I don’t think the ambitions of this chapter require me to weigh in on these questions. What matters for my argument going forward is that this is an activity in which the other person, the occupant of the perspective, is an essential participant.

²² See the discussion in O’Neill (2015) and Markovits (2013: 133–6).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

14

Kenneth Walden

Before moving on, I should address two objections. First, there is the problem of “unoccupied” perspectives, like a perspective from the dark side of the moon a hundred years ago. On the third-personal model, there is no special mystery about such perspectives. Their contents exist; they’re just harder to retrieve since there’s no one at hand to get them. On the secondpersonal model, however, “unoccupied perspective” is something like an oxymoron, since perspectives are constitutively had by persons. So what is it to consider an unoccupied perspective? My answer is straightforward. When I consider an “unoccupied” perspective, like one from the dark side of the moon, I am imagining a hypothetical person (or persons) and thinking about what they would say, do, or judge. My interest is an interest in a conversation with a merely possible person. But—one might reply—isn’t there something incoherent or defective about an interest in a conversation with a non-actual person? I don’t think so. Witness, for example, our attitudes about fictional characters. I can desire to go on an adventure with Philip Marlowe, I can fear for his safety when he’s blackjacked by a goon, and I can be absorbed by his soliloquies. My desire can never be satisfied, my fear involves a state of affairs that never will obtain, and the object of my absorption doesn’t exist. And yet I have these attitudes all the same.²³ By the same token, it is perfectly coherent for me to be interested in having certain kinds of conversation with Marlowe. And this, I am suggesting, is what an interest in his perspective amounts to. The second objection is that the second-personal construal of scrutiny implies that every perspective-haver is someone we can reason with. This may be challenged: we might think that a fish has a perspective even though it lacks the capacities that are distinctive of persons and could certainly never be our partner in conversation. I disagree on the first point. When we talk about a fish’s perspective, we are anthropomorphizing it. We are introducing a fiction on which the fish is more like a person than it actually is,where it has opinions, preferences, and other attitudes. Insofar as the fish has a perspective, it is the perspective of a hypothetical person who shares certain key properties with the fish. (If you think I’m being unfair to fish, you can turn the argument around: fish do have perspectives and could, with the right aid, be our conversation partners.) (4) For any judgment subject to the scrutiny of reason, that scrutiny ought to be unrestricted in the sense that all perspectives are relevant to it. In the passage above Kant says that reason’s scrutiny must be unbounded and unchecked—that nothing can be “exempted from” reason’s “searching review and inspection.” I am going to assume he is right about this and ²³ Gendler and Kovakovich (2005).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Reason and Respect

15

address a related proposition. What I want to argue is that the scrutiny of reason is unrestricted along another axis. For any judgment subject to the scrutiny of reason, every perspective is in principle appropriately consulted as part of this scrutiny. One might demur from this claim in a few ways. First, they might say that it is ridiculous to demand agents complete the laborious, probably endless task of consulting all perspectives—including extremely exotic ones—in order to count as having reasoned about a given judgment. I agree, but my claim isn’t that we should attempt such a comprehensive consultation in practice. I agree that that would be impossible. I am instead claiming that every perspective is in principle relevant to reasoning about a judgment. This doesn’t yield a minimal requirement for reasoning, but a regulative ideal. How we try to live up to this ideal given our cognitive limitations is an important and difficult question, but not one we need to answer before concluding that it is an ideal. Second, one might insist that certain perspectives are simply too remote from the subject matter to be relevant, even in principle. Suppose that it appears to me that there is a teacup before me, and I set about reasoning about this. Some perspectives are clearly relevant to this endeavor: the perspective from the other side of the table, the perspective of someone in the room before I arrived who might’ve seen an elaborate hoax constructed. But others are more questionable. How could the perspective from (say) a person standing on the dark side of the moon a thousand years ago be relevant to this question? Our standard for relevance ought to be whether it is possible that a given perspective may bear on our question. And it is plainly possible that the perspective from the dark side of the moon a thousand years hence could have some probative value. This may be the one time and place that principles of optics my visual experience depends on are falsified. It may be the one time and place where a powerful goblin reveals his plan to deceive me. That these are remote possibilities means that such a perspective is of low priority in practice, but not that it is wholly irrelevant. Third, one might say that perspectives of other people are in principle irrelevant to certain kinds of reasoning. A version of this objection might come from a certain kind of Humean. It makes sense to consider the perspectives of other people in the case of theoretical reasoning, they might say, since we are trying to uncover some subject-independent reality that is common to ourselves and others. By contrast, practical reason does not aim at the representation of some objective reality, but to produce rational action. And all that is required for this is internal coherence. Thus, the thought goes, the kind of scrutiny an agent applies to a practical judgment will come from the point of view of their own practical

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

16

Kenneth Walden

commitments and no others. We find a version of this thought in Sharon Street’s rejection of certain reflective questions: According to the Humean constructivist, eventually (at least in theory, if we pursue our reflections far enough) we get to a point where we have arrived at a coherent web of interlocking values. At that point, one can ask: “But why should I endorse this entire set of normative judgments? What reason do I have to endorse this set as opposed to some other set, or as opposed to no set at all?” The proper answer at this point, according to the Humean constructivist, is that the question is ill-formulated. One cannot coherently step back from the entire set of one’s interlocking normative judgments at once, and ask, from nowhere, whether this set is correct or incorrect, for on a constructivist view there are no independent standards to fix an answer to this question.²⁴

Here Street seems to acknowledge that her question—“What reason do I have to endorse this entire set of normative as opposed to some other set?”—sounds forceful. It sounds like it merits an answer. But in fact, she says, it is “ill-formulated” because asking it requires us to “step back” from all standards of evaluation that might be called upon to answer it. But I don’t think this is right. Even if we are constructivists and agree that there are no mind-independent normative standards, it doesn’t follow that this is a question that can only be asked from “nowhere.” It is a question that can be asked from the point of view of another person. That is, even if my normative commitments form a coherent, interlocking whole, I can scrutinize those commitments, in toto, from the point of view of other people and their normative commitments. I can ask: what would Philip Marlowe make of this comprehensive set of normative judgments? That this question sounds forceful and doesn’t end up ill-formulated in the way Street suggests is evidence that the question is forceful—that it is a question we are required to answer as part of the process of scrutinizing our normative judgments. This reply suggests a more general argument for the unrestrictedness of reason’s scrutiny. For any set of attitudes and judgments that are internally coherent, we can ask, “but why these judgments?” This question has prima facie intuitive force: it is something that seems, in principle, to merit an answer. On the conception of reason I have developed here, the only way to provide that answer is with an appeal to a perspective beyond the ones from which these judgments are made. The intuitive force of these questions, then, is evidence of the relevance of a corresponding perspective to our reasoning. This point is usually expressed as an objection to constructivism, but it’s only an objection to those versions of constructivism, like Street’s Humeanism, that would impose a limit on which perspectives are employed ²⁴ Street (2012: 51–2).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Reason and Respect

17

in our construction procedure. The “openness” of these normative questions is perfectly compatible with a similarly open-ended constructivism.²⁵

1 . 2 . R E AS O N I N G A S A C O L L E C T I V E AC T I V I T Y In the previous section I argued that (1) reasoning is an activity of critique and scrutiny, (2) that this scrutiny involves consulting other perspectives, (3) that this consultation is second-personal in the sense that it depends on the participation of another person, and (4v) that all perspectives are relevant to this scrutiny. These claims entail that reasoning is something I do with others, lots of others. It is something that I do with the occupants of every other perspective. It is of course equally true that you need to rely on other people, including me, in your reasoning. And this reasoning is both ubiquitous and ongoing. Individual episodes of reasoning will naturally overlap with each other: if I change my mind after reasoning with x, that will affect my reasoning with y. This suggests a more economical way individuating episodes of reasoning. Rather than thinking of there being a glut of discrete episodes of reasoning—x with y and z, y with x and z, z with x and y, and so on—we can think of reasoning as involving a single, massive, ongoing collective activity in which x, y, and z are all partners.²⁶ This thesis raises some obvious questions. As we observed before, most of the perspectives I might consider in reasoning are not occupied by actual persons. And I only occasionally find it necessary to converse with the occupants of those perspectives that are occupied. Much of my reasoning goes on in perfect solitude, without any overt conversation at all. Much of it involves no thought of other people at all, but only impersonal things like principles of inference. How should we square these facts with the conception of reasoning on offer? If my proposal is correct, then much of what we commonly regard as the practice of reasoning is actually a simulation of the real thing. The cogitations we call reasoning frequently involve the imaginative projection of oneself into the point of view of another person in hopes of simulating their half of a conversation. If I want to scrutinize my judgment from Marlowe’s perspective, I do my best to act out his side of an imagined conversation. This enterprise will obviously vary in difficulty and likelihood of success. It’s fairly easy for me to simulate what me-in-two-minutes would ²⁵ I elaborate on this point in Walden (2015). ²⁶ Is Philip Marlowe a partner in this activity? Obviously not, since he doesn’t exist. Nonetheless, the activity is one that has an interest in doing the impossible, in making Philip Marlowe a partner.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

18

Kenneth Walden

think about my intention to leave the house in a rainstorm without an umbrella. It’s much harder for me to say what a person from a very different system of values would say about my proposed resolution of a moral dilemma. Sometimes this simulation will be a good one, and sometimes it won’t be. But we plainly do engage in this kind of simulation, and some of our most intense episodes of reasoning play out as imagined conversations between ourselves and a persistent interlocutor. As I said before, in principle every perspective is fair game for the scrutiny of a judgment, but in practice we can by no means have a conversation, real or simulated, with the owner of every perspective. So we employ shortcuts. One shortcut involves the formation of perspectives into representative amalgams. My future self encompasses many slightly different points of view. But when I am scrutinizing some judgment according to the demands of prudence, I am likely to suppress these differences and consider the perspective of a single representative of these future selves if I believe the differences between them are immaterial. I may also develop heuristics to stand in for conversations. I know my wife well enough to know how a conversation about my intention to buy our toddler an illustrated edition of Naked Lunch will go, so I can rely on a summary of that conversation instead of simulating it. The activities Clarkeans and Lockeans identify as the essence of reasoning can be modeled through a combination of these strategies. The obvious truths that Clarkeans say reason allows us to perceive can be understood as those things we judge to be manifest from every perspective—to be such that only the “extremest stupidity of mind, corruption of manners, or perverseness of spirit, can possibly make any man entertain the least doubt concerning them.”²⁷ Similarly, the rules that a Lockean associates with reasoning can be understood as those that are “valid” across all perspectives. These rules will codify those functions F such that if it seems that p from someone’s perspective, it will also seem that Fp from their perspective. This reconstruction would allow us to capture much of what is attractive about the Lockean and Clarkean conceptions while also doing justice to the unique authority of reason that the critical conception does so well with.

1 . 3 . C O L L E CT I V E A CT IV I T Y A N D R E S P E C T I have sketched a conception of reason, and I expect readers will recognize something pro-social, even nascently moral in the sketch. There are a few ²⁷ As himself Clarke puts it (Raphael 1991: 296).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Reason and Respect

19

different ways to make these features more explicit. Here I will do so by relying on a thesis about the structure of collective activities: participation in collective activities is conditioned on certain forms of respect for one’s partners in those activities. I cannot mount a complete defense of this thesis here, but I will try to motivate it with examples. The respect that collective activities presuppose is a kind of recognition respect.²⁸ It is a recognition of the authority our partner has in virtue of their being our partner. Suppose you and I are baking a cake together. For us to be baking together I must recognize your standing to make certain suggestions about how the activity unfolds. In many cake-baking scenarios, this will include your standing to suggest what kind of cake to bake, how to divide our labor, what order to perform certain tasks in, and what substitutions to make to the ingredients. We could even talk about your having rights as part of this activity. You may have a right to veto an almond cake because of your nut allergy. If I wander away halfway through to watch the baseball game, you have a right of rebuke.²⁹ Call the duties of respect I have in virtue of my participation in a collective activity activity-specific duties of respect. What these duties amount to will depend on the nature of activity naturally. If I am the owner of a bakery and you are my employee, then our baking a cake together may not accord you the same standing to make suggestions about what kind of cake we bake or how we divide the labor. But even in this case, it seems clear that I do have some activity-specific duties of respect. I should recognize your objections about nuts and allergies, or hear you when you say that the dough is too wet, or simply not push you away from your work area. If we imagine a superficially similar activity that lacks even these minimal forms of recognition, it is hard to say that it is a genuine example of collective activity, rather than two activities in parallel, or even an example of one person using another like a tool.³⁰ Where do these duties of respect come from? This is an important question but beyond what I say here. I am sympathetic to the view that collective activities depend on a “plural” kind of subjectivity or agency—that

²⁸ Darwall (2006: 140ff.). ²⁹ Compare Gilbert (1996 and 2014). ³⁰ One might object that in even more asymmetrical relationships—like that between slave and captor—this respect is completely absent. Slavery is a complicated and heterogenous relationship, and I think different replies are appropriate for different cases. Some slavery does feature non-negligible respect for the slave (Cicero and Tiro). Other instances do not but also lack genuinely collective activity. Still others may involve collective activity and so presuppose a kind of respect but systematically deny that respect; these are pathological but not a counterexample to my claim.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

20

Kenneth Walden

they are performed by a distinctive we—and that the duties of respect that structure those activities reflect the conditions of this plural subject’s integrity.³¹

1 .4 . R E A S O N A N D RE S P E C T We now have the first two premises of our central argument: (1) reasoning is a collective activity and (2) all collective activities entail activity-specific duties of respect. Combining these yields the claim that reason involves activity-specific duties of respect. But what are these duties? The activityspecific respect associated with, say, cake-baking is not very impressive. For one thing, it is conditional on my participation in the activity, so if I tire of respecting you as my co-cake-baker, I can simply exit that activity. For another, the duties of respect I owe you are highly circumscribed. They only concern our baking together. It seems that I can respect you as my co-baker while maintaining total insensitivity to your pain, while blackmailing you, and even while planning to murder you as soon as we’re done. Reasoning is very different from baking, though. Recall our fixed point: the authority of reason over a wide range of our judgments is unconditional. Capturing this idea was the primary motivation behind adopting the critical conception of reason in the first place. But now it is relevant in a different way. For it means that reasoning-specific duties of respect are neither conditional nor circumscribed. I cannot opt out of the reasoning-specific duties of respect because I cannot opt out of the demands of reason. And these duties of respect are relevant to every attitude that is subject to the scrutiny of reason—that is, to nearly every attitude. This is the third premise of the argument. Combining it with the first two, we have the result that I have an unconditional obligation to recognize the authority of every person to weigh in on nearly any question I might entertain. This is the sense in which respect for persons is implicit in reason itself. Establishing this conclusion is the principal ambition of the chapter. It’s a schematic conclusion, though, so we might wonder, with some skepticism, what exactly it means for us. What am I required to do in light of this universal duty of respect? One possibility is that we can derive particular obligations and prohibitions directly from the conclusion. We could do this by arguing that certain act types or species of maxim constitutively involve a failure to recognize the standing of others. In performing such acts or adopting such maxims, I am thereby refusing to acknowledge your authority ³¹ Gilbert (1989) and Korsgaard (2009: ch. 9).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Reason and Respect

21

over certain questions. I am dubious about this strategy because I doubt that there are any act or maxim types that necessarily involve a problematic refusal to recognize. More generally, I am inclined to doubt that concrete, actionguiding imperatives can be extracted from the principle I have defended. The constraints introduced are simply too vague to produce this kind of result. In this respect my ambitions are more modest than other Kantian constructivists: I do not think I have shown that “enlightenment morality is true.”³² Of course, this doesn’t mean that the duty of respect is completely empty. Its force is simply more mediated. My model for this control comes from Kant’s theoretical philosophy. There he distinguishes between constitutive and regulative principles. Constitutive principles give us specific guidance on how we must represent nature. They have authority because they codify the conditions of experience. Regulative principles prescribe ideas for us to aspire to—e.g. the idea of systematic unity—but do not issue specific commands. Regulative principles are irremediably indeterminate because their content is fixed by an open-ended process of rational self-scrutiny. The principle of respect I have defended is regulative in this sense. It offers an ideal of our attitudes being subject to a massive conversation to which every person is a party. How agents should operationalize this idea of reason will depend on further features of those agents, just as their operationalization of other ideas will depend on their specific cognitive constitution and circumstances. The “cash value” of my conclusion will therefore be a complicated question that involves reckoning with many other factors that shape a given agent’s normative landscape.³³

References Bonjour, L. (1998). In Defense of Pure Reason. New York: Cambridge University Press. Brandom, R. (2009). Reason in Philosophy: Animating Ideas. Cambridge, MA: Harvard University Press. Cavell, S. (1969). The Claim of Reason. New York: Oxford University Press. Darwall, S. (2006). The Second-Person Standpoint. Cambridge, MA: Harvard University Press. Enoch, D. (2006). “Agency, Shmagency,” Philosophical Review 115(2): 169–98.

³² Korsgaard (1996: 123). ³³ I am grateful to audiences at the 2018 Madison Metaethics Workshop and Binghamton University for helpful questions and discussion as well as to two anonymous referees for the same. I owe special thanks to Julia Markovits and Timothy Rosenkoetter for extensive comments.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

22

Kenneth Walden

Ferrero, L. (2009). “Constitutivism and the Inescapability of Agency” in R. ShaferLandau (ed), Oxford Studies in Metaethics, vol. 4. New York: Oxford University Press. Gendler, T. S. and Kovakovich, K. (2005). “Genuine Rational Fictional Emotions” in M. Kieran (ed.), Contemporary Debates in Aesthetics and the Philosophy of Art. Malden, MA: Wiley-Blackwell. Gilbert, M. (1989). On Social Facts. Princeton: Princeton University Press. Gilbert, M. (1996). “Walking Together” reprinted in her Living Together. Lanham, MD: Rowman & Littlefield. Gilbert, M. (2014). “Mutual Recognition and Some Related Phenomena” reprinted in her Joint Commitment. New York: Oxford University Press. Habermas, J. (1984). The Theory of Communicative Action. Boston: Beacon Press. Haslanger, S. (1992). “Ontology and Pragmatic Paradox,” Proceedings of the Aristotelian Society 92: 293–313. Korsgaard, C. M. (1996). The Sources of Normativity. New York: Cambridge University Press. Korsgaard, C. M. (2009). Self-Constitution: Agency, Identity, and Integrity. New York: Oxford University Press. Laden, A. (2014). Reasoning: A Social Picture. New York: Oxford University Press. Langton, R. (2007). “Objective and Unconditioned Value,” Philosophical Review 116(2): 157–85. Manne, K. (2016). “Democratizing Humeanism” in E. Lord and B. Maguire (eds), Weighing Reasons. New York: Oxford University Press. Markovits, J. (2013). Moral Reason. New York: Oxford University Press. McHugh, C. and Way, J. (2018). “What is Reasoning?” Mind 127(2): 167–96. Nagel, T. (1997). The Last Word. New York: Oxford University Press. O’Neill, O. (2000). “Four Models of Practical Reason” in her Bounds of Justice. New York: Cambridge University Press. O’Neill, O. (2015). “Constructivism in Rawls and Kant,” reprinted in her Constructing Authorities. New York: Cambridge University Press. Parfit, D. (2011). On What Matters. New York: Oxford University Press. Pinkard, T. (1996). Hegel’s Phenomenology: The Sociality of Reason. New York: Cambridge University Press. Raphael, D. D. (ed.). (1991). The British Moralists. Indianapolis: Hackett. Regan, D. (2002). “The Value of Rational Nature,” Ethics 112(2): 267–91. Ridge, M. (2005). “Why Must we Treat Humanity with Respect? Evaluating the Regress Argument,” European Journal of Analytic Philosophy 1(1): 57–73. Scanlon, T. M. (2013). Being Realistic about Reasons. New York: Oxford University Press. Schafer, K. (2015). “Realism and Constructivism in Kantian Metaethics I and II,” Philosophy Compass 10(10): 690–701. Schafer, K. (2019). “Rationality as the Capacity for Understanding,” Noûs 53(3): 639–63. Setiya, K. (2007). Reasons without Rationalism. Princeton: Princeton University Press. Setiya, K. (2014). “What is a Reason to Act?,” Philosophical Studies, 167(2): 221–35.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Reason and Respect

23

Silverstein, M. (2015). “The Shmagency Question,” Philosophical Studies, 172(5): 1127–42. Street, S. (2012). “Coming to Terms with Contingency” in J. Lenman and Y. Shemmer (eds), Constructivism in Practical Philosophy. New York: Oxford University Press. Theunissen, L. N. (2018). “Must we be Just Plain Good? On Regress Arguments for the Value of Humanity,” Ethics 128(2): 346–72. Velleman, J. D. (2009). How we Get Along. New York: Cambridge University Press. Walden, K. (2015). “The Euthyphro Dilemma,” Philosophy and Phenomenological Research 90(3): 612–39.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

2 The Phenomenal Appreciation of Reasons (Or: How Not to Be a Psychopath) M. Coetsee 2 . 1. I N T R O D U C T I O N Stealing is wrong.¹ Huck knows it’s wrong, and he believes that by helping Miss Watson’s slave Jim escape to freedom, he is stealing—and so doing something wrong. But Huck does it anyway. Going against what he explicitly believes he has decisive moral reason to do, he decides to help Jim escape to the North.² Why do we credit Huck for his deed when it constitutes a failure to respond to the moral reasons he explicitly believes he has? Huck’s choice to help Jim corresponds with what he has objective moral reason to do, but our esteem for Huck isn’t warranted if his choice only happens to accord with the moral reasons that apply to him, and isn’t also—at least in some capacity—made for the sake of those reasons.³ To earn our esteem, thus, Huck must appreciate and respond to considerations speaking in favor of helping Jim as moral reasons. But then, if Huck is worthy of our esteem, there must be some implicit way of appreciating and responding to considerations as moral reasons that does not involve explicitly believing that those considerations are moral reasons. ¹ My conception of a “psychopath” (as in the subtitle) is taken from philosophical folklore, which holds that a psychopath is someone who is by constitution utterly imperceptive of moral considerations, considered as such, and numb to their reasongiving force. This should be sharply distinguished from any actual psychiatric condition recognized by the American Psychiatric Association—e.g. anti-social personal disorder, to which it may bear little resemblance. See Jalava and Griffiths (2017). ² For discussion of Huck’s case see e.g. Arpaly (2002), Bennett (1974), and Markovits (2010). ³ I assume that for an agent to be morally praiseworthy for φ-ing, her φ-ing must at least be traceable to a prior appreciation of moral reasons.

M. Coetsee, The Phenomenal Appreciation of Reasons: (Or: How Not to Be a Psychopath) In: Oxford Studies in Metaethics Volume 15. Edited by: Russ Shafer-Landau, Oxford University Press (2020). © M. Coetsee. DOI: 10.1093/oso/9780198859512.003.0002

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

The Phenomenal Appreciation of Reasons

25

In this chapter I argue that an agent like Huck can implicitly appreciate a consideration as a moral reason to φ without explicitly believing it to be a moral reason by presenting it under the light of a particular phenomenologically-mediated mode of presentation: one that presents the relevant consideration via the light of a felt directive force “pointing” towards φ-ing—lending weight to it, or soliciting it—in a particular way. Thus, I suggest, Huck may be understood on analogy with a young jazz piano virtuoso. She may appreciate that the G-seventh chord having been played just so constitutes an aesthetic reason for her to ease into the C-majorseventh chord just so—despite lacking explicit beliefs about what she has aesthetic reason to do—by virtue of experiencing the former as pointing or directing her to the latter. Like her, I propose, Huck may appreciate the considerations speaking in favor of helping Jim as moral reasons to help Jim—despite lacking explicit beliefs about what he has moral reason to do— by virtue of experiencing them as pointing or directing him to help Jim. My argument proceeds as follows. In the first two sections of the chapter I consider four major approaches to explaining implicit moral reasonsappreciation and outline difficulties faced by each. In the third section, I propose my own—more intuitively plausible, I argue—account of implicit moral reasons-appreciation. In the fourth section, I suggest that my account of implicit moral reasons-appreciation may also provide insight into the nature of ordinary moral reasons-appreciation. In general, I treat moral reasons-appreciation as that kind of (non-factive) representation of considerations as moral reasons that is—at least in non-derivative cases—relevant to moral agency and evaluations thereof.⁴ My main aim is not to give a definitive, differentiating description of the phenomenology relevant to such appreciation, but rather to show that there is a critical phenomenal component to it that bears greater investigation than it has so far received.⁵

2 .2 . D E R E A C C O U N T S O F I M P L I CI T A P P R E C I A T I O N One natural way to try to accommodate cases of implicit appreciation of moral reasons like Huck’s is to lower the cognitive bar on moral reasonsappreciation: to do away with representation of moral reasons altogether and allow mere de re contact with moral reasons to underwrite appreciation. Arpaly (2002), for instance, claims that an agent is creditably responsive to ⁴ See the conclusion for more discussion of how my proposal relates to issues of agency. ⁵ See n. 18 for a discussion of related proposals, including proposals that link moral reasons-appreciation to emotions.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

26

M. Coetsee

moral reasons if (roughly) she responds to the genuinely moral reason-giving features of the world out of an intrinsic de re desire that the course of action that has those features be taken.⁶ Along similar lines, Dreyfus and Dreyfus (1991) claim that exemplar moral agents comport themselves with the moral reason-giving features of their circumstances in a direct and spontaneous way, without the baggage of intermediary de dicto representations of those features as reason-giving. Drawing on these discussions, we might propose: (1) DE RE: An agent implicitly appreciates a consideration c as a moral reason r to φ iff c is a moral reason to φ and her intrinsic concern about c leads her to desire that φ-ing be done.

Unfortunately, this proposal won’t work: for DE RE can’t distinguish the perspective of an implicit moral agent on her reasons from that of, say, a lucky psychopath. Consider Juliet and her psychopathic cousin Silas: Juliet: Juliet’s family is caught in a bitter rivalry with another family, characterized by repeated cycles of violent revenge that are prodded by the premium that her society puts on machismo and social status. Juliet sees her cousin Tybalt punch Benvolio, a member of the other family on her family’s “turf,” right in the eye. She sees that Benvolio is badly injured and realizes he’s in great pain. Uneasy and disturbed about Benvolio’s being in pain, she tries to coax Tybalt to stop. “Why? What’s wrong with you?” Tybalt asks, “Don’t you care about our family’s honor?” Juliet is not sure what to say. “No, I care,” she sighs, “I guess I’m just a softie who can’t bear seeing people in pain.”

We can imagine that Juliet implicitly appreciates Benvolio’s pain as a moral reason for Tybalt to stop, even though she lacks the conceptual resources to explicitly characterize that consideration as a moral reason. Now consider, in contrast, her psychopathic cousin Silas: Silas (De Re): Silas is Juliet’s fussy psychopathic cousin. Silas sees Tybalt punch Benvolio and realizes that he is in great pain. Silas becomes very concerned about it, since Silas has a fetishistic preoccupation with pain. This preoccupation is of such a nature that it leads him to desire to stop the pain. What it is like for Silas is what it would be like for one of us if we developed a fetishistic preoccupation with purple balloons, and this preoccupation was such as to lead us to desire to pop purple balloons whenever we saw them.

Here, there is a consideration c (Benvolio’s pain), c is a moral reason to φ (to try to stop Tybalt), and Silas’s intrinsic concern about c leads him to desire to φ. So, if a de re account of implicit moral reasons-appreciation is right ⁶ See also Smith (1994: 71–6).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

The Phenomenal Appreciation of Reasons

27

then Silas and Juliet should equally appreciate Benvolio’s pain as a moral reason to try to stop Tybalt. However, this is clearly not the right result. The problem is that though Silas is intrinsically concerned about Benvolio’s pain, that concern is of the wrong sort: it does not construe Benvolio’s pain as being bad—perhaps at all—and definitely not in the way relevant to appreciating Benvolio’s pain as a moral reason to φ.⁷ Nothing pertaining to the moral weight of painfulness makes any difference to Silas’s perspective on pain, and so there’s nothing about the moral badness of pain that affects Silas’s motivation to φ. As a result, a de re account of implicit moral reasonsappreciation is inadequate. 2. 3. D E D I C T O A C C O U N T S O F I M P L I C I T APPRECIATION We need to say something about the kind of perspective that Juliet has on Benvolio’s pain that a lucky psychopath like Silas does not. To distinguish Juliet’s perspective from Silas’s, however, it seems we need to say something about how she de dicto construes Benvolio’s pain. Moral theorists excessively focused on reflective deliberation have tended to assume that the kind of de dicto representation of moral reasons that occurs in reflective deliberation is the only kind of de dicto representation of moral reasons there is.⁸ If one makes this assumption, then it’s natural to think that one must dispense with de dicto representations of moral reasons to make sense of Juliet’s (or Huck’s) case and opt for a DE RE account of implicit appreciation instead. But the kind of de dicto representation that occurs in reflective deliberation is not the only kind of de dicto representation that there is. Below I consider and reject three alternative de dicto accounts of representation before proposing my own account.

2.3.1. A Discursive Account of Implicit Appreciation Let’s look more closely at the kind of de dicto representation of moral reasons that implicit agents apparently lack: Dez’s Deliberation: Dez receives a request from a student for extra credit. She’s not inclined to grant it at first, but remembers that she ⁷ If one substitutes Benvolio’s pain with the concepts of the correct moral theory (e.g. MAXIMIZE HAPPINESS), this point also provides an objection against Arpaly and Schroeder (2013). ⁸ Markovits (2010) and Arpaly (2002) motivate their accounts by denying that an agent must believe that φ-ing is right to be creditable for φ-ing, and seem to assume that belief requires discursive representation.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

28

M. Coetsee recently allowed another student to do extra credit. She realizes that both students are in the same class, failing, and now asking for the same opportunity. Given the parity of the cases, she realizes that considerations of fairness apply. Given that they do, she infers that she has moral reason to grant the request.

Dez presents the relevant considerations of parity as together constituting a moral reason to grant the extra credit request by making use of an inferentialist grasp she has of a moral reason-giving concept: fairness. That is, Dez is able to pick out these considerations as morally salient because she knows that (i) they constitute an application condition of the concept of fairness and (ii) the concept of fairness, once applied, supports inferences about moral reasons. I’ll say that Dez’s inferentialist grasp of her reasons underwrites discursive appreciation of those reasons, since it allows her to talk herself through her decision in a deliberative, step-by-step way, and also to articulate an account of her reasons to others. Applying a discursive account of appreciation to the implicit case, we get: (2) DISCURSIVE: An agent implicitly appreciates a consideration c as a moral reason r iff her representation of c plays the inferentialist role it is characteristic of moral reasons-representations to play— underwriting inferences from salient application conditions to conclusions about (say) moral obligation—and she is able to articulate these inferences (at least with Socratic prodding).⁹

DISCURSIVE has more resources for explaining implicit moral reasonsappreciation than it may initially appear. As Dez continues to teach, she will become more adept in her assessments of fairness, and she may ultimately be able to bypass deliberation altogether: she may eventually learn to intuit or “see” what to do in new cases without formulating any discursive thoughts about what her reasons are. If she does this, her old inferentialistically characterized concept of fairness will not thereby necessarily have been made obsolete. It may still be fundamentally responsible for how she sorts through information and draws conclusions, perhaps operating via the influence of cognitive penetration on her experience. If this is indeed the way that Dez’s appreciation of her reasons develops, it will be evidenced by her continued and non-confabulatory ability to articulate an account of her reasons when called on to do so. Many of those interested in understanding non-deliberative modes of reasons-responsiveness have emphasized cases like Dez’s—cases of articulate skilled agency, where agents “see” what to do without first needing to ⁹ See Jackson and Pettit (1995) and, for a related account, Wedgwood (2007). See also Peacocke (1992).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

The Phenomenal Appreciation of Reasons

29

explicitly reason about it, but are nevertheless also poised to answer Anscombean “what” and “why” questions about their actions.¹⁰ However, cases of implicit moral reasons-appreciation like Huck and Juliet’s can’t be explained in the same way as such cases of articulate skilled agency. Unlike the skilled Dez, agents like Huck and Juliet don’t have a tacitly operative inferentialist grasp of their reasons for φ-ing. Huck, for instance, does not have an inferentialistically characterized grasp of the concept of (say) free and equal moral agency, and so can’t articulate even a post facto account of his φ-ings that adverts to it.

2.3.2. A Causal Tracking Account of Implicit Appreciation Philosophers of mind have conventionally offered two theories of mental representation that may help account for agents’ non-discursive, but still de dicto, appreciation of moral reasons: causal theories and functional theories. According to causal theories of representation, a mental representation m represents p by virtue of being a token of a type of representation that stands in a suitably reliable causal tracking relation with p-types of things.¹¹ Drawing on a causal theory of representation, we might propose: (3) CAUSAL TRACKING: An agent implicitly appreciates a consideration c as a moral reason r to φ iff her representation of c is of a type that causally tracks moral reasons for φ-ing.¹²

Causal tracking may be necessary for implicit moral reasons-appreciation, but it does not also appear to be sufficient. This is because it seems there will always be deviant ways of tracking relevant considerations. In particular, so long as moral reasons supervene on natural properties—or, for that matter, on any subvening base properties—it seems it will always be possible for an agent to track those properties without taking any notice of their practicalmoral import. Consider again a lucky psychopath like Silas: Silas (Causal Tracking): Silas sees Tybalt deal Benvolio a blow to the eye and realizes he’s in great pain. The pain is represented under the aspect of the same type of mental representation that, because of (say) the fortuitous forces of species evolution, is also deployed when Silas sees a murder, theft, assault, etc. Like in other cases where that ¹⁰ See Annas (2011: 20) and Railton (2009). ¹¹ Thus, the rings on a tree trunk might represent the years a tree has lived because the former reliably causally “track” the latter. See Dretske (1988), Fodor (1990), and Millikan (2004). ¹² Railton (2014: 816, 829–3) and Allman and Woodward (2008) partially rely on CAUSAL TRACKING.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

30

M. Coetsee representation is deployed, Silas is weirded out by the represented target. He tries to coax Tybalt to stop.¹³

If causal tracking is sufficient for moral reasons-appreciation, then Silas appreciates his moral reasons just as well as Juliet does because he, like her, represents Benvolio’s pain via a moral reasons-tracking representation. But clearly, Silas does not appreciate Benvolio’s pain as a moral reason as Juliet does. So, CAUSAL TRACKING is insufficient to explain implicit moral reasons-appreciation.

2.3.3. A Functional Role Account of Implicit Appreciation Another way to try to distinguish Silas from Juliet is by drawing on a functional theory of mental representation. According to this theory, a mental representation m represents p by virtue of being a token of a type of representation that plays a certain internal causal role in a subject’s cognitive economy—that is, whatever causal role it is characteristic of prelated representations to play.¹⁴ Drawing on this theory to give an account of implicit moral reasons-appreciation, we might say: (4) FUNCTIONAL ROLE: An agent implicitly appreciates a consideration c as a moral reason r to φ iff her representation of c plays the internal causal role that it is characteristic of moral reasonrepresentations to play.¹⁵

Though some argue that phenomenology itself can be understood in functional terms, functionalist theories of representation typically avoid appeal to phenomenology. Given that they do, the possibility of a spectrum inversion prevents them from adequately distinguishing Silas from Juliet. Take your favorite dispositional profile for moral reasons-representations and match it with the wrong phenomenal feel; I submit that you will get a state that intuitively fails to constitute an instance of moral reasonsappreciation. Silas (Functional a): When Silas represents Benvolio’s pain, he experiences—associated with that representation—a feeling of itchiness. The itchy feel of the representation prompts Silas to try to stop ¹³ This draws on Millikan’s account. ¹⁴ Thus, the mermaid Ariel represents a fork as a hairbrush because her representation of the fork plays the functional role that hairbrush-representations typically play—causing her to (e.g.) brush her hair with the fork and to think of the fork when she wishes to untangle her hair. See Block (1986) and Harman (1973). ¹⁵ See e.g. Blackburn (1998: 57–8), Gibbard (1992: 75), and Arpaly and Schroeder (2013). I discuss non-cognitivism further in the last section.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

The Phenomenal Appreciation of Reasons

31

the pain (which he senses will relieve the itch), and so Silas tells Tybalt to stop. Silas has the same experience with other representations of people being in pain: they feel itchy, and cause him to take steps to relieve the pain.

We can suppose that Silas’s representation of Benvolio’s pain plays the same functional role as Juliet’s: Silas rushes equally quickly to aid Benvolio, attends to all the same features as she does, displays the relevant punishing behaviors towards Tybalt, and so on. Indeed, we can “wire” into Silas, at birth, all the dispositions he needs to duplicate the causal-functional profile of a typical moral agent, but then invert the phenomenal feels normally associated with these dispositions: Silas (Functional b): The dispositions associated with our representation of a moral reason not to φ, instead of being harnessed to feelings of indignation and the like, are harnessed to feelings of unseemliness and being weirded-out. The dispositions associated with our representation of a moral reason to φ are harnessed to feelings of arousal and titillation rather than to feelings of moral admiration and the like.

I grant that FUNCTIONAL ROLE may capture a necessary condition on implicit moral reasons-appreciation. But if it also offers a sufficient condition on such appreciation, then Silas (Functional a and b) and Juliet will equally well appreciate their moral reasons. Clearly, however, this is not the right result, and so a functionalist account of implicit moral reasonsappreciation is inadequate. Like the de re account and the causal account of implicit appreciation, it fails to differentiate lucky psychopaths from genuine moral agents.

2 .4 . A P HE N O M E N O L O G I C AL A C C O U N T O F I M P LI CI T A P P REC IA T IO N

2.4.1. A Cognitive Phenomenology Account of Implicit Appreciation Functionalist and causal tracking theories of representation avoid appeal to phenomenology in part because the computational theory of mind that originally motivated them aimed to disentangle investigation into the mind’s operations from discussions about the nature of consciousness.¹⁶ Recently, however, philosophers of mind have taken renewed interest in the role of phenomenology in mental representation. Some have proposed ¹⁶ Fodor (1987).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

32

M. Coetsee

a cognitive phenomenology account of representation, according to which a representation m represents p by virtue of being a token of a type of representation that is associated with a distinctive type of phenomenal feel q.¹⁷ Whatever its merits as a general theory of representation, I suggest that this account can help specify a necessary condition on implicit moral reasons-appreciation: (5) PHENOMENOLOGY: An agent implicitly appreciates a consideration c as a moral reason r to φ only if her representation of c is (at least in paradigmatic and non-derivative cases) associated with phenomenal feel qM.

I’ll assume a weak version of PHENOMENOLOGY on which (i) qM may be either sui generis or reducible to a “cocktail” of other phenomenal feels (sensory, affective, etc.), and (ii) not every instance of appreciative representation must be accompanied by qM. As a phenomenologically-laden presentation of red might establish a representational kind that can be re-tokened—perhaps via anaphoric reference—without the agent’s having to re-experience the relevant phenomenology with each re-tokening, so also, I’ll assume, a phenomenologically laden instance of implicit appreciation may establish a representational kind that can be re-tokened without the agent’s each time having to re-experience that phenomenology. We’ve already seen preliminary support for PHENOMENOLOGY. We’ve seen Silas entertain Benvolio’s pain under the aspect of a feeling of a fetishistic preoccupation, itchiness, and an unseemliness that weirds him out, and it seems that part of what is “off” about Silas’s representation of Benvolio’s pain in each of these cases is the phenomenology associated with it. If that’s right, there’s already a weak version of PHENOMENOLOGY that is true, in which qM refers to some restricted range of—perhaps disjunctive—phenomenological feels. Ideally, however, we’d be able to find some more unified set of phenomenal feels for qM to refer to. To do that, it helps to start by considering cases of moral reasons-appreciation that are more familiar to our imagination than the cases of implicit appreciation we’ve focused on so far. Once we’ve used these cases to focus attention on the relevant phenomenology, we can return to assess its applicability to the implicit case. To begin, consider Vegetarian: Vegetarian: You’re at the grocery store and are excited to see your favorite cut of steak on sale. In the shelf next to it is a display of tofu.

¹⁷ See Horgan and Tienson (2002), Horgan and Graham (2013), Pitt (2004), and Bayne and Montague, (2011).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

The Phenomenal Appreciation of Reasons

33

You love the taste of meat and don’t have any appetite for tofu, but you’ve recently learned about the agricultural industry’s inhumane practices towards animals. Upon considering the inhumane practices you’ve learned about, you now feel compelled to pass up the steak and reach for the tofu instead, since you realize there are excellent moral reasons to do so.

When you realize there is a moral reason to buy the tofu rather than the steak, you feel a force directed at buying tofu, lending weight to it and soliciting you to do it, in a particular authoritative way—even despite your lack of an appetite for tofu. This force is anchored in the consideration you appreciate as a moral reason: the inhumane treatment of animals presents itself with an authoritative directive force “pointing” you to purchase non-meat products. Similar phenomenology is also present in other paradigm cases of appreciation of moral reasons. You experience, say, the consideration that your elderly neighbor is lonely and loves baseball as lending weight to taking her to the game; you see someone being harassed on the street and experience a felt demand to step in; or perhaps you consider the categorical imperative, and feel solicited to be honest in your tax papers.¹⁸ This experience of authoritative solicitation appears to be a good candidate for qM, since it seems to be able to distinguish Juliet’s perspective on Benvolio’s pain from Silas’s. QM plausibly inheres in Juliet’s experience when she sees Tybalt injuring Benvolio: when she sees that Benvolio is in pain, it plausibly strikes her as authoritatively calling out for, or directing her, to try

¹⁸ Mandelbaum (1955), Hampton (1998: ch. 3), Horgan and Timmons (2005, 2008, n.d.), and Kriegel (2008) offer important discussions of moral phenomenology, but haven’t argued for its crucial importance vis-à-vis other representational modes as a basis for either implicit or ordinary moral reasons-appreciation. Some philosophers also argue that emotions can represent evaluative properties (see Deonna and Teroni (2011) for an overview). But their theories aren’t yet adequate to support an account of moral reasons-appreciation. First, their theories are concerned with the representation of evaluative properties, generally, and so face the attitude-specification problem that I discuss in section 2.5.1. Second, it seems that an impassible God or a Spock-like, stoic Kantian may be able to appreciate moral reasons, even if she does not experience emotions. While an emotions-based account of moral reasons-appreciation would seem to prematurely close off this possibility, my account allows for it, since both figures are still phenomenally conscious. Third, prominent psychological accounts of emotion characterize them as involving motivation or “action-readiness” (Frijda 2007), but some may reject the idea that moral reasons-appreciation necessarily involves motivational components—see e.g. the evil demon case in section 2.5.2. Finally, what an “emotion” is—and how it relates to moods, feelings, and other affects—is not yet well-defined; indeed some psychologists hold that phenomenological experience is not really intrinsic to emotion (Adolphs and Andler 2018). Given these problems with an emotions-based account of moral reasonsappreciation, it is safer to hold to my phenomenological account.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

34

M. Coetsee

to stop Tybalt. On the other hand, Silas seems to lack that phenomenology. When Silas experiences Benvolio’s pain as itchy, unseemly, or the target of a fetishistic preoccupation, he may feel incited, pushed, or “triggered” by that pain to stop Tybalt, but that is distinct from Juliet’s experience of the pain as summoning her via an authoritative, governing mandate to step in.¹⁹ This understanding of qM also helps make sense of Huck’s case. Huck’s decision to aid Jim is plausibly not based on considerations that have impressed him as neutral stimuli, but on a kind of visceral impression of (something like) Jim’s equal moral personhood, that “calls out” in a governing way for him to support Jim in his journey to freedom.²⁰ Since we in the analytic tradition are not accustomed to reflecting on subtle differences in phenomenology, it may help strengthen our grasp of qM—and also reinforce its applicability to the implicit case—to situate it within a wider spectrum of non-discursive, experiential representations. QM, I’ll now suggest, is one among a variety of such representations that have modal contents.

2.4.2. A Non-Discursive Experiential Representation? To begin, consider our representational capacities as they pertain to a broad class of possibilities concerning how physical states of affairs in the world are, might, or will be. When I see a log teetering at the edge of a cliff-side, I can see that it is precariously perched, and so might possibly fall; then I see, as it teeters, that it is almost certainly going to fall, and then that it is now about to and is definitely going to fall. Here, I experience what we might categorize as nomic modalities. These are experiences as of nomic realities that we don’t need sophisticated discursively articulable concepts to represent. Thus, a small child who lacks an inferentialist grasp of the concept of

¹⁹ These phenomenal qualities aren’t just arbitrarily paired with contents; instead, they plausibly help account for the relevant experience having the representational content it does. Consider the difference between (i) judging that it is not good that you feel itchy (since it is distracting) and yet feeling itchy, and (ii) judging (perhaps by rationalizing) that you aren’t morally required to give food to someone begging on the street, and yet experiencing a felt mandate to give him charity. There is plausibly a kind of cognitive dissonance that occurs in (ii) but not in (i). Like when you look at a Müller-Lyer illusion and it appears to you that the lines are different lengths, but you judge that they are not, in (ii) you experience a cognitive dissonance between how the world is experientially presented to you and your considered judgment. This is evidence that the felt mandate has representational content that the itch doesn’t. ²⁰ Arpaly suggests that there’s a distinct quality to Huck’s moral vision (2002: 76). In doing so, she seems to anticipate my account.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

The Phenomenal Appreciation of Reasons

35

nomic possibility can nevertheless see that the log might fall or now definitely will fall, and so on.²¹ A second set of modalities that it seems we can non-discursively, experientially represent are dynamic modalities—that is, modalities involving possibilities and necessities that involve us as agents. Here, on the side of possibility, you can (for instance) look down a path, see obstructions, but also see that the path can be traversed. Following the work of J. J. Gibson (1986), psychologists have termed this kind of representation of a possibility for agency in an environment an “affordance.” Thus, they say, a chair affords sitting, a door affords opening, and so on. In addition to experientially presenting φ-ing as something that (dynamically) can possibly be done, we can also experientially present φ-ing as something that (dynamically) may well be done or must be done.²² Thus, someone who feels an inclination to get fresh air may experience a door as something that may well be opened, whereas a claustrophobic person who feels a compulsion to get out of the room may experientially represent the door as something that must be opened. As before, experiential representations of these dynamic modalities don’t seem to require discursive competence on the part of the representing subject. Plausibly, a young claustrophobic child can present the door as something that must be opened, for instance, without having the inferentialist conceptual competences to talk about the dynamic necessity of opening doors. In these cases more so than in the nomic ones, the distinctive phenomenology of the experiential representation seems to contribute to determining its contents. For instance, it seems to be by virtue of the kind of phenomenal feel that a compulsion has—a feeling of being propelled or “pushed” to φ—that φ-ing is represented as something that has to be done.²³ In addition to having experiences that present it as being the case that something (nomically) might or will happen, or that something (dynamically) can or has to be done, we also have experiences that present it to be the case that something normatively may be done or is required to be done. Here we have experiential representations on the side of possibility in (say) cases

²¹ Phenomenology may or may not help determine these experiences’ representational contents. One reason to think it might is that there seems to be a phenomenal contrast between walking through solid granite cliff faces—where nothing is going anywhere— and (say) walking through the world of Dali’s paintings—where anything could happen. ²² Siegel (2014)’s discussion of “felt mandates” doesn’t distinguish dynamic from deontic experiential representations. ²³ If you feel a compulsion to φ but judge that φ-ing isn’t necessary, you experience cognitive dissonance in a way that you plausibly don’t with, say, an accompanying feeling of nausea. This is evidence that the compulsion has representational content that nausea does not. See also n. 19.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

36

M. Coetsee

of arbitrary choice. Consider a case where two paths diverge in a wood, and you see what appears to be a nice shady place to rest under a tree down one path. Your representation of this consideration may have a feeling of felt relevance associated with it as you consider whether you might want to sit. But then, when you decide that there’s no reason to sit down right now, that felt relevance comes to naught: it provides no soliciting force “pointing” or “directing” you to go down that path rather than the other. Contrast this case with cases where we have a stronger experience of deontic necessity. You see someone being harassed on the street and have an experience as of being mandated to step in; or an artist, coming close to the end of her performance, feels compelled by the music—aesthetically demanded—to perform a grand finale of this sort. These considerations— that someone is being harassed, or the flow of the music up to this point— don’t merely trigger a felt propulsion or “push” to φ (though they may also have the effect of stimulating such a feeling), rather, they summon you to φ by virtue of a kind of commanding authority. Other cases fall between these experiences of arbitrary choice and deontic necessity. Thus, I may experience the consideration that (say) my elderly neighbor is lonely and loves baseball as having a felt relevance to the question of whether to bring her to a baseball game, and as lending weight to an affirmative response to that question, without also experiencing it as mandating me to take her to the game. Whether a consideration is experienced as merely lending weight to φ-ing in this way or otherwise as generating a felt demand to φ, however, the experience of being “called to” φ is still marked by the sense of being under the jurisdiction of a governing authority. Thus it is that I “feel the force” of the consideration as a reason to φ.²⁴ As in the nomic and dynamic cases, an agent doesn’t need a discursive grasp of relevant deontic concepts to experientially represent deontic contents. Consider Jasmine: Jasmine: Jasmine is a young jazz piano virtuoso. Listening intently to the flow of the music, she hears the particular way in which a Gseventh chord is played by the band. Jasmine doesn’t know any formal jazz theory and doesn’t have any discursive grasp of the concept of an “aesthetic” reason, but when she hears that chord, she experiences it as having a kind of directive force “pointing to” and soliciting her to ease into a C-major-seventh chord just so.

²⁴ Epistemic modality may also be experientially presented. Consider Locke’s discussion of “evident luster” and Descartes’s discussions about “clarity” and “distinctness,” referenced in Plantinga (1993: 57–9). See also Conee and Feldman (2004: 64–7).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

The Phenomenal Appreciation of Reasons

37

Jasmine plausibly appreciates the consideration that the G-seventh chord was played just so, as an aesthetic reason to ease into the C-major-seventh chord just so via her experiential representation of the former as directing or soliciting her towards the latter, without employing (or being able to employ) any discursive representation of the former as an aesthetic reason for the latter. So also, it’s plausible that a consideration can strike an agent like Huck or Juliet as a moral reason to φ via their experiential representation of that consideration as directing or soliciting them to φ, even though they don’t (or can’t) employ any discursive representation of that consideration as a moral reason to φ.

2.4.3. A Distinctive Moral Phenomenology? We now have a fuller framework for distinguishing Juliet’s experiential presentation of Benvolio’s pain from Silas’s. While Juliet’s presentation of Benvolio’s pain via the phenomenological aspect of a felt mandate falls in the range of non-discursive presentations of deontic reasons, Silas’s presentation of Benvolio’s pain via the phenomenological aspect of (say) a fetishistic preoccupation involves a feeling of being incited, propelled, or driven to φ, and so plausibly falls in the spectrum of non-discursive presentations of dynamic facts. But what, if anything, distinguishes Juliet’s experiential presentation of Benvolio’s pain as a moral reason, from the experiential presentation of that pain as some other kind of deontic reason—say, an aesthetic or prudential one? Phenomenal qualities are by nature difficult to describe, and they become even more so as the contrasts between them become more subtle. Even when descriptions are elusive, though, we can discern phenomenal contrasts by “pointing” to illustrative examples. I’ll focus for now on this illustrative approach, and offer only a preliminary description of the relevant moral phenomenology. I leave a more detailed description for a later study. Consider two pious religious parents who own an artefact they regard as sacred. They take its sacred status to generate certain moral demands—for instance, it must be put in a place of honor, on a certain pedestal in the living room. The parents have a son, Sid, who has learned to exactly mimic the patterns of treatment they take to be morally required. When he countenances the artefact, however, he does not share their inner experience of it. At first, Sid’s patterns of behavior with respect to the artefact—for instance, his careful placement of the artefact on its designated pedestal— feel like rote and wholly unmotivated habits. He is numb to the force of any reasons to treat it with special regard. But now consider three different ways

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

38

M. Coetsee

in which he may come to appreciate a reason to φp: move the artefact to its pedestal from where it has mistakenly tumbled to the floor. Practical Purpose: After a long day, Sid collapses onto the comfortable living room sofa. He notices the artefact on the floor, and realizes that if his parents walk in the room and see the artefact there, they will chide and criticize him for not picking it up. Though Sid feels sluggish and is not otherwise inclined to get up from his comfortable place on the sofa, he experiences φp-ing as fitting or suited to the attainment of his desire to avoid this result. Cognizant of this fit, he feels directed to φp. Aesthetic Completion: Years later, Sid returns home after a long day of taking art classes at college. He collapses onto the sofa, but then notices the artefact on the floor. Gazing at the artefact, he considers how it might look up on the pedestal. He is struck by how the placement of the artefact on the pedestal would “complete” the picture, aesthetically—supporting a delicate, unified coherence of presentation that is now lacking. Though Sid feels sluggish and is not otherwise inclined to get up from his comfortable place on the sofa, he now also experiences a sense of being directed to effect that perfect completion.²⁵

In Practical Purpose and Aesthetic Completion, Sid shifts from being numb to reasons for φp-ing to appreciating two different kinds of deontic reasons to φp, and it seems that this shift is attended with a shift in the phenomenal quality of his experience: what it is like to feel compelled by the “fit” of aesthetic completion is different from what it was like to feel compelled by the “fit” of means well-suited to his ends. Now imagine a final, moral transformation: Moral Appreciation: As an adult, Sid undergoes a religious conversion. After a long day at work, Sid collapses onto the sofa. But then he notices the artefact on the floor. Contemplating the artefact, he now feels the weight of the special kind of moral authority its sacred status exerts. Though Sid feels sluggish and is not otherwise inclined to get up from his comfortable place on the sofa, he now also experiences a sense of being solemnly directed to place the artefact in its deserved place on the pedestal.

However we describe the details of the adult Sid’s experience of appreciating a moral reason to φp, it is plausibly distinct from his experiences as a youngster, of appreciating aesthetic or practical reasons to φp. Feeling directed by the “fit” of moral rightness is new to him in this context; his ²⁵ See Beardsley (1958: 527–8).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

The Phenomenal Appreciation of Reasons

39

previous experiences of feeling directed by the “fit” of means well-suited to an end and the “fit” of aesthetic completion did not acquaint him with what it’s like to appreciate the moral reasons to φp. Sid’s example is especially useful because it’s not difficult to imagine the phenomenal quality of his experience of the artefact changing even while his other, non-phenomenological dispositions stay more or less constant from the original rote habits he picked up as a kid. His distinct experience of the artefact as imparting moral—verses aesthetic or prudential—reasons, however, is also present in other cases of moral reasons-appreciation. Consider, for instance, three people at a bar who feel directed to keep their distance from a stranger—but in different ways: one anticipating the practical repercussions of a negative reaction, another in awe of the stranger’s pristine aesthetic beauty, and a third who feels the governing force exerted by the moral import of her dignity. Or three hikers who gaze out over a landscape of rolling hills of forest—one, a conqueror, entertaining it as a means wellsuited to her economic ends, another in wonder of its natural beauty, and a third apprehending it as a living community that compels her respect.²⁶ In each case, the person who appreciates the relevant intentional object as affording moral reasons plausibly has an experiential encounter with it that the others miss out on. Indeed, that person may reasonably complain that the others’ experiential perspectives on the object fail to do justice to the moral force of the reasons that apply. The same lesson also seems to pertain to Juliet and Silas. Suppose Silas is a prudent politician, who encounters Benvolio’s pain as ill-suited to the end of procuring his vote, or a pure aesthete—an overly avid actor who sees the whole world as a stage and experiences Benvolio’s pain as a tasteless distraction to the “flow” of the dramatic narrative. Even if Silas’s non-phenomenal dispositions somehow manage to fortuitously mirror those of Juliet’s, the quality of his experience of Benvolio’s pain will still be worryingly “off.” Juliet could rightly complain that he fails to “pick up on” what really matters, in virtue of the felt quality of his experiential encounter with Benvolio’s pain. In everyday experience, phenomenology is often not as well compartmentalized as it is in these examples—and even when it is, we may not be paying close enough attention to “register” its distinctive qualities. But these examples help illuminate contrasts between the phenomenology of appreciating a consideration as a moral reason, verses a prudential or aesthetic one. After all, there doesn’t seem to be just one generic experience of deontic reasons that all the agents in all our scenarios have—so that what could differentiate them is only (for instance) distinct dispositional profiles that are

²⁶ See Leopold (1970: 204).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

40

M. Coetsee

arbitrarily paired with one generic kind of deontic experience. If anything, it seems that it is variations in the felt quality of the agents’ experience that would motivate differences in dispositions downstream, by way of underwriting the appreciation of different kinds of reasons. Supposing that the above examples successfully highlight a contrast between the phenomenology of moral reasons-appreciation and the appreciation of other deontic reasons, what preliminary description of the former might we offer? To begin, it’s worth noting that the phenomenology of moral reasons-appreciation isn’t necessarily distinguished by how vivid or intense the experience of the felt demand is. A committed aesthete may feel the force of aesthetic reasons more vividly than he feels the force of moral reasons. Additionally, the phenomenology of moral reasons-appreciation isn’t necessarily distinguished by the felt priority of relevant demands. The weight of the moral reasons speaking in favor of taking my elderly neighbor to a baseball game, for instance, may not be experienced as overriding my prudential reasons to (say) complete my errands or my aesthetic reasons to (say) finish playing through the orchestral masterpiece I’ve started. Instead, what seems distinctive of the experience of a moral reason is the peculiar kind of felt independence of the authority the reason feels to exert. This felt independence is a feature of moral experience that both cognitivists and non-cognitivists have noticed and tried to accommodate²⁷—but it is difficult to specify just what it consists in. Prudential and aesthetic reasons of course also involve some degree of felt independence. They also feel to govern over me, and so aren’t experienced as being entirely at my beck and call. Still, in the case of my appreciation of a consideration as a moral reason to φ, the source of the demand to φ seems to be anchored “outside” of me in some more profound and robust way than in the practical and aesthetic case. I experience the bounds of these latter reasons’ jurisdiction over me as being more actively constrained by my idiosyncratic desires and sensibilities than in the moral case.²⁸ When I experience there as being a prudential reason to (say) take my car to the mechanic today—because this is well-suited to my end of enjoying a long drive tomorrow—I experience the authority of this prudential reason as in some respect resting on, or drawing its normative power from, my own higher-order desire to go for a long drive tomorrow. ²⁷ While some cognitivists claim that this felt independence is identical with the experience of being subject to a mind-independent moral order, others—including sensibility theorists and non-cognitivists—insist that it is compatible with the kind of “independent” authority of moral reasons that comes from those reasons being identified with the opinions or attitudes of the best or most decent sensibility. See Kirchin (2003), Horgan and Timmons (2008), Blackburn (1993: 298–304), and Gibbard (1992: 171–88). ²⁸ This doesn’t necessarily imply that they are more constrained.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

The Phenomenal Appreciation of Reasons

41

Since the authority of the prudential reason is experienced as in this respect dependent on my higher-order desire to go for a long drive, it also feels that if I no longer desired to go for a long drive, I could gut that practical reason of its binding force. Not so for moral reasons. The authority of the moral reasons to take my neighbor to the baseball game isn’t experienced as drawing its normative power from, and so being dependent on, my higher-order desires. As a result, it also doesn’t feel that I could gut the moral reasons to take my neighbor to the baseball game of their reasongiving force just by (say) changing my desire to be kind. Analogous considerations apply in the aesthetic case. When I experience there as being an aesthetic reason to (say) finish my orchestral masterpiece on the right note, I do not experience the authority of that reason as depending directly on my desires, as in the prudential case; instead, I experience the aesthetic reason to finish on the right note as proceeding from the internal norms of coherence and completeness exerted by the piece. Nevertheless, I plausibly experience the authority of these norms of coherence and completeness as itself being dependent on some deeper resonance between those norms and my own aesthetic sensibilities. The authority of these norms to demand me to φ feels to rest in part on some intimate cooperation between these norms and my basic musical sensitivities: it feels that I could have “stepped outside” the jurisdiction of these norms if I had had the radically altered sensibilities of (say) a sophisticated Tuvan throat singer. Things are once again different in the moral case. The authority of a moral reason to direct my action doesn’t feel to depend in the same crucial way on the resonance of any associated moral norms with my own idiosyncratic sensibilities. When I appreciate the force of the reason to take my elderly neighbor to the baseball game, it doesn’t feel like I could have “stepped outside” of that reason’s jurisdiction just by virtue of having had the thoroughly altered sensibilities of (say) a sophisticated Nazi.

2 . 5 . P H E N O M E N OL O G Y AN D O R D I N A R Y M O R AL R E AS O N S- A P P R E C I A T I O N So far I have argued that an agent who implicitly appreciates a consideration as a moral reason to φ must present that consideration via the light of a particular phenomenologically-mediated mode of presentation—one that presents the relevant consideration via the light of a felt directive force “pointing” towards φ-ing, lending weight to it, or soliciting it, in a particular authoritative way. To close, I’ll suggest that the phenomenal component of implicit moral reasons-appreciation may also provide insight into the nature

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

42

M. Coetsee

of ordinary moral reasons-appreciation, since it can serve to enhance both cognitivists’ and non-cognitivists’ accounts of such appreciation.

2.5.1. Non-Cognitivists and the Attitude Specification Problem According to non-cognitivists, an agent appreciates a consideration as a moral reason to φ if and only if she has certain pro-attitudes towards φ-ing in light of that consideration.²⁹ Non-cognitivists face the challenge, however, of specifying which kinds of pro-attitudes support moral reasonsappreciation.³⁰ Some non-cognitivists suggest that relevant pro-attitudes may be distinguished by their functional roles,³¹ but the spectrum inversion argument used earlier against the functional account of implicit moral reasons-appreciation undermines the plausibility of this proposal. (After all, that argument targeted the functional account’s ability to explain the moral, rather than the implicit, aspect of such appreciation.) It behooves a non-cognitivist, then, to consider making reference to pro-attitudes’ phenomenological features, and not just their functional roles: (6) PHENOMENOLOGICAL PRO-ATTITUDE: An agent appreciates a consideration c as a moral reason r to φ iff her representation of c underwrites an appropriate pro-attitude a towards φ-ing, where a is specified in part by reference to its phenomenal features.

We can highlight one phenomenal feature that a must have by revisiting the kind of case where it is absent: Silas’s Compulsion: When Silas thinks about Benvolio—who is by now in the hospital—as being in pain, it triggers in him a brute compulsive desire to send a get-well card. The desire, he feels, could be resisted, but it feels totally inexplicable and out of the blue. What it is like for Silas is what it would be like for one of us if hearing a highpitched sound suddenly triggered a brute compulsive desire to stand up and jump up and down. The drive to φ feels to be caused by the stimulus, but φ-ing doesn’t feel to “make sense” in light of it.

Silas’s brute compulsive desire may play the functional role that appreciation of a moral reason would, but it nevertheless fails to underwrite such appreciation. This is because, though the desire involves a “pro”-presentation of φ-ing as to-be-done, it presents φ-ing as to-be-done in the wrong manner: in a manner that fails to connect the to-be-donedness of φ-ing in the ²⁹ Blackburn (1998) and Gibbard (1992). ³⁰ Though Railton (2009, 2014) and the philosophers of emotion discussed in n. 18 aren’t non-cognitivists, they also face this challenge. ³¹ See n. 15.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

The Phenomenal Appreciation of Reasons

43

right way to Benvolio’s pain. For Silas to construe Benvolio’s pain as a reason to send a get-well card, his pro-attitude must feel different from this compulsive desire: it must make the to-be-donedness of φ-ing, feel not merely causally triggered by the representation of Benvolio’s pain, but rather made worthwhile, warranted, or—most broadly—as something that “makes sense” to do, in light of his pain. But that means the pro-attitude must also construe Benvolio’s pain in a certain light. Plausibly, that pro-attitude must construe the pain so as to make it feel connected to sending a get-well card by “calling out” for or speaking in favor of sending a get-well card in some way. This “calling out” for φ-ing, moreover, can’t be of just any sort. If the pro-attitude construes Benvolio’s pain as calling out for φ-ing, but this calling out is underwritten by (e.g.) Silas’s aesthetic feeling of the demands of dramatic narrative flow, it still won’t serve to construe Benvolio’s pain as a moral reason to φ. Instead, the pro-attitude must build into itself a different kind of feeling—one that construes Benvolio’s pain as “calling out” for sending Benvolio a card via the light of that kind of authoritative solicitation relevant to the phenomenal appreciation of moral reasons. If this is right, then non-cognitivists will need to draw on my phenomenal account of moral reasons-appreciation to solve the attitude specification problem. Their account of ordinary moral reasons-appreciation will thus need to include the phenomenological component I have discussed.

2.5.2. Cognitivism and Moral Motivation Cognitivists’ account of ordinary moral-reasons appreciation may also benefit from incorporating reference to a phenomenological component. The inadequacy of causal and functionalist accounts of implicit moral reasonsappreciation stemmed from their failure to account for the moral rather than the implicit aspect of such appreciation, so these accounts are also inadequate to fully explain ordinary moral reasons-appreciation. The discursive account of implicit moral reasons-appreciation, however, ran aground specifically because of the implicit aspect of that appreciation, and so we still need to consider whether it can deliver cognitivists an account of ordinary moral reasons-appreciation that doesn’t make reference to phenomenology. It appears that it cannot. For consider: Silas (Discursive): Silas receives and believes testimony from a moral expert giving him a complete specification of the correct moral theory. His patterns of inference mirror the network of inferences relevant to the moral reason-giving concepts of that theory, and he comes to act as a pure ratiocinator with respect to it. When you present him with a

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

44

M. Coetsee consideration—like Benvolio is in pain—he cranks it through his theoretical algorithms and infers the appropriate conclusions about what he has moral reason to do. But, though he believes the inferred claim that helping Benvolio is the moral thing to do, he doesn’t experience that as in any way demanded of him, or feel the weight of Benvolio’s pain as bearing at all on the practical question of what to do.³²

Silas has a discursive representation of Benvolio’s pain as a moral reason to stop Tybalt, but he appreciates Benvolio’s pain only (at best) as an epistemic reason to believe helping Benvolio is the moral thing to do, and not as a practical moral reason to help Benvolio. So what is missing? A phenomenal account of moral reasons-appreciation provides a plausible answer: Silas must also present Benvolio’s pain under the light of a felt solicitation demanding him to φ. It is by presenting Benvolio’s pain under the light of this sort of felt solicitation that Silas will come to feel the weight of that consideration as having practical relevance to the question of what he should do, and so—intuitively—will come to appreciate Benvolio’s pain as a moral reason to help Benvolio. Since it now appears that none of the major theories of mental representation cognitivists could utilize can explain ordinary moral reasons-appreciation without some supplementary appeal to a phenomenal component, it seems that cognitivists’ account of ordinary appreciation will (like that of non-cognitivists) need to draw on my phenomenal account of moral reasons-appreciation. A phenomenal account of moral reasons-appreciation may also support cognitivists’ in giving a plausible account of moral motivation. Relative to other descriptive judgments, judgments about what one has moral reason to do have a marked tendency to motivate action, but cognitivists typically have difficulty explaining why this is so since they can’t (as non-cognitivists do) fall back on the idea that moral judgments are expressions of proattitudes. One way cognitivists can try to explain moral motivation is to say that moral judgments motivate in the same way that other descriptive judgments do: by being appropriately paired with standing de dicto desires. Just as my judgment that My car won’t drive motivates me to bring it to the mechanic because I have a standing de dicto desire to drive my car, so also (cognitivists may suggest) my judgment that I have moral reason to φ motivates me to φ because I have a standing de dicto desire to do the moral thing. Critics, however, argue that this “standing desire” account of moral motivation makes moral agents objectionably “fetishistic” about morality. If it is right, after all, the motivation of someone like Juliet to help Benvolio is

³² See Oddie (2005: 48).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

The Phenomenal Appreciation of Reasons

45

just a byproduct of her generic, abstract de dicto desire to do the right thing. But, it is complained, a good moral agent’s motivation should not be derivative in this way. Instead, critics claim, the right picture of Juliet as a good moral agent must depict her motivation as stemming directly from her concern about the moral reason-constituting features of the world—e.g. Benvolio’s pain.³³ A phenomenal account of moral reasons-appreciation can help cognitivists avoid this criticism while remaining faithful to their descriptivist understanding of moral representation. On the one hand, moral solicitations may be understood descriptively and as separable from pro-attitudes. Perhaps an evil demon can consider Benvolio’s pain under the aspect of a representation that “asks for” or solicits her to help without having any pro-attitudes towards helping, if she just continually dismisses or refuses to answer that call. On the other hand, though feeling a moral solicitation to φ may be separable from having a pro-attitude towards φ-ing, such a solicitation is intuitively the sort of thing that is well-poised to motivate. After all, a solicitation does solicit a response, and answering or “saying yes to” the call of a solicitation to φ is naturally understood as constituting a motivating desire to φ.³⁴ If cognitivists understand moral motivation in these terms, moreover, they need no longer appeal to a generic de dicto desire to do the moral thing. Benvolio’s pain itself calls out to Juliet, authoritatively demanding her concern, and it is this consideration—thus appreciated as a moral reason—that directly and underivatively gives rise to her desire to help Benvolio.

2.6. CONCLUSION What it’s like to be a moral agent in the world is more than what it’s like to feel pain or see red; there’s a way it feels to walk in the space of moral reasons, and if you lack this feeling, you lack something important about that walk. Starting with cases of implicit agency like Huck and Juliet’s, I have tried to show what this feeling is—what, in particular, it is for an agent to appreciate a consideration as a moral reason. I have suggested that (at least in non-derivative cases), appreciation of a consideration as a moral reason requires that the agent presents that consideration via the light of a

³³ Smith (1994: 75–6). ³⁴ The Stoics thought that motivation was constituted by assent to hormetic impressions, which are much like felt solicitations. See Inwood (1985).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

46

M. Coetsee

felt authoritative force “pointing” towards φ-ing—lending weight to it, or soliciting it—in a particular authoritative way. If correct, this account of moral reasons-appreciation sheds light on several important discussions about moral agency and understanding. First, it provides resources for developing a better understanding of a variety of forms of “marginal” moral agency that involve the implicit appreciation of moral reasons—for instance, the agency of subjects who are victim to hermeneutical injustices,³⁵ those who face cognitive developmental challenges,³⁶ and even that of non-human animals. Second, it suggests that— contra the claims of some³⁷—robots cannot be full-fledged moral agents unless and until they are also phenomenally conscious. Third and finally, it may urge us to recognize a more significant role for moral experience in underwriting inter-personal moral understanding. In particular, if phenomenology is critical to moral reasons-appreciation, then the development of inter-cultural moral understanding will plausibly require first-personal moral experience and not just academic study. We will need to know what it is like to have relevant considerations “call out” to us—as they do to our interlocutors—with a solicitive force directed at φ-ing. I leave further investigation into these important potential payoffs of a phenomenal account of moral reasons-appreciation for later study.³⁸

References Adolphs, R., and D. Andler (2018). “Investigating Emotions as Functional States Distinct from Feelings,” Emotion Review 10(3): 191–201. Allman, J., and J. Woodward (2008). “What are Moral Intuitions and Why should we Care about Them?” Philosophical Issues 18(1): 164–85. Annas, J. (2011). Intelligent Virtue. Oxford: Oxford University Press. Arpaly, N. (2002). Unprincipled Virtue: An Inquiry into Moral Agency. Oxford: Oxford University Press. Arpaly, N., and T. Schroeder (2013). In Praise of Desire. Oxford: Oxford University Press. Bayne, T., and M. Montague, eds (2011). Cognitive Phenomenology. Oxford: Oxford University Press. Beardsley, M. C. (1958). Aesthetics. Indianopolis: Hackett. Bennett, J. (1974). “The Conscience of Huckleberry Finn,” Philosophy, 49(188): 123–34.

³⁵ Fricker (2007: 164). ³⁶ Shoemaker (2015: 75–7, 97–9). ³⁷ McLaughlin (n.d.); Sullins (2011). ³⁸ Thanks to the audience at the 2017 MadMeta Conference, the Rutgers Epistemology Dissertation Group, Ruth Chang, and Carolina Flores for helpful discussion.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

The Phenomenal Appreciation of Reasons

47

Blackburn, S. (1993). Essays in Quasi-Realism. Oxford: Oxford University Press. Blackburn, S. (1998). Ruling Passions: A Theory of Practical Reasoning. Oxford: Oxford University Press. Block, N. (1986). “Advertisement for a Semantics for Psychology” in P. A. French, T. E. Uehling, and H. K. Wettstein (eds), Midwest Studies in Philosophy, x. Minneapolis: University of Minnesota Press, 615–78. Conee, E., and R. Feldman (2004). Evidentialism. Oxford: Clarendon Press. Deonna, J., and F. Teroni. (2011). The Emotions: A Philosophical Introduction. New York: Routledge. Dretske, F. (1988). Explaining Behavior: Reasons in a World of Causes. Cambridge, MA: MIT/Bradford. Dreyfus, H. L., and S. E. Dreyfus. (1991). “Towards a Phenomenology of Ethical Expertise,” Human Studies 14(4): 229–50. Fodor, J. A. (1987). Psychosemantics: The Problem of Meaning in the Philosophy of Mind. Cambridge, MA: MIT Press. Fodor, J. A. (1990). A Theory of Content and Other Essays. Cambridge, MA: MIT/ Bradford Press. Fricker, M. (2007). Epistemic Injustice: Power and the Ethics of Knowing. Oxford: Oxford University Press. Frijda, N. (2007). The Laws of Emotion. Mahwah, NJ: Lawrence Erlbaum. Gibbard, A. (1992). Wise Choices, Apt Feelings: A Theory of Normative Judgment. Cambridge, MA: Harvard University Press. Gibson, J. (1986). The Ecological Approach to Visual Perception. Mahwah, NJ: Lawrence Erlbaum. Hampton, J. E. (1998). The Authority of Reason. Cambridge: Cambridge University Press. Harman, G. (1973). Thought. Princeton: Princeton University Press. Horgan, T., and G. Graham (2013). “Phenomenal Intentionality and Content Determinacy” in R. Schantz (ed.), Prospects for Meaning. Berlin: De Gruyter. Horgan, T., and J. Tienson (2002). “The Intentionality of Phenomenology and the Phenomenology of Intentionality” in D. J. Chalmers (ed.), Philosophy of Mind: Classical and Contemporary Readings. Oxford: Oxford University Press. Horgan, T., and M. Timmons (n.d.). “Illuminating Reasons.” MS. Horgan, T., and M. Timmons (2005). “Moral Phenomenology and Moral Theory,” Philosophical Issues 15(1): 56–77. Horgan, T., and M. Timmons (2008). “What does Moral Phenomenology Tell us about Moral Objectivity?” Social Philosophy and Policy 25(1): 267–300. Inwood, B. (1985). Ethics and Human Action in Early Stoicism. Oxford: Oxford University Press. Jackson, F., and P. Pettit (1995). “Moral Functionalism and Moral Motivation,” The Philosophical Quarterly 45(178): 20–40. Jalava, J., and S. Griffiths (2017). “Philosophers on Psychopaths: A Cautionary Tale in Interdisciplinarity,” Philosophy, Psychiatry, and Psychology, 24(1): 1–12. Kirchin, S. (2003). “Ethical Phenomenology and Metaethics,” Ethical Theory and Moral Practice 6(3): 241–64.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

48

M. Coetsee

Kriegel, U. (2008). “Moral Phenomenology: Foundational Issues,” Phenomenology and the Cognitive Sciences 7(1): 1–19. Leopold, A. (1970). A Sand County Almanac. Oxford: Oxford University Press. McLaughlin, B. (n.d.). “Could an Android be Sentient?” MS. Mandelbaum, M. (1955). The Phenomenology of Moral Experience. Baltimore, MD: Johns Hopkins University Press. Markovits, J. (2010). “Acting for the Right Reasons,” Philosophical Review 119(2): 201–42. Millikan, R. G. (2004). Varieties of Meaning. Cambridge. MA: MIT Press. Oddie, G. (2005). Value, Reality, and Desire. Oxford: Oxford University Press. Peacocke, C. (1992). A Study of Concepts. Cambridge, MA: MIT Press. Pitt, D. (2004). “The Phenomenology of Cognition, or, What is it Like to Think that ‘P’?” Philosophy and Phenomenological Research 69(1): 1–36. Plantinga, A. (1993). Warrant: The Current Debate. Oxford: Oxford University Press. Railton, P. (2009). “Practical Competence and Fluent Agency,” in D. Sobel and S. Wall (eds.), Reasons for Action. Cambridge: Cambridge University Press. Railton, P. (2014). “The Affective Dog and its Rational Tale: Intuition and Attunement,” Ethics 124(4): 813–59. Shoemaker, D. (2015). Responsibility from the Margins. Oxford: Oxford University Press. Siegel, S. (2014). “Affordances and the Contents of Perception” in B. Brogaard (ed.), Does Perception have Content? Oxford: Oxford University Press. Smith, M. R. (1994). The Moral Problem. Oxford: Oxford University Press. Sullins, J. (2011). “When is a Robot a Moral Agent?” in M. Anderson and S. L. Anderson (eds), Machine Ethics. Cambridge: Cambridge University Press. Wedgwood, R. (2007). The Nature of Normativity. Oxford: Oxford University Press.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

3 Who’s on First? Daniel Wodak

Abbott: Strange as it may seem, they give ball players nowadays very peculiar names. Costello: Funny names? Abbott: Nicknames, nicknames. Now, on the St. Louis team we have Who’s on first, What’s on second, I Don’t Know is on third— Costello: That’s what I want to find out. I want you to tell me the names of the fellows on the St. Louis team. Abbott: I’m telling you. Who’s on first, What’s on second, I Don’t Know is on third— Costello: You know the fellows’ names? Abbott: Yes. Costello: Well, then who’s playing first? Abbott: Yes. Costello: I mean the fellow’s name on first base. Abbott: Who.

(From Abbott and Costello’s skit, “Who’s on First.”) “X-Firsters” hold that some normative feature is fundamental to all others. This is a common view. As Mark Schroeder wrote in Slaves of the Passions, for centuries [metaethicists] have characterized their subject matter as being everything which ultimately involved claims about what was good, or . . . what was right, or . . . what someone ought to do. All of these views claim that what it is for a property or concept to be normative, is for it to be ultimately analyzable in terms of some basic normative property or relation or concept. They merely disagree about what this basic property or concept is. (2007: 81)

Much ink has been spilled in the “mere” disagreements about what this basic property is. This is the internecine debate between X-Firsters, especially between Reasons Firsters (e.g. Parfit 2011; Scanlon 1998), Values Firsters Daniel Wodak, Who’s on First? In: Oxford Studies in Metaethics Volume 15. Edited by: Russ Shafer-Landau, Oxford University Press (2020). © Daniel Wodak. DOI: 10.1093/oso/9780198859512.003.0003

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

50

Daniel Wodak

(e.g. Moore 1903; Maguire 2016), and Fittingness Firsters (e.g. Chappell 2012; McHugh and Way 2016). To some, this internecine debate resembles an Abbott and Costello skit. In saying this I do not intend to cast aspersions on interesting, insightful work that has developed problems and solutions for specific X-First views. Rather, I intend to draw attention to something about the debate itself. It takes as a starting point that some normative feature is fundamental to all others. Little to nothing is said about whether or why we should think that. That said, to my knowledge there has been no clear argument against the X-First program,¹ and no clear articulation of an alternative to it either. Hence my two main goals in this chapter. First, I provide a fairly simple argument that one shouldn’t be an X-Firster about the normative domain. The central move is to show that X-First theories have dubious merits when applied to a range of analogous domains. Second, I offer an alternative to X-First views. I develop an approach—taking normativity to be a determinable—that provides a stark contrast with X-First views, especially in how it treats the structure and unity of normativity.

3 . 1 . W HA T I S I T F O R S O M E T H I N G TO BE “ O N F I R S T” ? Before we proceed to these two main goals, it’s worth covering our bases. What exactly are X-Firsters committed to? I’ll start with Reasons First, zoom out to characterize X-First itself, then offer two clarifications. Reasons First is, roughly, the view that reasons are “the only fundamental elements of the normative domain, other normative notions such as good and ought being analyzable in terms of reasons” (Scanlon 2014: 2). Selim Berker argues that this “widely popular” approach is most aptly framed in terms of two theses about grounding: that “reasons are first,” or all other normative “facts are grounded in facts about reasons, that it is in virtue of the facts about reasons that these other normative facts obtain”; and that “reasons are not tied for first,” or “it is not because of any facts about other normative categories that facts about reasons obtain” (2017: 15–16). Generalizing, the X-First program takes some X, and declares it to be prior to all Ys. What’s X? A paradigmatic normative property (or relation or

¹ Though skepticism has been aired: see e.g. Cuneo (2007: 64–5; Väyrynen (2011: 203).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Who’s on First?

51

concept or . . . ). The three most common candidates are reasons, fittingness, and value. But others are possible—e.g. ought.² What are all the Ys? All of the other normative properties (or relations or concepts or . . . ). Here we obviously need a demarcation of the normative. I suspect X-Firsters have a narrow demarcation in mind,³ but since little is typically said on this score I won’t discuss it further. Note also that X can be prior to all the Ys even if some Z is prior to X. Reasons Firsters can allow that all facts about reasons are fully grounded, so long as they are fully grounded in non-normative facts (as Berker notes 2017: 16). More generally, X-First is compatible with both naturalism and non-naturalism. Modulo questions about what terms like “priority” mean here—it’s compatible with expressivism. Consider Gibbard: “W I regard as the basic normative concept, the conceptual atom that renders molecular concepts like  normative” (2006: 196–7). Two further theses are sometimes classed with X-First. The first concerns local questions of explanatory priority, such as: Is the good prior to the right, or vice versa? The good can be prior to the right (or vice versa) without being prior to all other normative properties. X-Firsters sometimes count as allies historical figures who only answer local questions, but those figures seem to be playing a different ballgame. Gibbard’s quote above points to the second thesis. For Gibbard,  is not just the atomic normative concept out of which other normative concepts are constructed; it is what renders those other concepts normative. More generally, X-Firsters might be committed to there being some X that other normative properties have in common, in virtue of which they are normative. Call this the thesis that some X is “the mark of the normative.”⁴ Like others,⁵ the Schroeder quote that we started with entangles both issues (“what it is for a property or concept to be normative, is for it to be ultimately analyzable in terms of some basic normative property”). But they ² See also Wodak (forthcoming) on views like Broome‘s (2013)—which may just answer a local question in the sense described below. It’s also noteworthy here that X-Firsters standardly take a commonsensical (rather than some recherché) normative notion to be what’s on first. Thanks to Francois Schroeter for discussion. ³ X-Firsters focus on what we can call “authoritative” normative properties; they’re silent on how we explain “merely formal normativity” (McPherson 2011; Wodak 2019). ⁴ To accommodate views like Raz’s (discussed later), the “mark of the normative” thesis needs to be framed somewhat broadly: for all Y, Y is normative iff Y is related in “the right way” to X. It’s unclear how to cash this out. (What does Raz mean by “otherwise related to reasons”?) Thanks to Selim Berker and Gabriel Shapiro for helpful discussions here. ⁵ Consider Snedegar (2016: 156): a central “appeal of the reasons first program is that it promises to explain what it is that makes the normative normative”; but this “really only motivate[s] taking some single notion to be normatively basic.” This assumes a great deal:

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

52

Daniel Wodak

should be separated. For instance, Raz has endorsed the view that “the normativity of all that is normative consists in the way it is, or provides, or is otherwise related to reasons” (1999: 67); but he has also endorsed the view that facts about practical reasons are explained in terms of facts about value.⁶ These commitments don’t seem inconsistent. So while Raz is often classed with Reasons Firsters, this is a mistake. The commitments about explanatory priority are definitive of the X-First program, I think. That said, there’s no use fighting over terminology. The crucial point is the distinction between the claim that some normative X is on first and the claim that some normative X is the mark of the normative, so defined. I’ll discuss both claims, but mostly focus on the first. 3 . 2. I S S O M E T HI N G “ O N F I R S T ” ?

3.2.1. The Goal My first goal is to argue that one shouldn’t be an X-Firster. But that’s ambiguous between two claims. X-F ’ F: We should not assume that there’s some X that’s “on first” for normativity. X-F  F: It’s false that there’s some X that’s “on first” for normativity.

I am somewhat sympathetic to the second claim, and some of what I’ll say can be marshalled in its defense. My case for the first claim is stronger. Some may think the first claim is a weak target. But it’s dialectically significant if I can provide a good case that X-F ’ F is true. The X-First debate takes as a starting point that some normative feature is fundamental to all others. More specifically: the X-First debate proceeds via arguments for preferred X-First views (e.g. Reasons First) that are just arguments against rival X-First views (e.g. Values First). These arguments obviously rely on the assumption that some X is on first. If we shouldn’t assume that, the whole debate is built on sand. Some X-Firsters might find this characterization of the debate to be unfair. Some X-Firsters explicitly say they aren’t assuming that some X is on first.⁷ And some may say that they are defending that claim, not assuming it. that we can explain what it is that makes the normative normative only by taking there to be a mark of the normative, and that the mark of the normative must be on first. ⁶ For discussion and references, see Heuer (2004), especially at 133: for Raz, “we have reasons to act in certain ways because so acting is an appropriate response to value.” ⁷ See e.g. McHugh and Way (2016: 577): “Why expect that there will be some basic normative or evaluative property from which the rest of the normative and evaluative

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Who’s on First?

53

Two responses. One, X-Firsters sometimes hedge, but that doesn’t fit with the conclusions they draw. Consider Howard (2019). Howard argues for Fittingness First by arguing that Reasons First and Values First are unacceptable. If Howard grants that we have no good reason from the onset to think that some X is on first—or, more specifically, that reasons or values or fittingness is on first—his argument can’t be for The Fundamentality of Fit (which is his title). It can only be for the less interesting conditional claim: if reasons or values or fittingness is fundamental then fittingness is fundamental. Either the antecedent of that conditional is assumed to get to a stronger conclusion, or X-Firsters’ conclusions aren’t as interesting as their titles suggest. Pick your poison. My second response may initially seem like a curve ball, but bear with me. Alonzo Church once considered a view he called “ontological misogyny.” One form of this view would explain all facts about women in terms of facts about men plus the being a father of relation. As Church noted, this view may give us adequate resources to express all facts about women, but if so that doesn’t seem like a good argument for the doctrine.⁸ Now compare this to X-Firsters’ arguments views for their preferred views. Chappell’s main argument for taking fittingness to be “primitive” is that this provides “adequate conceptual resources for us to express any expressible normative truth” (2012: 686). Does that really show that fittingness is “The Sole Normative Primitive”? Or that there is any “Sole Normative Primitive”? No. The point here is simple. X-Firsters often aim to show that we can put some X on first for normativity (without losing expressive power or extensional adequacy). But so what? That we can does not mean that we should. If ontological misogyny lets us express any expressible truth about gender, that does not mean that men are “on first” with respect to gender, or indeed that there is any Sole Gender Primitive. (If Church’s example is too out of left field, we could make similar points concerning e.g. logical concepts. See McSweeney (2019) on this issue: we can take ‘∀’, ‘&’, and ‘~’ to be the sole logical primitives without losing expressive power or extensional adequacy; that doesn’t mean we should !) Let me put the preceding points differently. The stronger conclusion X-F  F rejects every X-First view. The weaker conclusion X-F ’ F rejects the X-First debate. As Schroeder put it, everyone in the debate agrees that that there’s some basic normative property and “merely disagrees” about what it is; so that starting point is some X is on domain is constructed? We do not assume from the outset that this is so.” (I discuss the “significant attractions” they attribute to this “hypothesis” in n. 20.) Cf. Rowland (2017: 213, emphasis mine), who assumes that “We should accept an account of the basic normative property that enables us to analyze the evaluative in terms of the normative.” ⁸ For references for Church’s 1958 talk, see Inwagen (2004: 123).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Daniel Wodak

54

first, and we argue from there to work out whether X is reasons or values or whatever. If X-F ’ F, X-Firsters owe us much more.

3.2.2. The Argumentative Strategy There are domains where X-First views seem plausible. Take rectangles. For any shape, if it’s a rectangle, this is in virtue of its being equiangular-andquadrilateral. That conjunctive property is “on first” for rectangles. Perhaps normativity is like rectangles. But many domains that are analogous to normativity don’t look like rectangles in this respect. This is the basis for my master argument against X-Firsters: P1. P2. C.

We should not have an X-First debate for Z. If P1, we should not have an X-First debate about normativity. So, we should not have an X-First debate about normativity.

There are different instances of this argument, depending on which domain we focus on—i.e. what candidate for Z we consider.⁹ I’ll consider a few different candidates below. I don’t need each of them to succeed, obviously; if one version of the argument is sound, that’d suffice. But it’s valuable to consider a range of candidates to see how much the X-First debate in metaethics seems like an aberration in philosophy. This sociological observation will do some work in the argument below. Here’s the first candidate for Z: the mental. Considering P1 helps us elucidate a curious feature of X-First theories. You might think they are simply monistic theories: they just require a single property to be fundamental to normativity. But that’s not right. They require a single normative entity to be fundamental to normativity. The equivalent view in philosophy of mind is that there’s a single mental property that’s fundamental to the mental. This involves something like a Beliefs First theory of the mental. There are theories in this ballpark. Maybe ideas or intentionality are “on first.” But there are well-known and widely accepted theories, like functionalism, which don’t have this structure.¹⁰ We have theoretical resources— like Ramsey sentences—to explicate the intuitive idea that the most basic mental concepts like  and  might be interdependent, such that no mental X is “on first.” It would be a mistake to set aside all views like ⁹ I’m using “domains” roughly in the way that Scanlon (2014) does, but nothing hangs on the term. It may be more apt to speak of normativity as a category. (Interestingly, for “the highest kinds” in Aristotle’s Categories, like qualities, no X is “on first”—or so I’m told.) ¹⁰ One way of thinking about functionalism is that to be a mental property is to be a realizer of a certain functional role that’s specified via a Ramsey sentence. But the single entity “on first” (a) isn’t mental, and (b) is a recherchésecond-order property. Cf. n. 15. Thanks to Jan Dowell and Laura Schroeter for helpful discussion here.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Who’s on First?

55

functionalism and construct a central debate in philosophy of mind to only be between theories like Beliefs First. Now consider P2. One point in its favor: if we need not be X-Firsters about the mental, why must the normative be so different from the mental? X-Firsters have an explanatory debt here. (I’ll expand on this below.) Here’s another point in its favor. Considering views like functionalism helps us set aside a wide set of considerations that might seem like putative advantages of being an X-Firster about normativity: that such views are simpler, or more conducive to a naturalistic reduction (e.g. Schroeder 2007: 81). Is functionalism less simple, or less reductionist-friendly, than views like “Intentionality First”? If so, is that a good ground to reject functionalism and agree from the onset to only consider views like Intentionality First? Once we attend to theoretical positions that don’t fit the strictures of an X-First program, it’s hard to see why general theoretical virtues (or an appetite for reduction) should be a good motivation for that program. Let’s turn to a more obvious candidate for Z: descriptivity. By this I mean the facts, properties, etc. that fall on the “is” side of Hume’s is/ought gap. This is a motley crew. Numbers and narwhals, sensations and supernatural beings, photons and pharmacies, Germany and gravity: if they all exist, they are all descriptive. Must there be some descriptive property, X, that is prior to all other descriptive properties (some descriptive X that is “on first”), and that all other descriptive entities have in virtue of which they are descriptive (that is, a “mark of the descriptive”)? If so, what would X be? What is the equivalent of being equiangular-and-quadrilateral for numbers, narwhals, photons, and so on? Any candidate for X would be highly contentious. Consider dualism: the mental and the physical are plausibly both descriptive, but dualists say they’re fundamentally different. Can an X-Firster about descriptivity respect that commitment? If the mental and the physical are like isosceles and oblique triangles, are they really fundamentally different? A similar concern arises for abstracta and concreta (Rosen 2017; Thomasson 1999).¹¹ Say I’m right about P1. Why believe P2? Because normativity and descriptivity are meant to carve an important joint in reality. If the heterogeneous stuff on one side, descriptivity, does not fit the mold of an X-First theory, why should the heterogeneous stuff on the other? Even if we could entertain some X-First theories for descriptivity, we shouldn’t construct a debate comparing them and taking the best of the bunch to be the true theory of descriptivity. So why have that debate for normativity?

¹¹ Consider e.g. the view that the essential descriptive feature is having causal powers. Adopting such an Eleatic Principle is hard to square with abstracta and dualist views.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

56

Daniel Wodak

I suspect that most objections to this argument will target P2. You may think, for instance, that it’s a mistake to posit a descriptive property that’s “on first,” but only because being descriptive is primarily a feature of sentences or propositions, rather than of properties or relations. But similar motivations for taking descriptivity to be primarily a feature of sentences or propositions can be marshalled in favor of taking normativity to be primarily a feature of (uses of) sentences or propositions. Why should X-Firsters treat normativity and descriptivity so differently? A better way to reject P2 is to argue that “descriptivity” is just the grabbag of whatever is not normative. If this is so, of course P2 will be false! It’s like saying rectangles need not have any essential feature because there is no essential feature for the heterogenous collection of non-rectangles. But is it so? I’m not sure. Something can be rectangular or nonrectangular, but not both. But it’s at least coherent to take something to be normative and descriptive. So the distinction between the descriptive and the normative doesn’t seem akin to the distinction between rectangles and non-rectangles: “descriptive” means something other than “not normative.” Even if one sympathizes with such objections, however, the X-Firster isn’t home. The objections only target one version of the argument. They don’t help with the mental. Nor do they help with our final candidate for Z: the physical. Being physical is clearly not a feature of sentences or propositions in the way that being descriptive might be. And the domain or category of the physical is, it’s safe to say, not some grab-bag of leftovers. So, should we have an X-First debate about the category of the physical? Is there some physical X that’s prior to all other physical properties and relations (some physical X that’s “on first”)? There’s some precedent for this view. Thales of Miletus—perhaps the first philosopher—held that everything is water. On this Water First view, water is a “primary principle” of all things like the sun, the stars, and the cosmos. But this Water First theory is false. And it’s not just false because water itself is not fundamental to physics—that is, we didn’t just learn it was false when we discovered that water is H₂O. One way to see this is to think of some more contemporary counterparts to Thales’ view about physics. A Photons First ontology for the physical may not be threatened by a discovery like Lavoisier’s, but we still know that it’s a non-starter. Why? For one thing, physical entities include material things like molecules and non-material things like forces. If there are “fundamental physical forces,” the prospects for anything like a Photons First ontology are dim.¹² (This is part of what led materialists to become ¹² Some may endorse a Particles First or Fields First ontology of the physical. But the correspondence to an X-First theory here is superficial. Being committed to a plurality of fundamental particles (photons, bosons, whatever) or a plurality of fundamental fields

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Who’s on First?

57

physicalists in philosophy.) A similar point applies to the view that there some X that all physical entities have in virtue of which they are physical (i.e. some X that’s “the mark of the physical”). There’s a debate about what makes entities physical (see Stoljar 2017: §§11–12). But it does not resemble the X-First debate—there’s no agreement from the onset that some X is “the mark of the physical.” This is why the case for P1 here is strong: X-First theories about the physical are implausible, and we certainly shouldn’t assume that one is right. The case for P2 is just as strong as before. The distinction between the normative and the descriptive is often used because of a taxonomical consideration in the naturalism vs. non-naturalism debate. One way of denying the fundamentality of the normative is to claim that normativity is ultimately explained in terms of the supernatural, which is descriptive but not physical. But much of what drives naturalism about normativity drives one towards physicalism (at least, ceteris paribus); a distinct realm of sui generis normative stuff raises similar concerns to a distinct realm of sui generis supernatural or mathematical or mental stuff. If so, the normative and the physical are playing similar roles as philosophical domains.

3.2.3. The Sociological Observation These three candidates—the descriptive, physical, and mental—hardly exhaust the categories which are similar to the normative, but don’t have anything resembling the X-First debate in metaethics. (For instance: once vitalism was rejected, it has become unclear if there’s any essential feature that all animate beings have in virtue of which they’re animate.) In this sense, the X-First debate in metaethics is something of an aberration in philosophy.¹³ And this sociological observation warrants an explanation: why does the X-First debate have pride of place in metaethics if there aren’t analogous debates about these analogous philosophical categories? Here are three possible explanations for this sociological observation: (i) Pessimism about X-First Views: The X-First debate is built on sand. (ii) (ii) Normative Exceptionalism: We have to be X-Firsters about normativity, but not about these other philosophical categories. (electromagnetism, maybe gravity, some poorly named ones) is being committed to the kind of fundamentally disjunctive account that X-Firsters want to reject. See section 3.3. ¹³ It’s not quite an anomaly. Attempts to reduce all mathematical entities to sets or categories are arguably similar to X-First theories about normativity. (Though I do not think sets were thought to be “the mark of the mathematical” in this literature.) That said, there was no assumption in this literature that some view like Sets First must be true.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

58

Daniel Wodak

(iii) Revisionism about the Non-normative: We need, and sorely lack, X-First debates about these other philosophical categories. I think the best explanation is the first: we should think either that X-First views are false, or (more modestly) that the X-First debate is misguided. Just as there may be an essential mental feature, there may be an essential normative feature. But there’s no cause to agree from the onset that there is an essential normative feature and merely disagree about what it is. X-Firsters may wish to take the second option. But if so, they really need to step up to the plate and give us arguments that explain what makes the normative exceptional. This requires more than throwaway remarks. We need an explanation for why whatever motivates being an X-Firster about normativity doesn’t motivate being an X-Firster across the board. If we baldly declare that X-First theories are always preferred on grounds like simplicity, we take on odd commitments about e.g. functionalism. I find the third option to be the least appealing. Though if it’s the right way to go, my argumentative strategy won’t have yielded the fruit that I expected, but it will have led to something surprising and significant nonetheless: there’s an enormous amount of new work to be done in almost every field. This observation, then, poses a simple challenge for X-Firsters. Note that it’s not a direct challenge to the arguments that preoccupy the literature (whether Fittingness First is a better view than Reasons First or Values First, and so on). It’s a challenge to an inference from such arguments to the conclusion that the preferred X-First theory about normativity is true. Unless more is said, it’s unclear if those arguments show us much more than the argument that Photons First is a better theory than Water First, or that Intentionality First is a better theory than Berkeleyan idealism.

3. 3. WH A T I F NOTH ING IS “ O N F I R S T ” ? If I’m right about the X-First debate, how should we theorize about normativity instead? Currently we lack any systematic work on what alternatives to X-First views about normativity can and should look like. There’s plenty of space for many different rivals to X-First views. The alternative I present below is by no means the only way to go, or indeed the way we should go about all philosophically interesting categories. I’m going to go in to bat for it because it provides a particularly sharp contrast with XFirst research programs, and in doing so offers an interesting, fruitful way to think about some of the considerations that I suspect are driving metaethicists towards embracing X-First theories. I’ll initially present a clearer, albeit

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Who’s on First?

59

more committed, version of this view, then show how we might be able to water down some of its commitments.

3.3.1. Two Kinds of Categories Let’s start by contrasting two kinds of categories. Some are genera. Following Rosen (2010: 124–7), it is definitive of genera and species that: For all x, that x is a member of species y is fully explained in terms of x having the genus’s essential feature(s) and species’s differentia.

Recall the example of rectangles: they all have one essential feature in common, namely being equiangular-and-quadrilateral. ABCD is member of a species of this genus, a square, in virtue of having this essential feature plus the requisite differentia (being equilateral). But not all categories are genera. Some, for instance, are determinables. Here’s how Rosen (2010: 127) contrasts genera and determinables: [D]espite the similarities between the determinable–determinate relation on the one hand and the genus–species relation on the other, there is this difference: the determinate grounds the determinable, but the species does not ground the genus.

Take a scarlet letter. Scarlet is a determinate of red, which is in turn a determinate of color. So the letter is red in virtue of being scarlet, and colored in virtue of being red. It is not scarlet in virtue of having some essential feature that colored things share, plus some differentia. On a common—though by no means universal—approach, the facts about a determinable are fully explained in terms of the facts about the disjunction of its determinates (see Wilson 2017: §3.4.1).¹⁴ So on this approach, we can have fundamentally disjunctive accounts of determinables such as color. Three quick asides. First, I’m going to assume this common approach to determinables, though little hangs on it; the main alternative doesn’t help X-Firsters.¹⁵ Second, I’m going to play fast and loose with talk of explaining F (e.g. explaining the determinable color) and explaining the facts about ¹⁴ I should note that Wilson rejects this view (2012, 2014: 557); though cf. Bennett (2017: 23–4), who notes that Wilson most plausibly points out that it is not universally true that determinables are explained in terms of the disjunction of their determinates. ¹⁵ On the most salient alternative, “each determinable property F of ordinary individuals is associated with a second-order property of properties: the property of being an F-determinate” (Rosen 2010: 129). If normativity is a determinate, for x to be normative would be for x to instantiate the second-order property of being a normativity determinate. This fits the letter but not the spirit of an X-First theory: there is a single second-order property that’s “on first,” but it’s not a paradigmatic normative property like being valuable; it’s highly recherché. So doesn’t really vindicate the aims of X-Firsters.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

60

Daniel Wodak

F (e.g. explaining the facts about color); you’ll live. And third, I’m going to speak of taking normativity to be a determinable below; take such talk as a convenient shorthand for taking some more precise property—e.g. being normatively valanced—to be a determinable.¹⁶

3.3.2. The Stark Alternative Genera and determinables don’t exhaust the kinds of categories. But they provide a stark contrast that’s useful for our purposes (though accepting an X-First view isn’t identical to accepting that normativity is a genus).¹⁷ My proposal is that normativity is a determinable, and it reduces to the disjunction of its determinates. On this view, no single normative feature is “on first” (red and blue are explained in terms of different more specific colors), and no X is “the mark of the normative” (explanations of why something is normative, like explanations of why something is colored, will bottom out in different specific determinates of the determinable). Call this Stark Alternative. Views in the ballpark have been endorsed by Tappolet (2004) and Oddie (2005), and suggested by Jessica Wilson (2014: 547), who noted that we could formulate normative naturalism as the view that “normative state types and/or tokens stand in something like the determinable/determinate relation to naturalistic goings-on.” This brings out that the determinates of normativity could be naturalistic. But they could also be other normative state types and/or tokens. In order for the view to be compatible with naturalism or non-naturalism, Stark Alternative—like X-First—must remain somewhat schematic. To put some flesh on the bones of this schema, here’s a version of the view. It’s inspired by Susan Hurley’s discussion of “centralism”: A feature common to many philosophical accounts of ethical concepts is that the general concepts, right and ought, are taken to be logically prior to and independent of the specific concepts, such as just and unkind. According to such accounts, the general concepts carry a core meaning . . . that also provides the specific concepts with reason-giving status. . . . I shall refer to accounts that take ¹⁶ When someone says color is a determinable, they don’t mean that a ball is color in virtue of being red. Likewise, when I say normativity is a determinable, I don’t mean that an act is normativity in virtue of being morally right. That said, being normatively valanced may not be the best candidate for the determinable here. Inter alia, there’s a thorny issue of whether being permitted should be among the relevant class of determinates. (This issue is thorny because of the role it plays in a preposterous argument against error theory.) ¹⁷ For one, there could be an essential non-normative property or relation that is essential to normativity: that could vindicate the view that normativity is a genus, but could not vindicate X-Firsters. For another, the relevant essential feature and differentia could both be normative, which would mean that no single normative notion is “on first.”

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Who’s on First?

61

the general concepts in some category to be logically prior to and independent of the specific as centralist. (Hurley 1985: 56)

Hurley suggests that instead of being centralists we should take “discrete particular values as a starting point” (1985: 56). Following her use of examples like justice and kindness, we could take specific virtues as determinates of normativity. This gives us a Virtues First account. Hurley’s remarks only touch on moral normativity, but one could generalize: the view dovetails nicely with virtue-theoretic approaches to epistemology, for instance. This may seem to resemble Reasons First, but the resemblance is superficial. Virtues First is disjunctive: Hurley’s whole point is that we do not take a “general concept” like  to be what’s “on first”; rather normativity will bottom out in a list of “specific concepts” like  and .¹⁸ I don’t know whether Virtues First is the best option to take for those who are attracted to Stark Alternative. I’m just using it to illustrate the kind of options that become available on this schematic approach. According to Virtues First, the explanation for why something is normative will resemble the explanation for why something is colored. Being courageous is a specific way of being practically virtuous and being conscientious is a specific way of being epistemically virtuous, just as being scarlet is a specific way of being red and being azure is a specific way of being blue.

3.3.3. Why Accept Stark Alternative? Good question, subheading. Here’s your answer. Stark Alternative better accounts for what motivates X-Firsters in how it explains how normativity is both structured and unified.¹⁹ One motivation for X-Firsters is that normativity is structured: relations of explanatory priority or relative fundamentality obtain between reasons and oughts and values and so on.²⁰ Of course, X-Firsters disagree about the

¹⁸ Interestingly, the Virtues First view is neither Aristotelian nor neo-Aristotelian. See Hirji (forthcoming) on the structure of each of those views: neither puts the aretaic “on first.” It is also not Hurley’s view. Hurley embraces a kind of Rawlsian coherentism. ¹⁹ There are also some off-the-shelf arguments that I could appeal to in motivating Stark Alternative, e.g. Armstrong (1997: 50) about asymmetric necessitation; see also Wilson (2012: 8). I won’t appeal to those arguments. I don’t like them. ²⁰ That consideration is clear in Schroeder (2018)’s reconstruction of the motivation behind Reasons First: since Ross (1930), many thought that reasons explain oughts, then generalized from the “core case” of ought: reasons are prior to the rest of normativity. The same consideration crops up with Fittingness Firsters like McHugh and Way (2016: 577): since “the normative and evaluative domain seems highly interconnected, the hypothesis that there is one basic normative property has significant explanatory promise.” They say

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

62

Daniel Wodak

direction of those relations of explanatory priority. But they agree that these explanations all bottom out in the same place: some highly general feature like reasons or fittingness or value. However, Stark Alternative also accepts that normativity is structured. By analogy: color is a determinable which is explained in terms of its determinates like red and blue, which are themselves determinables that are explained in terms of their determinates like crimson and cerulean. So according to Stark Alternative, normative explanations bottom out in different highly specific places—e.g. on the Virtues First view, in courage and conscientiousness. So far this only shows that a motivation for X-First is neutralized, and in a way that further supports a point I made earlier: the good can be prior to the right without having to be prior to every other normative property. Further, I want to argue that Stark Alternative is a more plausible account of the structure of normativity than X-First. Our views about how normativity is structured should be in sync with our views about the pattern of normative explanations. If some X is “on first,” the pattern of the explanation for why x is morally right or aesthetically good is like the pattern of explaining why ABC is an isosceles triangle or an oblique triangle: we start with one general property (being trilateral) then build to the specific properties. By contrast, if normativity is a determinable, the pattern of normative explanations is like the pattern of explaining color facts: we move from the specific to general. The latter seems much more promising. That is: an act is normatively valanced (colored) in virtue of being prima facie wrong (red), and prima facie wrong (red) in virtue of being unjust (scarlet) or being harmful to others (crimson) or whatever. And an act can also be normatively valanced (colored) in virtue of being prudentially good (blue), and prudentially good (blue) in virtue of being an achievement (teal) or being pleasurable (aqua). These explanations start with the specific properties, and they don’t all start in the same place. Another motivation for X-First is that normativity is unified. Concerns about whether a category is unified are often run together with concerns about whether it is a genus. This occurs in the literatures on causation and grounding (see Schaffer 2015, 2016: 152–3). And it occurs in metaethics.²¹ The thought seems to be: Unless there is a single essential feature of normativity, how can normativity be one thing? Since the essential feature nothing about why it is attractive to explain these interconnections by taking one X to be on first; they don’t seem to countenance that interconnectedness is otherwise explicable. ²¹ Here’s Kearns and Star (2009: 215): “Philosophers have distinguished between species of reasons in a number of ways (moral/prudential/aesthetic, practical/theoretical etc.), but it is commonly thought that no unified and informative analysis of the genus is possible.” Why must reasons be a genus? Why must its varieties be species of reasons?

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Who’s on First?

63

of normativity is presumably normative (i.e. a feature like reasons or values or fittingness), X-Firsters are home free. Or so it may seem. It may also seem that Stark Alternative is hopeless on this front. After all, the view holds that normativity is fundamentally disjunctive. A common, recurring line of thought in metaethics has been that this entails disunity: If what it is to be a reason were fundamentally disjunctive, then reasons would have about as much in common as pieces of jade, which, beyond being called “jade”, have no more distinctively in common than do moons of Jupiter and natural numbers smaller than 17. (Schroeder 2007: 69; see also 60)²²

Two points here. First, I’ll play defense. Arguments like Schroeder’s are meant to show that any fundamentally disjunctive analysis is disunified. But consider the rhetorical use of gruesome, gerrymandered disjunctions like jadeite-or-nephrite or, to use an example from Michael Smith (2017: 102), “numbers-or-dogs.” The problem with jadeite-or-nephrite—that is, NaAlSi₂O6-or-Ca₂(Mg,Fe)₅Si₈O₂₂(OH)₂—and numbers-or-dogs is what gets disjoined, not the mere presence of “or.” (Indeed, X-Firsters can’t say the disjunction is gruesome just because of the “or”; in a crucial sense that’s often ignored, their views are disjunctive too.²³) Intuitively, an analysis of color in terms of red-or-blue-or- . . . , and of red in terms of crimson-or-scarlet-or- . . . , does not look at all like an analysis of anything in terms of jadeite-or-nephrite or numbers-or-dogs. This idea is common ground among those who think that determinables are explained in terms of the disjunction of their determinates. As Rosen (2010: 128) notes, not “any old disjunction of properties suffices to define a determinable with the disjuncts as determinates.” What makes the disjunction of determinates different from any old disjunction is that they exhibit sufficient objective yet inexact similarity: “objective” because the similarity in question is not just located in something about us (e.g. we call distinct minerals “jade”), but “inexact” because the similarity is not explained in terms sharing a single more fundamental property. This might sound opaque, but there are ways of cashing it out. Determinates are more specific than determinables. But that’s not all that they are. Red and square is not a determinate of red, even though it is more specific ²² I point to some recurrences of this line of thought below. A kind referee noted another: Schroeder (2015: 381) on analyzing practical and epistemic normativity. ²³ Why? Consider Fittingness First. It can’t hold that to be normative is to be explained in terms of fittingness: after all, fittingness is normative, and can’t explain itself. So the view must be that for x to be normative is for x to be fittingness or for x explained in terms of fittingness. The same holds for Reasons First and Values First. Thanks to Daniel Fogal and Chris Howard for helpful discussion of this point. See Shapiro (n.d.) for an independent (and better) argument that X-First accounts must be disjunctive.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

64

Daniel Wodak

than red: “For an object to have a determinate property is for that object to have the determinable properties the determinate falls under in a specific way” (Funkhouser 2006: 548). One way of being red (e.g. scarlet) will always differ from another (e.g. crimson), but it will differ along certain dimensions: hue, brightness, saturation. Squareness has nothing to do with these dimensions, so being red and square isn’t a way of being red.²⁴ This talk of determination dimensions is central to how Funkhouser unifies determinables. The dimensions that can determine a determinable generate regions of “property space,” and that gives us an account of objective but inexact similarity (proximity in property space), as well as conditions under which a disjunction is a determinable (2006: 554–6; Fine (2011) takes a similar approach in terms of regions of “state space”). Applied to normativity, the task would be to specify the relevant determination dimensions, which would help us pick out the relevant determinates. (We’d also need to specify the levels of determination.²⁵) The Virtues First view sketched above is one option for proponents of Stark Alternative. I’ve already implicitly suggested another (in terms of normative valences). There are more. But I want to stick to the big picture of how Stark Alternative accounts for the unity of normativity, which is via appealing to how the determinates that are disjoined exhibit objective yet inexact similarity, and thereby form a region of property space. Even if you reject this appeal to “objective yet inexact similarity,” you should acknowledge that metaphysicians are confident that some view can be plugged into its place. As Wilson notes, “determinables clearly terminologically, metaphysically and formally unify their determinates,” and this is a consensus point in the literature even though “philosophers very commonly assume that determinables are reducible to disjunctions of determinates” (2014: 568). This suggests that the view that disjunctiveness entails disunity is, at least, highly metaphysically contentious. It’s not something that metaethicists should take for granted, or defend via rhetorical appeals to gerrymandered disjunctions like numbers-or-dogs. So Stark Alternative can account for the unity of normativity and thereby neutralize this motivation for being an X-Firster. But it can also do more: it can offer a more satisfying account of the unity of normativity. My case for this is similar to why some have proposed that other philosophically interesting categories are best understood in terms of ²⁴ That said, conjunctive properties like being red and square are still partly explained in terms of their conjuncts, and in that sense are still in a rough sense color properties. The same holds for conjunctive properties where one conjunct is normative and the other is non-normative. Thanks to Mark Schroeder for a very helpful discussion of this point. ²⁵ Thanks to a referee for noting this problem. I wish I knew how to solve it.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Who’s on First?

65

determinables. For example, some suggest that taking the grounding relation to be a determinable could capture how it is both “unitary and variegated” (Bliss and Trogdon 2016: §1). Stark Alternative similarly promises to account for how normativity is unified while capturing the deep variegation in the varieties of the normativity. To tease this out, return to how X-Firsters unify normativity. Say we’re Reasons Firsters, and we say normativity is unified insofar as it is the “domain of reasons.” That seems straightforward. But it isn’t. The view doesn’t unify normativity unless we can unify reasons—or, more aptly, unless we can unify the reasons relation.²⁶ And here challenges arise. Consider the arguments for the view that some reasons are internal (or desire-dependent) and for the view that some reasons are external (or desireindependent). Reasons Firsters claim that we cannot accept both: that’d make their account of normativity fundamentally disjunctive, and hence disunified. It is in this context that we get Schroeder’s rhetoric about jadeiteor-nephrite. Similarly, Sobel argues that, with respect to reasons internalism, “the reasons provided by desires in matters of mere taste are the thin end of the wedge.” If reasons were internal-or-external we’d end up with a “fundamentally disunited” account (2016: 297).²⁷ There are other similar challenges. Smith’s rhetoric about numbers-ordogs crops up in his discussion of putative differences between practical and epistemic reasons. An account on which the two turn out to be fundamentally different would, he says, make normativity a “ragbag.” The general point is that there may be deep variegation in the varieties of reasons relations, or more broadly in the varieties of normativity. Internal and external reasons and practical and epistemic reasons aren’t the only examples. Consider the deontic, the evaluative, and the aretaic; the moral, the prudential, the aesthetic, the rational; and subjective (belief- or evidencerelative) and objective (fact-relative) standards. They strike me as being significantly and deeply distinct, in a way that the varieties of triangles (obtuse and acute; equilateral and isosceles) do not. X-Firsters struggle to explain such differences; that’s why Schroeder and Sobel think that because they accept internal reasons, they must deny the existence of external ²⁶ As I noted in my (2019), unifying causes (e.g.: they’re all events) does not unify causation. The relatum is not the relation. Metaethicists sometimes focus on unifying reasons (e.g.: they’re all propositions) but this unifies a relatum, not the relation itself. ²⁷ Some adopt such disjunctive accounts. On Chang’s hyrid view, “there is no univocal answer to the question, What metaphysically makes a fact have the normativity of a reason?” (2013: 177). But even some defenders of this view concede that it is “metaphysically unsatisfying” because it is not a “unified account” (Behrends 2015: 172). See also Cuneo (2007: 64) for a different disjunctive approach to key normative notions.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

66

Daniel Wodak

reasons. But Stark Alternative makes space for such deep variegation in the varieties of normativity because it can explain how normativity is unified while also being fundamentally disjunctive.

3.3.4. Objections to Stark Alternative I’ve offered an alternative view that is schematic, but well-motivated: it better explains two central considerations that were thought to favor the XFirst program. There are, of course, objections to Stark Alternative. I won’t discuss them all; some are technical issues that don’t strike at the core of the view under consideration.²⁸ I’ll focus on the two deepest problems. First, entailment. A common view is that determinates entail the relevant determinable.²⁹ If this is right, it poses problems for various ways of spelling out Stark Alternative. Take a view on which normativity is a determinable whose determinates are verdictive properties like rightness, and rightness is in turn a determinable whose determinates are right-making properties. The problem is that the presence of right-makers doesn’t entail rightness; they can be defeated or disabled. (This mirrors a long-standing objection to Ought First views about reasons.³⁰) One way to avoid this problem might be take all determinables and determinates to be either prima facie or pro tanto. But even then, problems could recur once we factor in how the absence of enablers and the presence of disablers can block entailments. It might be, then, that our model of the contributory properties must be holistic to preserve such entailments (à la Fogal 2016). Second, exclusion. A common view is that determinates at the same level of determination exclude each other. A ball can’t be fully red and fully blue at the same time. But it can be both morally and aesthetically good for Peter

²⁸ For instance, a determinate property must be attributed to the same object as its determinable (see e.g. Berker 2017: n. 35), but reasons, fittingness, and value are not attributed to the same objects. I think we can specify the relevant properties in such ways that they are attributable to the same objects, but I admit that this requires work. ²⁹ One could also have the view that complex determinables generate a reverse entailment: any object that instantiates the determinable (sound) must instantiate each determinate (pitch, volume, timbre) at least to some degree. (Thanks to Jan Dowell and Pekka Väyrynen here.) If so, this poses a similar problem. Perhaps some things can be normatively valanced by being to some degree right and good, without being to any degree virtuous or vicious. I think this problem might just arise from over-generalizing from the case of sound. Something can have a taste by being sweet and/or salty, without being to any degree sour or bitter. For examples, see most British culinary inventions. ³⁰ Cf. Toulmin (1950: ch. 11) and Schroeder (2007: 35–6), and for contemporary responses to the problem see Nebel (forthcoming); Alvarez (2010); Broome (2004).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Who’s on First?

67

Singer to gracefully scoop a drowning child out of a shallow pond: these properties don’t exclude each other, even though they are putative determinates at the same level of determination. This strikes me as a harder challenge. But I’m not sure if it’s ultimately a successful one. I suspect that the challenge arises only when we bring in totality facts: a ball can be red and blue, but it cannot be fully red and fully blue. (What does “fully” contribute beyond specifying that the ball is red, and that’s it?) Once we add in totality facts in the normative case, the problem disappears. If it is aesthetically good for Singer to gracefully scoop up the drowning child and that’s it, then it cannot also be morally good for him to do so. I’m not sure whether these responses to the problems above are adequate. And they only scratch the surface of the challenges that we face in taking normativity to be a determinable in the way that I’ve proposed. So let me offer three further points about the project at this incipient stage. First, these problems might beset some versions of Stark Alternative but not others, depending on how we specify the determination dimensions and so on. This is one reason why I’ve kept the account schematic. To get a sense of some of the variety of options here, consider a different example of a determinable: “Pitch, timbre, and loudness may be properties in their own right, but many think they also combine to form a unity—the property sound” (Funkhouser 2006: 553). One option for Stark Alternative would be to take deontic, evaluative, and aretaic valences to be properties in their own right which combine to form normativity. This is very different from the Virtues First account sketched earlier. Just as an objection to Ought First need not rule out X-First per se, an objection to (e.g.) Virtues First need not rule out Stark Alternative per se. Second, these problems might beset all versions of Stark Alternative, but not all views in the ballpark. Stark Alternative says normativity is a determinable. Certain commitments about the determinable–determinate relation may generate potential problems for the view. But perhaps we can water down those commitments while keeping its motivations intact. Consider, for instance, Bennett’s discussion of how to explain the unity of the grounding relation by appealing to “resemblance classes”: [R]esemblance classes are more straightforward and better understood than talk of determinables or genuses. Indeed, I’m not entirely sure what the difference is between a determinable property and the disjunctive property that is the result of disjoining the members of a reasonably natural resemblance class. (2017: 20)

What would be an example of a reasonably natural resemblance class whose members are unified when disjoined? You can probably guess:

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

68

Daniel Wodak

Consider the colors. Whatever we take colors to be, and whatever exact kind of unity is in play in this case, we can all agree that the colors form a unified family. That’s why we have the general label “the colors”. But . . . [n]o one would say that the determinable property being colored is more fundamental than the more determinate color properties like being red, being blue, and being green. Nor would anyone say that really there is just one property—Color with a capital “C”?—which is very general, can be had in different ways by quite different things, and whose pattern of instantiation makes true all color sentences, like “this mug is red” and “my shirt is black”. (2017: 23–4)

The colors form a resemblance class, even though there is no X—like being colored, or Color, or whatever—that is “on first.” Given this, in virtue of what do the colors form a resemblance class? On her view, “a family of properties and relations are unified” as a resemblance class when they are “objectively similar to each other”; or, more aptly, they are more or less unified depending on whether they are more or less objectively similar (2017: 19). This relies on a notion of objective, inexact similarity, as before. But taking normativity to be a “natural resemblance class” may let us avoid commitments that are thought to be definitive of determinables, without putting just one property—or Normativity with a capital “N”—on first. Third, even if all views in the ballpark of Stark Alternative face decisive objections, I still don’t think we should all just agree that there’s some X “on first” and merely disagree about what it is. The challenge from the previous section remains: there seem to be many philosophically interesting categories that don’t obviously allow for, let alone insist upon, an X-First approach. Taking normativity to be a determinable or a resemblance class is just one way to depart from the X-First program.

3 . 4 . C ON C L U S I O N The “Who’s on first?” debate has pride of place in contemporary metaethics. It’s high time we ask why. Why should we think that some normative property or relation or concept is “on first”? And if nothing is “on first,” it’s high time we provide alternatives to the X-First program. I’ve tried to make progress on both fronts. But I want to conclude with a methodological point about how the debate should proceed. The best path is to see what we want from a general theory of normativity, then see what models best fit those motivations. I’ve argued that we should want a model on which normativity is like color in two ways: normative explanations go from the more specific to the more general, and normativity can be unified despite deep variegation between the varieties of normativity. I’ve argued

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Who’s on First?

69

that these motivations point towards Stark Alternative, or at least something like it. My main complaint about the X-First program is that it is Procrustean. It does not start by investigating the nature of the phenomenon we want to explain and looking at what models provide the best fit. It starts by assuming that some X must be on first, then tries to chop and change normativity to fit the theory.³¹

References Alvarez, M. (2010). Kinds of Reasons. Oxford: Oxford University Press. Armstrong, David M. (1997). A World of States of Affairs. Cambridge: Cambridge University Press. Behrends, Jeff (2015). “Problems and Solutions for a Hybrid Approach to Grounding Practical Normativity,” Canadian Journal of Philosophy, 45(2): 159–78. Bennett, Karen (2017). Making Things Up. Oxford University Press. Berker, Selim (2017). “The Unity of Grounding,” Mind 127(507): 729–77. Bliss, Ricki, and Kelly Trogdon (2016). “Metaphysical Grounding,” Stanford Encyclopedia of Philosophy. Broome, John (2004). “Reasons” in R. Jay Wallace (ed.), Reason and Value: Themes from the Moral Philosophy of Joseph Raz. Oxford: Oxford University Press, 28–55. Broome, John (2013). Rationality through Reasoning. Malden, MA: Wiley. Chang, Ruth (2013). “Grounding Practical Normativity: Going Hybrid,” Philosophical Studies 164 (1): 163–87. Chappell, Richard Yetter (2012). “Fittingness: The Sole Normative Primitive,” The Philosophical Quarterly 62(249): 684–704. Cuneo, Terence (2007). The Normative Web: An Argument for Moral Realism. Oxford: Oxford University Press. Fine, Kit (2011). “An Abstract Characterization of the Determinate/Determinable Distinction,” Philosophical Perspectives 25(1): 161–87. Fogal, Daniel (2016). “Reasons, Reason, and Context.” in Errol Lord and Barry Maguire (eds), Weighing Reasons. Oxford: Oxford University Press. Funkhouser, Eric (2006). “The Determinable-Determinate Relation,” Noûs 40 (3): 548–69. Gibbard, Allan (2006). “Moral Feelings and Moral Concepts,” Oxford Studies in Metaethics 1: 195–215. Heuer, Ulrike (2004). ‘Raz on Reasons and Values’ in R. J. Wallace, P. Pettit, S. Scheffler, and M. Smith (eds), Reason and Value: Themes from the Moral Philosophy of Joseph Raz. Oxford: Oxford University Press.

³¹ Many thanks to Selim Berker, Jan Dowell, Daniel Drucker, David Faraci, Daniel Fogal, Sukaina Hirji, Chris Howard, Barry Maguire, Katy Meadows, Nala, Samuel Preston, Mark Schroeder, Laura and Francois Schroeter, Gabriel Shapiro, Jack Woods, Pekka Väyrynen, audiences at The Madison Metaethics Workshop and the University of Melbourne, two referees, and more people than I can name within the word count.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

70

Daniel Wodak

Hirji, Sukaina (forthcoming). “What’s Aristotelian about NeoAristotelian Virtue Ethics?” Philosophy and Phenomenological Research. Howard, Christopher (2019). “The Fundamentality of Fit” in Russ Shafer-Landau (ed.), Oxford Studies in Metaethics, 14. Oxford: Oxford University Press. Hurley, Susan (1985). “Objectivity and Disagreement” in Ted Honderich (ed.), Morality and Objectivity: A Tribute to J. L. Mackie. New York: Routledge & Kegan Paul. Inwagen, Peter van (2004). “A Theory of Properties” in Dean W. Zimmerman (ed.), Oxford Studies in Metaphysics, 1. Oxford: Clarendon Press, 107–38. Kearns, Stephen, and Daniel Star (2009). “Reasons as Evidence,” Oxford Studies in Metaethics 4: 215–42. McHugh, Conor, and Jonathan Way (2016). “Fittingness First.” Ethics 126(3): 575–606. McPherson, Tristram (2011). “Against Quietist Normative Realism,” Philosophical Studies 154: 223–40. McSweeney, Michaela Markham (2019). “Following Logical Realism Where it Leads,” Philosophical Studies 176(1): 117–39. Maguire, Barry (2016). “The Value-Based Theory of Reasons,” Ergo, an Open Access Journal of Philosophy 3. http://dx.doi.org/10.3998/ergo.12405314.0003.009. Moore, G. E. (1903). Principia Ethica. Cambridge: Cambridge University Press. Nebel, Jacob M. (forthcoming). “Normative Reasons as Reasons Why We Ought,” Mind. Oddie, Graham (2005). Value, Reality, and Desire. Oxford: Oxford University Press. Parfit, Derek (2011). On What Matters, vol. 1. Oxford: Oxford University Press. Rosen, Gideon (2010). “Modality: Metaphysics, Logic, and Epistemology” in Bob Hale and Aviv Hoffmann (eds), Modality: Metaphysics, Logic, and Epistemology. Oxford: Oxford University Press. Rosen, Gideon (2017). “Abstract Objects,” Stanford Encyclopedia of Philosophy. Ross, William David (1930). The Right and the Good. Oxford: Oxford University Press. Rowland, Richard (2017). “Reasons or Fittingness First?” Ethics 128(1): 212–29. Scanlon, Thomas (1998). What We Owe to Each Other. Cambridge, MA: Harvard University Press. Scanlon, T. M. (2014). Being Realistic about Reasons. Oxford: Oxford University Press. Schaffer, Jonathan (2015). “Grounding in the Image of Causation,” Philosophical Studies, 173(1): 49–100. Schaffer, Jonathan (2016). “Ground Rules: Lessons from Wilson” in Kenneth Aizawa and Carl Gillett (eds), Scientific Composition and Metaphysical Ground. New Directions in the Philosophy of Science. London: Palgrave Macmillan UK, 143–69. Schroeder, Mark (2007). Slaves of the Passions. Oxford: Oxford University Press. Schroeder, Mark (2015). “Is Knowledge Normative?” Philosophical Issues 25(1): 379–95. Schroeder, Mark (2018). “Getting Perspective on Objective Reasons,” Ethics 128(2): 289–319.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Who’s on First?

71

Smith, Michael (2017). “Parfit’s Mistaken Metaethics,” in Peter Singer (ed.), Does Anything Really Matter? Essays on Parfit on Objectivity. Oxford: Oxford University Press. Snedegar, Justin (2016). “Reasons, Oughts, and Requirements,” Oxford Studies in Metaethics 11: 155–81. Sobel, David (2016). From Valuing to Value: A Defense of Subjectivism. Oxford: Oxford University Press. Stoljar, Daniel (2017). “Physicalism.” Stanford Encyclopedia of Philosophy. Tappolet, Christine (2004). “Through Thick and Thin: Good and its Determinates,” Dialectica 58: 207–21. Thomasson, Amie L. (1999). Fiction and Metaphysics. Cambridge: Cambridge University Press. Toulmin, Stephen Edelston (1950). An Examination of the Place of Reason in Ethics. Cambridge: Cambridge University Press. Väyrynen, Pekka (2011). “A Wrong Turn to Reasons?” in Michael Brady (ed.), New Waves in Metaethics. Basingstoke: Palgrave-Macmillan. Wilson, Jessica (2017). “Determinables and Determinates.” Stanford Encyclopedia of Philosophy. Wilson, Jessica M. (2012). “Fundamental Determinables,” Philosophers’ Imprint 12. Wilson, Jessica M. (2014). “No Work for a Theory of Grounding.” Inquiry 57(5–6): 535–79. Wodak, Daniel. (2019). “Mere Formalities: Normative Fictions and Normative Authority,” Canadian Journal of Philosophy 49(6): 828–50. Wodak, Daniel (2019). “An Objectivist’s Guide to Subjective Reasons,” Res Philosophica 96(2): 229–44. Wodak, Daniel (Forthcoming). “Redundant Reasons,” The Australasian Journal of Philosophy.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

4 Excuse without Exculpation The Case of Moral Ignorance Paulina Sliwa 4 . 1. I N T R O D U C T I O N As epistemically limited agents, we are prone to mistakes. Sometimes these mistakes are about what we are morally required or permitted to do. Such mistakes about moral matters can come about in two ways. They can result from ignorance about nonmoral features of the world—features on which the moral status of our action supervenes. I may be oblivious that the cake I am offering you contains poison and that’s why I believe it’s permissible for me to pass you a slice. But even knowing all relevant nonmoral facts does not eliminate the possibility of error. Many moral questions are hard and rife with opportunities for mistake. We may fail to recognize facts as morally relevant, misjudge their significance, fail to reason properly about how various competing morally relevant factors weigh up, or be guided in our deliberations by false moral principles. Let’s reserve the term moral ignorance to refer to this second kind of moral error—moral error that does not derive from ignorance about nonmoral facts. I am construing ignorance and error broadly, to cover both false belief and absence of true belief, though my focus will be on the former. It’s uncontroversial that nonmoral ignorance can function as an excuse. When accused of poisoning my friend, I can appeal to the fact that I didn’t know that the cake contained poison to defend myself. And insofar as I really did not know and my ignorance reflects neither recklessness nor negligence on my part, that defense is a good one. But what if my wrongdoing resulted from moral ignorance—I knew that the cake contained poison but thought it permissible to feed it to my friend regardless? This question invites considerable disagreement. According to

Paulina Sliwa, Excuse without Exculpation: The Case of Moral Ignorance In: Oxford Studies in Metaethics Volume 15. Edited by: Russ Shafer-Landau, Oxford University Press (2020). © Paulina Sliwa. DOI: 10.1093/oso/9780198859512.003.0004

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Excuse without Exculpation

73

Gideon Rosen and Michael Zimmerman, cases of moral and nonmoral ignorance are exactly analogous.¹ If my failing to know that the cake contained poison can function as an excuse, so can, in principle, my failing to know that it’s wrong to poison. According to others, there is a deep asymmetry between nonmoral and moral ignorance: only the former can ever function as an excuse.² In this chapter, I want to home in on an important assumption that frames the current debate: that excuses are all-or-nothing affairs and that to have an excuse is to be blameless. I argue that we should reject this assumption. Excuses are not binary but gradable: they can be weaker or stronger, mitigating blame to greater or lesser extent. I explore the notions of strength of excuses, blame mitigation, and the relationship between excuses and moral responsibility. These ideas open up some principled middleground between the two positions staked out in the literature. Moral ignorance may well excuse but it does not exculpate.

4.2. THE LIMITS OF OUR INTUITIONS A B O U T C AS E S Let me start by charting the terrain of the present debate. One camp maintains that the morally ignorant wrongdoer is blameless, as long as the ignorance itself is non-culpable—that is, the result of neither negligence nor recklessness. Rosen appeals to our intuitions about various cases to support this conclusion. He invites us to consider a slaveholder in ancient times who harbors no illusions about the fact that slaves are humans just like him. He also believes it’s really bad luck to be a slave. Nevertheless, he believes that it is permissible to own slaves. We can imagine filling in the details so that, in reaching his belief, he has been neither negligent nor reckless. Slavery is, in his circles, not a controversial practice. It may be possible to arrive at the conclusion that it is morally abhorrent but doing so would take extraordinary moral vision. We expect each other to be morally competent, not to be moral visionaries. According to Rosen, once we grant and attend to the fact that the slaveholder’s moral ignorance is non-culpable, we will also regard him as blameless for owning slaves. Rosen argues: insofar as [the biblical slaveholder] acts from blameless ignorance, it would be a mistake for us to blame the slaveholder—to feel anger or indignation directed at him

¹ See Rosen (2003, 2004, 2008), Zimmerman (1997). ² See e.g. Harman (2011 and 2015), Arpaly and Schroeder (2013), FitzPatrick (2008).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

74

Paulina Sliwa

for his action. If the historical situation is as we have supposed, then the appropriate attitude is rather a version of what Strawson calls the “objective” attitude.³

Rosen also considers the case of a ruthless capitalist who, like everyone around him in the enclave of the very rich, has been groomed to believe that it is right to exploit the poor. Again, he suggests that once we are clear in our minds that the ruthless capitalist is non-culpable for his moral-epistemic predicament and his resulting moral ignorance, we will not regard him as blameworthy for his wrongful exploitation of the poor, either: I contend that if you are careful to bear in mind the stipulation that in reaching his conclusion our capitalist has not been reckless or negligent in the management of his moral opinion, you will find it plausible that his moral ignorance is not his fault. I further contend that if you bear this thought fully in mind, you will find it equally compelling that it would be a mistake to blame him for the wrong he does.⁴

The second camp—Elizabeth Harman, Nomy Arpaly and Timothy Schroeder, William FitzPatrick—deny that moral ignorance can constitute an excuse. Arpaly and Schroeder consider the case of an ancient Roman who goes to the circus, arguing that his false moral beliefs do nothing to excuse his participation in and enjoyment of the grisly spectacle. His moral ignorance fails to excuse, even when it is non-culpable: Consider the ancient Roman who goes to the circus because he heartily enjoys watching people thrown to the lions. We think this person is blameworthy for going to the circus. Enjoying other people’s suffering in this manner speaks ill of the agent’s will even if the enjoyment in question is encouraged by a corrupt and corrupting society, and even if there is no moral theory available that disagrees.⁵

Elizabeth Harman points to structurally similar cases involving mob and gang members: Max works for a Mafia “family” and believes he has a moral obligation of loyalty to the family that requires him to kill innocents when it is necessary to protect the financial interests of the family. This is his genuine moral conviction, of which he is deeply convinced. If Max failed to “take care of his own” he would think of himself as disloyal and he would be ashamed. Gail is a gang member who believes that she has a moral obligation to kill a member of a neighboring gang as revenge after a member of her own gang is killed, although her victim was not responsible for the killing. This is her genuine moral conviction, of which she is deeply convinced. If Gail failed to “take care of her own” she would think of herself as disloyal and she would be ashamed.

³ Rosen (2003: 66). ⁴ Rosen (2004: 305). ⁵ Arpaly and Schroeder (2013: 182–3).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Excuse without Exculpation

75

She suggests that we do not consider agents like Max and Gail to be blameless for their actions: I claim that Max and Gail are paradigm cases of agents blameworthy for their wrongful actions. They know that they are killing innocent people; this is sufficient for the agents to be blameworthy.⁶

Harman concludes that moral ignorance can never be non-culpable. Hence, not only does it not constitute an excuse but it functions as an aggravating factor. Why? According to Harman, an agent is morally blameworthy if her action manifests a lack of responsiveness to right-making reasons. There are two ways of being responsive to right-making reasons: motivationally and cognitively. The wrongdoer who knowingly commits a wrong is cognitively responsive to right-making reasons—she recognizes them as evidence on which to base her moral beliefs—yet fails to be responsive to them in her motivations. The morally ignorant wrongdoer, on the other hand, is insensitive to right-making reasons twice over. Not only is she not moved by them, she also fails to believe the moral truth on the basis of them. So much for the dispute. What are we to make of it? It is difficult to see a path forward. The cases both camps appeal to are structurally identical—the ancient slaveholder and the Roman who enjoys going to the circus, the ruthless capitalist and the gang member. Yet the two camps report opposite intuitions about them. We have arrived at an impasse. How do we proceed? I suggest that we home in on a hitherto unexamined assumption that frames the debate: that to ask whether moral ignorance constitutes an excuse is to ask whether the agent is blameless for whatever she has done.

4 .3 . AN UN E X A M I N E D A S S U M P T I O N A B O U T EX C U S E S Let’s think about this for a moment. Why think that, if moral ignorance excuses, it renders the wrongdoer blameless? Behind this assumption lies a general view about the nature of excuses that is not confined to this particular debate. This view conceives of excuse as a binary notion. Excuses are considerations that switch off blame; they are blame-negators.

⁶ Harman (2015: 65). FitzPatrick (2008) cites a similar example, involving a ruthless business man who believes there is nothing wrong with exploiting the poor.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

76

Paulina Sliwa

Rosen is explicit about the fact that this is the conception of excuses he is working with; he tells us that an excuse is a “consideration that defeats the standing presumption of blameworthiness.”⁷ But Rosen is hardly alone— indeed, this view of excuses is close to orthodoxy in moral philosophy. For an illustration, consider: Calhoun: A good excuse gets one off the evaluative hook. To be excused is to have no reason to think badly of oneself or for others to think badly of oneself.⁸ Owens: to recognize that someone has an excuse . . . is to recognize that blame is inapt already.⁹ Murphy: if resentment and forgiveness are to have an arena, it must be where such wrongdoing remains intact—i.e., neither excused nor justified.¹⁰ Baron: to excuse is to say that what the agent did was wrong, or at least untoward, but that it would be unfair to blame him for the action.¹¹ Wallace: Excuses . . . aim precisely to challenge the claim (or suspicion) that S was morally responsible for x; they adduce conditions that make it unfair to hold S morally responsible for x. Now to hold a person responsible for a particular action x that is morally wrong is to regard the person as having done something blameworthy; so excuses . . . may be considered “blameworthiness inhibitors”.¹²

All these authors subscribe to the view that, if an agent has an excuse, she is not to blame for the wrong she committed. The idea that excuses are all-ornothing affairs is thus very common. This rare moment of philosophical consensus should strike us as puzzling. For, on the face of it, it is very natural to think of excuses as admitting of different strengths. After all, we typically think of blame as coming in different amounts. We can deserve more or less of it, we can be more or less to blame for whatever it is we have done. It makes sense to wonder how much someone is to blame for some bad outcome or consequence. It also makes sense to ask comparative questions: whether Max or Sam is more to blame for the sad demise of their friendship, for example. Plausibly, thus, blame admits of a measure— not a cardinal but an ordinal one. And insofar as blame admits of different

⁷ Rosen (2004: 298). ⁸ Calhoun (2015: 28–9). ⁹ Owens (2012: n. 25). ¹⁰ Murphy (1982: 506). ¹¹ Baron (2007: 22–3). ¹² Wallace (1994: 121). Other examples: Allais (2008: 34–5). Rosen (2008: 609), Baron (2007: 27), Hieronymi (2001: 530).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Excuse without Exculpation

77

amounts, we would naturally expect that excuses come in different strengths—the stronger the excuse, the less the agent is to blame. Second, when we think of the various considerations that constitute excuses, these typically come in degrees: emotional upset, such as stress or fear, physical discomfort, such as nausea, pain, or tiredness, can all vary in their intensity. Coercion can plausibly vary in strength, depending on what is at stake for the coerced party. Given that the considerations which constitute excuses come in varying intensities, it seems natural to think that excuses themselves will admit of degrees. Again ordinary experience lends support to this idea: if you are to miss your friend’s birthday party, a debilitating migraine makes for a stronger excuse than a slight headache. This suggests an alternative conception of excuses, on which excuses are considerations that mitigate rather than negate blame. On this conception, excuses lessen how much an agent is to blame for a given wrong. They come in different strengths: some excuses excuse more than others. Let’s be clear on what exactly negating and mitigating accounts of excuses agree and disagree on. Both accounts allow that excuses can be partial. An excuse is partial when it excuses some of the wrong that someone has done without excusing all of it. Sometimes our actions violate several moral norms at once. Jones’s remark may have been both racist and sexist. He may have an excuse for its being sexist—perhaps he was speaking in a foreign language and wasn’t sensitive to the sexist overtones of his words—without having an excuse for its being racist—that it was racist was plain enough, even to him. What mitigators and negators disagree about is whether given a particular wrong W—say Jones’s broken promise to Smith that he will come to his birthday party—some excuses have greater excusing power than others. Another way of putting the disagreement is this. Negators say that excuses vary along one dimension only: in which wrongs they excuse. Mitigators, in contrast, maintain that excuses can vary along a further, independent dimension: in how strongly they excuse any particular wrong. 4.4. A DEFENSE OF BLAME-NEGATION? You might object that I am mischaracterizing what the negators are up to. When they say that excuses negate blame and moral responsibility, they do not mean to rule out that there can be considerations that only lessen blame, without negating it all the way. They are simply making a terminological choice to not refer to such considerations as excuses. Thus, negators are not making a substantive claim about our ordinary notion of excuses; rather they are stipulating that the term “excuse” will be reserved for those considerations that lessen blame maximally: namely, by negating it.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

78

Paulina Sliwa

But even if we grant that negators are stipulating, rather than advancing a substantive claim about excuses—and I am not sure that all of them are charitably interpreted as merely stipulating—there are two responses to make.¹³ First, this concession does not affect the substance of my argument in this chapter. For it remains true that the discussion of moral ignorance has centered solely on whether moral ignorance negates blame. The question whether it may mitigate it remains open and unexplored, as do questions about how to make sense of blame mitigation. Second, stipulations are not immune to criticism. Poorly chosen, they can distort theoretical inquiry. When they concern concepts that are deeply embedded in our everyday practice, stipulations need to be responsive to how those terms function in this practice. In particular, they need to preserve the central features of those concepts. Otherwise, stipulative choices risk obscuring theoretical options and distorting the phenomena under investigation. Negators have taken a concept that functions as gradable in our moral practice and offer a stipulation that is binary. This goes beyond mere philosophical sharpening. It obscures a central feature of excuses, namely that they lie on a spectrum from weak to strong. It distorts inquiry by presenting excuse and blame mitigation as two distinct normative phenomena, to be theorized separately, rather than in one unified framework. And, in the case of moral ignorance, it has obscured the possibility that moral ignorance may mitigate blame, without negating it. Can the negator respond to this charge? She may argue that her stipulation is grounded in our moral practice after all. In particular, she may point to the conceptual connections between blame, excuses, and forgiveness. Given these connections, she may argue, it does seem plausible that excuse functions as a binary term. What is excused is not to be forgiven, what is to be forgiven cannot be excused. Forgiveness renders blame inapt. Excuses preempt forgiveness. The most natural explanation for this is that excuses negate blame, so that forgiveness is rendered moot. Thus, recall Murphy’s quote from above: if resentment and forgiveness are to have an arena, it must be where such wrongdoing remains intact—i.e., neither excused nor justified.¹⁴

Murphy goes on to suggest that there’s a bit of a conceptual muddle going on in the New Testament; Jesus on the cross ought to have chosen his words more carefully: ¹³ Rosen is the only writer who is explicit about the fact that he is stipulating. See his (2004: 298). ¹⁴ Murphy (1982: 506).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Excuse without Exculpation

79

“Father forgive them for they know not what they do” would go better as “Father excuse them for they know not what they do.”¹⁵

Similarly, Owens argues: To think it appropriate to forgive an offence is not to think that blame for that offence is inapt already. Rather it is to think that you ought to render blame for that offence inapt. By contrast, to recognize that someone has an excuse (or that time has passed or that they have already been forgiven, etc.) is to recognize that blame is inapt already.¹⁶

Hieronymi suggests that making excuses is incompatible with asking for forgiveness: To ask you to understand things from my point of view is to hope for an excuse, not to ask for forgiveness.¹⁷

I find this line of defense underwhelming. It does not ring true that our moral practice treats the relationship between excuse and forgiveness as a strict either/or. For there is nothing unusual or jarring about offering an excuse and asking to be forgiven at the same time: “I have been really anxious about something at home. I’m really sorry I snapped at you.” Or: “I’m so sorry. I didn’t stand you up intentionally—the appointment completely slipped my mind. Please forgive me.” The fact that there is nothing at all jarring or unusual about such requests presents the negator with a dilemma. She can deny that, in those cases, the agent is really offering an excuse. But that’s implausible: the fact that an action was unintentional, done from anxiety or stress or from nonmoral ignorance, are typical examples of excuses and they change our moral assessment of the wrong. A second possibility is to deny that in such cases we are really asking for forgiveness. But absent further explanation, this, too, seems unattractive. In contrast, the mitigator has a straightforward way of accommodating the combination of excuse and plea for forgiveness: excuses, she says, mitigate blame. Thus, having an excuse does not mean that there is nothing left for forgiveness to do. At the same time, however, the presence of the excuse makes it easier to forgive the wrongdoer. Forgiveness is a matter of forswearing blame; that gets easier the less blame there is to forswear. Pairing excuses with apology thus makes good sense. The conceptual connections between forgiveness, blame, and excuses thus do little to bolster the negator’s claim (or motivate her stipulation) ¹⁵ Murphy (1982: 506). ¹⁶ Owens (2012: n. 25). ¹⁷ Hieronymi (2001: 554). See also Allais (2008), who stresses that the core notion of forgiveness comes into play only where wrongdoing is “unexcused and unjustified.”

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

80

Paulina Sliwa

that excuses are binary. On the contrary, our moral practice lends support to a mitigating account. That leaves us with the task of working out its details. Since on the mitigating account excuses bear on how much an agent is to blame for a given wrong, we need to spell out how to make sense of this more and less. How do we put a measure on being to blame?

4 .5 . N O R M A T I V E F O O T P R I N T S A N D AM O U N T S O F B L AM E Let me give you a sketch of the central idea first. Wrongdoing leaves normative footprints: it changes the normative landscape in characteristic and systematic ways by creating new claim-rights, obligations, and permissions. Depending on the nature and gravity of the wrong in question, these normative changes can be more or less extensive. How much an agent is to blame for a wrong corresponds to the size of the wrong’s normative footprint; that is, to the extent to which the wrong has modified the normative landscape. We can now unpack the central elements of this idea. We engage in wrongdoing when we perform actions that violate our moral obligations and/or infringe on someone’s moral rights. By engaging in wrongdoing, we place ourselves under new obligations and we give others new rights against us. Thus, if I renege on my promise to bake you a cake, I may thereby acquire a duty to acknowledge that I have let you down, to explain myself, to apologize, and to make it up to you in some way. Wrongdoing alters the normative landscape in three characteristic ways. First, it creates reparative rights and duties. These include the duty to acknowledge the wrong one has done, the duty to listen to complaints about one’s actions, the duty to explain one’s actions and motivations, the duty to apologize, the duty to compensate or otherwise make amends. Reparative duties are often directed duties, they are owed to the wronged party, who holds the correlate set of claim rights.¹⁸ Thus, you owe an apology or compensation to someone, namely the person you have wronged. But there can also be undirected reparative duties, which are not owed to ¹⁸ See Thomson (1990: 84–98). Thomson argues that violating someone’s right leads to “moral residue.” Similarly, Kramer (2005: 313) offers a detailed defense of the “Remedy Principle”: “If and only if Y owes X a moral duty not to Q, Y’s Q-ing will place Y under a moral obligation to X to remedy the resultant situation in some way.” My proposal is in the same spirit but goes further: the normative changes induced by one’s wrongdoing extend beyond remedial obligations.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Excuse without Exculpation

81

anyone in particular. Suppose there’s an undirected duty to protect the environment. If you violate this duty, you commit a wrong but without wronging anyone. And you may thereby acquire a further, undirected duty to make up for it. Thus, if you litter, you can be required to clean up after you. Second, wrongdoing changes feeling rights and duties.¹⁹ Like our actions, our emotional lives are constrained by a matrix of permissions, rights, and obligations. Social context, personal relationships, past actions—both yours and mine—affect what I may, should, or must not feel towards you.²⁰ Wrongdoing alters these norms. It can entitle the wronged party to feel anger, resentment, frustration, sadness, or disappointment. If, as a good friend, I reveal a secret you confided in me, you are entitled to be mad at me. To say that you are entitled is not to say that you must—you may be too exhausted or distracted by other things or decide it’s more prudent to remain calm. Still, you have a right to be angry even if you do not exercise it. The right came into existence because I have wronged you. And there are limits on just what kind of emotional response you are entitled to: unless the betrayal was very grave indeed, you are not entitled to murderous rage. If it was a minor indiscretion, you may be entitled to some annoyance but not to weeks and weeks of seething fury. Wrongdoing changes feeling norms for the wronged party but also for the wrongdoer. If I have knowingly betrayed your trust, I should feel ashamed and remorseful for what I have done.²¹ If I have unintentionally harmed you, I should feel regret and sorrow. For the wrongdoer these feelings are generally not merely optional; they are required. Wrongdoing thus creates characteristic feeling-rights, feeling-duties, and feeling-permissions. Third, wrongdoing changes relationship norms: for example, the right to someone’s trust, time, help, support.²² The fact that, as a friend, I betrayed your secret makes it permissible for you to withdraw your trust, to stop ¹⁹ Thus, I agree with Strawson (1962) that blame is related to reactive attitudes, such as resentment. But unlike Strawson and others, I do not think that this is the only or most central element of blame. ²⁰ See Hochschild (1975: particularly pp. 288–92). ²¹ Philosophers tend to think of emotions as governed by norms of appropriateness or fittingness, rather than by deontic norms. But the idea that that there are things we can be obligated, permitted, or prohibited from feeling—and that we regulate our emotions in light of these norms—is widespread in sociology, where it has been championed by Hochschild (1975, 1979, 2015). ²² This echoes the suggestion of Scanlon (2008), who argued that blame is connected to our relationships with others. To regard someone as to blame for an action is “to claim that the action shows something about the agent’s attitudes toward others that impairs the relations that others can have with him or her” (p. 128). Unlike Scanlon, however, I don’t think we can give an account of blame wholly in terms of impairments of relationships.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

82

Paulina Sliwa

checking in with me, to leave me off invitations for future birthday parties. The fact that your new colleague treated you badly may mean that she loses the right to your help and support—the kind of help and support that, as a more junior colleague, she would otherwise be entitled to. Let’s call the way in which a given instance of wrongdoing modifies the reparative rights and duties, feeling norms, and relationship norms its normative footprint. The normative footprint of a wrong can vary in size. Thus, not all instances of wrongdoing will modify relationship norms. Whether and how a wrong bears on relationship rights depends on its nature and gravity, and the context and relationship in which it has been committed. A minor wrong in the context of a close relationship—a late arrival to an afterwork drink, an ill-considered remark, a forgotten chore— will typically not have any repercussions for the relationship norms. Such isolated, minor wrongs do not normally licence one to withdraw one’s trust, to “cool off ” the friendship or to break it up altogether. But all instances of wrongdoing will generally induce normative changes of the first two kinds. Even a minor wrong entitles the wronged party to a bit of annoyance and a quick apology.²³ I suggest that to ask how much someone is to blame for a wrong is to ask about the size of her wrongdoing’s normative footprint. Of course, this assumes that we can make good sense of the size of a normative footprint. Are changes in the normative landscape something we can put a measure on? I think we can. Setting aside wrongdoing, we can often make comparative judgments about the extent of our rights and duties. The following statements seem all true: We owe our children more than we owe our friends. You have more obligations towards a friend than towards a stranger.

This is not to say that these judgments are terribly exact or that we can put a precise number on the set of rights and obligations in question. The measure on sets of rights and obligations will be a cardinal one, not an ordinal one. But if we can make such comparative judgments about rights and duties in general, there is no reason why we should not be able to make them about the rights and duties created by instances of wrongdoing. This chimes with our moral practice. When it comes to a particular wrong that was jointly committed by several wrongdoers, we can make comparative judgments about which culprit is more to blame than the other. We can also make comparative judgments about whether, given the presence or absence of particular factors, the culprit is more or less to blame for

²³ I discuss the issue of moral luck at the end of this chapter.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Excuse without Exculpation

83

the wrong in question. We can ask whether Jan or Marie is more to blame for their acrimonious divorce. We can ask whether, in Jan’s case, the fact that his father was dying of cancer at the time makes a difference for how much he’s to blame for it. But many other comparisons do not seem terribly sensible: is Jan more to blame for the acrimonious divorce than Marie is to blame for her mishandling of a work project? Is the NRA more to blame for the rise in school shootings than Jill for her traffic accident? It’s hard to see how there could be a fact of the matter here. I have argued that the idea of a normative footprint helps us to spell out what it is to be more or less to blame for a wrong. Let us now see how excuses fit into the picture.

4 .6 . E X C U S E S A S B L A M E - M I T I G A T O R S An excuse is a consideration that diminishes the normative footprint of a wrong. Excuses are thus normative difference-makers: they make a difference to the reparative duties, feeling rights, duties, and permissions, and relationship norms that ensue as a result of one’s wrongdoing. Excuses may make a difference to the feelings you are entitled to— annoyance or frustration, instead of anger. They may also bear on what kind of apology the wrongdoer owes: a quick, simple “I’m sorry” or a large, public gesture. If I broke my promise to bake you a cake because I had to look after an ill toddler, I may only owe you a sincere apology but not also an additional pan of brownies. You may perhaps be disappointed but not furious with me, nor are you entitled to permanently cut me off. We can measure the strength of an excuse by its impact on the normative footprint of a wrong. The stronger the excuse for a wrong, the smaller its normative footprint. How do excuses differ from justifications? Justifications, too, bear on the normative footprint of a wrong. But unlike excuses they do so by disputing the wrongness of the act in question. A justification shows the action to have been permissible, after all.²⁴ If your action is justified, no excuse is needed. In contrast, excuses are considerations that bear on the size of the normative footprint of the wrong without bearing on its moral status. That the secret slipped your tongue while you were severely sleep-deprived does not make the indiscretion any less wrong. Still, it bears on the scope of ²⁴ As Austin (1957: 2) argues: “You dropped the tea-tray: Certainly, but an emotional storm was about to break out: or, Yes, but there was a wasp. In each case the defence, very soundly, insists on a fuller description of the event in its context; but the first is a justification, the second an excuse.”

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

84

Paulina Sliwa

your reparative duties, on how you should feel about what you have done, on its implications for your relationship with the person who entrusted you with the secret.²⁵ Since excuses mitigate blame, they can make it easier to forgive a wrong. We can think of forgiveness as removing the normative footprints created by a wrong. Forgiveness releases the wrongdoer from their outstanding reparative obligations, erases the changes in feeling norms, and/or reverses changes to relationship norms. On this view, forgiveness is a normative power.²⁶

4 . 7 . T H E C O N T R A ST IV E M E T H O D With an account of excuses as blame-mitigators on the table, let us now return to moral ignorance. Is moral ignorance an excuse? This is now a question about whether moral ignorance makes a difference to the normative footprint of a wrong. Posing the question in this way naturally suggests a contrastive method. We compare two cases of wrongdoing, keeping the nature of the wrong constant but altering the agent’s moral-epistemic situation. We judge whether the normative consequences in one are more extensive than in the other. What test cases should we use? I want to steer clear of the biblical slave owners, Mafia members, and Roman circus-goers that have dominated the literature. Relying on these cases raises serious methodological worries. This is because they all involve stipulating that the wrongdoer lacks “easy” moral knowledge, i.e. knowledge of moral truths that we regard as completely obvious: that slavery is wrong, that it is reprehensible to watch religious minorities being chased and eaten by lions, that killing innocents to protect one’s criminal profits is impermissible. The fact that such stipulated moral ignorance strikes us as beyond the pale raises worries about “epistemic egocentrism.” This is a well-documented family of cognitive biases that reflects limitations in our ability to take each other’s epistemic perspective. Small children’s failure to pass the false belief test is a familiar example of an epistemic egocentrist bias. It is less appreciated that the underlying tendency to impose our own epistemic perspective on ²⁵ It is a further question which considerations constitute excuses and why excuses bear on the normative footprint in this way. Answering those is the task for a theory of excuses. This lies beyond the scope of this present chapter but I present and defend such a theory in Sliwa (2019). ²⁶ Forgiveness can thus be partial. The idea that forgiveness is a normative power has also been suggested by Warmke (2016) and Nelkin (2013).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Excuse without Exculpation

85

others persists into adulthood. In particular, we systematically fail to set aside knowledge we know to be privileged—knowledge that we know the other party lacks—for the purposes of predicting their actions: this bias is evocatively named “the curse of knowledge.”²⁷ Its influence is not accessible by introspection, nor is it under conscious control. Individuals are not aware when they are affected by it, nor can they avoid it by being made aware of its potential influence on their judgments. And insofar as we have great difficulties to set aside privileged knowledge about, for example, the value of a product, for the purposes of predicting someone’s buying behavior, it would be surprising indeed if we did not have difficulties to set aside privileged moral knowledge for the purposes of morally evaluating their actions. This presents a challenge to Harman’s and Arpaly and Schroeder’s argumentative strategy. If we share their intuitive judgment that the gang member and slave owner are blameworthy for their killings and enslavings, we now have to contend with two competing explanations for these intuitions. The explanation that Harman and Arpaly and Schroeder favor is that our attributions of blame are independent of our attributions of moral knowledge—that is, we think the gang member and slave owner are blameworthy regardless of whether they know that killing and enslaving is wrong. But an equally plausible alternative is that we simply resist the stipulation that the gang member and slave owner are morally ignorant; we think they are blameworthy partly because we take them to knowingly act wrongly. Since we succumb to epistemic egocentrism unwillingly and unwittingly, these two rival explanations are impossible to tell apart “from the inside.”²⁸ We can sidestep these methodological complications by focusing on cases of moral ignorance that are closer to home—cases where the agent’s moral mistake is neither crude nor blatant.

4.7.1. Against Harman Consider the following two scenarios: Tom and Sara are planning a wedding and both of their families have offered to contribute money towards it. Sara’s family, who are less wealthy than Tom’s, offered a certain sum, which will cover less than half of the expenses. The couple is now wondering whether it would be ²⁷ See Royzman, Cassidy, Baron (2003) for an excellent overview, including some discussion of how egocentric bias may bear on other discussions in ethics that appeal to intuitions, in particular, Nozick’s experience machine experiment. While, to my knowledge, there have been no systematic studies of egocentric bias involving moral knowledge, it would be surprising if the empirical findings did not carry over. ²⁸ Royzman et al. (2003: 60).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

86

Paulina Sliwa permissible for them to ask Tom’s family for a greater contribution. They worry that it wouldn’t be fair of them to ask one set of parents for more. They decide to ask a friend whose judgment they trust. Anna is a journalist who is preparing to go on a reporting trip to a dangerous and conflict-ridden area. She has to tell her family that she will be away but she really doesn’t know how much to tell them. If she tells them where exactly she’s going and why, they will be extremely worried. On the other hand, she worries that by evading the questions she would be lying and that that would be wrong. She goes back and forth but cannot decide what the right thing to do is and eventually decides to ask a friend whose judgment she trusts.²⁹

First, imagine that Sara and Tom’s chosen advisor tells them that it is entirely permissible to ask Tom’s family for a higher contribution to their wedding. Unfortunately, the advisor gets it wrong: doing so would be unfair. Sara and Tom believe their advisor; they have no reason to doubt her judgment. They ask Tom’s family for a greater contribution to their wedding. They make a moral mistake but unwittingly so. Similarly, Anna’s generally reliable advisor has an off day and leads her astray. She tells her it’s permissible for Anna to be so evasive as to deceive her family about her whereabouts. Anna trusts her advisor and goes along with her advice. In doing so, she makes a moral mistake. Is their moral ignorance non-culpable? I think it is. Tom and Sara, and Anna are initially uncertain about what the right thing to do is, not because they lack some crucial bit of nonmoral information, but because the questions they are facing are tricky, with competing considerations at stake. They are simply not sure how to strike the right balance between them. In such cases, it’s very reasonable to outsource one’s moral belief to someone one trusts and believes to be reliable. Relying on trustworthy moral advice will decrease one’s chance of error—but, if the advisor is human, it will not altogether eliminate it. That’s the limitation that Tom, Sara, and Anna come up against. Tom and Sara and Anna have done everything to discharge their epistemic obligations: they were genuinely conflicted, sought out a friend they had every reason to trust, and were nevertheless misled. Compare these cases to the following: Tom and Sara’s advisor tells them correctly that expecting Tom’s parents to contribute more would be unfair. They do not have any reason to doubt their advisor’s judgment and they form the relevant belief. But then they decide to ask for a greater contribution anyway. They knowingly act wrongly. Similarly, Anna’s advisor tells

²⁹ Both cases are taken from Sliwa (2012: 177–8).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Excuse without Exculpation

87

her—correctly—that being evasive would be lying; she should tell her family the truth. Anna believes her advisor but nevertheless proceeds to deceive her family. She acts wrongly, and deliberately so.³⁰ There are many reasons for which Tom and Sara, and Anna may have chosen not to act in line with their moral knowledge: selfishness, a mere whim, a desire to be contrarian. But no matter how we spell out the details—as long, of course, as we do not build in considerations such as duress, coercion, hypnosis, which themselves are excusing—we judge the agents who knowingly act wrongly more harshly than their merely mistaken counterparts. We think that in the case of deliberate wrongdoing, Tom’s parents are entitled to take greater offence for being treated unfairly, that they are entitled to feel more upset (as well as, perhaps, disappointed, taken advantage of, etc.). We are more inclined to think that Tom and Sara themselves should feel worse about what they have done—they should be ashamed of themselves—in the case where their unfair treatment is deliberate. Similarly, we judge Anna more harshly for being deceitful despite her knowledge that she should tell the truth than when she is merely acting on mistaken advice. Her family is entitled to greater disappointment and sense of betrayal, she owes them a more serious and heartfelt apology, there may be a real rupture of trust that she will have to repair. These judgments go against those who think that moral ignorance is no excuse. Arpaly and Schroeder think that how much an agent is to blame is only determined by the nature and content of their desires. This makes it impossible to differentiate the merely mistaken agent from the deliberate wrongdoer in any principled way. Both ultimately fail to be motivated by things they should be motivated by—e.g. the features of the situation which make it unfair to ask Tom’s family for a greater contribution. Indeed, the desires motivating the agent—Sara and Tom’s desire to have a lovely wedding, Anna’s desire to have peace from her anxious family—may be exactly the same when they make a moral mistake and when they knowingly act wrongly. The only difference is that in the former the agents believe their actions to be permissible, when in the latter they know their actions to be wrong. The intuitive judgments elicited by the contrastive method also decisively tell against Harman, who thinks that moral ignorance always constitutes an aggravating factor rather than an excuse. On Harman’s view, agents who act on misguided moral advice are more to blame than those who go against the advice. That’s because the mistaken agent fails to be both motivationally and

³⁰ I am assuming here that to knowingly act wrongly is sufficient for deliberately or intentionally acting wrongly. See Holton (2010).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

88

Paulina Sliwa

cognitively sensitive to the right-making reasons. They fail twice where the wrongdoer who knowingly acts wrongly fails only once. Arpaly and Schroeder and Harman are committed to the claim that moral ignorance never excuses. But the contrastive method suggests otherwise. Cases like Sara and Tom and Anna show that, at least sometimes, knowingly committing a wrong has a greater impact on the normative landscape than committing the same wrong by mistake. It follows that moral ignorance can diminish the normative footprint of a wrong.

4.7.2. Against Rosen The contrastive method also spells trouble for Rosen’s claim that moral ignorance negates blame altogether. Rosen invites us to think of wrongdoing that arises from non-culpable moral ignorance as a “mere event” rather than a genuine instance of wrongdoing.³¹ Recall, Rosen tells us that the appropriate attitude to take towards the wrongdoer acting from moral ignorance is the Strawsonian objective stance: insofar as [the biblical slaveholder] acts from blameless ignorance, it would be a mistake for us to blame the slaveholder—to feel anger or indignation directed at him for his action. If the historical situation is as we have supposed, then the appropriate attitude is rather a version of what Strawson calls the “objective” attitude.³²

This gives moral ignorance the same normative standing as considerations such as psychosis, an epileptic seizure, being drugged, being under the influence of hypnosis, etc. Moral ignorance negates both blame and moral responsibility. Let’s see if this is borne out by the contrastive method. We compare the case in which Anna acts on mistaken moral advice—just as in our discussion above—to a variant in which her moral-epistemic faculties are momentarily gravely compromised. Thus, imagine that as Anna grapples with her moral uncertainty, she is slipped a drug that induces a psychotic episode in her. She begins to hear voices commanding her to deceive her family. In the grip of this psychotic episode, she deceives her family about her plans. If Rosen is correct, the cases of morally ignorant Anna and of momentarily gravely impaired Anna should strike us as morally equivalent. But I don’t think they do. When Anna acts from moral ignorance and comes to realize her mistake, it remains apt for her to apologize and ask for forgiveness for lying to her family. And it is appropriate for her family to expect such contrition: an acknowledgment on Anna’s part that she has ³¹ Rosen (2008: 609).

³² Rosen (2003: 66).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Excuse without Exculpation

89

fallen short of the relevant moral expectation, that she resolves to make up for it, and to do better in the future. This is not true of the case where Anna is acting in the grip of a psychotic episode. Once the circumstances of her behavior have been explained, it would be inappropriate for her family to insist on an apology or expressions of contrition. In this case, something like the objective stance seems clearly warranted: once it is understood that Anna was “out of her mind,” that’s that. The culprit who slipped her the drug that induced her temporary psychosis is the one who has something to answer for. Now, Rosen concedes that it would be “unseemly” for the morally ignorant wrongdoer to “do anything but ‘accept’ responsibility for his behavior.”³³ But he insists that the fact that the wrongdoer should “accept responsibility” does not show that he is to blame for his wrongdoing: To accept responsibility for one’s behaviour is to acknowledge that it was wrong and that one is therefore under some obligation to apologize and make amends. This sort of responsibility does not entail moral responsibility in our sense—i.e., blameworthiness.³⁴

This position strikes me as puzzling. For what could be the source of the morally ignorant wrongdoer’s obligations “to apologize and make amends” unless she is to blame for the wrong? More importantly, this concession does not address the asymmetry revealed by the contrastive method. In the case where Anna lies to her family on the basis of mistaken moral advice, we do not just think that it would be impolite of her to refuse to apologize—we think she is under moral pressure to do so. But there is no such moral pressure in the case of psychosis. Indeed, once Anna’s family understands that she was psychotic, it would be “unseemly” for them to insist that she should apologize and make amends. The contrastive method thus does not bear out Rosen’s assimilation of moral ignorance to hypnosis, psychosis, and other conditions that give rise to profound agential impairment. For our intuitive judgments about the latter differ from our judgments about cases involving moral ignorance. Profound agential impairment negates blame whereas moral ignorance merely mitigates it.³⁵ ³³ Rosen (2003: 68). ³⁴ Rosen (2003: 68). ³⁵ Could Rosen respond by saying that he is happy for his conclusion to have revisionary implications for our ordinary practices of blaming: namely that insofar as we do blame the morally ignorant agent more than the gravely impaired one, we are making a mistake? No, for the contrastive method’s criticism is not that the conclusion is revisionary but that the argument for it does not work. This argument is based on the claim that, once we are clear-eyed about a wrongdoer’s non-culpable moral ignorance, we will intuitively judge her as blameless and the wrongdoing as a “mere event.” It’s this intuitive

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

90

Paulina Sliwa

Let’s take stock. On a mitigating account of excuses, excuses make a normative difference to how much an agent is to blame for a wrong. I suggested that we can test for normative difference-making by using a contrastive method. The results of this method spell trouble for both sides of the dispute as it has been playing out in the literature. Contra Arpaly and others, moral ignorance can make a difference: it can diminish the normative footprint of a wrong. Contra Rosen, wrongs committed from moral ignorance are not morally on a par with harmful behavior under grave impairments of agential capacities.

4 .8 . M O R A L V E R S US N O N M O R A L IG N O R AN C E I want to end by addressing some questions about the scope and limitations of the argument.

4.8.1. How Far-Reaching is its Conclusion? The argument establishes that moral ignorance, when it is non-culpable, can mitigate blame without negating it. This is compatible with moral ignorance sometimes making a negligible difference or even no difference at all. The conclusion I argue for is thus quite modest. The argument presented here does not aim to give a formula for how much moral ignorance bears on a wrong’s normative footprint. This will plausibly vary from case to case. It will depend on the nature and seriousness of the wrong, on the harm that was done, as well as the extent and subject matter of the moral ignorance. This lack of formula should not trouble us. Equally, there is no straightforward way of telling, for each feature of a wrong, how much that feature contributes to the “total amount” of blame an agent is on the hook for.

4.8.2. What does the Argument tell us about Agents who act from profound Moral Ignorance? As noted above, philosophical inquiry into the relevance of moral ignorance has focused on cases of profound moral ignorance: the slave owner, the ruthless business man, the Mafia boss, the gang member, the Roman circusenthusiast. I suggested that to answer the general question whether moral ignorance can excuse while avoiding the “curse of knowledge,” we set those premise that the contrastive method casts doubt on. I thank an anonymous reviewer for suggesting this response on behalf of Rosen.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Excuse without Exculpation

91

aside. Still, we might hope for some guidance on cases where moral ignorance is more profound. The present argument helps with those cases in two ways. It suggests that agents who are profoundly morally ignorant are to blame for the wrongs they commit to some extent. And, it helps us sharpen what is at stake in trying to determine the extent to which they are to blame. The question is what these agents owe—or would owe, were they still alive—in terms of reparative duties, how their wrongdoing has changed the feeling rules in play, whether they are (would be) deserving of trust. Their moral ignorance will have to be balanced against other morally relevant considerations, such as the precise nature of the wrongdoing and its consequences. That will have to be settled by substantial moral inquiry.

4.8.3. How does Moral Ignorance as an Excuse Compare to Nonmoral Ignorance? There is widespread agreement that non-culpable nonmoral ignorance negates blame. The question about moral ignorance is then often framed as one about symmetry. Rosen argues that the cases of moral and nonmoral ignorance are symmetrical—both negate blame. Harman and others argue for asymmetry—they think only nonmoral ignorance negates blame. I suggest that there is broad symmetry between moral and nonmoral ignorance.³⁶ Nonmoral ignorance, too, merely mitigates blame; it does not extinguish it. Normative footprints remain. And just as in the case of moral ignorance, how much a difference nonmoral ignorance makes depends very much on the circumstances—the nature of the wrong, the harm done, the consequences unfolding. This view is at odds with the received wisdom—that nonmoral ignorance exculpates—and so, it requires defense. I won’t be able to do full justice to this task here but let me offer a sketch. We can start by looking at one of the cases discussed by Rosen at length: Goldberg has been charged with investigating financial impropriety at Acme Corp. The evidence so far points to Himmelfarb as the guilty party, and it is predictable that things will go badly for him if he is named. Goldberg conducts a scrupulous inquiry, which ultimately sustains the case against Himmelfarb. At this point, Goldberg has done everything a responsible investigator ought to do, and his evidence amply supports the conclusion that Himmelfarb is a crook. In response to this evidence, he forms a belief to this effect and accuses Himmelfarb of

³⁶ The whole story is likely to be more complicated. The broad symmetry is compatible with differences in detail: moral ignorance may well generally be a weaker excuse than nonmoral ignorance. See Sliwa (2019: 50–54).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

92

Paulina Sliwa

embezzlement. In fact, Himmelfarb has been framed by a malicious colleague. This means that in filing his report, Goldberg accuses an innocent person of a crime, and that is wrong. But if there is nothing more to the story, this is a straightforward case of blameless wrongdoing from excusable ignorance.³⁷

Since Goldberg’s false belief is epistemically justified, his wrongdoing is blameless, Rosen claims. I think this overstates the case. If Goldberg abided by his procedural epistemic obligations, his ignorance furnishes him with an excuse. But it does not follow that this excuse absolves him from all blame. That’s because his ignorance does not altogether eliminate the normative footprint of his wrongdoing. To see this, suppose that, based on Goldberg’s report, Himmelfarb is convicted and sentenced to many years in prison. After a while, his malicious colleague’s evildoing comes to light. Himmelfarb remains in jail, mounting a costly and uncertain appeal process. In light of these facts, it is appropriate for all of us to be moved by Himmelfarb’s fate. Any decent human being should feel sorry for him and outraged at his malicious colleague. But how should Goldberg feel about his role in Himmelfarb’s injustice? Goldberg is not merely a witness or bystander. He played a central role in getting Himmelfarb to where he is now, namely behind bars. This contribution gives rise to special moral obligations: he ought to feel deep sorrow and regret for accusing an innocent man, he ought to reflect on whether he had overlooked anything, and he owes Himmelfarb whatever support he can provide in Himmelfarb’s quest for exoneration. It would not be out of line for him to contact Himmelfarb to apologize for his role in the wrongful conviction. Of course, the fact that Goldberg had, at the time, acted in light of his (misleading) evidence makes a huge normative difference to what Goldberg owes, compared to a scenario in which his accusation of innocent Himmelfarb had been negligent, reckless, or worse yet, intentional. He need not be ashamed of himself. He does not deserve our nor Himmelfarb’s disdain and indignation. He owes it to Himmelfarb to set the record straight and do what he can to get him out of jail, but he does not owe him compensation on top of that. Goldberg’s nonmoral ignorance thus excuses—it bears on the normative footprint of his accusation of an innocent man—but it does not exculpate him. The wrong he has done saddles him with normative demands and obligations

³⁷ Rosen (2008: 601; highlight mine).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Excuse without Exculpation

93

towards Himmelfarb that go well beyond those of mere bystanders. These special obligations are a mark of moral responsibility: he has them because he is (partly) to blame for Himmelfarb’s fate.³⁸ There is tragedy in Goldberg’s situation. What led him to ruin the life of an innocent man was bad epistemic luck. For this reason he deserves our pity. He is fully entitled to resent the one who placed him in this situation: the actual criminal, Himmelfarb’s malicious colleague. But the fact that Goldberg is to blame for accusing an innocent man remains.³⁹As Bernard Williams poignantly writes in his analysis of Oedipus: Someone may simply have ruined his life . . . If that has happened, then it is something that has happened to him, but at the same time it may be something that he has brought about. What has happened to him, in fact, is that he has brought it about. That is the point of Oedipus’s words at Colonus. The terrible thing that happened to him, through no fault of his own, was that he did those things.⁴⁰

The tragedy of Oedipus—and Goldberg—lies precisely in that he is both unlucky and yet to blame. Just like moral ignorance then, nonmoral ignorance excuses without exculpating. Sometimes we leave normative footprints through bad moral or epistemic luck. That they are left through bad luck can make them smaller. But it does not make them any less ours.

References Allais, Lucy (2008). “Wiping the Slate Clean: The Heart of Forgiveness,” Philosophy and Public Affairs 36(1): 33–68. Arpaly, Nomy, and Timothy Schroeder (2013). In Praise of Desire. Oxford: Oxford University Press. Austin, John L. (1957). “A Plea for Excuses,” Proceedings of the Aristotelian Society 57: 1–30. Baron, Marcia (2007). “Excuses, Excuses,” Criminal Law and Philosophy 1(1): 21–39. Calhoun, Cheshire (2015). Moral Aims: Essays on the Importance of Getting it Right and Practicing Morality with Others. Oxford: Oxford University Press. FitzPatrick, William J. (2008). “Moral Responsibility and Normative Ignorance: Answering a New Skeptical Challenge.” Ethics 118(4): 589–613.

³⁸ For a much more detailed defense of this claim, see Sliwa (2019: 27–35). ³⁹ Kramer (2005) puts forward a similar argument directed against Hart’s “oughtimplies-can”-principle. ⁴⁰ Williams (2008: 70). Williams’s argument further develops the line he put forward in his (1981) article on moral luck.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

94

Paulina Sliwa

Harman, Elizabeth (2011). “Does Moral Ignorance Exculpate?” Ratio 24(4): 443–68. Harman, Elizabeth (2015). “The Irrelevance of Moral Uncertainty,” Oxford Studies in Metaethics 10: 53–79. Hieronymi, Pamela (2001). “Articulating an Uncompromising Forgiveness,” Philosophy and Phenomenological Research 62(3): 529–55. Hochschild, Arlie Russell (1975). “The Sociology of Feeling and Emotion: Selected Possibilities,” Sociological Inquiry 45(2–3): 280–307. Hochschild, Arlie Russell (1979). “Emotion Work, Feeling Rules, and Social Structure,” American Journal of Sociology 85(3): 551–75. Hochschild, Arlie Russell (2012). The Managed Heart: The Commercialization of Human Feeling. Berkeley, CA: University of California Press. Holton, Richard (2010). “Norms and the Knobe Effect,” Analysis 70(3): 417–24. Kramer, Matthew (2005). “Moral Rights and the Limits of the Ought-Implies-Can Principle: Why Impeccable Precautions are No Excuse,” Inquiry 48(4): 307–55. Murphy, Jeffrie (1982). “Forgiveness and Resentment,” Midwest Studies in Philosophy 7(1): 503–16. Nelkin, Dana (2013). “Freedom and Forgiveness” in Ishtiyaque Haji and Justin Caouette (eds), Free Will and Moral Responsibility. Newcastle upon Tyne: Cambridge Scholars Press, 165–88. Owens, David (2012). Shaping the Normative Landscape. Oxford: Oxford University Press. Rosen, G. (2003). “IV—Culpability and Ignorance,” Proceedings of the Aristotelian Society 103(1): 61–84. Rosen, G. (2004). “Skepticism about Moral Responsibility,” Philosophical Perspective 18(1): 295–313. Rosen, G. (2008). “Kleinbart the Oblivious and Other Tales of Ignorance and Responsibility,” The Journal of Philosophy 105(10): 591–610. Royzman, Edward B., Kimberly Wright Cassidy, and Jonathan Baron (2003). “ ‘I Know, you Know’: Epistemic Egocentrism in Children and Adults,” Review of General Psychology 7(1): 38. Scanlon, Thomas (2008). Moral Dimensions: Permissibility, Meaning, Blame. Cambridge, MA: Harvard University Press. Sliwa, Paulina (2012). “In Defense of Moral Testimony,” Philosophical Studies 158 (2): 175–95. Sliwa, Paulina (2019). “The Power of Excuses,” Philosophy and Public Affairs 47(1): 37–71. Strawson, P. F. (1962). “Freedom and Resentment,” Proceedings of the British Academy 48: 1–25. Thomson, Judith Jarvis (1990). The Realm of Rights. Cambridge, MA: Harvard University Press. Wallace, Jay (1994). Responsibility and the Moral Sentiments. Cambridge, MA: Harvard University Press. Warmke, Brendon (2016). “The Economic Model of Forgiveness,” Pacific Philosophical Quarterly 97(4): 570–89.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Excuse without Exculpation

95

Williams, Bernard (1981). Moral Luck: Philosophical Papers 1973–1980. Cambridge: Cambridge University Press. Williams, Bernard (2008). Shame and Necessity, vol. 57. Berkeley, CA: University of California Press. Zimmerman, Michael J. (1997). “Moral Responsibility and Ignorance,” Ethics 107(3): 410–26.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

5 Resisting Reductive Realism N. G. Laskowski 5 . 1. I N T R O D U C T I O N In the second half of my graduate studies, I had a meeting with the Director of Graduate Studies (DGS). After confirming that my dissertation was progressing smoothly, I was asked to describe the project. I gave the pitch: There’s a family of views organized loosely around the possibility of explaining goodness, reasons, and other normative entities appealing only to the sort of entities to which natural and social scientists appeal. But a recently influential class of objections has led ethicists away from such views. I argue that the objections come up short in ways that reveal a slew of neat things about the concepts used to think about goodness, reasons, and the like. “So you’re defending reductive naturalism?,” the DGS asked. “Basically—I defend reductive realism in ethics, a wider family of views of which reductive naturalism happens to be the most prominent member,” I said. “Reductivism about all normativity?,” asked the DGS. “That’s right,” I replied. The DGS then reached for an e-cigarette. “Wow,” he said, through a plume of sweet-smelling vapor. “That’s crazy. But I wish you the best of luck. I’ll be here if you need anything.” The point of this anecdote is not to single out the DGS for vaping. It’s to highlight their reaction to my project. The DGS was shocked (“Wow”) to hear that I was defending reductivism. They found it incredible (“That’s crazy”) that I was attempting to do so. It’s not an uncommon reaction. Similar ones can be found in Nagel (1986: 138), Wiggins (1993: 311), FitzPatrick (2014: §§7, 8), and Scanlon (2014: 46).¹ My aim in this chapter is to better understand why ethicists resist reductivism. I want to know what

¹ This is a very small sampling. Unfortunately, references contribute to the 10,000word count to which each chapter in this volume is capped. See Laskowski (2019) for many more examples of ethicists expressing similar reactions toward reductivism.

N. G. Laskowski, Resisting Reductive Realism In: Oxford Studies in Metaethics Volume 15. Edited by: Russ Shafer-Landau, Oxford University Press (2020). © N. G. Laskowski. DOI: 10.1093/oso/9780198859512.003.0005

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Resisting Reductive Realism

97

leads ethicists like Enoch (2011: 4, my emphasis) to say things like “Normative facts are just too different from natural ones to be a subset thereof.” In section 5.2, I discuss the natural idea that ethicists resist reductivism simply as a result of following the arguments against reductivism where they lead. Seeing why this idea comes up short further clarifies the sense in which ethicists resist reductivism. It also points toward the idea that resistance to reductivism isn’t high-minded—that it has more to do with our psychology than any process of reflective reasoning. There are many different psychological explanations of why ethicists resist reductivism available. In section 5.3, I extract a concept-based, psychological explanation from contextualists who argue that normative concept use is interestingly flexible. I argue that the explanation fails but in a way that pushes the discussion in the right direction. Toward the end of section 5.3 through 5.4, I come down on the side of a different concept-based psychological explanation of why ethicists resist reductivism—one that is fully compatible with the truth of reductivism. Over the course of defending the explanation, I develop a novel kind of hybrid view of normative concepts on which using them depends on both cognitive and noncognitive aspects of our psychology. That resistance to reductivism traces to special features of our use of normative concepts rather than the falsity of reductivism isn’t a groundbreaking idea. Many philosophers of mind have gotten mileage out of a similar one in explaining resistance to reductive views concerning phenomenal consciousness.² In fact, the basic form of the idea has been imported to ethics already.³ But I argue that too much has been read into parallel forms of such resistance. As a result, I claim that a promising version of such an explanation has been dismissed too quickly. While resistance to reductivism can be found in various philosophical domains, I highlight what is distinctive of such resistance in ethics. Two disclaimers before proceeding. First, the main goal of this chapter might appear familiar—it looks like one that ethicists have been pursuing at least since Moore coined the phrase “naturalistic fallacy.” But much of the commentary in Moore’s wake focuses on the force of his charge against naturalism and reductivism more broadly. My project is more upstream. I’m interested in what might have led Moore to characterize reductivism as fallacious in the first place.⁴ Second, I’m concerned principally with explicating a psychological phenomenon—the way in which ethicists struggle to

² See Balog (2009) and the references therein. ³ See Chappell and Yetter-Chappell (2013) and Mehta (forthcoming). ⁴ A natural thought is that it was his open question argument that led him to view reductivism so unfavorably. I’ll be arguing that something like this is only part of the story.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

98

N. G. Laskowski

believe reductive theses.⁵ So, while it’s true that much of what I say undercuts a familiar inference to the falsity of reductivism from the fact that ethicists struggle to believe it, I don’t see that as my primary aim.⁶ Outspoken resistance to reductivism far outstrips the force of the arguments against the view.⁷ That makes ethicists’ relationship to reductivism somewhat peculiar among views in ethics.⁸ Everyone, including opponents of reductivism, should want to have a better grip on this psychological phenomenon regardless of which view of the metaphysics of ethics turns out to be true.

5. 2. STA G E S ET TING

5.2.1. Resistance to Reductivism, What I’ll be treating reductivism as the view that while normative entities⁹ figure in a metaphysical account of everything, none of them do so at the most fundamental level.¹⁰ This is a view that ethicists appear to resist. But they ⁵ Though psychological questions are amenable to empirical investigation, I won’t be employing experimental data. That’s in part because data bearing on the question of why ethicists struggle to believe reductive theses aren’t yet available. Ethics isn’t an exception. Chalmers (forthcoming) points out that such data are welcome but not currently available with respect to similar questions in the philosophy of mind regarding our attitudes toward reductivism about phenomenal consciousness. But even if such data were available, it could inform but not settle the question at hand. As we’ll see, part of what’s at issue is the not-recognizably empirical question of what it takes to even have the beliefs that ethicists struggle to form. Thanks to an anonymous referee for requesting clarification. ⁶ But it’s one of my primary aims in Laskowski (2019). ⁷ Schroeder (2005: 3) makes a similar observation. It’s not that there aren’t any objections to reductivism. Indeed, see Paakkunainen (forthcoming) for a litany of objections that based on the “just-too-different” slogan that’s at issue in this chapter. But objections usually amount to challenges rather than fatal problems for a view. You’d never get that impression from the hyperbolic language that so many ethicists use in dismissing reductivism. ⁸ True, ethicists describe some other views in ways suggesting that they have a similarly hard time believing them. Streumer (2017) calls his book on the error theory Unbelievable Errors for a reason. But Streumer arrives at his incredulity about the error theory after working through a complex argument. I’ll be arguing that ethicists arrive at their take on reductivism very differently. ⁹ I use the word “entities” broadly to remain neutral about whether reductivism is best understood as a view about facts, properties, and the like. However, I also use the word in such a way as to exclude mentalistic entities, such as concepts, attitudes, and the like. ¹⁰ There isn’t consensus over the nature of reductivism. One issue concerns whether reductive theses should be understood as stating identities rather than constitutive analyses. See Cuneo (2015) and the references therein. Another issue concerns whether reductive theses are understood strongly, as providing characterizations of normative entities that are

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Resisting Reductive Realism

99

don’t seem to do so in the sense of merely intuiting that reductivism is false. That’s not what Chappell (2019: 125, my emphasis) is getting at when they say that “Normativity is, intuitively, so fundamentally different in kind from natural phenomena that a reduction of the former to the latter may seem hopeless or even absurd.”¹¹ It’s more like ethicists have the intuition that reductivism is in some way incoherent rather than false, which prevents them from believing that the view is even possible.¹² This is what I call the sense of incredibility in ethics.¹³ Instead of providing a precise characterization of it, I characterize it ostensively as that shared kind of struggle among ethicists to believe that reductivism is at all plausible upon reflection. This is a rough characterization of the sense of incredibility. I’m eschewing the common philosophical instinct to offer a rigorous definition of it. A rigorous definition is a useful philosophical tool. But it’s not the only such tool available. Experiencing the sense of incredibility is also a useful way into understanding it. Try it. Take a moment to reflect on the reductive thesis that what it is for an action to be good is for it to maximize pleasure. If you feel like you’re experiencing the sense of incredibility at the moment, then I suspect that you understand the sense of incredibility well enough to see that it calls for explanation.¹⁴ But you’re not out of luck if you don’t find finitely specifiable, or weakly, as providing characterizations of normative entities that are not finitely specifiable. See Schroeter and Schroeter (2015) for discussion. There are also issues, among still many others, about what makes reductivism a form of realism. See Dunaway (2017). There isn’t enough space to adjudicate these disputes. And as an anonymous referee suggests, pluralism about reductivism might be the right way to go in the end. I’ll be treating the view as involving fundamentality and weak specification for the sake of exposition. ¹¹ I treat reductive naturalism (“naturalism”) as a precisification of reductivism. It’s the view that normative entities figure in a metaphysical explanation of everything, but not at the most fundamental level, because only natural entities so figure. Like reductivism, there’s no uncontroversial way to understand naturalism. I’ll be using the word “natural” to mean what others use the words “non-normative” or “descriptive” to mean. I’ll also be assuming that intuitions about what falls on either side of the distinction are firm enough not to weigh in on the controversy. See Dowell (2013) for discussion. ¹² Parfit’s (2011: 325, my emphasis) discussion is telling. He asserts that reductivism “could not be true” in the same way that claims like “rivers are sonnets” or “heat is a shade of blue” could not be true. Copp (2017: 29, original emphasis) reads Parfit in a similar way, writing that “I think that Parfit and many other non-naturalists may be driven to reject naturalism less by arguments than by the prior conviction that no natural fact could be normative. Their view is that normative naturalism is hopeless, not in detail, but in basic conception.” But see Howard and Laskowski (forthcoming) for an interpretation of Parfit’s remarks on which they serve as a premise in a sophisticated objection to reductivism. ¹³ I borrow the phrase “sense of incredibility” from Melnyk (n.d.). ¹⁴ It can be fruitful to compare your reaction to a different reductive thesis from outside of ethics, e.g. that what it is for a sample of liquid to be a sample of alcohol is for it to be sample of liquid consisting of an ionic compound of carbon, hydrogen, and oxygen.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

100

N. G. Laskowski

this exercise illuminating. Another, less direct way into understanding the sense of incredibility is to observe it at work in others. Consider, in particular, an exchange between a friend of reductivism, Schroeder (2005), and a foe of reductivism, Enoch (2011). In one of the first explicit attempts of its kind of which I’m aware, Schroeder offers a diagnosis of why ethicists are so thoroughly resistant to reductivism. According to Schroeder, reductive views are all about explanation—they’re put forward explicitly to account for the central phenomena constitutive of the relevant subject matter.¹⁵ From this perspective, reductive views are appealing only insofar as anyone uses them to explain such phenomena.¹⁶ The problem is that many of the most influential debates about reductivism in ethics, Schroeder suggests, take place at too high a level of abstraction for anyone to use them to explain anything in particular.¹⁷ Schroeder can be read as suggesting that ethicists experience the sense of incredibility in virtue of failing to develop plausible reductive views.¹⁸ In his discussion of reductive views in ethics generally, and Schroeder’s diagnosis of resistance to reductivism in particular, Enoch (2011: 104) acknowledges explicitly that one way of evaluating reductivism involves precisely the sort of work Schroeder encourages, namely, that of “engag[ing] in piecemeal evaluation of specific reductive proposals or the arguments for them.” As it happens, Enoch thinks various considerations count

¹⁵ “Reductions, after all, are at least in principle supposed to be theoretically fruitful . . . they are supposed to make the reduced domain simply less mysterious, by telling us a little bit of what it is about.” (2005: 3, original emphasis). ¹⁶ “The only way to see whether any particular reduction succeeds at capturing all of the important phenomena about the normative is to actually carry out the reduction and test it on various normative phenomena in order to see how much sense it can make of them.” (2005: 7). ¹⁷ Schroeder has discussions like Jackson’s (1998) in mind. Jackson’s influential argument from supervenience tells us that one part of normativity, morality, concerns descriptive properties picked out by the Ramsification of mature folk morality, but it doesn’t identify which particular descriptive properties morality concerns. ¹⁸ I read Schroeder’s diagnosis differently from Lutz and Lenman (2018), who claim that, on Schroeder’s diagnosis, ethicists resist reductivism as a result of a lack of analytic connections between normative and natural concepts. While it’s true that Schroeder is committed to such a lack of connections, I think the quotes in the footnotes above indicate that he is better read as claiming that such resistance stems from a lack of exposure to plausible reductive views. Indeed, in later work, Schroeder (2017: 682, original emphasis) writes explicitly that “no one has offered a proposed reduction that seems like it could possibly be true. High-level arguments that some reductive view must be true [e.g. Jackson’s arguments] simply do not address the source of skepticism that no such view could be. The only antidote to this—the only dialectically fruitful way forward—is to defend better views.”

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Resisting Reductive Realism

101

decisively against all extant reductive views.¹⁹ Tellingly, however, Enoch also suggests that anyone who already experiences the sense of incredibility is committed pre-theoretically to rejecting even merely possible reductive views that ethicists have yet to consider.²⁰ This is circumstantial evidence that ethicists don’t experience the sense of incredibility as a result of failing to develop plausible reductive views. It’s the other way around. Ethicists experience the sense of incredibility first and then conclude that reductive views can’t work. Some might still feel as though they don’t quite see that the sense of incredibility is something to explain, even after trying to experience the sense of incredibility directly and even after observing it indirectly in others. Hold tight if that’s the state in which you find yourself. I’ll work toward further discharging the assumption that the sense of incredibility deserves special treatment over the course of this chapter. In the meantime, I can report that I experience the sense of incredibility. In fact, I experience it even though I’m quite sympathetic to reductivism.²¹ I’d like to better understand how it could be that I struggle to believe reductivism despite my attraction to it.

5 .3 . P S Y C H O L O G I C AL , C O N C E P T - B A S E D E X P LA N A T I O N S Ethicists resist reductivism in the sense of experiencing the sense of incredibility—they find reductive views so obviously implausible that they struggle to believe that they’re even possible. I’ve argued against the idea that ¹⁹ Enoch (2011: 104) claims that currently available reductive views are either a priori or a posteriori. If they’re a priori, then he says that the “just-too-different” intuition or what I call the sense of incredibility, counts against them. If they’re a posteriori, then, according to Enoch, they’re vulnerable to familiar moral twin-earth style considerations. As for the former horn, Enoch doesn’t say why he thinks the sense of incredibility is incompatible with a priori reductive views. Enoch’s second horn is also unpersuasive. Arguments from moral-twin earth style considerations simply don’t have the right form to show that no a posteriori reductive view could work, as Schroeder (2017: 682, n.d.) observes. A more recent, stronger version of the argument from Sinhababu (2019) only further confirms Schroeder’s diagnosis. ²⁰ Enoch writes that “perhaps there are a priori reductions that we just haven’t thought of yet,” and goes on to suggest strongly that it wouldn’t matter. This is because, if we’re following Schroeder, reductive views have to explain the central phenomena of ethics, and “the list of things we pre-theoretically want to say may itself include a denial of naturalist reduction. Indeed, the just-too-different point is precisely an attempt to capture such a pre-theoretical desideratum.” (2011: 105–6). ²¹ I defend reductivism in Laskowski (2018a, 2018b, 2019). Copp (2018: 17), one of the most influential contemporary proponents of reductivism, also admits to experiencing it.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

102

N. G. Laskowski

they experience it in virtue of having reflected only on implausible reductive views. Indeed, the sense of incredibility seems to have little to do with any kind of process of reflective reasoning, which points toward the idea that it has a basis in arational, merely psychological mechanisms. There are various ways to develop such a psychological explanation of the sense of incredibility. In this section, I’m going to begin exploring whether it’s well explained by the psychological mechanisms employed in and enabling ethicists to have thoughts about reductive theses, which is to say the concepts that they use.²² For example, there are various concepts that ethicists use in reflecting on the reductive thesis that what it is for an action to be good is for it to maximize pleasure. This includes the paradigmatically normative concept ²³ and the natural concept  .²⁴ I’m going to explore the idea that concepts like the former might be special in a way that helps explains the sense of incredibility. There also happen to be a number of ways of developing the idea that such normative concepts explain the sense of incredibility. I’ll introduce and argue against one such explanation in this section before introducing and defending another.

5.3.1. Using Normative Concepts to Explain the Sense of Incredibility: Contextualism Start with the banal observation that you don’t think of things as just plain short. You don’t think of the 5’6” professional baseball player José Altuve as short, period. Rather, you think of them as short relative to a standard, like the average height among professional baseball players (6 feet). It’s plausible to many philosophers that the concept  resembles the nature of a gradable adjective for which an incomplete predicate analysis is best, in that its context of use has to be checked to uncover the full thought that it’s being used to form. A number of ethicists have argued that normative concepts like  are similarly flexible in virtue of their dependence on context.²⁵ You don’t think of things as plain good, but rather as good relative to whichever standard happens to be relevant or salient in context. Such flexibility in the use of normative concepts appears to be in tension with the aim of ambitious reductive ethical theorizing—the aim of sorting out ²² See Margolis and Laurence (1999) for a discussion of understanding concepts as psychological entities rather than abstracta. See Laskowski and Finlay (2017) for a discussion of the nature of normative concepts in particular that’s informed by Margolis and Laurence. ²³ I am following the convention of using small caps to denote concepts. ²⁴ It could be that   isn’t a natural concept. I’m only using it as an example. ²⁵ See Ziff (1960) and Thomson (1992), among many others.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Resisting Reductive Realism

103

what it is for something to be (say) good and hence the aim of sorting out what good things have in common across all contexts. Contextualists about normative concepts might exploit this tension to explain the sense of incredibility. Consider Finlay’s (2014) brand of contextualism about normative concepts to illustrate. According to Finlay,  has the structure of a relational predicate with contextually sensitive argument-places for objects and ends. On Finlay’s view, objects stand in a kind of probability-raising relation to ends. When you think that going vegan is good in a context where you’re discussing how to relieve suffering in the world, you’re thinking that going vegan raises the probability of relieving suffering. When you’re thinking that using a hammer is good in a context where you’re considering whether to hang a frame, you’re thinking that using a hammer raises the probability of hanging the frame. In other contexts, you’ll be thinking about raising the probability of still other contextually salient ends. The resulting picture is one on which it’s possible to think of things as good in at least as many ways as it’s possible for context to supply different standards. Imagine some ethicists reflecting on the question of what it is for an action to be good. On Finlay’s view, they’re really thinking about what it is for an action to raise the probability of end e, where “e” is whichever end is salient in context. Plausibly, whether any answer to such a question will strike them as plausible depends in part on whether they’re able to sort out which of their ends is salient in the context. But it can be awfully difficult to sort out which are salient for a variety of reasons. It could be due to the sheer number of ends. It could be indeterminate whether such ends are shared sufficiently. More interestingly, it could be that their ends just are their desires and that desires never rise to the level of conscious awareness to be recoverable from context.²⁶ In failing to recover their end(s) from context, the ethicists don’t uncover at least one of the elements constitutive of goodness (i.e. the ethicists’ ends). But that could leave them with the impression that reductivism is incredible, since the aim of ambitious reductive theorizing about goodness is to identify what all good things have in common.

5.3.1.1. Contextualism Doesn’t Explain the Sense of Incredibility There’s a problem with the explanation despite its promise. Continue to imagine the group of ethicists reflecting on the question of what it is for an action to be good. In doing so, however, don’t imagine them thinking about

²⁶ See Hulse et al. (2004).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

104

N. G. Laskowski

the question in the same way as before. That is, where the question of what it is for an action to be good is left incomplete (e.g. where adjunct prepositional phrases like “good for such and such” are not included), such that context supplies an end. Instead, imagine the ethicists stating explicitly that the question of interest to them is the question of whether what it is for action to be good across all contexts is for it to be an action that raises the probability of some contextual salient end. Such explicitness makes it implausible to maintain that the ethicists are being duped, as it were, by any covert contextual variation. But if not then the contextualist explanation of the sense of incredibility under discussion predicts that they wouldn’t experience the sense of incredibility. The problem is that it seems like they would.²⁷

5.3.2. Using Normative Concepts to Explain the Sense of Incredibility: Unanalyzability Of course, there are responses to this objection available. But instead of working through them, I’m going to explore a constructive reaction to the apparent failure of the contextualist explanation. In broad terms, contextualism seemed like a promising explanation of the sense of incredibility in virtue of the fact that it can be hard to reveal whatever analytic or otherwise a priori connections (“connections” from here on) there might be among normative concepts (e.g. ) and natural concepts (e.g.    ). And this seems like it engenders doubt about reductivism. But it doesn’t engender enough doubt because such a view doesn’t, as the explicit-ends case above suggests, make connections among normative and natural concepts hard enough to reveal. This points toward a way forward: appeal to views of normative concepts on which it’s much harder to reveal any connections between normative and natural concepts. One way to guarantee that it’s much harder to reveal connections between normative and natural concepts would be to appeal to views on which there just aren’t any such connections. Standardly, concepts are said to have connections to one another in virtue of having a definitional structure, wherein it is possible to list all the necessary and sufficient conditions for their application via analysis. The idea that concepts can be analyzable in such a way has been around a long time, of course. But there are wellestablished families of views in the general theory of concepts on which some concepts are unanalyzable.²⁸ ²⁷ Joyce (n.d.: 7–10) suggests a similar issue. ²⁸ I only need the weaker claim that they aren’t fully analyzable. But I use the word “unanalyzable” to mean “not fully analyzable” for the sake of convenience.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Resisting Reductive Realism

105

There are at least two such families. On the prototype theory, concepts are prototypes, which are stores of features observed together at a certain rate of statistical frequency.²⁹ Concepts are unanalyzable on this view because such features aren’t necessary or sufficient for a concept’s application—they’re just observed together at a high rate.³⁰ Another view of concepts on which they’re unanalyzable is conceptual atomism, which just is the denial of the possibility of concepts having a definitional structure, packaged with further views about the sort of relations such concepts have to stand in to the world to be associated with information at all.³¹ On either view, the result is that normative concepts are unanalyzable, which encourages the thought that when ethicists use normative and natural concepts to reflect on reductive theses, no amount of such reflection can reveal connections among these concepts.³² The view that normative concepts are unanalyzable builds on the insights of the contextualist explanation. The question now is whether it can explain the sense of incredibility. Consider, again, the reductive thesis that what it is for an action to be good is for it to maximize pleasure. Ethicists use natural and normative concepts in reflecting on this thesis. But if normative concepts are unanalyzable then even if it’s true that it’s part of the nature of good actions to maximize pleasure, it’s not accessible to ethicists that it is so. On such a view, no amount of reflection with the concept  can reveal any connection to  . That seems like it would leave ethicists struggling to believe the reductive thesis that what it is for an action to be good is for it to be an action that maximizes pleasure. In other words, the unanalyzability of normative concepts seems to provide an alternative explanation of why ethicists experience the sense of incredibility.³³ Nevertheless, despite its promise, there is still a lot of work to do to develop the explanation. In particular, I’m going to highlight several problems for the explanation and develop the view by showing how to solve them. Since the problems are challenging, I’m going to spend much of the rest of the chapter addressing them. But there will be an added payoff in ²⁹ Prototype theory hasn’t had much uptake in ethics. But see Goldman (1993) and Stich (1993). ³⁰ I only mean that prototypes aren’t analyzable in any standard sense. ³¹ Fodor (1998) is the locus classicus. ³² The thought is encouraged rather than forced because it might be possible to combine a prototype or atomistic view of concepts with a view of conceptual competence (perhaps e.g. a kind of inferentialist view) to forge some other kind of connection among normative and natural concepts that could in some way be revealed. ³³ While I’m using these claims to explain why ethicists struggle to believe reductivism, Moore used similar ones brought out by the open question argument to argue against reductivism. So, there’s a sense in which I’m turning the considerations that Moore enlists on their head.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

106

N. G. Laskowski

doing so. In addition to showing how this explanation of the sense of incredibility works more fully, I’m going to lay the groundwork for developing a new view of what it is to use a normative concept that’s of independent interest.

5. 4. RE FININ G TH E A CC OU NT

5.4.1. Addressing Two Problems The first problem is straightforward. The unanalyzability of normative concepts might explain why ethicists experience the sense of incredibility when reflecting on reductive views purporting to state connections among normative and natural concepts. But ethicists are well aware that reductive views aren’t always put forward as purporting to state analytic or otherwise a priori connections (“connections”) among normative and natural concepts. They’re well aware that reductive views often take the form of purporting to state a posteriori truths. Yet ethicists struggle to take both kinds of reductive views seriously. Thus, proponents of the view that the unanalyzability of normative concepts explains the sense of incredibility have to face up to the challenge of explaining the sense of incredibility in full, in light of the variety of reductive theses. The second problem gets going when you use your natural concepts to conceive of a world that is naturalistically identical to our own. That is, it gets going when you use your natural concepts to conceive of a world containing all of the same physical, biological, chemical, social, linguistic, and psychological entities with the same history of interaction among them as in our world. Conceiving of such a world involves conceiving of the same kind of events and actions that have taken place in our world. This includes conceiving of all of the same kind of events and actions involving slavery that have taken place in our world. If natural concepts have no connections to normative concepts whatsoever, as is true on the explanation of the sense of incredibility under consideration, then it seems possible to use your natural concepts to conceive of such a world in which slavery takes place without also using your normative concepts to conceive of slavery as wrong. After all, without such connections, it seems like you can conceive of slavery with all of its natural features and, as it were, stop there.³⁴ But it’s not possible to conceive of a world just like ours naturalistically without conceiving of ³⁴ Cf. Loar’s (1997) suggestion that the lack of analytic or otherwise a priori connections among phenomenal and physical concepts explains why failures of phenomenalphysical supervenience theses are conceivable.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Resisting Reductive Realism

107

slavery as wrong. In other words, the view under consideration seems to imply that failures of normative-natural supervenience theses that are intuitively inconceivable are conceivable.³⁵ It used to be that defending the conceivability of normative-natural supervenience failures would land an ethicist in hot water.³⁶ But those days are over. A surprising number of ethicists now argue explicitly that such supervenience failures are conceivable.³⁷ In light of such arguments, you might think that that the second problem isn’t much of one. If so, think of it this way. Even if normative-natural supervenience failures in ethics are conceivable, at least some of them are tremendously difficult to conceive of. I struggle to conceive of a world naturalistically like ours in which slavery is not wrong. I’d wager you do, too. But if, as is true on the explanation of the sense of incredibility under consideration, there aren’t connections among normative and natural concepts, then it seems like it would be easy to conceive of normative-natural supervenience failures in ethics (just “supervenience failures” from here on, unless otherwise stated). Thus, a proponent of the view that the unanalyzability of normative concepts explains the sense of incredibility still looks like they face a problem, even if supervenience failures in ethics are conceivable in principle. Addressing these two problems will clarify and elaborate the view that the unanalyzability of normative concepts has a role to play in accounting for the sense of incredibility. Accordingly, it will also be the focus of much of the rest of this chapter. I’ll start with the second problem, because my answer to it will carry an answer to the first problem along with it.

5.4.2. Why Supervenience Failures in Ethics are Hard to Conceive of Fortunately, I don’t have to go it alone in tackling the second problem, as several ethicists have recently set out to address it in different but related contexts. In particular, Hills (2009) and Hattiangadi (2018) suggest the same kind of explanation of why conceiving of supervenience failures can ³⁵ There are a variety of ways to precisify supervenience theses. But the problem I’m highlighting doesn’t turn on more than the basic idea of (strong) normative-natural supervenience—that there can’t be two worlds exactly naturalistically alike but normatively different. ³⁶ See Hare (1952: 145), Ridge (2007: 335), Shafer-Landau (2003: 78), and McPherson (2009: 77). ³⁷ See Hills (2009), Roberts (2017), Hattiangadi (2018), and Rosen (n.d.), the latter of whom has said in correspondence that he conceives of supervenience failures before he even has breakfast in the morning. Interestingly, a recent psychology study from Reinecke and Horne (2018) suggests that most people cannot engage in such conceptualizations.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

108

N. G. Laskowski

be difficult. Hills and Hattiangadi claim that some normative-natural supervenience failures are conceivable. They support this claim by appealing to similar cases. One such case involves a world that’s like ours naturalistically but where a different moral theory (e.g. consequentialism) is true from whichever one is true in our world (e.g. deontology). They, and especially Hills, suggest that not only is such a world conceivable—it’s easy to conceive of.³⁸ But both of them also acknowledge that not all supervenience failures are easy to conceive of. Hills and Hattiangadi appeal to cases involving Hitler to illustrate the point—it’s not easy to conceive of a world that’s like ours naturalistically but where Hitler’s actions are not wrong. They each suggest a similar explanation of why such a scenario is not easy to conceive of. According to Hills (2009: 172), conceiving of such a world could be “contaminating,” such that conceiving of it increases the risk of acquiring “bad habits.” Similarly, Hattiangadi (2018: 594) writes, “we are prevented” from conceiving of a normatively inverted Hitler scenario “by a powerful feeling of moral disgust.” When ethicists have trouble conceiving of worlds that are naturalistically but normatively different from our own, Hills and Hattiangadi might say, it’s because they want to avoid engaging in conceptual acts that are contaminating or disgusting.³⁹ Desires to avoid contaminating or disgusting actions might explain why ethicists struggle to conceive of a world that is naturalistically identical to ours in which Hitler’s actions are not wrong. But there are other worlds that are difficult to conceive of that aren’t well explained by such desires. For example, I have a hard time conceiving of a naturalistically identical world to ours that is normatively different, in which everyone with dark hair has an obligation to cartwheel on the third Sunday of each month. But it’s not plausible that I think of such an act as contaminating or disgusting.⁴⁰ My struggle to conceive of such a scenario is not plausibly explained by a corresponding desire to avoid contaminating or disgusting acts. At best, ³⁸ “Support” for their view, Hills (2009: 173, my emphasis) writes, “comes from reflection on different false moral claims . . . Suppose that you think (like the utilitarian) that it is not always wrong to tell a benevolent lie . . . Can you imagine that it is always wrong (as the Kantian believes), that it is a failure of respect for that person not to tell them the truth? Surely you can. Similarly, if you believe that the Kantian is right about this, you can imagine that she is not. But if so, then you can imagine two worlds with the same natural facts (someone tells another a proposition that they believe to be false in order to make her happy), one in which it is morally wrong, one in which it is morally acceptable. Perhaps in one world utilitarianism is true, in another Kant’s moral theory is true.” ³⁹ Moreover, according to Hills and Hattiangadi, the conceivability of supervenience failures is evidence for the further claim that such failures are metaphysically possible. ⁴⁰ Hattiangadi (2018: 595) notes that their explanation from disgust might only work to explain why it’s hard to conceive of normatively inverted “horrific” cases like Hitler’s.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Resisting Reductive Realism

109

the kind of explanation on offer from Hills and Hattiangadi has limited reach. It’s worth exploring whether a better explanation is available. Hills and Hattiangadi appeal to our desires or desire-like states to explain why it’s hard to conceive of a world that’s naturalistically identical to ours where Hitler’s actions aren’t wrong. This kind of noncognitive explanation makes an alternative explanation salient. Note that conceiving of a world that’s naturalistically identical to ours where Hitler’s actions aren’t wrong appears at least in part to be a kind of cognitive act. Perhaps ethicists have a hard time conceiving of Hitler’s actions as possessing their natural but not their normative features because conceiving of Hitler’s actions as possessing their natural features involves conceiving of Hitler’s actions as possessing their normative features. In particular, it could be that part of what it is to conceive of Hitler’s actions normatively as wrong is to conceive of Hitler’s actions as possessing some natural feature. As soon as ethicists use their natural concepts to conceive of Hitler’s actions as possessing their natural features, they might thereby count as using their normative concepts to conceive of Hitler’s actions as possessing their normative features. In other words, on this alternative explanation, ethicists have a hard time conceiving of a world that is naturalistically identical to our own where Hitler’s actions are not wrong because they can’t. Think of it this way. Hitler’s actions had a number of natural features or properties, such as their temporal and spatial properties. Plausibly, at least some of these natural properties are candidates for being identical with or fully constitutive of normative properties. For example, it’s plausible that Hitler’s actions had the natural property of failing to generate the weakest individual complaint, which is a candidate for being identical with or fully constitutive of the normative property of being wrong put forward in the tradition of Contractualism in normative ethics.⁴¹ If Hitler’s actions had the natural property of failing to generate the weakest individual complaint, then conceiving of his actions as possessing all of their natural features includes conceiving of them as possessing this property. Now suppose it turns out that the natural property of failing to generate the weakest individual complaint is identical with or fully constitutive of the normative property of being wrong. In conceiving of Hitler’s actions as possessing this natural property, ethicists are well on their way to conceiving of Hitler’s actions as possessing the normative property of being wrong. ⁴¹ The property of failing to generate the weakest individual complaint might not be a natural property—“complaint” has a normative ring to it. But the point I’m making doesn’t turn on the example I’m using to illustrate it. Feel free to substitute the property of failing to maximize pleasure or some other property that might be more clearly naturalistic.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

110

N. G. Laskowski

To be clear, I’m claiming that part of what it is to use a normative concept is to use a natural concept. Moreover, per our discussion above, I’m still pursuing the idea that normative concepts are unanalyzable concepts. Thus, I’m committing to the view that part of what it is to use a normative concept is to use an unanalyzable natural concept. This means that I’m not claiming that part of what it is to use the normative concept  is to use the analyzable natural concept       . That would be an implausible claim to make—not everyone competent with  is competent with or even possesses       . Rather, I’m claiming that you possess and are competent with a coextensive, unanalyzable natural concept the use of which is part of what it is to use the normative concept . As the discussion in section 5.3 illustrates, it could be that the concept is unanalyzable in virtue of being a prototype or atom. I’m not offering a view about this question, in part because I’m not offering a complete theory of normative concept use. I’m only saying enough about the nature of normative concepts to explain why it’s hard to conceive of a world that’s naturalistically identical to our own in which Hitler’s actions aren’t wrong. The idea is that conceiving of such a world involves conceiving of Hitler’s actions as possessing the natural property of failing to generate the weakest individual complaint. Conceiving of Hitler’s actions as possessing the natural property of failing to generate the weakest individual complaint involves using an unanalyzable natural concept. And using this unanalyzable natural concept is part of what it is to use the normative concept . That explains why it’s hard to conceive of a world that is naturalistically identical to ours in which Hitler’s actions aren’t wrong.⁴²

5.4.3. Explaining the Sense of Incredibility in Full Let’s take stock. By the end of section 5.3, I offered an explanation of why ethicists find reductive theses so obviously implausible that they struggle to believe them, i.e. an explanation of why ethicists experience the sense of ⁴² There are various ways to fill in the view to explain why it’s difficult to conceive of normatively inverted scenarios that are more innocuous, like the above case in which I struggle to conceive of a world that is like ours naturalistically in which everyone with dark hair has an obligation to cartwheel on the third Sunday of each month. One way would be to say that none of the natural concepts that I’m using in conceiving of such a world are among those natural concepts constitutive of using the normative concept . And that using whichever natural concept is constitutive of using  in conceiving of such a world would amount to conceiving of a world that is not naturalistically identical to our own.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Resisting Reductive Realism

111

incredibility. The idea is that normative concepts are unanalyzable and hence lack any connections to natural concepts, in virtue of being prototypes or atoms. But this explanation raised another issue. Without such connections, it might be implausibly easy to use natural concepts to conceive of supervenience failures, including conceiving of a world that is naturalistically identical to ours where Hitler’s actions are not wrong. I addressed this issue by offering a view of normative concept use. Part of what it is to use a normative concept is to use an unanalyzable natural concept. When ethicists conceive of Hitler’s actions as possessing the natural property of failing to generate the weakest individual complaint, they do so using an unanalyzable natural concept, and hence ipso facto use normative concepts to conceive of Hitler’s actions as wrong. The unanalyzability of normative concepts can be invoked to explain the sense of incredibility without worrying that supervenience failures become too easy to conceive of. That leaves the first of the two problems introduced in section 5.4.1. Recall that appealing to the unanalyzability of normative concepts looks like it explains why ethicists experience the sense of incredibility when reflecting on reductive theses purporting to state connections among normative and natural concepts. But ethicists also experience the sense of incredibility when reflecting on reductive theses purporting to state a posteriori truths not involving any connections among normative and natural concepts. The remaining issue is to find a way of supplementing the explanation from unanalyzability to explain the sense of incredibility in full. One way to address the issue is to continue developing the view advanced in the previous section of what it takes to use a normative concept. Reflect on the reductive claim that what it is for an action to be wrong is for it to fail to generate the weakest individual complaint, understood as stating a true a posteriori necessity. In reflecting on it, I claim that you’re using two natural concepts. One of those natural concepts is analyzable, namely,       . The other natural concept that you’re using is unanalyzable. I call this unanalyzable natural concept “gnorw.”⁴³ I also claim that using        is not constitutive of using , while using  is partially constitutive of using . Explaining why will reveal a full explanation of the sense of incredibility. ⁴³ I introduce the word “gnorw” for this unanalyzable natural concept in Laskowski (2019). It’s worth noting that some philosophers start becoming suspicious of the view when they hear the word “gnorw,” at least in correspondence. But this isn’t the right place for suspicion. That a word is needed to be introduced to talk about a concept that people already possessed is not unusual—I used to fear missing out on events like MadMeta and form thoughts about that fear-induced anxiety before I acquired the word “fomo” to talk about it.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

112

N. G. Laskowski

One explanation of why using  but not        is partially constitutive of using  is that the former but not the latter is unanalyzable. But that’s a lousy explanation. There are plenty of uses of unanalyzable natural concepts that aren’t partially constitutive of using normative concepts. For example, when I use the atomistic-demonstrative concept  to think of that dog over there I’m not thereby using a normative concept. So too would it be a lousy explanation to say that it’s because  but not        picks out the property of failing to generate the weakest individual complaint. Recall, both concepts pick out the same property. A better explanation of why using  but not using        is partially constitutive of using  is that only the former is hooked up with a corresponding desire or desire-like state. One way in which your use of  could be “hooked up” with a desire is in the sense that it activates or makes it occurrent.⁴⁴ And one way in which a desire could “correspond” to  is in the sense that  figures as part of its content, such as the desire to avoid performing actions that are gnorw.⁴⁵ The result of this illustrative package of claims is that you think of something as  when you think of it as  while actively desiring to avoid performing actions that are gnorw. By appealing to the way in which using  is related to a desire, I’m offering a kind of Humean answer to the question of why using the unanalyzable natural concept  is partially constitutive of using the normative concept . It’s an answer that carries with it a full explanation of why ethicists experience the sense of incredibility. Suppose that you’re reflecting on the thesis that what it is for an action to be wrong is for it to fail to generate the weakest individual complaint, which states a true a posteriori truth. Assuming the truth of the illustrative claims above, you’re using  and        in reflecting on the thesis. Your use of  activates or makes occurrent a desire to avoid performing actions that are gnorw whereas your use of        doesn’t. As such, the property that you’re thinking about in using  seems very different from the property that you’re thinking about in using    ⁴⁴ There are other views available of what it is for a use of a concept to be hooked up with a desire, such as the view that using the relevant concept disposes you to have the relevant desire. The view I’m plugging is meant to be illustrative. ⁴⁵ This view on what it is for a concept to correspond to a desire is also merely illustrative.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Resisting Reductive Realism

113

   , even though you’re using both to think about the same property. This results in the impression that it’s not possible to use  and        to think about the same property. It’s a mismatch between the way it seems to use these concepts that explains why ethicists experience the sense of incredibility when reflecting on a posteriori reductive theses involving them.⁴⁶ To be clear, I’m illustrating a distinctive kind of view of what it is to use a normative concept to explain the sense of incredibility in full. What it is to use the normative concept  is to use the unanalyzable natural concept  while actively desiring to avoid performing actions that are gnorw.⁴⁷ Part of what makes this a distinctive view of normative concept use is that it’s a distinctive hybrid view of what it is to use a normative concept. It’s a hybrid view in the sense that it explains normative concept use in terms of both a cognitive element (the natural unanalyzable concept ) and a noncognitive element (desiring to avoid performing actions that are gnorw). It’s a distinctive⁴⁸ hybrid view in the sense that it’s explicitly a view about normative concept use rather than the use of normative words or thoughts.⁴⁹ It’s also distinctive for its emphasis on the role of unanalyzable natural concepts.⁵⁰

⁴⁶ I’m using the word “seem” as another placeholder. I want to allow that it might seem like these concepts can’t pick out the same properties in the sense that using them feels phenomenologically different in virtue of activating a corresponding desire or desirelike state. But I also want to make space for a non-phenomenological explanation. Perhaps it might seem like these concepts can’t pick out the same properties in the sense that using them leads one to attend differentially to features of their awareness. In either case there would be the kind of structural mismatch in concept use that I conjecture is doing the work. Copp (2017: 48) offers an explanation in a similar spirit, which is expanded upon in Copp (2018). It’s worth noting that Copp’s view is more committal, built in part on a theory of vindicated reasoning that I’m not sure is strictly needed to explain the sense of incredibility. ⁴⁷ It’s a commitment of the view that everyone competent with  has the same desire. Some, like Schroeder (2009) and Perl (2018), have worried about this commitment. A full defense of it requires more space. But I will say that I don’t find it hard to talk myself into it. It strikes me as quite plausible to say that you don’t count as competent with  unless you desire to get along with others in your society, for example. ⁴⁸ But see also Finlay (2019), who now champions a related view explicitly. ⁴⁹ See Toppinen (2017) for an overview of hybrid views of thought and talk (but not concepts). See also Perl (2018), who develops a sophisticated account of normative thought that secures a wide range of theoretical goods on behalf of hybridists of a particular stripe. Perl’s view challenges other friends of hybridism to do more with their hybrid views. This chapter, along with Laskowski (2019), can be understood as making modest steps in that direction. ⁵⁰ But see Ridge (2015), who also suggests a role for unanalyzable concepts (demonstrative concepts) in his brand of hybridism. Mogensen (2018) can be read as suggesting that there’s something objectionably exotic about this element of Ridge’s view and

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

114

N. G. Laskowski 5 . 5 . C ON C L U S I O N

My goal has been to better understand why ethicists resist reductivism in the sense of experiencing the sense of incredibility, in the sense of finding reductive theses so obviously implausible that they struggle to believe them. After arguing against a contextualist explanation, I put forward the idea that the explanation involves the unanalyzabililty of normative concepts. Not only did this fail to explain the sense of incredibility in full, it also seemed to make it too easy to conceive of normative-natural supervenience failures in ethics. The keys to answering these challenges were found in claims about the nature of normative concept use. In particular, they were found in the claim that using normative concepts involves using an unanalyzable natural concept in the presence of a suitably related desire. I said at the beginning of this chapter that I’m not the only one who thinks resistance to reductivism in ethics can be understood in terms of normative concepts. Chappell and Yetter-Chappell (2013) and Mehta (2019) suggest similar ideas. They recognize that seemingly parallel forms of resistance to reductivism crop up in various philosophical domains, including, especially, philosophy of mind with respect to phenomenal consciousness. Chappell and Yetter-Chappell and Mehta also suggest that some of the explanations that have been offered by philosophers of mind to explain these analogous forms of resistance might help explain resistance to reductivism in ethics. In particular, Mehta suggests that something like conceptual unanalyzability can explain everything across these domains. But resistance to reductivism patterns very differently in ethics than in domains like the philosophy of mind. As I argued, it seems to be a problem for the view that unanalyzability explains the sense of incredibility that it makes it too easy to conceive of normative-natural supervenience failures in ethics. Such an issue would be much less of one in the philosophy of mind, where there is much more tolerance for the conceivability of supervenience failures (think “philosophical zombies”).⁵¹ This chapter tells in favor of the sort of “essentially piecemeal” solution to understanding resistance to reductivism in philosophy that Mehta dismisses.⁵²

perhaps the view on offer here. But section 5.3 makes it clear that unanalyzable concepts are an ordinary part of our cognitive economy. ⁵¹ Chappell and Yetter-Chappell (2013: 871) also flag this difference. ⁵² Special thanks to Alex Dietz, Stephen Finlay, Joe Horton, Nathan Robert Howard, Janet Levin, Mark Schroeder, François Schroeter, Ralph Wedgwood, and Daniel Wodak for providing numerous rounds of feedback on previous drafts. Special thanks for providing multiple rounds of feedback as well to Neil Roughley and the other members

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Resisting Reductive Realism

115

References Balog, K. (2009). “Phenomenal Concepts” in Brian McLaughlin, Ansgar Beckermann, and Sven Walter (eds), The Oxford Handbook of Philosophy of Mind. Oxford: Oxford University Press. Chalmers, D. (forthcoming). “The Meta-Problem of Consciousness,” Journal of Consciousness Studies. Chappell, R. (2019). “Why Care about Non-Natural Reasons,” American Philosophical Quarterly 56(2): 125–34. Chappell, R., and Yetter-Chappell, H. (2013). “Mind-Body Meets Metaethics: A Moral Concept Strategy,” Philosophical Studies 165(3): 865–78. Copp, D. (2017). “Normative Naturalism and Normative Nihilism: Parfit’s Dilemma for Naturalism” in Simon Kirchin (ed.), Reading Parfit: On What Matters. London: Routledge. Copp, D. (2018). “Just Too Different: Normative Properties and Natural Properties,” Online First. https://doi-org.csulb.idm.oclc.org/10.1007/s11098-0181189-1 Cuneo, T. (2015). “What’s to be Said for Nonnaturalism” in Kelly J. Clark (ed.), The Blackwell Companion to Naturalism. Malden, MA: Wiley-Blackwell. Dowell, J. (2013). “Naturalism, Ethical” in Hugh LaFollette (ed.), The International Encyclopedia of Ethics. Malden, MA: Wiley-Blackwell. Dunaway, B. (2017). “Realism and Objectivity” in T. McPherson and D. Plunkett (eds), The Routledge Handbook of Metaethics. London: Routledge. Enoch, D. (2011). Taking Morality Seriously: A Defense of Robust Realism. Oxford: Oxford University Press. Finlay, S. (2014). Confusion of Tongues: A Theory of Normative Language. Oxford: Oxford University Press. Finlay, S. (2019). “Defining Normativity” in Kevin Toh, David Plunkett, and Scott Shapiro (eds), Dimensions of Normativity: New Essays on Metaethics and Jurisprudence. Oxford: Oxford University Press. FitzPatrick, W. (2014). “Skepticism about Naturalizing Normativity: In Defense of Ethical Nonnaturalism,” Res Philosophica 91(4): 559–88. Fodor, J. (1998). Concepts: Where Cognitive Science went Wrong. Oxford: Oxford University Press. Goldman, A. (1993). “Ethics and Cognitive Science,” Ethics 103(2): 337–60. Hare, R. (1952). The Language of Morals. Oxford: Clarendon Press. of his Philosophical Anthropology and Ethics Research Group at Duisburg-Essen Universität, including, especially, Flavia Felletti, Stefan Mandl, and Hichem Naar. Thanks to audience members at the University of Groningen, including, especially, Daan Evers, Bart Streumer, and Herman Veluwenkamp. Thanks to Matti Eklund and other participants at his Higher Seminar at Uppsala University, including, especially, Katharina Felka and Nils Franzén. Last but not least, thanks for feedback from the audience at the 2018 Wisconsin Metaethics Workshop, two anonymous referees from Oxford University Press, and Russ Shafer-Landau for making it all happen.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

116

N. G. Laskowski

Hattiangadi, A. (2018). “Moral Supervenience,” Canadian Journal of Philosophy 48(3–4): 592–615. Hills, A. (2009). “Supervenience and Moral Realism” in Alexander Hiekeand Hannes Leitgeb (eds), Reduction, Abstraction, Analysis. Frankfurt: Ontos Verlag, 11–163. Howard, N. and Laskowski, N. (forthcoming). “The World is Not Enough,” Noûs. Hulse, D., Read, C., and Schroeder, T. (2004). “The Impossibility of Conscious Desire,” American Philosophical Quarterly 41(1): 73–80. Jackson, F. (1998). From Metaphysics to Ethics: A Defence of Conceptual Analysis. Oxford: Oxford University Press. Joyce, R. (n.d.). “Enough with the Errors: A Final Reply to Finlay.” Available at http://personal.victoria.ac.nz/richard_joyce/acrobat/joyce_2012_enough.with.the. errors.pdf Laskowski, N. (2018a). “Epistemic Modesty in Ethics,” Philosophical Studies 175(7): 1577–96. Laskowski, N. (2018b). “Reductivism, Nonreductivism, and Incredulity about Streumer’s Error Theory,” Analysis Reviews 78(4): 766–76. Laskowski, N. (2019). “The Sense of Incredibility in Ethics,” Philosophical Studies 176(1): 93–115. Laskowski, N., and S. Finlay (2017). “Conceptual Analysis in Metaethics” in T. McPherson and D. Plunkett (eds), The Routledge Handbook of Metaethics. New York: Routledge, 536–51. Loar, B. (1997). “Phenomenal States,” in N. Block, O. Flanagan, and G. Güzeldere (eds), The Nature of Consciousness: Philosophical Debates. Cambridge, MA: MIT Press. Lutz, M., and J. Lenman (2018). “Moral Naturalism” The Stanford Encyclopedia of Philosophy. https://plato.stanford.edu/entries/naturalism-moral/ McPherson, T. (2009). “Unnatural Normativity? Critical Notice of Ralph Wedgwood’s Nature of Normativity,” Philosophical Books 50(2): 63–82. Margolis, E., and S. Laurence (1999). “Concepts” in T. A. Warfield and S. P. Stich (eds), The Blackwell Guide to the Philosophy of Mind. Oxford: Blackwell, 190–213. Mehta, N. (2019). “Phenomenal, Normative, and Other Explanatory Gaps: A General Diagnosis,” Philosophy and Phenomenological Research 98(3): 567–91. Melnyk, A. (n.d.). “The Sense of Incredibility: A Physicalist Explanation.” Manuscript. Mogensen, A. (2018). “How Ecumenical Expressivism Confuses the Trivial and the Substantive,” Analysis 78(4): 666–74. Nagel, T. (1986). The View from Nowhere. Oxford: Oxford University Press. Paakkunainen, H. (2017). “The ‘Just Too Different’ Objection to Normative Naturalism,” Philosophy Compass 13(2): 1–13. Parfit, D. (2011). On What Matters, vols 1 and 2. Oxford: Oxford University Press. Parfit, D. (2017). On What Matters, vol. 3. Oxford: Oxford University Press. Perl, C. (2018). “A User’s Guide to Hybrid Tools,” Mind, fzy063, https://doi.org/ 10.1093/mind/fzy063 Reinecke, G., and Z. Horne (2018). “Immutable Morality: Even God Could Not Change Some Moral Facts.” https://doi.org/10.31234/osf.io/yqm48

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Resisting Reductive Realism

117

Ridge, M. (2007). “Anti-Reductionism and Supervenience,” Journal of Moral Philosophy 4(3): 330–48. Ridge, M. (2015). “Replies to Critics,” Analysis Reviews 75: 471–88. Roberts, D. (2017). “Why Believe in Normative Supervenience?” Oxford Studies in Metaethics 205. Rosen, G. (n.d.). “After Supervenience.” Manuscript. Scanlon, T. (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. Schroeder, M. (2009). “Hybrid Expressivism: Virtues and Vices,” Ethics 119(2): 257–309. Schroeder, M. (2017). “Normative Ethics and Metaethics,” in T. McPherson and D. Plunkett (eds), The Routledge Handbook of Metaethics. London: Routledge. Schroeder, M. (n.d.). “A Common Subject for Ethics.” Manuscript. Schroeter, L., and F. Schroeter (2015). “Review: T. M. Scanlon, Being Realistic about Reasons,” Ethics 125(4): 1225–30. Shafer-Landau, Russ (2003). Moral Realism: A Defence. Oxford: Oxford University Press. Sinhababu, N. (2019). “One-Person Moral Twin Earth Cases.” Thought: A Journal of Philosophy 8(1): 16–22. Stich, S. (1993). “Moral Philosophy and Mental Representation.” In M. Hechter, L. Nadel, andR. E. Michod (eds), The Origin of Values. New York: Aldine de Gruyer, 215–28. Streumer, B. (2017). Unbelievable Errors: An Error Theory about All Normative Judgments. Oxford: Oxford University Press. Thomson, J. (1992). “On Some Ways in Which a Thing Can Be Good,” Social Philosophy and Policy 9(2): 96–117. Toppinen, T. (2017). “Hybrid Accounts of Ethical Thought and Talk” in T. McPherson and D. Plunkett (eds), The Routledge Handbook of Metaethics. New York: Routledge. Wiggins, D. (1993). “Cognitivism, Naturalism, and Normativity: A Reply to Peter Railton” in J. Haldane and C. Wright (eds), Reality, Representation, and Projection. Oxford: Oxford University Press, 301–13. Ziff, P. (1960). Semantic Analysis. Ithaca, NY: Cornell University Press.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

6 Moral Realism and Philosophical Angst Joshua Blanchard

Philosophers express angst about the existence of God, the immortality of the soul,¹ freedom of the will, the nature of mind,² the metaphysics of modality,³ and other subjects. On such issues, some of us do not only try to figure out what is true, we worry about it. Consider the case of theism. Positions range anywhere from the view that life is meaningless without God, to the view that it is terrible with God, with room between those extremes.⁴ Less attention is devoted to similar questions in metaethics. Yet, just as someone might not only wonder, but worry about whether theism is true—so they might not just wonder, but worry about whether a particular metaethical theory is true. Given the importance of morality, there is much to explore in this area. This essay begins to give angsty metaethics the attention that it deserves.⁵ After sketching an account of philosophical angst, I describe and defend a particular instance of it: angst regarding moral realism. Following the convention in the literature on the value of God’s existence, in which “pro-theism” refers to the view that it is good if God exists, I will use

¹ Eli Hirsch (2009: 2, 20) says that death and radical deception are “ineffably horrible” and “undermine everything that I have ever cared about in my life.” ² Jerry Fodor (1989, 77) says that “if it isn’t literally true that my wanting is causally responsible for my reaching, and my itching is causally responsible for my scratching, and my believing is causally responsible for my saying . . . If none of that is literally true, then practically everything I believe about anything is false and it’s the end of the world.” ³ Robert Adams (1979) argues that modal realism justifies “moral indifference.” ⁴ See the essays in Klaas Kraay (2018). ⁵ See also Guy Kahane (2012)’s pioneering work on the “value question in metaphysics.” Kahane is more narrowly interested in evaluative judgments rather than full-fledged philosophical angst, but this chapter could be read partly as a metaethical contribution to the research program advanced by Kahane.

Joshua Blanchard, Moral Realism and Philosophical Angst In:Oxford Studies in Metaethics Volume 15. Edited by: Russ Shafer-Landau, Oxford University Press (2020). © Joshua Blanchard. DOI: 10.1093/oso/9780198859512.003.0006

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Moral Realism and Philosophical Angst

119

“pro-realism” to refer to the view that it is good if moral realism is true. According to pro-realism, the truth of realism is important for some good, and there are no adequate alternatives: it is realism or bust. The phenomenon of pro-realism is immediately recognizable, but it is less obvious how to understand and defend it. If moral nihilism is false, how could something of great normative significance hinge on which non-nihilistic metaethic is true? There are many non-realist theories that, by their own lights, vindicate the moral domain—so how could it be bad if one of those theories turned out to be right? This essay seeks to make angst about moral realism intelligible and commends some ways for pro-realists to make their case.

6 .1 . P HI LO S O P H I C A L AN GS T We may know philosophical angst when we see it, but it is useful to sketch a more exact account in order to isolate the angst-inducing features of any particular domain and morality in particular. Here is how I understand the phenomenon: Philosophical angst is a complex cognitive judgment combined with a conative component. The cognitive judgment is that something existentially important is also irreplaceable. Existential importance is positive importance to the meaning, intelligibility, or overall value of our lives and the world. If you think that a claim might be false but do not think that its truth is important, then you do not count as angsty. Irreplaceability means that there is no available surrogate for the important role in question. If you think that an existentially important claim might be false but that some alternative would do just as well, then you aren’t angsty. The conative component, licensed by the judgment, is an anxiety that what is important and irreplaceable might be unavailable. To be philosophically angsty, one need not believe the claim that one is angsty about, but the claim and its falsity must be live options. The most typical catalysts for philosophical angst are subjects including God, freedom, and death, rather than, say, the ontology of numbers or the semantics of counterfactuals. The former but not the latter subjects are especially significant for thinking about the meaning of our lives. But it is surprising that metaethics is not subject to more angsty reflection, as it concerns the metaphysics, epistemology, semantics, and psychology of arguably the most important domain of value that there is—at least, a very important one. Regarding angst, metaethics seems closer to the philosophy of religion than to the ontology of numbers. The angsty metaethicist who most readily comes to mind is probably Derek Parfit. Parfit says that, if metaethical non-naturalism is false, then

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

120

Joshua Blanchard

nothing matters.⁶ This declaration satisfies the cognitive conditions for philosophical angst, provided that it includes or presupposes the thought that it is existentially important that something matters, and that no rival theory secures an adequate replacement for non-naturalist mattering. But Parfit’s angst is really quite narrow. He thinks that views rival to his collapse into nihilism or something very close to it. Parfit does not offer a substantive proposal for why things matter only on his view; rather, their mattering is an immediate logical or conceptual consequence of the ways that he has defined the relevant terminology and concepts, as commentators have pointed out.⁷ Because he thinks that his rivals’ views collapse into nihilism, Parfit’s angst amounts to the worry that nothing matters if nihilism is true—a claim less controversial than it seemed prior to clarification. Moreover, it is not clear that Parfit thinks it would be a bad thing if nihilism were true. Indeed, in volume 3 of On What Matters, he distances himself from the claim that it matters whether something matters.⁸ These qualifications provide a good foil for my central concerns. In contrast to Parfit, we might be angsty about metaethics not because we think that most views collapse into nihilism, though that would be a reason to be angsty. We might instead worry that even many views that do not collapse into nihilism fail to accommodate something irreplaceably important to the meaning or intelligibility of our lives. This kind of angst accommodates more philosophical charity, because it evaluates theories on their own terms, allowing that they accomplish what they purport to. I am wondering about metaethical angst in this sense. I will discuss realism’s relationship to (moral) nihilism in its own right, but I am also interested in how realism compares to its non-nihilistic rivals—even when they succeed by their own lights. Compare: in thinking about freedom of the will, someone might be angsty if, in their view, only libertarianism about freedom secures something (in particular, libertarian freedom) that is both existentially important and irreplaceable. This is different from thinking that only libertarianism secures freedom in any sense. Someone (even a libertarian) may grant that compatibilist theories secure freedom of the will,

⁶ “If there were no such facts [facts that only ‘irreducibly normative claims’ could state], and we didn’t need to make such claims, Sidgwick, Ross, I, and others would have wasted much of our lives. We have asked what matters, which acts are right or wrong, and what we have reasons to want, and to do. If Naturalism were true, there would be no point in trying to answer such questions. Our consolation would be only that it wouldn’t matter that we had wasted much of our lives, since we would have learnt that nothing matters” (Parfit 2011: 367). ⁷ See Street (2017), Temkin (2017), and Driver (2017). ⁸ See n. 16.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Moral Realism and Philosophical Angst

121

but insist that such theories do not secure something valuable or existentially important about the nature of freedom. In short, angst about moral realism, or pro-realism, is the view that the truth of moral realism is necessary for something irreplaceable and existentially important to our lives, combined with anxiety that realism might be false. And this does not mean that the pro-realist must think that all alternative theories collapse into nihilism.

6 .2 . H O W T O B E A N G S T Y A B O UT M O R A L RE AL I SM I am thinking of moral realism in an ambitious sense, though it is a research project in its own right to figure out which among realist options best satisfies certain existential concerns.⁹ According to moral realism in this sense, there are objective moral facts, and these serve as the content of true beliefs just as descriptive non-moral facts do (for example, the facts that human beings evolved or that two and two make four). We can know and express moral reality propositionally, and we have, by-and-large, made progress in moral inquiry. The moral facts are necessary and radically response- or stance-independent, in that they are not fundamentally grounded in human minds or attitudes, whether individual, collective, actual, ideal, or otherwise hypothetical. This sort of picture, though loosely sketched, should be reminiscent of Enoch’s (2011) “robust” realism, Eklund’s (2017) “ardent” realism, and similar metaethical options. Eklund’s ardent realism is the view that “reality itself favors certain ways of valuing and acting.” The ardent realist “wants reality to undergird some ways of valuing over others,” such that there is even reason to favor a unique concept of “good” over alternatives.¹⁰ I agree with Eklund’s sense that “ardent realist intuitions and motivations are widely shared,” even if few would call themselves by that name. Pro-realism says that something about this picture irreplaceably plays an existentially important role in the meaning or intelligibility of our lives, and so the fact that realism might be false warrants anxiety.¹¹ It is not possible here to canvass all non-realisms in their dizzying ⁹ See Michael Zhao (2018) for the view that only a different kind of moral realism (of a naturalistic, “teleological” sort) secures the importance of morality. ¹⁰ Eklund (2017: 1, 5). ¹¹ My account of why the falsity of moral realism induces philosophical angst is friendly to the account of angst (or anxiety, or dread) that one finds in Jean-Paul Sartre’s Being and Nothingness and “Existentialism is a Humanism.” Sartre (1984) locates the origin of anxiety in the fact that human beings are radically responsible not only for their own actions but for the principles that guide them—there is no objective, independent moral reality that determines this in advance. Sartre didn’t have in mind contemporary

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

122

Joshua Blanchard

variety, so in what follows I will highlight three general contrasts: the contrast between realism and moral nihilism, between realism and antirealism, and, finally, between realism and quasi-realism. The reasons why it would be bad if moral nihilism were true are distinct from the reasons why it would be bad if any form of non-realism (including both anti- and quasi-realism) were true. Unlike moral nihilism, anti- and quasi-realist theories are vindicatory metaethics, in that they (by their own lights) justify moral discourse and practice. In short, they are not selfconceived as error theories.¹² The pro-realist, however, is not only hopeful that morality be given a vindicatory account, à la non-nihilism (a hope that anti- and quasi-realists may also satisfy), but that it be given a distinctly realistic account. It is not just morality per se that enjoys existentially important irreplaceability, but morality realistically construed.

6.2.1. The Badness of Moral Nihilism Consider a simple example of an unmitigated injustice. In the History of the Peloponnesian War, Thucydides gives us an exchange between some Athenian invaders and their victims, the Melians. The Melians have no chance of resisting conquest. Emphasizing the pointlessness of resistance, the Athenians accurately observe that “the strong do what they can and the weak suffer what they must.”¹³ What does this have to do with metaethics? If moral nihilism is true—that is, if morality is bunk—then there is no moral problem with this relationship between the strong and the weak. The dominance of the strong over the weak may be painful and regrettable, but it would not be morally unjustified or otherwise morally objectionable. This is a good ground for angst about the prospect of moral nihilism, because it seems bad that there would be no grounds for objecting morally to this pitiful state of affairs. Not morally bad—after all, nothing morally substantive follows from the truth of moral nihilism—but bad in a nonmoral sense; our lives (or, at least, the lives of the weak) are worse-off if they lack grounds for moral objection. It is appropriate to hope that morality is not bunk. This simple argument is intuitive, but is the hope really reasonable? In what non-moral sense is it “bad” to lack moral justification? As mentioned, metaethical categories and one cannot know whether he would have also rejected antirealism. Cf. Zhao (2018). ¹² Some nihilists want to vindicate moral discourse and practice in a revisionary sense—see Mackie (1977) and Joyce (2002). Thanks to an anonymous reviewer for pointing this out. ¹³ Thucydides (1996: 5.89, 416).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Moral Realism and Philosophical Angst

123

it cannot be that moral nihilism results in some morally bad situation, since nihilism results in no morally evaluable situations at all. Moreover, morality does not itself prevent the strong from doing what they can or the weak from suffering what they must; morality does not change the world. If morality does not change the world, then it is hard to see how nihilism would be bad even in a prudential or instrumental sense.¹⁴ What non-moral good does it do us that we can legitimately morally object to something, especially if it is just going to happen anyway? Nowadays it is common to distinguish between multiple domains of value, including but not limited to morality, meaning, aesthetics, and selfinterest. Careful attention to these distinctions and their interaction in deliberation is one of the significant accomplishments of contemporary normative theory. A life might be good along some of these dimensions but not others. A predominantly self-interested person might live a life abounding in prudential value (say, they always get what they want) but sorely lacking in moral value. The “moral saint” might live a life abounding in moral value but sorely lacking in meaning.¹⁵ The acknowledgment that moral value does not exhaust the values that we should (and do) care about is an important advancement beyond a myopic focus on moral value. But if philosophers of the past too often ignored the role that non-moral value plays in a good life, it is also important to see that moral value may contribute to, or even be required for, the fulfillment of the other kinds of value. It is true that a person’s life may be abounding in moral worth yet lacking in meaning, but it is also true that a life entirely lacking in moral worth may—precisely in virtue of that lack—suffer from meaninglessness. In answering the question about what kind of badness accrues to us if nihilism is true, the pro-realist must appeal to one of the non-moral domains of value to which moral value nevertheless contributes. The domain of meaning is well-suited to this purpose. What morality does, vis-à-vis the Athenian-Melian dynamic, is secure for us a particular normative standing in virtue of which we enjoy an

¹⁴ Here I am thinking of the potential moral goodness or badness of the truth of nihilism as having to do with what would be intrinsically good or bad about the nihilistic possibility itself. An anonymous reviewer suggests that if realism is true in the actual world, then the judgment “It would be bad if nihilism were true” could be true in the actual world, if by the true standard’s own lights the nihilistic possibility is worse. This suggestion is worthy of serious further exploration, though I will just flag here that I am not sure it is coherent: if not its intrinsic features, then in virtue of what could the nihilistic possibility be morally worse? It seems to me that the nihilistic world would at best be morally incommensurate with the actual world. See Kahane (2016) for discussion of related problems. ¹⁵ Wolf (1982).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

124

Joshua Blanchard

authoritative claim against injustices. It is much better that we have this kind of standing and claim, even when we have no way of enforcing it. It is non-morally good to be morally worth treating well. Unjustified harms and oppression cause one’s life to go poorly, but one’s life is better if one has available an authoritative protest, even when it is ineffective. Such a protest secures meaning in the midst of harms; it reflects that one’s condition is intelligible as morally unjustified, and objectionable on these grounds. There is no adequate, non-moral replacement for this standing and claim if morality is bunk. The death of morality may leave in its wake various instrumental and epistemic norms, but these are not even inconsistent with the aims and values that the Athenians press against the Melians. Moral standing differs from other kinds of standing not in degree, but in quality. Over and above instrumental, epistemic, and other norms, moral standing provides the strong with a reason not to harm the weak. This defense of angst about nihilism locates something that morality delivers for us and appeals to its irreplaceability and non-moral importance in our lives. My approach is consonant with some provocative, albeit narrower, remarks by Thomas Nagel regarding the importance of inviolability in any account of moral rights. Nagel writes, “What actually happens to us is not the only thing we care about: What may be done to us is also important, quite apart from whether or not it is done to us—and the same is true of what we may do as opposed to what we actually do” (2007: 108). Again, “not only is it an evil for a person to be harmed in certain ways, but for it to be permissible to harm the person in those ways is an additional and independent evil.” And finally, We can distinguish the desirability of not being tortured from the desirability of its being impermissible to torture us; we can distinguish the desirability of not being murdered from the desirability of our murder’s being impermissible; we can distinguish the desirability of not being coerced from the desirability of its being impermissible to coerce us. These are distinct subjects, and they have distinct values. To be tortured would be terrible; but to be tortured and also to be someone whom it was not wrong to torture would be even worse. (2007: 111)

Being the sort of being whom it is permissible to torture is bad, even despite the fact that it is in another way worse to be “killed unjustly than, say, accidentally.” Nagel’s point is not that permissible torture or murder are themselves worse than their impermissible counterparts; rather, it is “being someone it is not wrong to torture” that is worse than being someone it is wrong to torture. What if a realist world is such that, even if or precisely because it has moral value, it is, on the whole, very bad for us—for example, if it is full of little more than intense and constant suffering, and the best normative

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Moral Realism and Philosophical Angst

125

theory says that the suffering is morally unjustified? Does the non-moral goodness of mere (violated) moral status really outweigh or otherwise trump such a state of affairs? Surely (this objection insists), a realistic possibility is only better than its nihilistic counterpart if the net value in the world, for us, turns out positive. In contrast to a very morally bad realistic world, an otherwise good or even so-so nihilistic world seems better for us. If moral nihilism is true, then what is otherwise unjust suffering is not morally bad at all; it is nothing, morally speaking. And it is plausible that a set of morally bad states is worse than a set of states that are neither morally good nor bad. So, perhaps it is not true as a general proposition that it is better if realism rather than moral nihilism is true.¹⁶ Consider one of this view’s implications. If an otherwise morally bad state is better if it is neither morally good nor morally bad, then finding out that one’s otherwise unjust suffering is morally neutral should be encouraging news. If a would-be morally bad state (say, the state of being betrayed) was better on moral nihilism, then finding out that moral nihilism is true should be a kind of appropriate comfort for a person experiencing betrayal. Such a discovery would not alleviate a person’s anger or unhappiness, but it would mean that they were not morally wronged, and that no particular reactions are morally justified. Any appeal to moral nihilism for comfort is wrongheaded. It is better not only that there is moral value; it is better even that morally bad things really are so. If one is deciding between having a set of morally neutral things (say, trips to the beach) or a separate set of moral evils (say, trips to the torture chamber), then, all else equal, it is reasonable to choose the neutral things. If the question of preferring realism over nihilism were analogous to such a decision, then it would be reasonable to prefer nihilism in some of these cases. It is from this fact, I suspect, that the objection derives its plausibility. But this is not the right way to think about comparing metaethical possibilities. The would-be goods and evils must be held fixed in the comparison: this bit of would-be mercy, this instance of would-be wrongful treatment, etc.

¹⁶ This thought seems to motivate Kahane (2016). It is also given expression at various points by Parfit in vol. 3 of On What Matters, e.g. “It is a difficult question whether and how it matters whether anything matters. If we believe that suffering matters greatly, we may regret this fact. We might try to believe that, as Nihilists claim, nothing matters, because we have no reason to care about anything.
We might then conclude in despair that Nihilism is false, because some things, such as suffering, really do matter.” Again, “I don’t know whether I would be . . . very disappointed if I came to believe that nothing matters in [the] reason-involving sense. I am not glad, for example, that suffering matters. But since I believe that we have reasons to care about suffering, and that we have other, weaker reasons to care about some other things, I am trying to understand these reasons better.” Unlike Parfit, I am glad that suffering matters.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

126

Joshua Blanchard

We are considering not just whether we have certain goods or evils period, but what it would be like if the very things we already rightly believe to be good or bad were not really so. In this way, contrasting moral nihilism and non-nihilism bears some similarity to contrasting first-order normative theories that have different evaluative upshots, with the aim of figuring out which would be better. When contrasting Kantianism and Utilitarianism, for example, one need not radically change what the world is like when one imagines the truth of the rival theory—e.g. one need not toggle whether or not there are pleasure and pain states. Rather, one holds fixed the world as it otherwise is— including the existence of pleasures, pains, and rational capacities—and imagines the implications of Kantianism and Utilitarianism for this world. To illustrate the general view on offer with an example, not only is torturing innocent uncles for fun morally bad, but we rightly do not want it to be otherwise. The question of whether torturing for fun is good or bad is not itself an evaluatively idle question. It is non-morally good for uncles, and for us, that it is impermissible to torture them for fun. A world in which torturing innocent uncles is morally permissible, let alone good—holding everything else fixed—is worse than one in which it is morally evil. That is to say, it is to some extent better that torturing innocent uncles is morally bad rather than morally good. The intuition that the permissibility of torture would itself be bad is related to a popular normative objection to simplistic versions of divine command theory, according to which one ought to do whatever God happens to command. The objection says that such a theory allows God to make it the case that actually evil actions are good, and that this counts against the appeal of the theory. Returning to the Melian dialogue, the point here is not that the Melians should be happy that their moral claim and status are violated, nor should they wish that their violation was in fact permissible. Rather, the point is that they should prefer having a moral claim and status, given the facts that actually constitute the violation, to not having them. Ultimately, one need not be convinced by the details either of Nagel’s view or my view regarding, respectively, what is important about either a deontological first-order theory of morality or a non-nihilistic metaethic. Whatever one thinks are the goods (including non-moral goods) conferred by morality, these are vindicated by any non-nihilistic metaethical theory. In other words, take whatever goods you think are delivered by the best firstorder moral theory. These goods are, in turn, vindicated by the true nonnihilistic metaethic, and it is to these that we can refer in justifying an angsty aversion to the possibility of nihilism. Notice that in order to reject this argument, you must think that morality makes no net-positive non-moral

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Moral Realism and Philosophical Angst

127

contribution to our lives. If, per impossible, we could choose whether to live in a morally nihilistic or non-nihilistic world, and could not appeal to moral values in making the decision, the person who rejects the argument of this section must say that we have no positive reason to choose the non-nihilistic world. And that is very implausible.

6.2.2. The Badness of Anti-Realism To consider moral nihilism is to consider the possibility that there are none of the goods of morality; accordingly, these goods are dialectically available for licensing evaluative judgments about the possibility of moral nihilism. Moral anti-realism is trickier, because to consider anti-realism is to consider whether there are a great number of the goods of morality, but understood in an anti-realistic way.¹⁷ Speaking loosely, angst about moral nihilism is concern about the existence of morality, but angst about anti-realism is concern about the nature of morality. The rest of this subsection applies to anti-realist theories other than quasirealism, due to complexities arising from the pan-expressivist or minimalist semantic program partly constitutive of quasi-realism. In going expressivist or minimalist at every level of discourse about the moral, the quasi-realist makes it peculiarly difficult to say how quasi-realism differs from realism in its characterization of the moral domain, which in turn makes it difficult to say what is at stake between the two theories. Because this issue arises for none of the other varieties of anti-realism, it is useful to categorize and treat them separately. What is angst-inducing about (non-quasi) anti-realism? Why might it induce angst in someone if the moral badness of, say, murder is not realistically construed? Why might it induce angst in someone if moral facts are, say, constructions grounded in collections of subjective attitudes? Or if moral truths are contingent commitments that human beings have endorsed over time, for evolutionarily specified purposes? Or if moral facts are, fundamentally, facts about the results of idealized decision-procedures among fully informed agents? Standard, longstanding disagreements between realists and anti-realists are suggestive of answers having to do with objectivity, non-contingency, and the like. I suspect that there is something potentially angst-inducing in ¹⁷ There is an unfortunate terminological parallel between “pro-realism” and “antirealism.” With “anti-realism” I refer to anti-realist metaethics, not an angsty aversion to moral realism. I’m inclined to employ the somewhat unhappy label, “anti-moral realism,” to refer to the angsty aversion to moral realism. In correspondence Selim Berker suggests “con-realism” as a slightly happier unhappy label.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

128

Joshua Blanchard

the anti-realistic interpretation of each of these aspects of morality, but the answer that I want to pursue here has to do with what I take to be the radically independent character of moral justification if moral realism is true. Following Sharon Street, I regard some kind of mind- or responsedependence of moral phenomena to be constitutive of moral anti-realism. Here is Street: According to the anti-realist, if an agent has normative reason to X, then this conclusion must somehow follow from within her own practical point of view: if the conclusion that she has reason to X is not entailed from within the standpoint constituted by her own set of evaluative attitudes, then she does not have that reason.¹⁸

Street elaborates in a footnote, “the point of contention between realists and anti-realists about normativity is the answer to the central question of Plato’s Euthyphro (in rough secular paraphrase): are things valuable ultimately because we value them (anti-realism), or do some things possess a value that holds independently of us and our attitudes (realism)?”¹⁹ Street locates her own constructionist view on the anti-realist side of this “Euthyphronic” account of the debate. (Street’s own exclusion of quasirealism in this part of the discussion is what permits the Euthyphronic account, since quasi-realists emphatically deny that things possess moral value in a response-dependent way.) Street’s characterization is a bit too narrow for my purposes, since it counts idealizing theories as realistic. It is also focused on an agent’s normative reasons, rather than moral value and truth generally. The version of the realist/anti-realist distinction that will I utilize counts our hypothetical, idealized, and collective responses as part of “us and our attitudes.” So, here is the modified characterization of anti-realist theories that I will employ in what follows: According to anti-realism, moral truths hold in virtue of us and our attitudes, broadly construed: if a putative moral truth is not grounded in our evaluative attitudes—whether actual or idealized—then it is not a moral truth.

Following this taxonomy, if moral realism is true, then it is not just the case that our moral reasons are external in the traditional sense of being objectgiven, or intrinsic to the situations or objects that our reasons are about or for—after all, these very claims can be true in virtue of facts about moral agents qua valuing beings. And it is not just the case that moral properties are this way—after all, versions of anti-realism may claim that torture is ¹⁸ Street (2016): 295.

¹⁹ Street (2016: n. 9).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Moral Realism and Philosophical Angst

129

wrong because it causes unjustified pain, not because anyone thinks it is wrong. The metaphysical grounds of such truths are at issue. On realism of the “ardent” sort, we may have moral reasons, and things may have moral properties, in virtue of a system of moral truths that is itself independent of persons and their attitudes. Various metaethical parties offering positive accounts of morality may agree to some version of the claim that, for example, the suffering and oppression of the weak at the hands of the strong generates an object-given reason for us to care and do something, and that these reasons are reasons to stop something that is itself morally bad whether or not anyone thinks so. But there is an additional kind of independence involved in the true moral verdict, realistically construed: morality itself, the system of moral truths itself—or as Eklund says, “reality” itself—declares in favor of the weak against the strong, ruling the justification of their oppression a normative impossibility, with no reference to human moral attitudes in the grounds of this impossibility. When we think rightly, we concur with this verdict that did not come from us, either literally in our needing to issue it, or metaphysically in our independent or collective attitudes being needed to ground it. Is angst on this basis intelligible? One way of beginning to understand it is to appreciate that realistic independence is qualitatively rather than quantitatively different from anti-realistic independence. There is, the pro-realist should concede, a kind of anti-realistic independence that is a matter of degree or interpersonal distance, which I will call the quantitative sense of independence. Consider, for example, a view on which moral claims are true in virtue of their being the subject of hypothetical agreement among idealized agents. The pro-realist should not deny that there is a recognizable kind of independence involved in such an account. After all, this kind of view explicitly and literally distances moral truths from the actual responses of individuals. But notice that the view does not make moral truth at all independent of individual responses in what I call the qualitative sense. After all, on this kind of anti-realism, the responses of idealized agents are partly constitutive of the moral truth, and so the truth is still, at bottom, normatively shackled to responses. The same goes for other response-dependent theories. For instance, any view on which moral claims are true in virtue of one’s robust, consistent attitudes over time, would be a view on which there is independence, but it is independence of the anti-realist kind that is explicable in terms of degrees of distance from the actual responses of individuals. Anti-realist moral truth is always grounded in some way that leaves it captive to actual, individual responses. Realistic independence is qualitatively different. It is not just one more step on the ladder up and away from the actual attitudes of individuals. From the pro-realist point of view, for the system of moral truths to be fully

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

130

Joshua Blanchard

independent is for it to be independent of our attitudes in any sense— wholly free in its grounding from actual, individual responses. This is what I meant by the “radical independence” of moral truth—an independence that differs qualitatively from anti-realistic independence. After all, even the rational agreements among idealized agents would be themselves subject to the strictures and demands of a realistic morality. From the realist point of view, rational agreement among idealized selves might always produce true judgments, but such consensus should be understood as the non-trivial success of tracking, rather than constituting or constructing, the independent truth. I have made some effort to characterize the difference between realistic and anti-realistic independence with an eye toward making intelligible why someone might be angsty about only having the latter, but I have not yet attempted to explain directly what would be so good (and irreplaceable) about realistic independence. First, a caveat: someone who finds moral realism of the sort described to be itself unintelligible will struggle to see how there can be anything other than degreed independence from individual responses. In a way, such a person is not the target of my argument, any more than the person who claims not to understand the concept of God is a target of arguments that it is good if God exists. On the other hand, I do want to show that, given that someone thinks that there is such a possibility as moral realism, it is at the very least rationally intelligible that they would not find anti-realist independence a fitting surrogate for realist independence. Fitting surrogates must be relevantly similar in kind to their originals, and I hope this discussion makes it plausible that anti-realist independence is not such a surrogate for realistic independence. For the pro-realist, the radically independent nature of moral justification is good, but not merely good; it plays an existentially significant role in the meaning and intelligibility of life. This is because it militates against a kind of normative lonesomeness or moral bootstrapping, in which—in one way or another, at bottom—agents and their attitudes are responsible (either causally or constitutively) for providing our own moral vindication.²⁰ Again, I allow that pro-realist angst might be intelligible only from within the perspective of someone who grasps what it would be like to see morality in a realist way, in which context radically independent moral truth and justification is available. Given a backdrop of such radical independence, there seems to be little difference, vis-à-vis moral bootstrapping, between the different degrees of quantitative independence available on anti-realism.

²⁰ Cf. Wright (1995: 226), who makes a somewhat related point (though one whose import he ultimately rejects) about external sanction.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Moral Realism and Philosophical Angst

131

Consider an elaboration of the point via an analogy to circular reasoning.²¹ Relative to a realist framework, increasing the degrees of quantitative independence is like increasing the length or complexity of a bit of circular reasoning for an existentially important conclusion. Imagine that you take yourself to have an “independent” (now, in the sense of non-circular) argument for an existentially important claim (say, that human beings are worthy of respect). You then encounter a group of theorists (these are the analogues to anti-realists) who give a variety of circular arguments for the important claim. If circular reasoning is all these theorists think there could ever be, then, perhaps, length and complexity is the most you could want or hope for in terms of independence. “What more could you possibly want?” such theorists may ask, incredulously. But once one sees—or takes oneself to see—an independent, non-circular route to the existentially important claim, it makes sense that the circular routes, including the longest and most complex among them, will pale in comparison. In the relevant respects having to do with existential concern, antirealistically interpreted moral justification is to realistic justification as circular reasoning to an important claim is to non-circular reasoning for that claim. Crucially, the analogy is not supposed to establish that antirealistic moral justification is somehow formally circular—it isn’t. Rather, the claim is that, contrasted with moral justification realistically construed, its grounds are similarly self-referential, in a sense that includes but is broader than formal circularity. More can be said to defend the intelligibility of pro-realist angst. Although the considerations that follow do not depend on the analogy to circular reasoning, they are similar in spirit. Perhaps surprisingly, the perspective on offer here is sympathetic to (but not quite the same as) an argument due to the quasi-realist Simon Blackburn, who is hardly a friend of ardent moral realism. Blackburn argues that some anti-realist views face a problem of moral relativism. For example, he offers the following objection to the constructivist metaethic associated with Christine Korsgaard, according to which each person is rationally bound to self-legislate moral norms in virtue of an inescapable practical identity. Korsgaard’s view, he says, fails to preclude “pluralities of self-legislating persons whose identities are happily bound up in various constraints they set themselves under, but who unfortunately find these constraints in entirely different places.”²² How does this ²¹ Thanks to Ram Neta for suggesting that I think about such an analogy, even though he has no memory of doing so. ²² Blackburn (1999: 219). It must be noted that Blackburn also criticizes several theses associated with realism on the same grounds: in particular, McDowell’s sense theory and neo-Aristotelian accounts that ground morality in a conception of human flourishing (218–20).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

132

Joshua Blanchard

relate to pro-realism? The pro-realist may adapt Blackburn’s worry into the present discussion by saying that what would be bad or regrettable about Korsgaardian constructivism—even if it turned out to be true—is precisely that, between any two different persons or communities of persons who locate normative constraints “in entirely different places,” there is no radically independent source of authority to adjudicate the dispute, to vindicate either one side or none. This is reminiscent of the concern about moral bootstrapping that I expressed with the analogy to circular reasoning: if everyone can appeal to “constraints they set themselves under,” then they are ultimately justified self-referentially, by concerns that others need not share. To illustrate the point by another route, consider yet another challenge by Blackburn, this time aimed squarely at the moral realist. Blackburn demands that the moral realist explain what more it is important to “say” about moral evils beyond the moral condemnation to which anti-realists are equally entitled. “What is wrong with [the Taliban in its relation to women] is that the men oppress the women, impoverish their lives, and keep them in a state of ignorance and inactivity. Why should we feel any urge to say more than that? Isn’t it bad enough?”²³ The pro-realist should reply: the urge is not that there be something else that we might say, but that there be more than just “we” saying it. The traditional realist answer was that the Taliban is not just wrong, but objectively and robustly so. The angsty realist does not deny that these are appealing aspects of the realist picture. But the really important point is that it is not the saying of such a thing that matters. It is rather that there is an independent verdict issued, as it were, by reality itself. What is “said” can remain the same. Earlier I suggested that true moral statements “concur” with an independent verdict. The point is that true moral assertions are repetitions, restatements of a reality already partly constituted by the truths that the assertions express, much like two and two make four and no sphere has corners are, plausibly, reiterations of radically response-independent realities. Finally, it is worth considering a somewhat peculiar objection. The objection is worth considering not because it is necessarily powerful in itself, but because it helps to illustrate one further (and final) way of bolstering the intelligibility of pro-realist angst. Some readers might suspect that there is a kind of crypto-theism lurking in the background of the perspective here advanced, and they might want to launch a debunking argument against pro-realism on that basis. Perhaps pro-realism is just a watered-down version of angst about the “death of God”—of God who provided a kind of robust, external reference point for morality and meaning.

²³ Blackburn (1999: 223).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Moral Realism and Philosophical Angst

133

In a way, I think that the suspicion is right, but it actually provides confirmation of the overall framework defended here. Think of the familiar existential desires associated with theism: to “have God on our side,” or not to be “alone in the universe,” or to achieve and to see “cosmic justice.”²⁴ These desires plausibly derive from fundamental religious attitudes for which realist value is also fitting. (Incidentally, they also provide additional support for David Killoren’s recent argument that robust moral realism is a kind of “religion.”²⁵) Even if this connection may in turn provide grounds for a debunking argument against a certain style of religiously infused moral theorizing, that is a matter of what is true, not what would be better—and irreplaceable.²⁶ (It would, of course, be disappointing for pro-realists if the very grounds of their angst could be employed as defeaters for realism itself!) Notice, too, that the sorts of replies that the angst-free may give to prorealists are manifestly unsatisfying in their analogous religious form. To the pro-theist who desperately wants not to be “alone in the universe,” it is hardly comforting that some naturalistic construal of God—say, that God is identical to humanity’s highest ideals—also gives us a kind of company in the universe, in the form of our ideals being with us. If anything, the linkage to the religious case reinforces the intelligibility of angsty dissatisfaction with the falsity of realism. None of this is to say that moral realism is true; the point is that it is intelligible to regard its truth as preferable to its falsity, and that we can come to appreciate the reasons why someone might so prefer. Put another way, were we—per impossible—deciding in advance what kind of nature the moral domain is to have, we would have some good grounds to select from ardently realist options. There is reason to want moral justification and vindication to be wholly untethered from our responses.

²⁴ The terms in quotation marks are pulled from the ether, not any particular source. ²⁵ Killoren (2016). ²⁶ Nietzsche is probably the canonical debunker of this sort. Anscombe (1958) is widely regarded as offering a debunking argument along similar lines. According to a common reading, Anscombe argues that, because modern moral philosophy is inextricably bound up with a Christian, legalistic conception of the world, and modern thought is non-theistic, we moderns ought to abandon modern moral philosophy for a tradition continuous with pre-Christian conceptions, especially ancient virtue ethics and an attendant moral psychology. However, because Anscombe was a devout Catholic who philosophized as such and in fact doubted the prospects of virtue ethics, this almost certainly is not the correct reading. Here I agree with Driver (2014), who reads Anscombe as offering “a modus tollens argument intended to establish the superiority of a religious based ethics.” See Mavrodes (1986) for a more explicit modus tollens in favor of a theistic ethic on the grounds of the necessity of theism for morality, in response to Mackie’s “queerness” argument.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

134

Joshua Blanchard

6.2.3. The Inexpressible Badness of Quasi-Realism Anti-realists attempt to accommodate our moral discourse and practice without the characteristic metaphysical commitments of realists, that is, without commitments to a realistic construal of moral facts, properties, truths, and the like. The radical independence of moral truth, accommodated only by realism, can therefore play a role in establishing an evaluative difference between realism and anti-realism partly because it amounts to a kind of independence that the standard-issue anti-realist explicitly denies. Things are not so easy with the quasi-realist, and this section seeks to briefly explain (or to kvetch about) why. Like the standard-issue anti-realist, the quasi-realist attempts to accommodate our moral discourse and practice without the characteristic metaphysical commitments of realism. This very fact suggests that whatever goes for pro-realism vis-à-vis its metaphysical differences with standard-issue anti-realism should in turn apply to its differences with quasi-realism. But there is a problem: the quasi-realist is an accommodationist in the extreme, seeking to, as a common boast goes, say anything that the realist can say. Anything? Yes—including statements that emit the strong aroma of realism, for example, “It is a responseindependent, objective fact that torturing for its own sake is wrong.” If this project succeeds, then realists are precluded from identifying anything in the first-, second, or nth-order moral domain that distinguishes realism from quasi-realism. And if we cannot distinguish realism from quasi-realism descriptively, it seems that we cannot say what distinguishes it evaluatively. If there is no way to say how realism would be different from quasi-realism, then there is certainly no way to say why it would be better.²⁷ Due to quasirealism’s reliance on the minimalist notion of truth, according to which predicating truth of a proposition adds nothing substantial to the assertion of the proposition itself, the problem has become known as “the problem of creeping minimalism.”²⁸ The difficulty, in brief, is that if the anti-realist adopts a minimalist conception of truth that avoids identifying truth with what Crispin Wright calls “a property of intrinsic metaphysical gravitas,”²⁹ ²⁷ Given that quasi-realism adds nothing metaphysically to anti-realism, and that realism is certainly metaphysically different from anti-realism, it seems that there must be some difference—we just cannot say what it is. ²⁸ See Dreier (2004) for the classic contemporary statement of the problem. Dreier laments that “It’s not as if one side had better be able to come up with something clever to say about how to distinguish realism from [quasi-realism] or else the other side wins. It’s rather that those of us who feel confident that there is some difference between the two meta-ethical camps should be concerned that we don’t know how to say what that difference is” (31). My argument in this section says that we have reason to be unconcerned that we don’t know how to say what the difference is. ²⁹ Wright (1995: 213).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Moral Realism and Philosophical Angst

135

then the sorts of truth-claims that otherwise distinguish the realist from the anti-realist can no longer do their work.³⁰ While there is not room here to delve too far into this well-trodden territory, I will raise a suspicion about the state of the dialectic between quasi-realism and realism. This will in turn shed light on what pro-realists should say about quasi-realism—or more to the point, what they should say about not being able to say much at all. The suspicious fact is that what makes quasi-realism and realism allegedly metaphysically indistinguishable is what one might take to be a purely semantic maneuver—namely, the adoption of expressivism at every level of ethical and metaethical discourse.³¹ Such a maneuver only works (for the purposes of accommodating everything that realism accommodates) if reality is always exhausted by what we are able to say. To illustrate, imagine two figures: a conventional moral realist and an anti-realist who is expressivist about only first-order moral discourse. (The early emotivists can, without too much interpretative harm, be thought of in this way.) Both the conventional realist and first-order expressivist may agree in asserting any number of first-order moral sentences, for example, “Torturing for its own sake is morally wrong.” However, they can also agree that the underlying metaphysic for each theory is different. Accordingly, they may disagree about various meta-moral, metaphysical sentences, for example, “The fact that torturing innocent uncles for fun is wrong is an objective, mindindependent fact in all possible worlds.” They can then point to the object of the disagreement—the nature of moral facts—and ask whether it would be better for such facts to be realistic or anti-realistic. Notice, too, that the first-order expressivist can, consistently with his own semantics, acknowledge that the kind of realistic independence highlighted in the previous section is not countenanced in his theory. He may even lament that fact. But suppose that the first-order expressivist has an unusual conversion: he comes to endorse expressivism at all levels of discourse about the moral: he becomes a pan-expressivist. What becomes of the previous disagreement over the sentence, “[T]he fact that torturing innocent uncles for fun is wrong is an objective, mind-independent fact in all possible worlds”? Merely by

³⁰ Many metaethicists take the problem of creeping minimalism to be a problem for quasi-realists rather than realists: once quasi-realists adopt the minimalist program, it is they who have no way of explaining why they are not full-fledged realists. If that is right, then there is little sense in a pro-realist being worried about the truth of quasi-realism, since quasi-realism is just disguised realism. I am pursuing the other horn of an implicit dilemma—that the problem of creeping minimalism is really a problem for realists, who have no way of explaining what they add to quasi-realism. If one thinks that it is quasirealists who chiefly face this burden, then so much the better for pro-realism. ³¹ Here I am departing somewhat from the literature in identifying what we might call a problem of creeping expressivism rather than creeping minimalism.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

136

Joshua Blanchard

undergoing a change of semantics (but in a way that somehow changes nothing in his metaphysics), the pan-expressivist now putatively agrees with the realist. Moreover, he can taunt the realist as follows: “You realists claim to be endorsing something above and beyond what we anti-realists endorse—but what?” This challenge—“but what?”—is unanswerable in the dialectical context, since there is no expressible “what” for which the pan-expressivist cannot give an expressivist reading. And so he may also taunt the pro-moral realist, “You can’t tell me what the difference is between our realism and quasi-realism, so why are you so angsty about it?” What has the pan-expressivist done, dialectically speaking? I submit that he has simply taken away the realist’s license to make distinctly realistic assertions—but only in the context of confronting pan-expressivism. Unfortunately, if the realist cannot make such assertions, then the pro-realist cannot go on to say, in light of some expressible difference, that it is better if realism is true, even if it is! Notice, crucially, that there is no good reason to think that the limits on what can be appropriately said are likewise limits on what the world is like. Even so, the realist is pushed into a kind of context-specific, involuntary mysticism. By “mysticism” I mean a position that says there are truths that we are unable to express.³² Such mysticism is narrow, because the pro-realist is coerced into it only when engaging a panexpressivist. It is involuntary, because there are no commitments or independent considerations on the side of the pro-realist that force or even so much as motivate it. For example, there need be no explicit thought, characteristic of independently motivated mysticism, that moral reality does outstrip our expressive capacities, though such a view may be very well-motivated in general.³³ For these reasons, in the dialectical context of contrasting realism with quasi-realism, the pro-realist is coerced into going mystical about both the difference between realism and quasi-realism, and therefore any evaluative judgment about that difference. To illustrate that this is an instance of a general problem and not unique to metaethics, consider a much more familiar source of angst: God. Worries about God are likely to induce philosophical angst, especially as religion often plays an existentially important a role in human life akin to moral value. Although debates between traditional theists and atheists are most familiar to philosophers and non-philosophers alike, there is also a dispute between supernaturalist religion (henceforth, supernaturalism) and naturalist ³² This is a much thinner notion of mysticism than one finds in e.g. the canonical discussion by James (1997: 281–3). ³³ For examples of mysticism that is independently motivated in this sense, see Thomas Hofweber (n.d.), “Are there Completely Ineffable Aspects of Reality?”; Fodor’s (1983:120ff.) discussion of “epistemic bounded[ness]”; and Chomsky (1988: 151–2).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Moral Realism and Philosophical Angst

137

religion (henceforth, naturalism). Naturalists deny supernaturalist metaphysics, but seek to provide a positive account of religious discourse and practice.³⁴ Provided that something natural can play the role of the supernaturalist God, the naturalist may give naturalistic readings of various kinds of religious propositions—for example, that God exists or that God loves us. But even if the naturalist succeeds at this project of accommodating first-order religious discourse and practice, it is manifest both that naturalists and supernaturalists differ at the level of metaphysics, and that this has significant evaluative upshot. As long as the naturalist remains a second-order descriptivist, the supernaturalist and naturalist can agree that their metaphysics at the first-order level are different. But suppose the naturalist adopts a religious semantics in which all first-, second-, and nth-order religious discourse can come out true even without the truth of traditional theism. This robs the supernaturalist of his dialectical license to offer any first-, second-, or nth-order religious sentences that would otherwise distinguish the two theorists’ metaphysics. That there is no God in the traditional sense is not just an evaluatively indifferent idea; it is something one might intelligibly want not to be true. But it is impossible to appropriately explain what is so bad about this prospect to a naturalist who can agree with any sentences uttered by the supernaturalist. Yet, who thinks that there really is no difference—either descriptively or evaluatively—between the truth of supernaturalism and naturalism? Those of us who are either uncommitted to either view, or committed naturalists with a semantics that can express supernaturalism, or committed supernaturalists, can express perfectly well what is lacking on naturalism: God. Yet, we are robbed of the dialectical license to offer such a sentence in the context of arguing with a naturalist who adopts some kind of pan-naturalist semantic program. While the situation of the pro-realist in the presence of the quasi-realist is one of philosophical frustration, the reason for the frustration is clear enough. If necessary, the pro-realist can find innocent companions (like the angsty pro-theist) in coerced mysticism, companions who fall victim to the pan-expressivist gambit or something similar. In such cases, it seems reasonable for the pro-realist’s counterpart to remain angsty, despite being robbed of the expressive power to say why. On this basis, the pro-realist should remain confident that it is better if realism rather than quasi-realism is true, even if the realist lacks the dialectical license to say it to the quasirealist’s face.

³⁴ See especially Johnston (2009) and Dworkin (2013).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

138

Joshua Blanchard 6 . 3 . C ON C L U S I O N

My overarching goal has been to commend the research program of angsty metaethics. To this end, I defended the claim that it is better that moral realism rather than any of its rivals is true. There is considerably more work to be done. To mount a truly comprehensive defense of pro-realism, it is necessary to address possible reasons for thinking that the truth of realism would be bad, as well as reasons for preferring that particular versions of anti-realism are true. Here I have only considered realist and non-realist views in broad strokes. As noted above, one might also question whether “ardent” or “robust” realism really is the best form of realism for quieting metaethical angst. It is clear that, as far as metaethical angst is concerned, we have much to worry about.³⁵

References Adams, Robert (1979). “Theories of Actuality,” Nous 8(3): 211–31. Anscombe, Elizabeth (1958). “Modern Moral Philosophy,” Philosophy 33(124): 1–19. Blackburn, Simon (1999). “Is Objective Moral Justification Possible on a QuasiRealist Foundation?” Inquiry 42(2): 213–27. Chomsky, Noam (1988). Language and Problems of Knowledge: The Managua Lectures. Cambridge, MA: MIT Press. Dreier, James (2004). “Meta-Ethics and the Problem of Creeping Minimalism,” Philosophical Perspectives 18: 23–44. Driver, Julia (2014). “Gertrude Elizabeth Margaret Anscombe” in The Stanford Encyclopedia of Philosophy, ed. Edward N. Zalta, . Driver, Julia (2017). “Contingency and Constructivism” in Simon Kirchin (ed.), Reading Parfit: On What Matters. New York: Routledge. Dworkin, Ronald (2013). Religion without God. Cambridge, MA: Harvard University Press. Eklund, Matti (2017). Choosing Normative Concepts. Oxford: Oxford University Press. Enoch, David (2011). Robust Moral Realism. Oxford: Oxford University Press. ³⁵ For helpful discussion and feedback on this chapter I’d like to thank Spencer Case, Terence Cuneo, Raff Donelson, Luke Elson, David Faraci, Krasi Filcheva, Eli Hirsch, Zoë King, Barry Maguire, Ram Neta, Derek Parfit, Ryan Preston-Roedder, Geoffrey Sayre-McCord, Russ Shafer-Landau, Keshav Singh, Larisa Svirsky, Silvan Wittwer, and Susan Wolf. I am also grateful to audiences at the 2017 Rocky Mountain Ethics Congress and the 2018 Great Lakes Philosophy Conference, as well as two anonymous reviewers for Oxford University Press.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Moral Realism and Philosophical Angst

139

Fodor, Jerry (1983). The Modularity of Mind: An Essay on Faculty Psychology. Cambridge, MA: MIT Press. Fodor, Jerry (1989). “Making Mind Matter More,” Philosophical Topics 27(1): 59–79. Hirsch, Eli (2009). “Diabolical Mysticism, Death, and Skepticism,” Philosophic Exchange 39: 2–27. Hofweber, Thomas (n.d.) “Are There Completely Ineffable Aspects of Reality?” James, William (1997). The Varieties of Religious Experience. New York: Touchstone. Johnston, Mark (2009). Saving God. Princeton: Princeton University Press. Joyce, Richard (2002). The Myth of Morality. Cambridge: Cambridge University Press. Kahane, Guy (2012). “The Value Question in Metaphysics.” Philosophy and Phenomenological Research 85(1): 27–55. Kahane, Guy (2016). “If Nothing Matters,” Nous 50(2): 1–27. Killoren, David (2016). “Robust Moral Realism: An Excellent Religion,” International Journal for Philosophy of Religion 79: 223–37. Kraay, Klaas (ed.) (2018). Does God Matter? Essays on the Axiological Consequences of Theism. New York: Routledge. Mackie, J. L. (1977). Ethics: Inventing Right and Wrong. Harmondsworth: Penguin. Mavrodes, George (1986). “Religion and the Queerness of Morality” in William Wainwright and Robert Audi (eds), Rationality, Religious Belief, and Moral Commitment. Ithaca, NY: Cornell University Press, 213–26. Nagel, Thomas (2007). “The Value of Inviolability” in Paul Bloomfield (ed.), Morality and Self-Interest. Oxford: Oxford University Press, 102–14. Parfit, Derek (2011). On What Matters, vol. 2. Oxford: Oxford University Press. Sartre, Jean-Paul (1984). Being and Nothingness. New York: Simon & Schuster. Street, Sharon (2016). “Objectivity and Truth: You’d Better Rethink it” in Russ Shafer-Landau (ed.), Oxford Studies in Metaethics, vol. 11. Oxford: Oxford University Press. Street, Sharon (2017). “Nothing ‘Really’ Matters, But That’s Not What Matters” in Peter Singer (ed.), Does Anything Really Matter? Essays on Parfit on Objectivity. Oxford: Oxford University Press. Temkin, Larry (2017). “Has Parfit’s Life Been Wasted? Some Reflections on Part Six of On What Matters.” In Peter Singer (ed.), Does Anything Really Matter? Essays on Parfit on Objectivity. Oxford: Oxford University Press. Thucydides (1996). History of the Peloponnesian War, ed. Robert B. Strassler, tr. Richard Crawley. New York: Touchstone. Wolf, Susan (1982). “Moral Saints,” Journal of Philosophy 79(8): 419–39. Wright, Crispin (1995). “Truth in Ethics,” Ratio 8(3): 209–26. Zhao, Michael (2018). “Meaning, Moral Realism, and the Importance of Morality,” Philosophical Studies, https://doi.org/10.1007/s11098-018-1198-0.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

7 Getting a Moral Thing into a Thought Metasemantics for Non-Naturalists Preston Werner

Non-naturalism is the view that normative properties are responseindependent, irreducible to natural properties, and causally inefficacious.¹ Sharon Street and Richard Joyce, by arguing that a non-naturalism entails skepticism, have ushered in a virtually insurmountable literature on nonnaturalist moral epistemology and evolutionary debunking arguments.² These concerns are not new.³ Extensions of the so-called “Benacerraf problem” for mathematical entities to the moral realm have also attempted to undermine non-naturalism.⁴ And there remains no consensus on how— or whether—these epistemological challenges can be met. A less discussed question for non-naturalism concerns the metasemantic connection between normative beliefs and the normative facts. This underexplored question concerns the metasemantics of normative terms. Ideally, the non-naturalist could remain ecumenical. As has been noted, however, this is not possible.⁵ So the challenge is for the non-naturalist to find some independently motivated metasemantic view that fits well with nonnaturalism. Let’s call this challenge—which is developed in more detail below—the metasemantic challenge. The widely discussed epistemological challenges and the metasemantic challenge are at least superficially related. They both raise questions about the status of our normative beliefs if non-naturalism is true. Epistemological challengers argue that non-naturalism threatens to make all normative beliefs unjustified. Metasemantic challengers argue that non-naturalism ¹ Ridge (2014). ² Joyce (2001, 2006), Street (2006). ³ See e.g. Ruse and Wilson (1986). ⁴ See e.g. Benacerraf (1973), Harman (1977), Liggins (2010), Enoch (2011: ch. 7), Clarke-Doane (2017). ⁵ Schroeter and Schroeter (2014), Suikkanen (2017).

Preston Werner, Getting a Moral Thing into a Thought: Metasemantics for Non-Naturalists In: Oxford Studies in Metaethics Volume 15. Edited by: Russ Shafer-Landau, Oxford University Press (2020). © Preston Werner. DOI: 10.1093/oso/9780198859512.003.0007

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Getting a Moral Thing into a Thought

141

makes it impossible for us to have beliefs about normative properties and facts at all, since it is impossible for our beliefs to fix onto non-natural properties as their referent. This chapter focuses on the second challenge. My suggestion is that nonnaturalists endorse an epistemic account of reference determination of the sort recently defended by Imogen Dickie, with some modifications.⁶ An important implication of this account is that, if correct, a fully fleshed out moral epistemology will simultaneously rebut metasemantic objections to non-naturalism. Thus, the two challenges in effect amount to one. In section 7.1, I recap two influential epistemological challenges to nonnaturalism, as well as the less discussed metasemantic challenge. In section 7.2, I first review a few standard metasemantic theories, illustrating why they spell trouble for the non-naturalist (7.2.1–7.2.2). I then discuss Laura Schroeter and François Schroeter’s “normative connectedness model” to see whether it does any better; I conclude that it does not, and for reasons that will generalize to any internalist metasemantics for non-naturalism (7.2.3–7.2.4). In section 7.3, I assess whether the doctrine of reference magnetism may supplement any of the previous accounts, especially given that some recent work on metasemantics for moral realism has attempted to make use of them for responding to a variety of objections. In section 7.4, I develop my positive epistemic theory, which is well-suited to avoid all of the problems with the traditional metasemantic theories when extended to non-natural normative properties. Section 7.5 draws out three ways in which the epistemic theory accords nicely with widely discussed features of non-naturalism.

7 .1 . E P I S T E M O L O G I C A L AN D M E T AS E M AN T I C OBJECTIONS TO NON-NATURALISM The so-called Benacerraf problem goes back at least to Paul Benacerraf ’s (1973) “Mathematical Truth.”⁷ While Benacerraf was concerned with the case of mathematical knowledge, a similar sort of problem arises for nonnaturalist moral realists, according to which moral properties are not causally efficacious.⁸ This parallel was first noticed by Gilbert Harman (1977). As in the mathematical case, even while causal conditions on knowledge ⁶ Dickie (2015, 2016). ⁷ Field (1989), Cheyne (2001), and Liggins (2010) for discussions of this point and developments of Benacerraf ’s problem. ⁸ E.g. Heathwood (2015: 3), McGrath (2014: 186), and Scanlon (2014). Oddie (2005) is an exception to this general rule.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

142

Preston Werner

have fallen out of favor, there is still a widespread sense that explaining our epistemic access—in terms of responsiveness to the moral facts—is a serious concern for the non-naturalist moral realist. Another more specific sort of undercutting defeater for moral knowledge comes in the form of evolutionary debunking arguments (EDAs). At the most abstract level, the idea of an EDA is to provide a genealogy of our moral beliefs or belief-forming mechanisms in terms of the fitnessenhancing evolutionary nature of those beliefs or belief-forming mechanisms. Such an explanation, the debunker argues, will make no appeal to any mind-independent moral facts. And the lack of explanation here strongly suggests skepticism which serves to undercut any justification we initially had about stance-independent moral facts. I won’t say more about how the details of these arguments are best worked out. Instead, let’s turn now to the metasemantic challenge. David Enoch provides a concise statement of the problem: [According to non-naturalists] the word “good” (to pick one example) refers—at least in some of its occurrences—to the property goodness. And this property is, on [non-naturalist views], causally inert and response-independent. How is it, then, that our word manages to latch onto that property—rather than some other property, or perhaps no property at all?⁹

We’ll consider more specific metasemantic theories in the next section. For now, just notice the intuitive thought behind the worry. I have a cat named Zooey. As a result, I have a bunch of beliefs about Zooey: that he’s lazy, has a cute nose, is mean to other cats, etc. Somehow these beliefs all latch onto Zooey, rather than my partner, other cats, or anything else.¹⁰ It’s the task of a metasemantic theory to tell us exactly how this works. But whatever the details are here, it’s intuitive that my Zooey-beliefs bear all sorts of relations to Zooey that don’t appear to hold in the normative case, if non-naturalism is true. Zooey is frequently part of my perceptual experiences, there is a causal chain between my Zooey-beliefs and Zooey, my linguistic community’s patterns of “Zooey”-tokenings converge on Zooey. But it is unclear that any of these sorts of relations hold (or even could hold) between our normative beliefs and any non-natural properties. So there is a prima facie metasemantic challenge for non-naturalists. And this prima-facie challenge is made more formidable by noting that many mainstream metasemantic theories make reference to non-natural properties (as they’re traditionally construed, at least) not just challenging, but impossible. ⁹ Enoch (2011: 177). ¹⁰ I use italics, here and throughout, to refer to properties and objects in the world, as opposed to words or concepts.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Getting a Moral Thing into a Thought

143

One clarification is in order. Metasemantic theories are ambiguous between two possible theoretical inquiries. First, a metasemantic theory could be attempting to answer the question of how words latch onto properties and objects. Second, a metasemantic theory could be attempting to answer the question of how concepts—that is, psychological representations—latch onto properties and objects. The metasemantic challenge to non-naturalism has been framed both ways, without clear differentiation.¹¹ I will be speaking in terms of mental content and psychological concepts in what follows. All of what I say could in principle be extended to linguistic content as well.

7 . 2 . N O N - N A T U R A L I S M A N D T R A D IT I O N A L METASEMANTICS Three popular families of metasemantic views are (a) causal/teleological theories, (b) conceptual role semantics, and (c) neo-descriptivism. I briefly explain why these traditional views are not promising for the non-naturalist (7.2.1–7.2.2), before discussing Schroeter and Schroeter’s “connectedness” model in more detail (7.2.3–7.2.4).

7.2.1. Causal and Teleological Theories and Non-Naturalism Causal theories of content all share a commitment to the thesis that concepts get their content in virtue of some causal relation(s) that hold between tokenings of the concept and some property, object, or individual in the world. More formulaically: SCT: For any concept C and any property F, C refers to F iff tokenings of C tend to be caused by F.

SCT is open to interpretation between a range of different causal relationships. However, none of these details need to be considered for the present purpose. Any causal theory of content will be a non-starter for the nonnaturalist, since the non-naturalist denies that normative properties are causally efficacious. In fact, the causal theory of content has been exploited by naturalist moral realists to ensure that the moral terms pick out natural (causally efficacious) properties.¹² We can set aside causal theories of all ¹¹ See e.g. Wedgwood (2007), Enoch (2011) (and the quote above), and Schroeter and Schroeter (forthcoming). ¹² Boyd (1988, 2003), Sturgeon (1998).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

144

Preston Werner

types—they will not provide a non-naturalist-friendly metasemantics without radical alteration in the commitments of non-naturalists. Close cousins of causal theories of content, teleological theories, are in principle better placed to avoid the above incompatibility with nonnaturalism. On teleological theories, “what a representation represents depends on the functions of the systems that produce or use the representation.”¹³ “Function” here is understood in terms of what the particular state or process was selected for. So, for example, a given concept C refers to poutine just in case (and because) it was selected for the advancement of my broader psychological systems’ poutine-related goals. Unlike with causal theories, functions need not refer to causally efficacious properties. For example, evolution selects for creatures who are able to grasp mathematical concepts such as addition and division. These concepts play a certain function in allowing us to reason in a variety of ways that increase our survivability. So far so good for the non-naturalist. But despite the slight improvement on causal theories, non-naturalists—or at least the vast majority of non-naturalists—should still reject a teleological metasemantics for normative concepts. To see why, consider how the teleologist will determine the content of a given normative concept. In order to answer this question, the teleologist will look at what best explains the function of normative thoughts. Neil Sinclair (2012) has perhaps done the most to flesh out what a teleosemantics might look like for normative thoughts.¹⁴ Sinclair argues that moral concepts have an evolutionary function, which is to coordinate interpersonal behavior.¹⁵ On this sort of story, morality evolves in order to solve certain pervasive evolutionary bargaining problems. It should already be clear that this story about the function of our moral concepts, and any similar such story, will be incompatible with nonnaturalism. Supposing Sinclair’s story is right, the function of moral concepts is to encourage certain fitness-enhancing behavior. A behavior’s being (or not being) fitness enhancing is surely a natural property. For this reason, it is unsurprising that no non-naturalists have appealed to teleosemantics in their metasemantic theorizing.¹⁶

¹³ Neander (2012: introduction). ¹⁴ Of course teleologists have discussed weaker notions of normativity for a very long time (see e.g. Millikan 1984, and esp. Millikan 1995). I hope the difference between a normative thought related to an organism’s survival and a robustly moral thought is clear enough. Certain kinds of naturalists won’t see a difference in kind between these two notions of normativity, but rather a difference of degree. But we can set this aside, since the present chapter is only concerned with non-naturalists. ¹⁵ But see also Millikan 1995: section 5) for a relevant precursor. ¹⁶ Notice here that this forecloses a strategy analogous to third-factor responses to epistemological objections to non-naturalism. Showing a co-extension between the

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Getting a Moral Thing into a Thought

145

7.2.2. Conceptual Role Semantics Conceptual role semantics for normative terms has received the most discussion in the metaethical literature, largely because of Ralph Wedgwood’s influential book The Nature of Normativity, as well as his subsequent work.¹⁷ The basic idea is explained concisely by Laura Schroeter and Francois Schroeter: Like expressivists, Wedgwood thinks the open question argument suggests that the central element in the meaning of moral terms is their action-guiding role. Indeed, according to Wedgwood, grasping the action-guiding role of those terms is all there is to understanding their meaning. But Wedgwood embraces cognitivism: he thinks that the conceptual role of moral terms provides the resources to single out genuine properties as their semantic value . . . the action-guiding role of moral terms, he suggests, suffices to determine which property they pick out.¹⁸

Wedgwood takes the following to be a conceptual truth: (1) Necessarily, if one is rational, then, if one judges “I ought to ɸ,” one also intends to ɸ.

As a conceptual truth, anyone with a full grasp of OUGHT is in a position to know (1). More importantly, though, anyone with a full grasp of OUGHT will be disposed to make the relevant transitions from oughtjudgments to intentions, whether implicitly or explicitly. So this ensures, at the very least, that OUGHT is connected up with an action-guiding property. But this is not yet enough to connect OUGHT up with a nonnatural property, much less the correct one. It only shows that there is a certain relationship between “ought” judgments and intentions to act. Somehow it must be the case that this action-guiding concept connects up with a non-natural property, if Wedgwood’s story is to vindicate a nonnaturalist metaethics. A conceptual role semantics for non-natural properties must connect our concepts not just to the prescriptive role of the properties, but also their representational role. We can see this by considering an agent, Anya, who has some concept O*. O* meets the conditions of (1). Whenever Anya judges that she O* to ɸ, and she is rational, she forms the intention to ɸ. But furthermore, suppose that the evidence that Anya relies on to form her O* judgments are facts about causing as much pain and suffering to bunnies properties our concepts evolved to track and the non-natural properties does not secure reference onto the non-natural. ¹⁷ Wedgwood (2007). ¹⁸ Schroeter and Schroeter (2003: 191).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

146

Preston Werner

as possible. Anya may even take the following to be a conceptual truth—or at least something of a platitude: (2) Necessarily, if an action ɸ would result in more pain to bunnies than any alternative action, then I O* to ɸ.

We can call (2) the “Input” condition for O*,¹⁹ and (1) the “output” condition. Anya’s O* is perfectly coherent, but surely we don’t want to say that O* picks out the non-natural property of oughtness, unless we’re gravely mistaken about what our obligations are. But it’s not just that Anya’s concept is normatively problematic. Rather, it’s that (1) radically underdetermines the relevant property picked out by our OUGHT concept. As Neil Sinclair (2018) points out in a closely related context, “it is implausible to suppose that a concept refers to a worldly property when competence with that concept is compatible with no disposition whatsoever to be sensitive to any worldly property.”²⁰ The conceptual role semanticist must provide—or at least gesture toward—what I’m calling an input condition on our normative concepts. An input condition is one’s disposition to apply the concept in appropriate circumstances. (For example, the input condition for MODUS PONENS will be premises of the form {P, P!Q}.) And this input condition must be close enough to uniform across agents such that the OUGHT concept is shared amongst all competent users of the concept (lest we collapse into relativism).²¹

7.2.3. Neo-Descriptivism, Connectedness, and Non-Naturalism Descriptivist theories all share the core idea that a concept gets its content by its association with a set of application conditions, which are shared by any competent user of the concept. Call internalist any view such that the extension of a given concept for a given agent is determined wholly by (actual or counterfactual) mental states of the agent. The latter feature of descriptivism—that competent users must grasp the application conditions associated with a concept—renders descriptivist theories paradigmatically internalist. Notice that internalism should be initially attractive to the nonnaturalist. This is because it allows the normativity of our normative concepts to be built into their meanings. If, as non-naturalists think, ¹⁹ In the Conceptual Role Semantics for logical operators, what I’m calling the input condition is commonly called the introduction rule. ²⁰ Sinclair (2018: 113). ²¹ For compelling arguments distinct from (but related to) the argument of this paragraph, see Lenman (2010).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Getting a Moral Thing into a Thought

147

normative concepts refer to genuinely normative properties or nothing, then we don’t have to worry about our normative concepts fixing onto properties which are intuitively not action-guiding. As is widely known, traditional descriptivist theories have serious problems. So let’s just consider the neo-descriptivist theories that evolved to avoid these problems. The key strategy here is to make use of idealization—a subject’s concept refers to the property she would pick out once suitably idealized with respect to some relevant set of facts. Neodescriptivist theories like this have made some appearances in the metaethical literature, most notably perhaps in Jackson and Pettit’s (1995) naturalist-friendly “moral functionalism.” Perhaps the most sophisticated version of this kind of view is Schroeter and Schroeter’s “normative connectedness model.”²² Because this latter model is the most plausible neo-descriptivist theory for non-naturalists, and because the objection I give below will apply to any neo-descriptivist view that makes use of idealization, I will speak in terms of the normative connectedness model for simplicity. The theory that Schroeter and Schroeter propose to match their connectedness model attempts to shore up the problems with both neo-descriptivist and teleological theories by moving to a fundamentally relational account. As they summarize it: The connectedness model relies on a tradition-based [metasemantic] theory: the fundamental units to which semantic values are assigned are not token elements of thought considered in isolation . . . but rather an entire representational tradition . . . On this approach, the metasemantic theory seeks to assign a univocal semantic value to the tradition as a whole, taking into account the understanding, environment, and history of the entire diachronic and interpersonal tradition. Token elements of thought then inherit their semantic values from the traditions to which they are bound.²³

A token element of thought inherits its semantic value from its whole historical and social use. This captures the idea that a concept’s meaning in one person’s mouth should have a lot in common with previous uses of the concept and provides a check on a more radical semantic internalism. How, then, does the tradition help to determine the referent of some particular concept? Here is what Schroeter and Schroeter say: [T]he idea that the semantic value of normative concepts is fixed by some original baptismal event is highly implausible . . . The correct principles . . . are simply an ²² Schroeter and Schroeter (2014). ²³ Schroeter and Schroeter (2014: 12). They use “determination theory” for what I’m calling a metasemantic theory.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

148

Preston Werner

idealization of a subject’s own reflective methods for refining her understanding of the precise subject matter of her words and thoughts . . . In effect, we can build our [metasemantic] theory from the first-person reflective epistemology of the topic in question.²⁴

The difference between the normative connectedness model and traditional neo-descriptivism is that the idealization process includes information about the tokenings of the concept that the subject will count as tokenings of the very same concept that the subject sees herself as using. (This is the sense in which the metasemantic theory takes the semantic tradition as prior or at least essential to the determination of the concept’s referent.) This provides a non-trivial restriction on an idealized subject’s assessment of the referent of a given concept. Her assessment cannot be chauvinistic in the sense that her conceptual tradition was so radically epistemically and reflectively mistaken that their use of the concept was inherently defective. The idealized subject is then given constraints on her interpretation of the concept that are not present in other neo-descriptivist theories.

7.2.4. Neo-Descriptivism, Connectedness, and a General Lesson Whatever the strengths of the normative connectedness model, and neodescriptivist models more generally, the idealization strategy will fall into a principled problem for non-naturalists, or at least the vast majority of them. To see why, first notice that non-naturalists accept: Metaphysical Autonomy. The normative facts are irreducible to the natural facts, in the sense that there is no conceptual entailment or complete metaphysical explanation from the natural facts to the normative facts.

Metaphysical Autonomy is a way of cashing out the irreducibility claim that is so central to non-naturalism.²⁵ It is arguably closely related to Moore’s open question argument²⁶ and Hume’s is–ought gap,²⁷ as well as to the supervenience objection to non-naturalism.²⁸ Importantly for our purposes, the Metaphysical Autonomy implies the following:

²⁴ Schroeter and Schroeter (2014: 13–14). ²⁵ Thanks to Aaron Elliott for helpful conversation on how to frame Metaphysical Autonomy. ²⁶ Moore (1903). ²⁷ Hume (1975). ²⁸ Blackburn (1984), McPherson (2012), Elliott (2014).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Getting a Moral Thing into a Thought

149

Epistemological Autonomy. An agent could be wholly procedurally rational and fully informed about all the natural facts and yet ignorant or mistaken about the normative.²⁹

Given the lack of an entailment relation between two domains, it should be clear that knowledge of one domain is not guaranteed to improve knowledge of the other. In practice, this may seem puzzling—of course gathering more information about the natural facts assists us in drawing new and improved normative conclusions! But notice that this sort of knowledge will always be inferred from some purely normative claim, such as that suffering is bad. This is really just a lesson of Hume’s is–ought argument reiterated. Given the autonomy theses, we can show that the normative connectedness model and neo-descriptivist theories more generally face a dilemma. We need a story about what facts get fed into the idealization process. Either this idealization process will include the normative information or it will not. Notice that this idealization process must meet two constraints. First, the information in question should remove any ignorance about the information relevant to determining the concept’s referent. This information is relevant because it is required to illustrate the ways that non-idealized agents’ concepts can refer to a determinant property even in light of ignorance, in virtue of the fact that their concepts would fix on these properties once ignorance is eliminated. This can ensure that a subject’s WATER concept refers to H2O, even if her non-idealized self is disposed to mistake XYZ for H2O. Her concept refers to H2O just in case she would retract her judgments that instances of XYZ fall under her WATER concept under idealization. Second, the information included in the idealization process cannot make reference to the referents of the concept in question, on pain of circularity. So, for example, idealizing for determining the semantic value of WATER should not include facts about water (qua water), but merely facts about the distribution of clear potable liquids that fill the lakes and rivers, etc. Without this, the idealization process is just smuggling in the fact that water is the semantic value of WATER, which is circular. So first consider an idealization process which doesn’t include normative information. On this view, we feed into our idealization an ideal base-level description of all of the natural facts. While such an ideal base-level description will plausibly ensure a referent for natural properties, such a story won’t work for normative properties considered as non-natural. When it comes to ²⁹ As Hille Paakkunainen pointed out to me, on some views of what it takes to grasp the normative concepts, Metaphysical Autonomy is compatible with the rejection of Epistemological Autonomy (Setiya 2012). I think these views have their own problems, but such a discussion would take us outside of the scope of this chapter, so I set them aside.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

150

Preston Werner

a property whose truth conditions are not grounded in an ideal base-level description of the physical world, an idealization with respect to these facts will not remove any ignorance present in the non-idealized agent, and so won’t fix the concept onto a determinant property. And Metaphysical Autonomy entails that an ideal base-level description of the physical world will not (alone) determine the truth of normative facts. As such, given that the idealization base does not include any normative bridge laws, even wholly rational agents with full information about the physical world are not guaranteed to fix on a unique non-natural fact. Theories that rely on idealization of this sort—including neo-descriptivism and the connectedness theory—will all fail for this reason. Schroeter and Schroeter’s theory of de jure sameness may be able to show that a certain set of token instances of a concept all refer to the same normative property (if they refer at all), but it will leave open the question of which property is the concept’s semantic value. This objection suggests a very natural reply. The notion of an ideal baselevel description need not be restricted to a description of only physical facts. Such an understanding of what the ideal base-level description of the world would look like is only motivated by a pre-existing commitment to physicalism (the claim that all facts are reducible to physical facts). Furthermore, there is precedent for understanding the ideal base-level description as more than just physical facts. For example, property dualists would want to add phenomenological facts to the ideal base-level description.³⁰ So for the property dualist, fixing the semantic value of the concept of PHENOMENAL-RED requires acquaintance with a certain phenomenological fact, and so acquaintance with such a fact will be part of the ideal base-level description. So, it may be argued, the non-naturalist should embrace the same sort of strategy. The strategy involves including what I’ll call “base-level normative” facts in the ideal base-level description of the world. Thus, idealized agents would have access to the information they need to fully determine the referents of their normative concepts. A worry about this approach can be seen from considering exactly how this set of “base-level normative facts” might look. An obvious place to look is at the normative facts that metaethicists have argued are fundamental. One influential position in this literature is that reasons are the fundamental normative kind, and thus all other normative truths are grounded in reasons facts.³¹ Others have argued that it is not reasons but some other set of normative facts that are fundamental, but nothing I’ll say

³⁰ See e.g. Chalmers (2012: ch. 3).

³¹ See e.g. Scanlon (2014).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Getting a Moral Thing into a Thought

151

in what follows depends on which position is right.³² I speak in terms of reasons for simplicity. Suppose the base-level normative facts are just the facts about reasons described using the agent’s concept of REASON. Since the semantic value of the concept REASON is one of the very things we need a metasemantic theory to fix, building up the base-level normative facts from claims that involve the concept of REASON would result in obvious circularity. What’s required for idealization to work for the non-naturalist is some way of understanding a base-level description of the normative facts that doesn’t invoke one of the very concepts it’s intended to illuminate. And it’s unclear how to do so. I don’t deny that some way of characterizing a base-level description of normative information could avoid this problem. Notice that there is a parallel here in the case of natural facts. We will need a metasemantic theory of the concepts that figure in the base-level description of the physical world as well. At this point, one could go externalist, or perhaps one could interdefine a cluster of fundamental base-level concepts and then Ramsify over those concepts. However this might work in the case of natural facts, it won’t work in the normative case.³³ I haven’t given an argument to illustrate that avoiding this problem is impossible. But it appears to be a serious problem. What the non-naturalist needs, then, is either a solution to this problem, or an alternative metasemantic theory that avoids this problem as well as the problems for the other theories given above.

7.3. REFERENCE MAGNETS TO THE RESCUE? Before turning to my positive proposal, it’s worth briefly discussing reference magnets, which have a rich history of solving (or purporting to solve) metasemantic problems.³⁴ More specifically, several metaethicists have also made use of reference magnets to make theoretical progress.³⁵ The central theoretical role that reference magnets are supposed to play is to constrain reference in cases of (sometimes radical) underdetermination. Suppose a theory of reference entails that a particular concept has multiple eligible referents. A theory of reference may entail, for example, the facts don’t decide between greenness and grueness as the referent of GREEN. Reference magnets provide a further constraint on reference to eliminate all but one ³² ³³ ³⁴ ³⁵

See e.g. McHugh and Way (2016), Howard (forthcoming). See section 7.3 for discussion of this point. Merrill (1980), Lewis, (1984), Sider (2011). Suikkanen (2017), van Roojen (2006), Dunaway and McPherson (2016).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

152

Preston Werner

property as the eligible referent for a given concept and set of referencefixing facts. In this way, reference magnets are, as Ted Sider points out,³⁶ a further constraint on reference fixing: Reference magnetism is not a theory of reference itself, but a doctrine that can be (and perhaps must be) coupled with any theory of reference you have independently motivated. Just how do reference magnets function to take in multiple eligible referents and output one eligible referent? As Lewis explains: This constraint looks not to the speech and thought of those who refer, and not to their causal connections to the world, but rather to the referents themselves. Among the countless things and classes that there are, most are miscellaneous, gerrymandered, ill-demarcated. Only an elite minority are carved at the joints, so that their boundaries are established by objective sameness and difference in nature. Only these elite things and classes are eligible to serve as referents.³⁷

Lewis then weakens his claim a bit by allowing for matters of degree of eligibility of reference.³⁸ Others have followed this line of thought. What is important for our purposes is that, of the multiple eligible referents determined for a given concept in a given theory of reference, the doctrine of reference magnetism then kicks in to fix the concept onto the most fundamental, joint-carving referent that is eligible. In order for reference magnets to assist the non-naturalist, we first need a theory of reference that latches onto at least some eligible non-natural properties. If the theory of reference designates no eligible referents, then reference magnets have no set of eligible referents to pare down. It’s crucial, then, that the theory of reference the non-naturalist is supplementing with reference magnets already targets non-natural properties as eligible referents. This rules out the causal and teleological theories, since these theories don’t target any non-natural properties as eligible referents. A more plausible candidate theory here may be a conceptual role semantics, in which we take the cluster of inferential relations between the normative concepts and Ramsify over them.³⁹ On such a theory, the set—or some relevant subset— of the normative concepts is interdefined, and then reference magnets do the work to attach them onto the relevantly isomorphic joint-carving nonnatural properties. I concede that such a view could provide a positive metasemantics for the non-naturalist. However, it comes at a cost which, other things equal, we

³⁶ Sider (2011: section 3.2). ³⁷ Lewis (1984: 227). Lewis credits the idea to Merrill (1980). ³⁸ Lewis (1984: 227–8). ³⁹ This sort of proposal was suggested to me by Billy Dunaway, Caleb Perl, and Mark Schroeder. It is also gestured at in passing by Jussi Suikkanen (2017: 20).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Getting a Moral Thing into a Thought

153

should want to avoid. For notice that, in order for such a view to work, we have to hope (or better: argue) that two things hold. First, we must hope that all normative concepts can be interdefined in an asymmetric Ramsified set. Ramsification requires asymmetry. Lewis himself implicitly concedes this in his famous paper on Ramseyan Humility: We have assumed that a true and complete final theory implicitly defines its theoretical terms. That means that it must have a unique actual realization. Should we worry about symmetries, for instance the symmetry between positive and negative charge? No: even if positive and negative charge were exactly alike in their nomological roles, it would still be true that negative charge is found in the outlying parts of atoms hereabouts, and positive charge is found in the central parts.⁴⁰

As Lewis here assumes, symmetries will be problematic within a Ramseyan sentence if they can’t be eliminated. But the worry is that we can’t be confident that the distribution of the normative properties will be asymmetric in the relevant sense, at least without some independent (and presumably contentious) argument.⁴¹ Suppose, for example, that all normative concepts are reduced to GOODNESS and BADNESS. What would make it the case that GOODNESS refers to goodness rather than badness? Presumably, nothing more could do the work—the work that reference magnets could do has run out. And the argument for an asymmetry ought to be independent from the metasemantic considerations raised here—or anyway, it seems to me that allowing metasemantic considerations to rule out certain views about the structure and relationship between different normative concepts is getting the inquiry backwards. A second problem for the CRS + reference magnets view is that it assumes an isomorphism between the distribution and relationships between the normative concepts and the distribution and relationships between the normative properties. Let’s say we generate a Ramsey sentence with normative concepts A, B, C, . . . N. Start first with distribution. Now suppose that one of the concepts, C, is defective, in the sense that it fails to refer to any normative property. Because all of the other concepts are, directly or indirectly, defined partly in terms of C, this defectiveness bleeds into the rest of the conceptual system, and so it’s unclear whether any of the concepts continue to refer. Next turn to the relationships between the concepts. Suppose it is part of the inter-definition of A and B that [if x is A, then x

⁴⁰ Lewis (2009: 207). Thanks to Ryan Doody and Daniel Wodak for discussion here. ⁴¹ As Zoë Johnson King suggested to me, thick concepts may be able to help, because they can attach the valences to the relevant descriptive properties as a matter of conceptual truth. I think this can’t work for reasons that Parfit (2011: section 90) points out, but I can’t explore this issue here.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

154

Preston Werner

is not B]. Further, suppose that there are no two properties that have all of the other relations that A and B have but are also such that [if x is A, then x is not B]. A and B are thus not isomorphic with any normative properties. So they don’t clearly have referents. It’s not clear how much of a problem this second problem is. Surely the conceptual role semanticist should allow for some flexibility and error in a conceptual system, compatible with its concepts having referents. But it is not a trivial question to ask how much is acceptable, and whether we can be confident that normative concepts rise above this bar. Perhaps this is where reference magnets are supposed to help. However, it isn’t obvious that they can—for recall that the role of reference magnets is only to fix onto one reference from a set of eligible ones. So the conceptual role semanticist about non-natural properties must be sure that her theory allows for enough flexibility and error for our normative concepts to pick out the non-natural properties as eligible referents. This challenge may be surmountable, but it is not insignificant.

7.4. AN EPISTEMIC APPROACH TO METASEMANTICS Non-naturalism runs into problems for each of the traditional metasemantic theories, as well as for the “connectedness” model. It seems that the nonnaturalist metasemanticist is in serious trouble. Without a plausible explanation of how our moral thoughts could latch onto the non-natural, normative properties, non-naturalism faces a very serious metasemantic objection. I now defend a positive metasemantic theory for non-naturalism, closely related to Imogen Dickie’s “justification-based” theory of reference fixing.⁴² To make clear that, despite being influenced by Dickie’s view, my view involves important nuances. I will call the view below the epistemic theory of content. To motivate the epistemic theory, notice two central motivations for externalist theories. First, they capture the intuitive thought that, if we are able to think about some particular object or property, then we must bear some kind of special relationship to the object/property in question. This is important to avoid underdetermination problems. If I’m thinking about the gray mug in my cabinet, there must be some relationship between me and that mug that makes my thought fix on that mug rather than any of the thousands of qualitatively identical mugs in the world. Second, the ⁴² Dickie (2015, 2016).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Getting a Moral Thing into a Thought

155

connection that externalists posit as required for reference fixing explains our intuitions about twin-earth cases. What explains the fact that Oscar’s concept WATER refers to H2O, whereas Twin-Oscar’s refers to XYZ? Externalist theories can provide this explanation easily. As we saw above, it is these external-relational requirements that spell trouble for non-naturalism. Standard non-naturalists reject the claim that normative properties are causally efficacious, as well as the claim that representing non-natural properties fulfills some evolutionary or individualistic teleological function. But surely non-naturalists shouldn’t deny that our moral thoughts bear some relationship to the non-natural normative properties. Even setting aside the rough motivations for some kind of externalist theory of content, non-naturalists should endorse the claim that there is some relationship between our moral thoughts and the nonnatural normative properties in order to undergird justified moral beliefs or moral knowledge. This relation could come in any number of forms, depending on one’s favored non-naturalist moral epistemology.⁴³ These thoughts naturally suggest some kind of epistemic theory of reference fixing. The rough idea is that a concept C refers to some property F just in case the mode of justification for beliefs containing C non-luckily converges on F. Such a theory, if it could be made to work, would capture the intuitive motivations for externalist theories, but in a way that makes reference to non-causal properties such as non-natural normative properties in principle possible. This rough idea suggests two questions. First, can such a theory be made to work? Second, given the difficulties faced for nonnaturalist moral epistemologists, is this really an improvement for the nonnaturalist metasemanticist? I address each question in turn. Helpfully, a book-length defense of an affirmative answer to the first question has been given by Imogen Dickie.⁴⁴ I cannot hope to recap the entire argument of the book here. But, briefly, Dickie motivates her core idea on the basis of two premises, which she calls Principle connecting aboutness and truth and Principle connecting truth and justification: Aboutness and Truth: “A thought about an object (a thought attributing a property to an object) is true iff the object has the property.”⁴⁵ Truth and Justification: “Justification is truth-conducive: in general, and allowing exceptions, if a subject’s belief is justified, he or she will be unlucky if the belief is not true and not merely lucky if it is.”⁴⁶

⁴³ Bengson (2015), Cuneo and Shafer-Landau (2014) provide two such examples. ⁴⁴ Dickie (2015). ⁴⁵ Dickie (2015: 37). ⁴⁶ Dickie (2015: 38). This is Dickie’s approximate formulation of the principle, but the precise details are outside the scope of this chapter.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

156

Preston Werner

Aboutness and Truth should be relatively uncontroversial. Truth and Justification is, perhaps, slightly more controversial, but not much. Even if justification doesn’t aim at truth,⁴⁷ it is overwhelmingly plausible that, when things are going well, justified beliefs are, in virtue of their justificatory status, more likely to be true. I’ll assume that something like Truth and Justification is correct in what follows.⁴⁸ These two relatively mundane principles entail, for reasons explained in an appendix, a surprisingly interesting metasemantic thesis: Aboutness and Justification: S’s beliefs are about an object iff their means of justification converges on the object, so that, given how the beliefs are justified, the subject will be unlucky if they do not match the object and not merely lucky if they do.⁴⁹

First, a word about the notion of a “means of justification.” For Dickie, each concept has a “proprietary means of justification.”⁵⁰ To see what this comes to, imagine a case where there is conflict between two potential sources of justification. You’re looking in the fridge and you see that there is a package of tofu sitting on the shelf. You form the belief we have tofu in the house. Your roommate then shouts from the other room “We are out of tofu!” You now have testimonial evidence that conflicts with your perceptual evidence.⁵¹ For Dickie, a proprietary means of justification is the means of justification that you take—other things being equal—to trump in cases of conflict. In a case where you are directly perceptually linked up with some object, you will tend to take that to trump your testimonial evidence. But in other cases, such as cases in which an object is very far away and you know that your interlocutor is knowledgeable, testimonial evidence will trump perceptual evidence. Whichever piece of evidence tends to trump for some concept C will be C’s proprietary means of justification. And because of its potential relevance to the moral case, notice that a means of justification being proprietary does not entail that all or even that the majority of our beliefs containing the concept make use that means of justification. Here I want to briefly flag a worry about the notion of a proprietary means of justification. It’s unclear to me, contra Dickie, that there are any ⁴⁷ Contra Cruz and Pollock (2004). ⁴⁸ Dickie (2015: section 2.1) has a much more sophisticated defense of this principle. ⁴⁹ Dickie (2015: 37). Again, this is Dickie’s approximate version of the principle. ⁵⁰ Dickie (2015: 50–7). ⁵¹ Notice too that beliefs using the concept in question will still have the same reference even when they are formed using some non-proprietary means of justification. The proprietary means fix the reference of a concept, but the concept can be constituents of all sorts of beliefs that are not proprietary. Thanks to an anonymous referee for pointing out the ambiguity here.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Getting a Moral Thing into a Thought

157

such means of justification that will always serve the role of a proprietary means of justification. That is, it seems that, with respect to the very same concept, there will be contexts in which perceptual evidence will trump, and contexts in which testimonial evidence will trump. So what are we to say about such a case? There are a number of options here. On the one I prefer, proprietary means are fixed by what the fundamental mechanism of justification for beliefs about C is. This means that, if testimonial evidence ultimately traces back to the perceptual evidence of someone else, perception retains its status as the proprietary means of justification.⁵² But rather than get bogged down exploring this and the other options, I think it is worth noting that this is arguably less of a worry for the moral case. Insofar as there is some source of justification for moral beliefs (intuitions, rational insight, reflective equilibrium, etc.), it seems clear that this source will trump other sources, such as testimony.⁵³ While I grant that this is a contentious claim, it would take quite some time to defend. So instead, let me just grant that, if this idea of a proprietary means of justification cannot be defended or revised, the epistemic theory of content is in some trouble. Return to Aboutness and Justification. This principle is not quite what the non-naturalist metasemanticist needs, for two reasons. First, as stated, the principle is about fixing reference for objects, not properties. But altering the principle into one about properties is straightforward: S’s beliefs are about a property iff the (proprietary) means of justification converge on the property so that, given how the beliefs are justified, the subject will be unlucky if they don’t track the property and not merely lucky if they do.

The second problem is more substantive. Aboutness and Justification provides a biconditional relationship between a means of justification and the referent of beliefs that the means of justification generates. It doesn’t yet give us a complete metasemantic theory, since it doesn’t tell us whether it is the means of justification that determines the referent or vice versa. And in fact, Dickie argues that there is no priority here.⁵⁴ If the non-naturalist metasemanticist is going to use Aboutness and Justification (or something like it) to make theoretical progress, it must be that a particular means of justification fixes reference. Dickie rejects this

⁵² So e.g. this would entail that the proprietary means of justification for any mathematical belief is going to be a priori, whether or not I am disposed to defer to professional mathematicians’ testimony over my own a priori reasoning. ⁵³ As skepticism about forming beliefs based on moral testimony may help to show. ⁵⁴ Dickie (2015: 3.5).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

158

Preston Werner

position, on the grounds that “[o]ur grip of the kind of factor that justifies beliefs seems to rest on our grip of the kind of factor that—in most, or nearby, or, optimal circumstances—will result in formation of beliefs that are true.”⁵⁵ In other words, it appears that our conception of justification will ineliminably refer to modes of belief-formation that tend toward truth, in which case we need a grip on the truth conditions, in which case we need to know which objects and properties figure in the beliefs being formed before we can determine the conditions on justification. Using a mode of justification to fix reference won’t work, then: it helps itself to a concept’s referent, the very thing it is attempting to explain. This argument must fail: It proves too much. It generalizes far beyond an epistemic theory of content determination. To see this, consider first a prima-facie problem for the causal theory of reference. Any plausible theory of reference better entail that your concept DOG refers to dogs. But, of course, there are a number of circumstances in which your DOG concept will be tokened by non-dog things: large cats, small horses in the distance, stuffed dogs, and so on will often token your DOG concept. So the causal theorist is going to have to say something about why DOG doesn’t pick out dog-or-largecat-or-smallhorse-or-stuffeddog, but just dogs.⁵⁶ Causal theorists have had a number of things to say about how to do this.⁵⁷ But what is most important is that these ways do not, and need not, refer to the fact that DOG refers to dogs. So, at least in principle, there is no circularity in solving this problem. For similar reasons, we shouldn’t—at least without further argument—assume that a story about a mode of justification must assume the content of the justified beliefs that it feeds out. It is true that the path here is harder to hoe, because justification is itself a normative notion.⁵⁸ It will be hard to determine success conditions without knowing what facts the means of justification attempt to track accurately. But there is no reason to think this can’t be done. Notice that the situation is structurally similar to the causal theorist’s. The causal theorist must give an account to separate the causal conditions that determine content from those that do not. And this account must not help itself to the content that is being fixed. Similarly, the epistemic theorist must give an account to separate the beliefs formed on the basis of the mode of justification in

⁵⁵ Dickie (2015: 111). ⁵⁶ This is a problem originally flagged by Fodor (1984). ⁵⁷ See Adams and Aizawa (2017: section 3) for an overview. ⁵⁸ However, note that there is no circularity in having a normative notion in the metasemantic story. Unlike in the internalist case, we aren’t providing subjects with the justification-facts in order to fix their reference for a concept like JUSTIFICATION. (Thanks to Aaron Elliott for discussion here.)

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Getting a Moral Thing into a Thought

159

question that determine content from those that do not. She needs a theory of which beliefs are epistemically unlucky, in the sense that they are justified beliefs, but not ones that fix content. Once we set aside the epistemically unlucky beliefs, we can determine the content of the belief in question by seeing which object/property the non-lucky beliefs converge on.⁵⁹ How can the epistemic theorist distinguish the reference-fixing beliefs from those that are merely unlucky? After all, it may seem that the most natural ways of doing this will involve appealing to the content of the beliefs in question, which the epistemic theorist can’t help herself to on pain of circularity.⁶⁰ While a full solution of this problem lies outside the scope of this chapter, let me gesture at one that seems promising. (If this solution is ultimately untenable, that’s ok—what is important is that this problem mirrors a problem for any externalist metasemantics. So even a lack of a solution is not in itself a reason to reject the epistemic theory proposed here.) Take some proprietary means of justification, such as visual experience. Such a mode plausibly has (or would on a fully worked out theory) a canonical list of good conditions for justification-conferring uses of the mode—for example, good lighting, being awake, sober, and so on. Notice that, if we are careful, nothing on this list will itself appeal to the content of what is seen. When and only when a visual experience meets the conditions on the list, it will count as a reference-fixing instance for the concepts being deployed in the downstream belief(s). Now this won’t be quite enough, because even under good conditions, visual experiences can get things wrong. So there will remain some prima-facie reference-fixing beliefs that are nonetheless intuitively false. This requires, perhaps, some further modal condition: The token belief will count as reference-fixing just in case (a) it was formed in good conditions for its proprietary means of justification, and (b) it isn’t the result of an “epistemically” deviant chain.⁶¹ A similar story could be told for any proprietary means of justification. Let’s walk through how our normative concepts could fix onto nonnatural properties, according to the epistemic theory of content. Imagine a toy non-naturalist epistemology, rational insight. (It won’t matter how this gets fleshed out, so the reader should imagine this mechanism however she so desires.) On such a view, agents form their (pure) moral beliefs on the ⁵⁹ If they converge on nothing, then the concept is defective in some way. What to say about defective concepts is a vexed matter that I can’t address here. But notice that the possibility of a concept’s being defective is a positive feature of the epistemic theory. Surely any theory of reference fixing should allow for and explain the possibility of defective concepts. ⁶⁰ I thank Joshua Schechter for pressing me to say more here. ⁶¹ I take it that whatever gets said about what makes a causal chain deviant could also be said about epistemically deviant chains (some of which are surely causal).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

160

Preston Werner

basis of rational insight. When things go well, and the canonical conditions for this means of justification are met, rational insight converges on the facts about normative properties. An agent can’t—at least non-luckily—form justified beliefs on the basis of other means of justification, such as perception or testimony. These are not sources of fundamental non-natural information, and so they will not count as the reference-fixing modes of justification. Only the outputs of rational insight count as fixing the content, and so the non-naturalist who holds such a view doesn’t need to worry that beliefs will converge on some natural property.⁶² There is also no risk that the means of justification will converge on the wrong normative property: Part of what makes it the case that WRONGNESS targets wrongness, as opposed to, say, goodness, is that the means of justification make beliefs containing WRONGNESS non-luckily true, which can only be explained by WRONGNESS referring to wrongness.⁶³

7. 5. H O W T HE EPISTE MIC T H EORY H ELPS T HE N O N - N A T U R A L I S T My proposal is that the non-naturalist metasemanticist endorse Aboutness and Justification, or something like it, read as a theory about content fixing for (at least) our moral concepts. I think there are three important reasons why the non-naturalist should find the epistemic theory of content attractive. First, and most importantly, the epistemic theory explains how reference to non-natural normative properties is possible. This alone makes the epistemic theory better suited for non-naturalist metasemantics than the other views discussed above. Second, the epistemic theory helps to unify two independent objections to non-naturalism—on the one hand, that nonnaturalists cannot explain how beliefs in non-natural properties are possible, and on the other, how non-naturalist moral knowledge is possible. The epistemic theory of content, if correct, would entail that a wholly adequate non-naturalist epistemology would simultaneously rebut both metasemantic and epistemological concerns for non-naturalism. And finally, the ⁶² As Gunnar Björnsson points out, it could turn out that the normative concepts refer to natural properties on such a view, just in case there is some equally or more fundamental natural property that rational insight converges on. But that seems unlikely, at least on a standard non-naturalist view according to which the normative properties are fundamental (or close to it). ⁶³ Importantly, the theory of epistemic luck we endorse can’t itself make indispensable use of the referent in question, on pain of circularity. Thanks to Aaron Elliott and Bar Luzon for discussion here.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Getting a Moral Thing into a Thought

161

epistemic theory is ideally placed to capture Autonomy and “just too different” intuitions, the central motivations for non-naturalism in the first place. I’ll briefly discuss each of these advantages in turn.

7.5.1. Non-Naturalism and the Epistemic Theory of Content Virtually no non-naturalists are skeptics. So they’re committed to: Moral Knowledge. Human beings have some non-accidentally true justified moral beliefs.

Moral Knowledge entails that there is a source of justification for at least some of our moral beliefs. There is some belief-forming method we have—in Dickie’s phrase, a “means of justification”—that, if we’re not unlucky, converges on truths about non-natural moral properties. Just what is this means of justification? Dickie, who is concerned with means of justification for ordinary objects, discusses two potential means of justification: perception and testimony.⁶⁴ It’s possible for the non-naturalist to argue that these are the means of justification for non-natural properties as well, but this has not been the traditional approach of non-naturalists. What’s important is that the antiskeptical non-naturalist is committed to some proprietary means of justification. If their theory is plausible, it should turn out that in good cases the means of justification results in non-luckily true beliefs, and thus that the means of justification will, over time, converge on the non-natural properties. The fact that means of justification are only constrained by their tendency toward truth is central to avoiding the problems of the previous externalist metasemantic theories. Causal and teleological metasemantic theories are ill-placed for the non-naturalist metasemanticist precisely because they impose constraints on content-fixing that non-naturalist properties can’t meet. Even if we have direct epistemic access to certain properties or objects, on such views, so long as we aren’t causally or teleologically related to these properties, we can’t form beliefs about them. An epistemic theory of content can capture what’s good about these theories while remaining agnostic about the possibility of referring to other properties. Maybe we can’t non-accidentally track non-causal properties. But if we can, only the epistemic theory of content can move us from this nonaccidental tracking to content fixing. This is just what the non-naturalist needs. If the non-naturalist has an otherwise compelling theory of intuition, ⁶⁴ Dickie (2015).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

162

Preston Werner

rational insight, or moral perception, then they have a mode of justification that the epistemic theory of content can exploit to provide a metasemantics for the non-natural properties.

7.5.2. Metasemantic and Epistemic Objections to Non-Naturalism: One and the Same? There may seem to be an obvious but deep problem with what’s been just said. The epistemic theory of content is all well and good for domains where we have a clear epistemology. But the moral domain is not one in which we have a clear picture of how knowledge is possible. And the situation is especially dire for non-naturalists, given Benacerraf-influenced arguments and evolutionary debunking arguments, which try to show that nonnaturalism makes moral knowledge impossible. Isn’t appealing to the epistemic theory of content trying to solve a difficult problem by reference to an epistemological theory with an even more devastating problem? I admit, there’s something ironic here in appealing to non-naturalist epistemology to solve a problem for non-naturalism, given that epistemology is one of the biggest sources of concern for such a view. However, as noted above, the non-naturalist is already committed to there being some solution to these problems. The epistemic theory of content only aims to piggyback on whatever solution this might be, and use it to fix content onto the non-natural properties. A wholly adequate non-naturalist epistemology will already be committed to there being some means of justification that non-luckily converges on the non-natural normative facts.⁶⁵ The epistemic theory of content says that this is all that is needed to ensure that our moral beliefs pick out the non-natural properties. So it follows from commitments that the non-naturalist already has—controversial as they might be—that the epistemic theory of content can provide a proper metasemantics for nonnatural normative beliefs. Of course, none of this goes any of the way toward actually providing a theory of justification for non-naturalism. Such a project is non-trivial, but it is a task for the non-naturalist qua epistemologist, not qua metasemanticist. And this is tentatively good news for the non-naturalist, for two related reasons. First, it reduces two families of objections to non-naturalism—epistemic and metasemantic—to one. This means that, if it can be done, a positive epistemological theory for non-naturalism could rebut two potentially devastating sets of objections to non-naturalism in one fell ⁶⁵ This mode of justification also cannot presuppose that the metasemantic explanans is already met. Thanks to Bar Luzon for pointing this out to me.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Getting a Moral Thing into a Thought

163

swoop. Second, this is contingently good news for the non-naturalist given the fact that much more work has been done on addressing non-naturalist epistemology than on non-naturalist metasemantics. So the reduction of the two families of objections into one has the fortuitous upshot of rendering the success of non-naturalist metasemantics dependent on a problem that has already received an overwhelming amount of attention and work. Obviously, no consensus has emerged, even among those sympathetic to non-naturalism, about how to solve these epistemological problems. But there is hope.

7.5.3. Autonomy and the “Just Too Different” Intuition Finally, the epistemic theory of content fits perfectly with the autonomy claims, core motivations for non-naturalism. Recall Epistemological and Metaphysical Autonomy: Epistemological Autonomy. An agent could be wholly procedurally rational and fully informed about all the natural facts and yet ignorant or mistaken about the normative. Metaphysical Autonomy. The normative facts are irreducible to the natural facts, in the sense that there is no conceptual entailment or complete metaphysical explanation from the natural facts to the normative facts.

We saw above that these claims cause problems for certain attempts to provide a non-naturalist metasemantics. Non-naturalists claim that the metaphysical gap, sometimes expressed in terms of the “just too different” intuition, between natural facts and normative facts, is too wide to be crossed epistemologically. An epistemology for the purely normative facts won’t be assisted by gathering more non-normative facts. Our means of justification (in terms of the source of input) for normative facts is going to have to be fundamentally different from our means of justification for run of the mill natural facts, given their metaphysical status as independent from them. The epistemic theory of content nicely accommodates this line of thinking. To see why this is so, just recall from above that, for the non-naturalist, the proprietary means of justification for normative beliefs will be unlike the means of justification for their natural counterparts.⁶⁶ This fits perfectly

⁶⁶ Not all non-naturalists accept this claim (e.g. Seitya 2012), myself included, ironically (Werner 2016, 2018). But most do, and I think even those who do not accept that the sentence in the text is strictly true have reason to accept a variant of it so long as they’re committed to Metaphysical Autonomy.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

164

Preston Werner

with the line of thinking behind the Autonomy Theses. The Autonomy Theses are preserved, and thus the core motivation for non-naturalism is preserved—even explained in part—by the epistemic theory of content. A similar explanation will hold for any non-naturalist theory that aims to capture Epistemological Autonomy.

7 . 6 . C ON C L U S I O N Non-naturalists have done a great amount of work on the metaphysics and epistemology of irreducibly normative properties. Considerably less work has been done on their metasemantics. This is surprising, because many of the traditional metasemantic views rule out the possibility of referring to and having beliefs about non-natural properties. Thus an underexplored objection to non-naturalism remains unsolved. Non-naturalists may have believed that they could help themselves to other realist-friendly metasemantics for normative terms. The first goal of this chapter was to argue that that is mistaken: None of the traditional metasemantic theories, even those explicitly given to be realist-friendly, fit with non-naturalism, especially given considerations surrounding the Autonomy Theses. Thus, the pessimistic half of this chapter argued that non-naturalists really do face a metasemantic challenge. My second goal in this chapter defended a sketch of a positive metasemantic view, indebted to recent work by Imogen Dickie. On this view, what makes a given normative concept refer to a non-natural property is that its means of justification converge onto the facts that the property figures in. The view has powerful independent motivation. Furthermore, it avoids the problems that other metasemantic theories cause for non-naturalists, as well as according nicely with some of the central motivations for non-naturalism. The success of this theory of content at explaining how our normative beliefs pick out the non-natural properties depends on providing an adequate epistemology for non-naturalism. Since this task is notoriously difficult, non-naturalists are not out of the woods. If the epistemic theory defended above is correct, then, defending a positive epistemology for nonnaturalism is even more urgent. If it can be done, the non-naturalists will have made a significant amount of progress, not just epistemologically, but metasemantically as well.⁶⁷

⁶⁷ Thanks to an anonymous referee for pressing me to unpack the reasoning in the following Appendix.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Getting a Moral Thing into a Thought

165

APPENDIX

Aboutness and Justification: Dickie’s Arguments In the main body of the chapter, I claimed that Aboutness and Justification: S’s beliefs are about an object o iff their means of justification converges on the object, so that, given how the beliefs are justified, the subject will be unlucky if they do not match the object and not merely lucky if they do. is entailed by the following two principles: Aboutness and Truth: “A thought about an object (a thought attributing a property to an object) is true iff the object has the property.”⁶⁸ Truth and Justification: “Justification is truth-conducive: in general, and allowing exceptions, if a subject’s belief is justified, he or she will be unlucky if the belief is not true and not merely lucky if it is.”⁶⁹ But the proof here is not at all obvious. I here briefly walk through the proof of the biconditional; my discussion in this appendix is heavily indebted to Dickie’s own discussion.⁷⁰ Begin with the left-to-right conditional. The proof here is straightforward. Suppose 1.

S’s belief that is about o.

From 1 and Aboutness and Truth, we get: 2.

S’s belief that is true iff o is ɸ.

From 2 and Truth and Justification, it follows that: 3. Justification that renders S’s belief that unlucky if false and not merely lucky if true will make it the case that S’s belief is unlucky if o is not ɸ. Which thereby gives us the left-to-right conditional: 4. If S’s belief is about o, justification that renders S’s belief that unlucky if false and not merely lucky if true will make it the case that S’s belief is unlucky if o is not ɸ.

⁶⁸ Dickie (2015: 37). ⁶⁹ Dickie (2015: 38). This is Dickie’s approximate formulation of the principle, but the precise details are outside the scope of this chapter. ⁷⁰ Dickie (2015: ch. 2).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

166

Preston Werner

Let’s turn, then, to the right-to-left conditional. Here things get trickier. Here is the conditional to be proven: RtL Aboutness and Justification: If the proprietary means of justification for S’s beliefs converge on some object o, such that S will be unlucky if they do not match o and not merely lucky if they do, then S’s beliefs are about o. One reason the proof of this side of the conditional is trickier is that we need to rule out two alternatives to establish that S’s beliefs are about o. First, it may be that S’s beliefs are about some other object, o*. Second, it may be that S’s beliefs are about nothing at all—they may have no referent. Let’s consider each in turn. How can we rule out that S’s beliefs are about some other object, o*? In order for this to be so, we would need two objects, o and o*, such that all of their intrinsic and relational properties—or at least all of them accessible via the proprietary means of justification—are the same. In the case of ordinary objects, at least, this simply won’t happen, because some of the relational properties of o and o* have to do with their relations to the believing agent herself. And barring spatiotemporally overlapping intrinsic duplicates, this won’t happen.⁷¹ Now this reasoning only works for ordinary, physical objects, whereas in the chapter, I am concerned with non-natural properties. So a question can be raised about why one’s normative beliefs may not be about some other property, F*, such that F (the genuinely normative property) is not identical to F*. But in the case of properties, I submit that it is just impossible for there to be two distinct properties which share all of the same features, and so the problem just doesn’t arise. It is true that there may be some natural property F* which is extensionally equivalent to F, but F would still, assuming non-naturalists are right, have some second-order features that F* does not. The second possibility is that, while S’s proprietary means of justification converges on o, her beliefs are nonetheless about nothing. She has failed to secure reference because, presumably, there is some other condition on fixing reference that she has not met. This alternative is less plausible on its face. We would need some powerful argument to the effect that, even though S’s beliefs consistently and non-coincidentally track facts about o, nonetheless her beliefs are not about o. And it is hard to imagine how such an argument would go. I wholly realize that the above arguments leave some space for disagreement and rebuttal. But I hope to have done a good job of motivating the view without getting too far afield from the present project, which is to assume that Dickie’s theory is, broadly speaking, correct, and to show how it can be extended to non-naturalist metasemantics.⁷²

⁷¹ See Dickie (2015: ch. 2) for detailed reasoning along these lines. ⁷² For extremely helpful feedback and discussion on earlier drafts, I’m thankful to John Bengson, Gunnar Björnsson, Teresa Bruno Niño, Janice Dowell, Billy Dunaway, Kevan Edwards, Aaron Elliott, David Enoch, Nikki Fortier, Zoë Johnson King, Avi Kenan, W. Scott Looney, Bar Luzon, Hille Paakkunainen, Caleb Perl, Jared Riggs, Mark Schroeder, and Byron Simmons.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Getting a Moral Thing into a Thought

167

References Adams, Fred, and Ken Aizawa (2017). “Causal Theories of Mental Content” in Stanford Encyclopedia of Philosophy. https://plato.stanford.edu/entries/contentcausal/ Benacerraf, Paul (1973). “Mathematical Truth,” Journal of Philosophy 70(19): 661–79. Bengson, John (2015). “Grasping the Third Realm,” Oxford Studies in Epistemology 5: 1–38. Blackburn, Simon (1984). “Supervenience Revisited” in Ian Hacking (ed.), Exercises in Analysis: Essays by Students of Casimir Lewy. Cambridge: Cambridge University Press. Boyd, Richard (1988). “How to Be a Moral Realist” in G. Sayre-McCord (ed.), Essays on Moral Realism. Ithaca, NY: Cornell University Press. Boyd, Richard (2003). “Finite Beings, Finite Goods: The Semantics, Metaphysics and Ethics of Naturalist Consequentialism, Part 1,” Philosophy and Phenomenological Research 66(3): 505–53. Chalmers, David (2012). Constructing the World. Oxford: Oxford University Press. Cheyne, Colin (2001). Knowledge, Cause, and Abstract Objects. Dordrecht: Kluwer Academic Publishers. Clarke-Doane, Justin (2017). “What is the Benacerraf Problem?” in Fabrice Pataut (ed.), New Perspectives on the Philosophy of Paul Benacerraf: Truth, Objects, Infinity. Cham: Springer. Cruz, Joe, and JohnPollock (2004). “The Chimerical Appeal of Epistemic Externalism” in Richard Schantz (ed.), The Externalist Challenge. Berlin: De Gruyter. Cuneo, Terence, and Russ Shafer-Landau (2014). “The Moral Fixed Points: New Directions for Moral Nonnaturalism,” Philosophical Studies 171(3): 399–443. Dickie, Imogen (2015). Fixing Reference. Oxford: Oxford University Press. Dickie, Imogen (2016). “The Essential Connection Between Epistemology and the Theory of Reference,” Philosophical Issues 26(1): 99–129. Dunaway, Billy, and Tristram McPherson (2016). “Reference Magnetism as a Solution to the Moral Twin Earth Problem,” Ergo: An Open Access Journal of Philosophy 3. Elliott, Aaron (2014). “Can Moral Principles Explain Supervenience?” Res Philosophica 91(4): 629–59. Enoch, David (2011). Taking Morality Seriously. Oxford: Oxford University Press. Field, Hartry (1989). Realism, Mathematics and Modality. Oxford: Blackwell. Fodor, Jerry A. (1984). “Semantics, Wisconsin Style,” Synthese 59(3): 231–50. Harman, Gilbert (1977). The Nature of Morality: An Introduction to Ethics. Oxford: Oxford University Press. Heathwood, Chris (2015). “Irreducibly Normative Properties,” Oxford Studies in Metaethics 10: 216–44. Hume, David (1975). Enquiries Concerning Human Understanding and Concerning the Principles of Morals. Oxford: Oxford University Press.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

168

Preston Werner

Jackson, Frank, and Philip Pettit (1995). “Moral Functionalism and Moral Motivation,” Philosophical Quarterly 45(178): 20–40. Joyce, Richard (2001). “Moral Realism and Teleosemantics,” Biology and Philosophy 16(5): 723–31. Joyce, Richard (2006). The Evolution of Morality. Cambridge, MA: MIT Press. Lenman, James (2010). “Uggles and Muggles: Wedgwood on Normative Thought and Justification,” Philosophical Studies 151(3): 469–77. Lewis, David (1984). “Putnam’s Paradox,” Australasian Journal of Philosophy 62(3): 221–36. Lewis, David (2009). “Ramseyan Humility” in David Braddon-Mitchell and Robert Nola (eds), Conceptual Analysis and Philosophical Naturalism. Cambridge, MA: MIT Press. Liggins, David (2010). “Epistemological Objections to Platonism,” Philosophy Compass 5(1): 67–77. McConnell, Neil (2015). “The Deviance in Deviant Causal Chains,” Thought: A Journal of Philosophy 4(2): 162–70. McGrath, Sarah (2014). “Relax? Don’t Do it! Why Moral Realism won’t Come Cheap,” Oxford Studies in Metaethics 9: 186–214. McHugh, Conor, and Jonathan Way (2016). “Fittingness First,” Ethics 126(3): 575–606. McPherson, Tristram (2012). “Ethical Non-Naturalism and the Metaphysics of Supervenience,” Oxford Studies in Metaethics 7: 205–34. Merrill, G. H. (1980). “The Model-Theoretic Argument Against Realism,” Philosophy of Science 47(1): 69–81. Millikan, Ruth (1984). Language, Thought and Other Biological Categories. Cambridge, MA: MIT Press. Millikan, Ruth (1995). “Pushmi-Pullyu Representations,” Philosophical Perspectives 9: 185–200. Moore, G. E. (1903). Principia Ethica. Mineola, NY: Dover. Neander, Karen (2012). “Teleological Theories of Mental Content,” in Stanford Encyclopedia of Philosophy. Oddie, Graham (2005). Value, Reality, and Desire. Oxford: Clarendon Press. Parfit, Derek (2011). On What Matters, vols 1 and 2. Oxford: Oxford University Press. Ridge, Michael (2014). “Moral Non-Naturalism,” in Stanford Encyclopedia of Philosophy. Ruse, Michael, and Edward O. Wilson (1986). “Moral Philosophy as Applied Science,” Philosophy 61(236): 173–92. Scanlon, Thomas (2014). Being Realistic about Reasons. Oxford: Oxford University Press. Schroeter, Laura, and Francois Schroeter (2003). “A Slim Semantics for Thin Moral Terms?” Australasian Journal of Philosophy 81(2): 191–207. Schroeter, Laura, and Francois Schroeter (2014). “Normative Concepts: A Connectedness Model,” Philosophers’ Imprint 14(25): 1–26.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Getting a Moral Thing into a Thought

169

Schroeter, Laura, and Francois Schroeter (forthcoming). “The Generalized Integration Challenge in Metaethics,” Nous. Setiya, Kieran (2012). Knowing Right from Wrong. Oxford: Oxford University Press. Sider, Theodore (2011). Writing the Book of the World. Oxford: Oxford University Press. Sinclair, Neil (2012). “Metaethics, Teleosemantics and the Function of Moral Judgments,” Biology and Philosophy 27(5): 639–62. Street, Sharon (2006). “A Darwinian Dilemma for Realist Theories of Value,” Philosophical Studies 127(1): 109–66. Sturgeon, Nicholas (1998). “Moral Explanations” in James Rachels (ed.), Ethical Theory 1: The Question of Objectivity. Oxford: Oxford University Press. Suikkanen, Jussi (2017). “Non-Naturalism and Reference,” Journal of Ethics and Social Philosophy 11(2): 1–24. Van Roojen, Mark (2006). “Knowing Enough to Disagree: A New Response to the Moral Twin Earth Argument,” Oxford Studies in Metaethics 1. Wedgwood, Ralph (2007). The Nature of Normativity. Oxford: Oxford University Press. Werner, Preston J. (2016). “Moral Perception and the Contents of Experience,” Journal of Moral Philosophy 13(3): 294–317. Werner, Preston J. (2018). “Moral Perception without (Prior) Moral Knowledge,” Journal of Moral Philosophy 15(2): 164–81.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

8 The Metaphysics of Moral Explanations Daniel Fogal and Olle Risberg 8 . 1. I N T R O D U C T I O N What are the aims of first-order moral inquiry, or normative inquiry more generally? One aim is to specify which actions are right and which actions are wrong. But that’s not enough; another aim is to explain why the right actions are right and the wrong ones are wrong. How, more exactly, do such explanations work? On one natural view, the full explanation of why a particular action was wrong (or right or good or bad, etc.) involves two kinds of facts: (i) a particular “natural” or “descriptive” fact—perhaps it was a lie, for instance, or failed to maximize wellbeing—and (ii) a general moral fact—that it is wrong to lie, for instance, or to not maximize well-being. Call the latter facts moral principles. Despite the naturalness of this view, the precise nature and structure of the relevant principles, and the form of explanation involved, are not yet well understood. What’s more, the view that moral principles play an explanatory role has recently been attacked. Selim Berker (2019), for example, argues that moral principles can instead be viewed as mere summaries of the explanatory relations that obtain between particular moral and non-moral facts, with the principles themselves being explanatorily inert. Similarly, Mark Schroeder (2005: 3) argues that Ralph Cudworth’s objection against theological voluntarism—which crucially involves appeal to explanatory moral principles—threatens to generalize so as to rule out the possibility of “perfectly general explanatory moral theories.” Our aim in this chapter is to defend the explanatory role of moral principles by arguing that it best accommodates some intuitive claims about moral metaphysics. Indeed, while we focus on morality, parallel claims are plausible with respect to normative principles and explanations more generally. We begin by presenting the relevant “data points” (§8.2) before outlining, in general terms, how our favored view captures them (§8.3). We then argue that the view that moral principles aren’t explanatory in this way falls short

Daniel Fogal and Olle Risberg, The Metaphysics of Moral Explanations In: Oxford Studies in Metaethics Volume 15. Edited by: Russ Shafer-Landau, Oxford University Press (2020). © Daniel Fogal and Olle Risberg. DOI: 10.1093/oso/9780198859512.003.0008

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

The Metaphysics of Moral Explanations

171

(§8.4)—in particular, it fails to explain the supervenience of moral facts on natural facts. Next we discuss two competing accounts of what, more precisely, moral principles are like: the nomic view (§8.5) and moral platonism (§8.6). Finally, we consider the sense in which moral principles are metaphysically necessary (§8.7).

8. 2 . TH E D A T A We’ll begin by presenting three claims we think any fully satisfactory metaethical theory should explain, or otherwise accommodate in a principled manner. Although none are completely uncontroversial, theories that explain them will, other things being equal, enjoy an advantage over those that don’t (and even more so over theories which are incompatible with them). The first data point is that moral facts supervene upon purely natural (or descriptive or whatever) facts. Suppose that Matti is a good person. Besides being good, he also has numerous natural properties (including relational ones) that are connected to goodness in the following way: anyone who is descriptively just like Matti will also be good. Indeed, if Matti is good, it’s impossible for someone to possess all and only his natural properties without also being good. Generalizing, the supervenience relation that holds between “the natural” and “the moral” is standardly formulated as follows, where M is the family of moral properties, N is the family of natural properties, and ☐ is metaphysical necessity: Data point (i): Strong Supervenience (∀F in M)(∀x)[Fx → (∃G in N)(Gx & ☐(∀y)(Gy → Fy))]¹ In English: for every moral property F, if something is F, then that thing has some (possibly quite complex!) natural property G such that, by metaphysical necessity, everything that is G is also F.

Strong Supervenience has long been treated as something like a fixed point, though recently it’s been called into question.² We’ll nonetheless assume that its rejection comes as a cost. There are three things to note about Strong Supervenience. First, despite being standardly glossed (as we did above) as expressing a relation between ¹ Cf. McPherson (2015). Strong Supervenience is standardly assumed to hold of necessity, either metaphysical or conceptual (or both). We assume it at least holds of metaphysical necessity. ² See e.g. Rosen (2020). Rosen nonetheless remains committed to a form of supervenience that involves “normative” rather than metaphysical necessity, and so faces the analogous task of explaining why it holds.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

172

Daniel Fogal and Olle Risberg

“the moral” and “the natural,” the relevant pattern of covariation only concerns properties of particular things. (The higher-order quantifiers quantify over moral and natural properties while the first-order quantifiers quantify over particular bearers of those properties.) We’ll return to this point. Second, there’s controversy over how to characterize the supervenience base. At a minimum, we assume the relevant family of properties (N) is closed under property conjunction and property disjunction and restricted to repeatable properties (cf. Atiq forthcoming). In general, the goal is to strip base properties of their particularity. Although there are several additional complications, everyone should acknowledge that there is some truth in the neighborhood, though how exactly one formulates it may depend on one’s other commitments. Third, Strong Supervenience doesn’t entail that particular moral facts are explained by or obtain in virtue of natural facts. Indeed, supervenience claims in general are silent as to which, if any, explanatory relations obtain between the relevant kinds of facts.³ Nevertheless, the modal correlation specified by Strong Supervenience isn’t the only interesting relation that holds between particular moral and natural facts. An explanatory relation also seems to hold: when (e.g.) a person is good or an action is wrong, there are natural features of those entities that make the person good and the action wrong. Hence the second data point: Data point (ii): Particular moral facts are at least partly (and at least ultimately) explained by particular natural facts.

We say “at least ultimately” because some particular moral facts may obtain in virtue of other particular moral facts. In such cases, those other moral facts are in turn explained (at least ultimately) by particular natural facts. Some additional clarifications are in order. First, what is meant by “particular moral fact”? A particular moral fact is a moral fact about a particular (dated, non-repeatable) thing, such as a particular action, person, or state of affairs. Besides particular moral facts, there are also general, nonparticular moral facts, such as that pain is bad or that lying is pro tanto wrong. This intuitive distinction is surprisingly often overlooked. Second, what is meant by “explained”? The kind of explanation we have in mind is metaphysical explanation (or what is commonly called “grounding”). One distinguishing feature of such explanations is that they are noncausal. Suppose, for example, that we want to know why the barn is red.⁴ ³ On the difference between supervenience and explanation, see DePaul (1987), Bliss and Trogdon (2014), and Berker (2018). For dissent, see Kovacs (2019). ⁴ Cf. Glazier (2016: 11).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

The Metaphysics of Moral Explanations

173

There are at least two questions we might be interested in. The first is what made it the case that the barn is red. The fact that, say, someone painted it yesterday would help provide an answer—it would help causally explain why the barn is red. The second question concerns what presently makes it the case that the barn is red. Here historical facts are irrelevant. Instead, what matter are contemporaneous facts, such as the fact that the barn is crimson. This fact non-causally explains why the barn is red.⁵ We also take metaphysical explanation to be objective, roughly in the sense of being mind- or stance-independent. Impressionistically put, objective explanations involve explanatory relations—whether they be causal, nomic, metaphysical, or something else—that obtain “out there” in the world. This is different from the more familiar, pragmatic notion of an explanation, understood as the sort of thing we standardly ask for and provide concerning a variety of subject matters (and which can in that sense be causal or non-causal), and whose success depends on “facts about the interests, beliefs or other features of the psychology of those providing or receiving the explanation [or] the ‘context’ in which the explanation occurs” (Woodward 2014: §6.1). In slogan form, we can distinguish between “explains” in the sense of makes it the case and “explains” in the sense of makes sense of why it’s the case.⁶ Although knowing what made or makes something the case will often help make sense of why it’s the case, objective and pragmatic explanations are not neatly aligned. It might be objected that focusing on metaphysical explanation prejudges certain metaethical disputes, such as whether expressivism is true. Insofar as there’s a conflict, however, the problem lies with those views. After all, everyone needs a story about the metaphysical-seeming judgments we make about moral matters, including judgments about what makes actions right, wrong, etc. If expressivists can provide such a story, the seeming conflict disappears. If they can’t, that’s a problem for their view. The third clarification concerns what is meant by “partly explains.” The relevant contrast here is the notion of a full explanation. Almost all the explanations we actually provide—i.e. pragmatic explanations—are partial rather than full. If you purchase a pet, and someone asks you why, you might say that you’re lonely. While sensible as a reply, the fact that you’re lonely doesn’t by itself explain your purchase. Instead, its explanatory import depends on a bunch of background facts that are taken for granted, such as ⁵ We count present facts about the past and timeless facts as contemporaneous. Although causal explanations are paradigmatically diachronic and non-causal explanations are paradigmatically synchronic, there may be exceptions. ⁶ More nuance is called for, but we lack space to provide it. Suffice it to say, ordinary explanation-talk is both messy and context-sensitive (cf. Lewis 1986; Jenkins 2008).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

Daniel Fogal and Olle Risberg

174

that you don’t want to be lonely, that you believe a pet will make you less lonely, etc. Your loneliness is thus only a part, and indeed a rather small part, of what fully explains your action. This illustrates a general feature of our explanatory practice: rather than providing a full explanation, we’re typically content to highlight one or two notable factors, trusting our interlocutors to fill in the rest. Providing the full explanatory story is typically laborious, if possible at all, and unnecessary. Here, finally, is our third data point: Data point (iii):

Moral principles are explanatory in nature.

This claim is motivated, in part, by the observation that a moral theory merely consisting of a list of all particular actions that are right or wrong is incomplete. We also want to know why they’re on the list. General moral principles answer that question. Some care is needed, however. For as Berker (2019) notes, moral principles might be explanatory in either of two ways. Suppose it’s always wrong to lie. According to Berker, this principle is explanatory only in the sense that it specifies a natural property—i.e. being a lie—such that any particular act with that property is wrong fully because it is a lie. On this view, moral principles can be viewed as mere summaries of patterns of particular explanatory relations, with the principles themselves being explanatorily inert. As we’ll put it, such principles are merely explanatory in content— they specify which particular natural facts explain which particular moral facts and that’s it. Berker’s view thus resembles “Humean” accounts of laws of nature, according to which natural laws are mere summaries of patterns among particulars. Yet the view is also non-Humean insofar as the relevant patterns involve instantiations of a hyperintensional explanatory relation, not mere co-occurrence of distinct facts or properties. This view—call it Hyperintensional Humeanism—is effectively what you get if you take data point (ii), which concerns the explanation of particular moral facts by natural facts, to be more fundamental than data point (iii), which concerns the explanatory nature of moral principles. The salient alternative is that the truth of the principle that lying is wrong itself plays a role in explaining particular moral facts involving lying, with the fact that the action was a lie only partly explaining why it’s wrong. On this view, moral principles are explanatory in role—they themselves help explain particular moral facts.⁷ Data point (iii) is neutral with respect to whether such principles are explanatory in role or in content.

⁷ Berker calls such principles “explanation-serving” and principles that are explanatory in content “explanation-involving.”

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

The Metaphysics of Moral Explanations

175

8.3. EXPLAINING THE DATA To recap, here are the data points: Data point (i): Strong Supervenience—for every moral property F, if something is F, then that thing has some natural property G such that, by metaphysical necessity, everything that is G is also F. Data point (ii): Particular moral facts are at least partly explained by natural facts. Data point (iii): Moral principles are explanatory in nature.

A unified account of these data points is desirable. The most natural way of providing one is to view some data points as more fundamental than others. But which ones? We doubt (i) is most fundamental. As is standardly recognized, supervenience is “not a ‘deep’ metaphysical relation” but “a ‘surface’ relation that reports a pattern of property covariation, suggesting the presence of an interesting dependency relation that might explain it” (Kim 1993: 167). Supervenience theses thus call for explanation rather than provide them. So ideally, (i) can be accounted for in terms of (ii) or (iii) or both. We think the best way of making sense of (i) is by opting for the package of (ii) together with the interpretation of (iii) on which moral principles are explanatory in role. On this view, explanations of particular moral facts involve three main ingredients: Explanans: particular natural fact(s) (e.g. a is a lie). Principle: general explanatory moral principle (e.g. lying is wrong). Explanandum: particular moral fact (e.g. a is wrong).

The explanandum is fully explained by the explanans together with the moral principle, though (as we’ll see) the exact role played by principles will depend partly on how they are formulated. But for now this outline of the tripartite structure of moral explanations will suffice. Alternative terminological choices might be made. Schaffer (2017a), for example, emphasizes the importance of the tripartite structure of metaphysical explanations by distinguishing between the “source,” “link,” and “result” in an explanation. As he notes, when “there are three roles involved, nothing but confusion can arise from insisting on only using two classificatory boxes” (20). Causal explanations illustrate this point: while causal laws plausibly help explain particular effects, laws aren’t themselves causes. Causes and laws play different roles in the full explanation of whatever is caused. We agree, although we’ve chosen to supplement the explanans/ explanandum ideology rather than jettison it.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

176

Daniel Fogal and Olle Risberg

The tripartite account of moral explanations incorporates data points (ii) and (iii): particular moral facts are explained by particular natural facts together with general moral principles.⁸ Given the substantive but plausible assumption that the fundamental moral principles obtain of metaphysical necessity (if at all), making sense of Strong Supervenience—data point (i)— is straightforward. (We’ll return to this assumption in §8.7.) We’ll call this the Divide & Conquer (or D&C) strategy. Essentially, Strong Supervenience states that the moral properties of particular things cannot differ unless their natural properties differ. On the tripartite account, the moral properties of particular things depend on two things: (a) which natural properties they have and (b) which moral principles obtain. Regarding (a), it’s trivial that two things cannot differ in their natural properties without differing in their natural properties. Regarding (b), it’s also trivial that two things cannot differ with respect to which moral principles obtain, because such principles can’t differ period—they obtain of necessity and so trivially supervene on everything. Hence, the moral properties of particular things depend on two things—(a) and (b)— both of which supervene on their natural properties for trivial reasons. It’s therefore no surprise that moral properties of particular things can’t differ unless their natural properties differ.⁹ We take the claim just explained to be the “core” supervenience claim. But as David Faraci (2017) emphasizes, Strong Supervenience also entails a necessitation claim: no particular thing can have a moral property unless it also has some natural property. Making sense of this isn’t difficult, however, as its falsity would require something to have a moral property without having any natural properties at all. And given the very broad sense of “natural” at issue, this seems impossible. Every action, for example, will at least have the natural property of being an action, however featureless it might otherwise be. Similarly, Cartesian souls, even if empirically inaccessible, will still have properties such as being conscious, or, indeed, being Cartesian souls. This suggests that every particular thing will have some natural property or other, and a fortiori, that every particular thing with a moral property will also have a natural property.¹⁰ It might be worried that an account of Strong Supervenience that posits non-contingent explanatory moral principles fails to make genuine progress, since such principles merely restate (in a slightly different guise) the ⁸ Our account thus resembles what Schroeder (2005) calls “the Standard Model” of normative explanations, although we think the tripartite account is more perspicuous. ⁹ At least in general, if X depends on Y and Y trivially supervenes on Z then it should be no surprise that X supervenes on Z. ¹⁰ We suspect Faraci’s real concern is why it’s “impossible for normative properties to be ungrounded” (2017: 315)—i.e. data point (ii), not (i).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

The Metaphysics of Moral Explanations

177

problematic truth that was supposed to be accounted for.¹¹ But that’s false: supervenience theses say nothing about explanation, so principles concerning the latter can’t be restatements of the former. Although several recent explanations of supervenience resemble the D&C strategy, none are fully satisfactory. Enoch (2011), for instance, takes the basic normative facts to be “norms” that hold with metaphysical necessity, and thereby explain Strong Supervenience. But Enoch says little about what norms are or how they explain particular normative facts. We’ll consider some answers to this question in §§8.5–6.¹² One account takes the fundamental moral facts to concern kinds rather than particulars. Both Skarsaune (2015) and, more briefly, Schroeder (2014: §6.6) appeal to this idea in explanations of supervenience, though neither pays sufficient attention to data points (ii) and (iii). Scanlon (2014), in contrast, holds that supervenience is explained by “pure” normative facts of the form “∀x(R(p, x, c, a)),” which reads: for all agents x, in circumstance c, fact p is a reason to do a. However, for reasons discussed in §8.4, the use of universal generalizations in formulating moral principles is problematic. While the proposals just considered are all pursued in defense of some form of non-naturalism, the D&C strategy itself is neutral regarding the naturalism/non-naturalism dispute, and indeed regarding most other metaethical disputes. In particular, even if it’s always possible to explain why a particular thing has a certain moral property, it doesn’t follow that the moral properties themselves—i.e. what it is to have a certain moral property—can be explained.¹³ That’s a point of contention dividing naturalists and nonnaturalists, and one to be decided on other grounds. More generally, the D&C strategy is neutral concerning the status, specific content, number, complexity, and explanation (if any) of the relevant moral principles. This is a feature, not a bug: it focuses the debate precisely where it should be (and to some extent always has been)—namely, on the fundamental moral principles, if such there be. Given the foregoing, the attention garnered by Strong Supervenience (and its ilk) in recent decades looks rather misplaced. It has often been held, for example, that non-naturalists have a hard time explaining moral supervenience. But if the D&C strategy is successful, that’s false. This is just one example of how focusing on supervenience while neglecting the more fundamental data points (ii) and (iii) can be distorting. Other examples are furnished by various attempted explanations of supervenience—many of

¹¹ Cf. McPherson (2012) on “bruteness revenge.” ¹² Enoch (2019) advocates a form of “grounding pluralism.” For reasons to be skeptical, see Enoch (2019: n. 20). ¹³ As Leary (n.d.) argues, this distinction was recognized by G. E. Moore.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

178

Daniel Fogal and Olle Risberg

which are naturalist-friendly—that are either silent about or, worse, in tension with the other data points. To illustrate: • Naturalistic identity-theses make sense of supervenience but are harder to reconcile with the data points about moral explanation—i.e. (ii) and (iii). Obviously things that differ morally must differ naturally if moral properties just are natural properties. But if rightness is identical to happiness-maximizing, for example, it’s hard to see how actions can also be right in virtue of being happiness-maximizing, since nothing explains itself. • Expressivist explanations of supervenience have often centered on the idea that our moral practice wouldn’t fulfill its “function” of e.g. coordinating behavior if our moral views didn’t respect supervenience (cf. Blackburn 1984). Even if this thesis makes sense of supervenience, it’s silent concerning data points (ii) and (iii). While there are various things adherents of such views might say in response, our main complaint concerns the focus of the debate. For as Berker (2018) argues, the notion of moral supervenience was first introduced and motivated with reference to claims concerning both co-variance—data point (i)—and explanation—data point (ii). In one of the earliest discussions, for example, R. M. Hare notes that if two things differ in goodness “there must be some further difference between them to make one good and the other not” (1952: 81; emphasis added). The modal formalizations of supervenience that subsequently became dominant replaced this dual-focus with a single-minded one—they express covariation claims and that’s it. This coincided with general philosophical suspicion of heavier-weight notions like metaphysical explanation, but times have changed. Both of Hare’s original motivations are relevant, and accounts that don’t make sense of moral explanations fail to capture what motivated the focus on supervenience in the first place. To sum up so far: taking moral principles to be explanatory in role helps provide a unified account of the data points. The next task is to investigate the nature and structure of such principles. Doing so is not straightforward—for as we’ll see, the most common way to formulate moral principles makes it hard to see how they could be explanatory in role, rather than merely in content.

8.4. MORAL PRINCIPLES ARE NOT MERELY E X P L A N A T O R Y IN CO N T E N T Moral principles are commonly formulated as universal generalizations. Berker (2019) defends this view, arguing that “the most naive way of

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

The Metaphysics of Moral Explanations

179

formulating moral principles . . . is also the best, namely one that uses no materials other than a wide-scope necessity operator, standard quantification, mundane indicative conditionals, and the full grounding [i.e. metaphysical explanation] relation” (929). The utilitarian principle is thus formulated as follows (“B” for “Berker”): (UB) Necessarily, an action is required if and only if, and fully because, it maximizes happiness. (Berker 2019: 929)

It’ll be useful to pay attention to some details of this formulation. While Berker says statements like this one are necessitated universal generalizations, or necessitated universally generalized indicative (bi)conditionals, (UB) doesn’t explicitly involve universal generalization. Rather, it uses the indefinite “an action” which (as we’ll further discuss in §8.5.1) can be used in importantly different ways. Let’s nonetheless assume that (UB) can be paraphrased as follows: (UB*) Necessarily, for all actions x, x is required if and only if, and fully because, x maximizes happiness.

(UB*) states a general fact—call it “[UB*]”—that is explanatory in content, but not role: it states that whenever some particular action is morally required, that fact is fully “grounded” in its maximizing happiness.¹⁴ Given the common assumption that universal generalizations obtain at least partly in virtue of their instances, [UB*] obtains (if at all) in virtue of this action being required because it maximizes happiness, that action being required because it maximizes happiness, and so on for each possible required action (perhaps together with a totality fact).¹⁵ [UB*] thus summarizes, and is explained by, instantiations of the grounding relation between particular facts about happiness-maximizing and particular facts about moral obligatoriness—[UB*] itself plays no explanatory role.¹⁶ Indeed, given that metaphysical explanations are irreflexive and transitive, [UB*] can’t explain the facts about moral obligatoriness. This gets us Hyperintensional Humeanism, per above. The problem, however, is that Hyperintensional Humeanism has trouble making sense of supervenience—i.e. data point (i). To see why, let D be the set of Matti’s natural properties. Assuming he’s good, Strong Supervenience allows us to infer that every other possible entity with the properties in D is ¹⁴ Following Berker (following Rosen), we’ll use square brackets to denote facts. We’ll also sometimes use them to denote fact schemata—context will disambiguate. ¹⁵ It’s actually a hard question what the “instances” of [UB*] are, though one we can’t address here. ¹⁶ This is true even when the relevant universal generalizations are “non-accidental” and necessitated. For this reason we are skeptical of various formulations floated by Rosen (2017).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

180

Daniel Fogal and Olle Risberg

also good. But if we take (ii) to be more fundamental than (iii), it’s not clear why that inference is sound. On Berker’s view, there are some natural properties in D such that the fact that Matti has those properties fully explains the fact that he’s good. Call those natural properties D*. Given grounding necessitarianism—i.e. the thesis that if some facts Γ fully explain the fact [Q], then it’s necessary that if Γ obtain then [Q] obtains—it follows that, necessarily, if Matti is D* then he is good. But it doesn’t follow that if, say, Folke has the properties in D (and thus the ones in D*) then Folke is good. That’s because Berker takes the grounding relation to hold between wholly particular natural facts (e.g. that Matti is D*) and wholly particular moral facts (e.g. that Matti is good). As a result, nothing entitles us to generalize from facts about the natural and moral properties of one particular entity to those of another. In other words, even if the properties in D are repeatable, the subjects instantiating them—i.e. Matti and Folke—remain particular and non-repeatable. And generalizations from facts about one particular to another is precisely what the supervenience thesis captures: if someone with the properties in D is good, then anyone with those properties is good. Of course, if a principle like (UB*) is true for goodness, it follows that goodness supervenes on the natural properties specified by that principle. But the point is that, given Berker’s underlying metaphysics, there’s no reason to expect there to be true principles of that form, and hence no reason to expect supervenience to be true. Whether going from the wholly particular to the general secures such principles depends on what the various patterns among wholly particular facts across possible worlds happen to look like. Hence, the Hyperintensional Humean’s principles (and thus supervenience) end up hostage to a kind of modal miracle. In other words, although principles like (UB*), if true, would secure supervenience, they would do so in the wrong way—rather than being made sense of in a principled fashion, it would still look like a mystery that the Humean mosaic necessarily turns out to give rise to them. This worry resembles some of the traditional complaints in the literature on moral supervenience, though the present situation is in one regard even worse.¹⁷ For while the traditional worry is that it’s mysterious why particular moral facts should necessarily align with particular natural facts in the way specified by Strong Supervenience, the worry now concerns not only those facts but also particular instances of the grounding-relation between them. In response, Hyperintensional Humeans might appeal to a principle like Formality (Rosen 2010: 131). Simplifying somewhat, this principle states

¹⁷ See e.g. Skarsaune (2015: 267).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

The Metaphysics of Moral Explanations

181

that [a is F] fully grounds [a is G] only if any fact of the form [x is F] fully grounds [x is G]. Given Formality, the fact that Matti is good because he is D* entails that Folke is also good because he is D*. However, the same problem seems to arise. For why should Hyperintensional Humeans expect Formality to be true? Given their other commitments, they cannot consistently claim that this general principle (or something like it) is itself explanatory in role. Instead, whether Formality is true would also seem to depend on what the modal mosaic happens to look like. Rather than vindicating Strong Supervenience, then, Formality is hostage to the same sort of modal miracle. These problems are avoided if moral principles are explanatory in role. On this view, Matti’s being D* doesn’t fully explain why he’s good—a general moral principle also plays a role. If particular facts about goodness are always partly explained by such principles, it’s clear why we can infer that Folke, who is also D, must likewise be good. So we not only get Strong Supervenience but also a counterpart of Formality—at least in the moral case—for more or less the same reasons. The arguments just given rely on several assumptions which could be rejected, such as that (necessitated) universal generalizations are explained by their (non-necessitated) instances, that (UB) is correctly paraphrased as a universal generalization, and so on. But once these assumptions are rejected, it becomes unclear what the Hyperintensional Humean view is, and whether the main ingredients in principles like (UB)—i.e. “standard” quantification, indicative conditionals, and so on—are really as “naive” as Berker claims them to be. We won’t consider those questions here. Instead, in what follows we’ll focus on two accounts that treat moral principles as explanatory in role, rather than merely in content.

8 . 5 . T H E N O M I C V I E W O F M O R AL E X P L A N A T I O N S The nomic view of moral explanations is modeled on what we’ll call the lawbased view of grounding explanations. By “grounding explanations” we mean cases in which a particular fact obtains in virtue of others (its “grounds”), with the latter being more fundamental than the former. Standard examples include: (i) Mental facts obtain because of neurophysiological facts. (ii) The fact that the ball is red and round obtains in virtue of the fact that it is red and the fact that it is round. (iii) Socrates was pale because he was this specific skin tone.

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

182

Daniel Fogal and Olle Risberg

The nature of grounding explanations is a matter of controversy, however. One issue concerns the existence of general metaphysical principles, or “laws,” and what role (if any) they play. According to the law-based view, metaphysical laws play an ineliminable role in grounding explanations. On this view, the full metaphysical account of, say, Socrates’ being pale extends beyond his having a certain skin tone—the general fact that having that skin tone makes one pale is also relevant. Even if metaphysical laws exist and play an explanatory role, the precise role they play is disputed. Should we treat laws as part of grounding explanations, for example, or instead as merely “underlying” or “backing” or “governing” such explanations? (And what’s the difference?) We’ll set such questions aside. What matters for our purposes is what unites lawbased theorists—namely, that metaphysical laws play some essential role in metaphysical explanations. The law-based view of grounding explanations can thus be seen as a generalized version of the tripartite view of moral explanations above: Grounds: particular fact(s). Metaphysical Law: general explanatory principle. Explanandum: particular fact.

Discussions about grounding explanations are complicated, however, by the fact that the word “ground(s)” is often used in different ways. For some, for A to ground B is for A to metaphysically explain B, whereas for others, for A to ground B is for there to be a metaphysical relation between A and B that “backs” explanations.¹⁸ What’s more, many theorists—including Berker— take the grounding relation to hold between particular facts. But on the lawbased view, it’s a mistake to focus exclusively on relations between particular facts, just as on law-based views of causal explanation it’s a mistake to focus exclusively on relations between individual events. In both cases laws also matter. To minimize confusion we’ll mostly avoid using “ground(s)” as a verb, preferring instead to use it as a noun to pick out the first ingredient of grounding explanations (per above) on analogy with the use of “cause(s)” as a noun. (This terminological issue will re-arise in responding to Berker in §8.5.1.) While the law-based view of grounding explanations is plausible, we won’t be defending it here.¹⁹ Our primary concern is instead with the nomic view of moral explanations. What motivates the nomic view is the idea that the general principles or “laws” that help explain moral facts should ¹⁸ Raven (2015) calls the first view “unionism” and the second view “separatism.” ¹⁹ Instead, see Kment (2014), Wilsch (2016), Glazier (2016), and Schaffer (2017a, 2017b).

OUP CORRECTED AUTOPAGE PROOFS – FINAL, 7/7/2020, SPi

The Metaphysics of Moral Explanations

183

be understood in the same way as those thought (rightly or wrongly) to figure in grounding explanations elsewhere. The main difference is that moral laws involve moral properties whereas non-moral laws don’t—a difference in content rather than form. Although the nomic view is a natural companion to the law-based view, neither entails the other. One might accept the nomic view while rejecting the law-based view in general; alternatively, one might accept the law-based view while rejecting the nomic view (by, for example, denying that moral explanations are grounding explanations). How, then, are metaphysical and/or moral laws to be understood? There are different ways of trying to capture their characteristic features, such as their generality and directionality, but the proposal we find most congenial is from Glazier (2016). He writes: [A metaphysical law] clearly has a sort of generality, but it is a general fact that is not explained by its instances. Since this sort of generality is not achieved through quantification, it must instead be achieved through another variable-binding operator. I therefore propose that we recognize a new operator “