Personal Value

Personal Value

This page intentionally left blank

Personal Value

Toni Rønnow-Rasmussen

1

3Great Clarendon Street, Oxford ox2 6dpOxford 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 inOxford New YorkAuckland Cape Town Dar es Salaam Hong Kong KarachiKuala Lumpur Madrid Melbourne Mexico City NairobiNew Delhi Shanghai Taipei TorontoWith offices inArgentina Austria Brazil Chile Czech Republic France GreeceGuatemala Hungary Italy Japan Poland Portugal SingaporeSouth Korea Switzerland Thailand Turkey Ukraine Vietnam

Oxford is a registered trade mark of Oxford University Pressin the UK and in certain other countries

Published in the United Statesby Oxford University Press Inc., New York

# Toni Rønnow-Rasmussen 2011

The moral rights of the author have been assertedDatabase right Oxford University Press (maker)

First published 2011

All rights reserved. No part of this publication may be reproduced,stored in a retrieval system, or transmitted, in any form or by any means,without the prior permission in writing of Oxford University Press,or as expressly permitted by law, or under terms agreed with the appropriatereprographics rights organization. Enquiries concerning reproductionoutside the scope of the above should be sent to the Rights Department,Oxford University Press, at the address above

You must not circulate this book in any other binding or coverand you must impose the same condition on any acquirer

British Library Cataloguing in Publication DataData available

Library of Congress Cataloging in Publication DataData available

Typeset by SPI Publisher Services, Pondicherry, IndiaPrinted in Great Britainon acid-free paper byMPG Books Group, Bodmin and King’s Lynn

ISBN 978–0–19–960378–7

10 9 8 7 6 5 4 3 2 1

To Elly,Anna, and Sofia

Preface

Values accrue to objects. They are not free-floating entities that exist independently ofother things: necessarily, they are ‘carried’ by objects in a wide sense of this term.Clearly, not all objects carry the same kind of value; and some (perhaps most) havemore than one kind of value. Some objects have only minor value, while others seemworth fighting hard for. Moral values such as justice or equality belong to this lattercategory. Some valuable objects—whether material in nature, such as the concertticket we have kept all these years because something extraordinary happened thatevening, or something less tangible, like having a certain memory whenever you hear aparticular song on the radio—are distinctly connected with one person. They carryspecial value for that person. Valuable objects like these are often historical in character,tying us to some particular event in our past, but they need not do so. What is valuablemight rather be a future state of affairs. It may be something you strive hard to realize(say, fame or clearing your name in a scandal); or again, it may be neither a materialthing nor an abstract entity (such as a state of affairs) but rather involve some person—your children, parents, spouse, or friends. Some of these persons and objects areobviously very important to you; others carry only minor value for you. The discoverythat your dog has eaten the concert ticket will probably have no real, or at least noenduring, impact on your life. Of course, people tend to react differently, and it isprobably true of many objects that most of us consider to have trifling value that theirloss or damage would make a major difference in a person’s life. Here we differ, andthere is obviously room for disagreement about how much we should value suchobjects. In fact, we might even think that something can be of great value to a personeven if it plays no actual role at all in that person’s life. I am inclined to believe that it isat least an open question what to think about a case like the following: unknown toyou, you have a living child (either because you never realized you were a parent, orbecause a child you thought had died is in fact alive). I imagine that, in many people’seyes, an unknown child would nevertheless be of value to its oblivious parent. Thereare certain patterns of behaviour, and certain attitudes, that we would expect the parentto display as a matter of what is ‘fitting’, or of what one has reason to do, when onelearns of the existence of a child of one’s own. So the impact of an object, or person, inone’s life is not necessarily (or so we may suspect) a measure of the value the objectactually has for someone. The role of values is different: they function, generallyspeaking, as reason-grounded incentives or disincentives to action and the formationof attitudes. Objects like those mentioned set more or less negotiable boundariesdictating what we must, should, may, or have reason to, do or not do with them. Insome cases we ought perhaps to help another person to achieve fame, or see to it that

someone’s name is cleared, or simply respect a person’s wish to live life as she thinks fit.It may be that we should have overcome our fear, stuck fingers into the dog’s mouth,and saved the ticket. Of course, we should not do so for the sake of the ticket but ratherfor the sake of the person (and perhaps for the sake of the dog, too) whose value wethink is at stake here. It may be that the thing to do for the sake of your friend is to buyher record, or to stop people talking behind her back. Personal values, as I will refer tothese sorts of value, are what this book is about.

Any attempt to analyse personal value will at some point have to face the following twoquestions: In what sense is personal value a kind of value? And (if it is a value) in whatsense is this value personal? I will suggest that personal value is in fact a kind of value.One way to elaborate this suggestion is to say that in ascribing personal value tosomething we are, as I will claim, actually evaluating. That is, we are employing aparticular notion of value, and one that we standardly express with the help of thefollowing bona fide locutions: value-for, good-for. Now, I might be mistaken aboutthis in more ways than I care to think. For instance, I may be making an evaluativeblunder in saying that the examples I will invoke are in fact examples of value-for.Since I will not be getting involved in substantive value theory, I will have to choosemy examples carefully so that they do at least appear to be plausible bearers of personalvalue—that is, bearers of something that most of us would be ready to say is good for orvaluable for themselves or someone else. Here I will have to trust my intuitions.(Needless to say, this will also be the case when it comes to impersonal values.)Equally, I might make another kind of mistake. I want to work with the assumption

that there is a kind of evaluation that is conducted in terms of notions such as good-for,but it might be objected that there is no such notion as good-for, and that thereforegenuine evaluation cannot make use of that notion. Whenever we speak about what isgood for person a we are merely reporting on what is good according to a, or from a’sperspective. We are not committing ourselves evaluatively, in the sense of agreeing ordisagreeing with what we take to be the view of a.Sometimes, certainly, good-for judgements are indeed like this; and when they are

they should perhaps be enclosed in scare quotes. But I am convinced that there areother times when we are genuinely evaluating, genuinely attributing value-for. Andthe two kinds of case must not be confused. The occasions on which we genuinelyevaluate what is good-for someone are not rare. The best way to show this, as far as Iknow, is to present a plausible analysis of the more questionable notion.A caveat is now required regarding the verb ‘to value’ and my use of it. What

precisely we are doing when we value something will in this book be left open. This is,of course, in some ways regrettable, but it would be even less satisfactory to set outfrom something that is not yet settled.1

1 This work is a contribution to mainly formal value theory. Traditionally such theory admits of tworelatively independent approaches. On the one hand, we have a number of semantic theories about the

PREFACE vii

What I will be proposing in this work, then, is that we should employ a certainpattern of value analysis, one which has recently attracted much attention and whichhas roots in the writings of the Austrian philosopher Franz Brentano. If we fine-tunethis kind of value analysis it will come to cover personal values, too. I believe thesegoals can be achieved. A fortiori such an achievement supports the idea that we aredealing with a genuine value (because, in important ways, the finely tuned analysisstructurally resembles the way in which values have been analysed according to aplausible pattern). Moreover, I think this approach has many advantages. Since myproposal is novel, my concern is less with evaluating rival proposals and more witharticulating and defending my own positive view.

In trying to understand value-for it will be helpful, as I mentioned, to have a bettergrasp of value, period. But anyone who has followed value theory in recent years willknow only too well that there are a number of more or less interrelated value concepts.Therefore, it will be necessary at the outset to outline the more important andfundamental ones. So before elaborating my views on the particular pattern of valueanalysis I endorse, I will introduce, and distinguish between, various kinds of value.

The presentation of a value taxonomy often leads to discussion of the nature ofthings that bear value. That there must be an object of some sort that value accrues to issomething over which value theorists, with very few exceptions, agree. Most philo-sophers would also say that many different things can be ascribed value. Thus, ineveryday talk we speak of physical objects such as paintings bearing value (e.g. beauty),but we also refer to events, facts (e.g. that it is more valuable that two persons feel wellthan that one person feels well), places, times, and so on, as having value. Theconsensus that exists today on this matter—i.e. the wide agreement that there is arelation between value and something that bears this value—does not, however,dissipate when we ask questions about the nature of these value bearers. While somephilosophers argue that just about everything can be a bearer of value, others havetaken a much more restrictive view, and, in effect, allowed only one kind of thing tobear value.

Investigation of the metaphysical nature of value bearers may at first glance seemrather inconsequential. However, I do think there is reason to more carefully look atthis issue, and not only to satisfy a philosopher’s curiosity. In fact, one of the things Iwill argue in Chapter 10 is that, by excluding certain types of value bearer, we limit thesubstantive views available. Equally, by endorsing certain types of substantive theory of

meaning of our value terms (e.g. versions of cognitivism, non-cognitivism, expressivism). But not everyformal value theory makes it a priority to improve our understanding of the ‘semantic meaning’ of valueterms. What I will be doing in this work falls, for instance, beyond the scope of these semantic theories.Rather, the general aim of work of the kind undertaken here is to reach an understanding of value (inparticular, final personal value). A good way of doing this is to exhibit structural connections that exist in thefield of value, and in particular those holding between our central evaluative concepts. In other words, thefocus is the logic of value: the relevant conceptual distinctions, the relations between different types of value,and the relations between value concepts and deontic (normative) notions such as ‘ought’, ‘reasons’, and‘fittingness’.

viii PREFACE

value, we will quite naturally be led to certain positions regarding the nature of valuebearers. I suspect both of these claims will be looked at with some scepticism.This work easily divides into three parts. First, there is the scene setting of Chapter 1.

This is followed by a much longer part, Chapter 2 to Chapter 8, which contains theanalysis, together with discussions more directly concerned with my proposal. Thethird part, consisting of Chapter 9 and Chapter 10, discusses some rather different ideaswhich have had a significant influence on my analysis. The notion of personal valueplays only a minor, indirect role in these last two chapters. Since I believe this book tobe somewhat dense, I shall now brief the reader on the main structure.In Chapter 1, I introduce some contrasting notions of value. G. E. Moore’s

influential views on intrinsic value, as well as his much employed, so-called isolationtest, are discussed in some detail. This test, as I shall explain, might have done more badthan good to value theory in general. At least, it may well have blinded value theoriststo a value notion that is central to both the chapter and the book—namely, the notionof final value, or what is valuable for its own sake. This kind of value plays an importantrole throughout the book. Chapter 1 also discusses the common claim that values aresupervenient features: the notion that, whatever values are, they are dependent onsomething else, some subjacent set of properties (which set is generally taken to containnatural properties). In particular, I disentangle two intuitions involved in the idea thatvalue is a supervenient feature which, regrettably, are often lumped together. These arewhat I call the ‘consistency intuition’ and the ‘dependence intuition’. Although closelyrelated, these intuitions seem to be different in nature. My discussion leads me, amongother things, to identify what is missing in Jaegwon Kim’s formulation of super-venience. In this work, supervenience plays a special role, in that it is one of the keyelements I employ in order to get clear about the often misconstrued distinctionbetween subjectivism and objectivism about values. Since I take it to be an advantageof my analysis of personal value that it is essentially neutral on this vexed issue, I want tomake clear at the outset just how I conceive of the distinction between subjectivismand objectivism. So much has been written on this topic that it would require a verysubstantial work or two to do it justice. I have decided, therefore, to set out, in as muchdetail as is possible in the present discussion, just how I conceive of the debate, and toleave the writing of the substantial books to someone else. This is not a very nice thingto do to my forerunners, and I can only hope that they will bear with me. The twocore ideas at play in this discussion are those of supervenience and value constitution.The notion of value constitution is a particularly tricky one, and I am painfully awarethat much work remains to be done here. The ‘Invariance Thesis’, the claim that whatis valuable for its own sake is invariant over possible worlds, is often thought to serve asa line of demarcation between subjectivists and objectivists. However, I argue that theInvariance Thesis, which often divides value theorist into two camps, is only reasonablegiven certain assumptions about the nature of the bearer of value. Once we take intoconsideration the fact that the supervenience base might contain non-essential proper-ties it becomes clear that the thesis should be rejected.

PREFACE ix

Chapter 2 opens with a discussion of the fitting-attitude analysis, or buck-passingaccount (as it is also called), that I will be employing.2 Besides commenting on two ofits more important architects, Franz Brentano and A. C. Ewing, I outline the centralfeatures of the analysis and consider some possible objections to it. One objectionstands out as being particularly troublesome to this pattern of value analysis. If we wantto distinguish the right kind of reason from the wrong kind (where only the formerjustify value ascriptions), which it seems we must do, we ought to be able to say whatmakes a reason right or wrong. Unfortunately this has proven to be a rather stubbornproblem. Chapter 3 is devoted to it. My suspicion is that fitting-attitude analysts face afurther, related problem—one that concerns whether we should regard value as areason-providing property on its own. At the end of Chapter 3 I explain why I thinkthis might, in effect, be the case. In the next chapter I formulate my analysis of value-for. I also consider some influential views on personal value. The first, and mostimportant, is G. E. Moore’s famous attack on the very notion of a personal good.Another approach that I examine is a competitor to my own position. In this alterna-tive, value-for is understood in terms of the de facto attitudes of its beneficiary (i.e. theperson whose value it is). In discussing these views I take the opportunity to clarify myown analysis on several important points.

The normative has primacy, according to my analysis, over the evaluative: value isexplained in terms of what we have reasons to favour. The reason element in theanalysis carries much of the weight in this view (but not all of it, as we shall see). Mysuggestion is that the analysis of personal values requires a special kind of attitudetype—what I refer to as a for a person’s sake-attitude. Such attitudes are both verycommon and important; they have received less attention than they merit. In Chapter5 I describe what I believe is a defining feature of these peculiar attitudes. To do this,I introduce a distinction between two roles that properties may play in the intentionalcontent of such attitudes.

The general aim of Chapter 6 is to deepen the analysis, and I seek to do this primarilyby discussing a range of issues which, I believe, my analysis accentuates. The firstconcerns the possibility that more than one sense, or notion, is associated with theexpression ‘good for’. One notion in particular interests me here, namely that whateveris welfare-constitutive is good for the agent. A second issue concerns the extent towhich the analysis permits us to identify intrinsic personal values. I also ask to whatextent personal values can be treated as ‘private’ values once they are understood interms of agent-relative reasons for action. Finally, I consider a more complex matter.This is the question whether universalizability seriously undermines an analysis likemine. This alleged problem has nothing to do with the fact that the notion of reasonemployed in the analysis is likely to be agent-relative. It is generally recognized that

2 There might be something in the idea that these two analyses differ in some respects. For instance, itmight be that, as a matter of fact, they have been applied to different things. However, in this work I will notseek to separate the two notions.

x PREFACE

agent-relative reasons are as universalizable as supposedly neutral ones. The obstacleseems to arise elsewhere, and in fact it turns out to be connected with the peculiarnature of the for a person’s sake-attitudes. Since at least some of these take non-fungible(i.e. non-replaceable) persons as their intentional objects, they are not obviouslyuniversalizable. There is no cause for alarm, though. The chapter concludes by high-lighting some plausible responses to this challenge. Chapter 7 is devoted to discussionof the views of Thomas Hurka, Michael J. Zimmerman, Donald Regan, and ConnieRosati. Hurka and Regan have launched separate attacks on what they take to be anumber of confused ideas about good-for. I try to respond to these. I also comment onRosati’s and Zimmerman’s recent analyses of good-for. In particular, I explain why Iwish to distance my views from Rosati’s suggestion as to how we should understandpersonal value. Moreover, I explain why I find Zimmerman’s account of good for interms of benefit not promising. In Chapter 8, I consider remaining objections to myanalysis. I also look at Stephen Darwall’s theory of welfare, and at an objection tofitting-attitude analyses of good-for recently raised by Chris Heathwood.It will probably prove useful in what follows to keep in mind that, throughout the

present book, I understand reasons as what are sometimes called ‘good’, ‘justifying’, or‘normative’ reasons for action or the adoption of an attitude. The contrast here is withexplanatory reasons for action and attitudes (and also with motivating reasons foraction). The precise nature of these normative reasons will be left open in this work.However, I will take them to be facts favouring action or the adoption of an attitude.In other words, I am inclined to share the views of the large number of philosophersthese days who think the notion of a normative reason defies non-circular analysis, andwho therefore treat it as a primitive. Of course, this does not mean that we cannot tryto improve clarity when it comes to normative reasons; and in fact on one particularissue I hope to do just this. Thus in Chapter 9 I argue—against what I myself held for along time—that, given plausible assumptions, all normative reasons are agent-relative.If I am right about this, a number of things follow about the dichotomy betweenpersonal and impersonal value. For one thing, obviously this distinction cannot now beaccounted for in terms of the distinction between agent-relative and agent-neutralreasons. We will have to look elsewhere if we want to understand these two kinds ofvalue. Hence my suggestion that the analysis of personal values requires the special kindof for a person’s sake-attitude.Prior to Chapter 10 I assume that very different kinds of thing can be valuable.

I think this assumption is reasonable, but some value theorists have expressed reserva-tions about such pluralism. In Chapter 10 I therefore present an argument that hasimplications for the question whether we should be monists or pluralists when it comesto the bearers of final non-derivative value. The idea is quite simple. The argument setsout from the notion that a formal analysis of value should be accessible to as manyplausible, substantive value theories as possible. A pair of commonly held, substantiveviews about value—preferentialism and hedonism—will be considered. Much speaksin their favour: pleasure seems quite obviously to be something of value and many

PREFACE xi

people would agree that the satisfaction of preferences is valuable. I argue that, givencertain plausible assumptions about the nature of pleasure and preference, the valueaccruing to pleasure (from a hedonist point of view) cannot be accounted for bypreferentialists. This will probably strike many observers as a rather incredible idea.However, my reasoning is quite straightforward: the two positions are most plausiblyregarded as ascribing non-derived final value to different kinds of value bearer (even ifthey ascribe derived final value to the same sort of bearer). Since these positions do notascribe non-derivative value to the same kind of value bearer, we have reason toendorse a formal view of value that is consistent with value-bearer dualism.

The focus of this book, to repeat, is not value in general but personal values. I shouldstress that by ‘values’ here I do not have in mind that which is valued (a new carperhaps, friendship, or medical care). I have in mind rather that which is ascribed tosuch things, namely value and value-for. Personal values accrue to objects in virtue ofthe fact that those objects have value-for, or are valuable to, someone—not in virtue ofthe fact that those objects have value simpliciter, or are valuable, period. The objects areconsidered not just valuable but valuable for someone. This suggests that the expression‘to value’ is subject to an ambiguity. As well as thinking that objects are good or bad,we may also think they are good-for or bad-for someone. Sometimes we value in terms ofthe impersonal ‘to value’ (examples include judging that something is good andjudging that something is bad). Sometimes—at least, this is the contention I shall tryto render credible in this work—we value in terms of what is ‘good-for’ or ‘bad-for’somebody.3 In the latter cases we are valuing something with someone in mind; in theformer we are valuing something with an eye to nothing but the object of value.

3 Valuing something need not involve so-called thin value concepts such as ‘good (for)’ or ‘bad (for)’;often it will involve only thick value concepts such as ‘admirable (for)’, ‘loveable (for)’.

xii PREFACE

Acknowledgements

A number of people made the work on this book a much more pleasant, albeit notnecessarily easier, endeavour than I had anticipated. One person in particular whoshould be mentioned is Wlodek Rabinowicz. For about a decade he has been myregular lunch companion, and the generosity with which, over the years, he hasserved me with new ideas, arguments, and philosophical know-how simply cannotbe overestimated. (My weight gain is the most tangible proof of who has been doingthe talking and who has been doing the learning.) So entangled are his views withmy own that I am quite certain they do not appear explicitly in enough places in thisbook. Were it not for my bad memory, and my equally bad custom of transformingclear ideas into less clear ones, I would have mentioned his name a good deal moreoften.I have been fortunate in other ways, too. David Alm, Johan Brännmark, and

Michael Zimmerman, who read and criticized earlier drafts of this book, providedme with invaluable help and encouragement. They forced me to clarify the expositionand sharpen many of my arguments. I also wish to thank Erik Carlson, Finn Collin,and Jan Osterberg, who were subjected to readings of a very early version. Their kindcomments were greatly appreciated. I have had a privileged working situation. Myoffice at the Department of Philosophy, Lund University, lies between the offices ofmy two good friends Bjorn Petersson and Dan Egonsson. Over the years I have spent atremendous amount of time in Dan’s room, discussing more or less every aspect of life.His original and down to earth way of approaching any subject in philosophy, andespecially my metaphysical problems, has been a constant reminder that I now and thenought to put on the brakes and reflect on what I am actually doing. Over the years Ihave also learned much from my discussions with Bjorn about many of the book’sthemes. His gentle way of managing the department makes Kungshuset such a splendidworking place.Core ideas of the book were presented at various places during a hectic 2004: thus I

want to thank the members of a workshop I attended at the Swedish Collegium forAdvanced Study in the Social Sciences, Uppsala, and in particular David Brax, JohanBrännmark, John Broome, Krister Bykvist, Erik Carlson, Sven Danielsson, MargaretGilbert, Jonas Olson, Wlodek Rabinowicz, the late Howard Sobel, and DanielSvensson; thanks are also due to Michael Stocker and David Chan, and the otheraudience members who attended my talk at the University of Wisconsin, StevensPoint. (David did a great job in organizing such a splendid conference.) Kind invita-tions from the University of Stirling and King’s College, University of Aberdeen, werefollowed by stimulating discussions, and for this I am especially grateful to Peter

Baumann, Michael Brady, Rowan Cruft, Adrian Haddock, Alan Miller, and othermembers of these audiences. In Krakow, in 2008, I attended the European Congress ofAnalytic Philosophy and a workshop arranged by Kevin Mulligan and WlodekRabinowicz. I received many acute comments on topics dealt with in the book frommembers of the audience, some of which I have already mentioned. In particular,however, Jonas Olson, Christian Piller, and Andrew Reisner gave me much to thinkabout. Later that year I was invited to give a talk at the meeting Metaphysics of Value,(the first Eidos-Thumos-Episteme meeting, convened in Geneva), and I am gratefulboth to the organizers and audience at that event. The latter gave me valuable feedbackon a couple of arguments that ended up in the book; I owe particular thanks to GrahamOddie, Olivier Massin, Kevin Mulligan, and Nishi Shah. I was also grateful for theresponses I received from Albert Musschenga, Robert Heeger, Simon Kirshin, MichaelQuante, and Andreas Vieth during a visit to Münster in 2010.

Lunches and dinners have played a most nourishing role in the work on this book.I had some very rewarding ones in Lisboa, and later on, with Jonathan Dancy, inSkanor. I am also grateful for the advice and encouragement I received from KevinMulligan during our occasional meetings—would that I possessed his knowledge of theAustrian–German value theorists! I learned much from correspondence with JohnSkorupski. I was honoured by his talk in Lund, and our discussion about reasonsover lunch in Copenhagen’s Glyptotek the day after it helped me steer clear of somebad ideas. Åsa Wikforss and I had some not-so-good sushi in Lund, but fortunately thedialogue was all the better; she gave me insights into areas of philosophy of which I amquite ignorant. Michael Zimmerman merits a special place in this list. My friendshipwith Michael goes back many years now, and I am really very thankful for his greatsupport and excellent advice, and for the arguments we have had on a wide range oftopics over the years.

I have learnt much from the people at my department, especially those who attendthe weekly Higher Seminar in Practical Philosophy. I have also benefited fromoccasional comments and email correspondence with a number of people over theyears. I cannot thank all who merit my gratitude, in part because of their number, butchiefly because my memory does not allow me to recollect all of the many times thatI have benefited in this way. But I can at least mention some: Gunnar Bjornsson, FredFeldman, Fritz-Anton Fritzson, Alan Goldman, Christopher Grau, Ish Hadji, ChrisHeathwood, Noah M. Lemos, Ia Maurin, Derek Parfit, Ingmar Persson, Douglas W.Portmore, Connie Rosati, Tim Scanlon, Mark Schroeder, Caj Strandberg, FolkeTersman, Alex Voorhoeve, and Heath W. White.

Comments sent by two anonymous readers from Oxford University Press were veryhelpful in the later stages of revision. Paul Robinson polished the English in this work,for which I am very grateful. My work was also supported by generous grants from TheBank of Sweden Tercentenary Foundation and the Swedish Research Council.

This work includes, with the kind permission of Springer Science and BusinessMedia, revised portions of my articles: ‘On For Someone’s Sake Attitudes’ (2009b),

xiv ACKNOWLEDGEMENTS

‘Normative Reasons and the Agent-Neutral/Relative Dichotomy’ (2009a), ‘Love,Value and Supervenience’ (2008b), ‘Buck-Passing Personal Values’ (2008a), ‘AnalysingPersonal Values’ (2007a), ‘Instrumental Values—Strong and Weak’ (2002b), and‘Hedonism, Preferentialism, and Value Bearers’ (2002a).

I dedicate this book to my wife and our two daughters.

ACKNOWLEDGEMENTS xv


Contents

Preface viAcknowledgements xiii

1. Types of Value 11.1 Preparing the ground—different value concepts 21.2 Isolating intrinsic value 31.3 Isolating extrinsic final value 51.4 Competing notions of intrinsic value 71.5 The source of value 91.6 Supervenience 101.7 The Invariance Thesis 121.8 Constitution 131.9 Subjectivism and objectivism 14

2. Fitting-attitude Analysis 192.1 The origins 202.2 A. C. Ewing 222.3 The primacy of the normative over the evaluative 242.4 Different value idioms 252.5 The FA analysis of final positive values 272.6 W. D. Ross’s objection 30

3. The Wrong Kinds of Reason 333.1 The wrong kind of reason problem 333.2 Dual-role attitudes 373.3 Solving the WKR problem? 383.4 Danielsson and Olson on the WKR problem 403.5 The biconditional buck-passing account 42

4. Mistaken Value Analyses 464.1 Moore’s objection to good-for 494.2 Instrumental value 514.3 A bad approach to personal value 53

5. For Someone’s Sake 555.1 Types of attitude 575.2 Sake, end, and objective 605.3 Discerning attitudes 635.4 Identifiers and justifiers 685.5 Two kinds of FSS attitude 735.6 Summing up 75

6. Examining the Analysis 776.1 Dropping an FSS attitude 786.2 Non-fungible persons and identity 806.3 Good-for and welfare 836.4 Two sorts of personal value 866.5 Personal intrinsic value 886.6 Agent-relativity and private values 916.7 Personal value and universalizability 92

7. Mo(o)re Objections 957.1 Hurka’s views on good for 967.2 Regan’s views on good for 987.3 Rosati and good occurring in a life 997.4 Personal and impersonal value 1047.5 Thin-thin and thin-thick conceptions 105

8. Problems and Possibilities 1098.1 Janus values 1108.2 Some counter-examples 1118.3 Darwall on welfare 1158.4 Different strategies 1188.5 Heathwood’s objection 1218.6 The good life and the argument from fetishism 122

9. One Reason Dichotomy Less? 1269.1 Introductory notes 1279.2 The essentialist sense 1289.3 The number approach 1319.4 Reason-for 1329.5 Meeting the challenge 1379.6 An overcrowded boat 1409.7 More replies to the challenge 1419.8 A positive argument 144

10. Value Bearers and Value Pluralism 15210.1 Extrinsic final values 15310.2 Hedonism and preferentialism 15610.3 Preferentialism 15810.4 Value bearers 15910.5 Value-bearer monism 16010.6 Simplicity 16110.7 Some reductions 16310.8 Separating the concrete from the abstract 16610.9 Recapitulating 168

Bibliography 172Index 181

xviii CONTENTS

1

Types of Value

Certain things we value are best described as possessors of impersonal value. That theworld is just, and that there are more happy people than there are unhappy people, arestates of affairs that are, in a clear enough sense, impersonally valuable: such states ofaffairs are valued, when they are, in themselves, without reference to anything else.Other things appear to carry a different kind of value: they are seemingly personalrather than impersonal in nature. They are examples of value-for rather than value,period. For instance, some time ago, tidying my desk at home, I found a poem in adrawer that Anna, my oldest daughter, wrote for me when she was just a young girl.It was a typical child’s poem. Despite the fact that my room was cluttered with paper,I could not throw it away. The piece of paper, with its few lines of Swedish, had somevalue for me, i.e. it was a value that related in some way to me as a person in a wayother valuable things do not. It was clear to me that although this poem carried value, itwas not the sort of value that people are ready to ascribe to, say, Baudelaire’s Les Fleursdu Mal or a sonnet by Shakespeare. Many people (whether they are right or wrong inthis is another question) take poems, television series, rock bands, special parts of ourplanet, to be valuable. Now it might, of course, be the case that what remains of theAmazon jungle carries personal value for someone, as my daughter’s poem does for me,but whether or not this is the case, it is also quite common to value nature that is largelyuntouched by man in an impersonal way. Wild nature, untouched by human beings, isjust valuable. So is the fact that more people feel well than feel bad, or that justice ratherthan injustice prevails. In contrast with my daughter’s poem, these facts do not carryperson-relative values—they are just valuable, period.This distinction between value, period, and value-for, or—as I shall also be referring

to it—between impersonal and personal value, is just one among many distinctions inthe area of value that philosophers have focused on. However, at this point it isprobably not wise to proceed directly to an analysis of personal value. To set thestage something needs to be said about the various types of value that have played animportant role in modern value theory; personal values will be subsumable under atleast some of these value types.

I will turn to this in a moment. First of all, however, I must enter a minor but notunimportant caveat: in what follows, I shall take the idioms value for and good for to referto personal values. Likewise, ‘value’ and ‘good’ will be terms for impersonal values.‘Good for’ will be taken to refer to a positive value for, ‘good’ to a positive value, ‘bad’and ‘bad for’ to negative values. (There are other, so-called ‘thick’ evaluative terms,such as ‘desirable’ and ‘admirable’. We shall consider these later, but for the moment Iwish to focus on ‘value’ and ‘goodness’.) Often the context in an example beingdiscussed makes it quite clear that ‘value’ refers there to ‘positive value’, but sometimesI shall not specify the sense in which I am using the word ‘value’. The benefits ofmaking it clear, on each occasion, what kind of value—positive, negative, or neutral—I have in mind are for the most part extremely limited. Little clarity is gained, I havefound, from such an approach.

‘Value for’ and ‘good for’ (and their negations) are expressions that can reasonably beexpected to carry more than one sense, and it is quite likely that the analysis I present inChapter 2 and on will fail to suit all the different senses of these expressions. Much lateron I will discuss a couple of these, as it were, errant senses; although at least some ofthem are, from a value-theoretical perspective, less interesting, they need to becommented on.

1.1 Preparing the ground—different value conceptsThere are many ways to sort values. In fact, philosophers of value in the nineteenthcentury concerned themselves a great deal with sorting tasks. Values were organizedinto more or less important sets of subject matter, such as moral, aesthetic, cultural, andnatural values; they were ranked and placed along one scale or another. For instance,higher values would typically be those which (in contrast to lower values) play, orought to play, an important role in our lives. However, none of these tasks will be minein this work. My concern is a different one.

The value notions that I want to draw attention to next are in some respectsphilosophers’ phrases of art. Notwithstanding this, many of these notions do play animportant role in people’s everyday reasoning about what is valuable. Parents (andgrown-up children, for that matter), are likely to recognize, for instance, the perhapsmost common of all value distinctions—namely, that between something being valuableas a means and something having final value. The child who asks why she must brush herteeth, and who is not satisfied with ‘It is what every good boy or girl ought to do!’ willsoon turn the conversation into a search for the final value, or end, that makes it nolonger necessary to ask, once again, ‘But why is that good?’ Why is having clean teethgood? Why is fresh breath good? Why is it valuable to be liked by one’s fellow men?

Typically (I know I did), people end up appealing to the pain connected withtoothaches, or the pleasures that come from having healthy teeth, or both.

The history of the distinction between something’s being valuable as a means tosomething—what is commonly referred to as ‘instrumental value’—and something’s

2 TYPES OF VALUE

being valuable, not as a means, but for its own sake, is impressive.1 Nonetheless, it is adistinction which, over the years, has been expressed in a number of ways. Consideragain the earlier example. Pleasure is one of the most common items invoked to put anend to iterated ‘But why is so-and-so valuable?’ questioning. At least many people doseem to think it is unnecessary to continue asking why experiencing pleasure issomething good. It seems pointless to search for some consequence of one’s feelingpleasure that would somehow justify our conviction that pleasure is valuable.2 In thelight of this axiologists3 have said that when we value pleasure in this sense we value it‘for its own sake’ or ‘for the properties it has in itself ’, or ‘as an end’, or ‘in its ownright’, instead of valuing it as a means to something else or for its consequences. In fact,recently philosophers have seriously pondered the possibility that these distinctionsshould be taken more seriously than has historically been the case. But it is not only‘end’ values that have been described differently. Instrumental values, also, though to alesser extent, have received different descriptions, as we shall see later on.

1.2 Isolating intrinsic valueOne major issue in recent value theory is whether something’s being valuable for its ownsake (that is, what I will refer to as final value) and something’s being valuable for theproperties it has in itself (what I shall call intrinsically valuable) are coextensive. Up untilrecently, the distinction barely attracted attention. One major reason for this is,perhaps, that G. E. Moore, in his seminal Principia Ethica (1903a), regarded this divisionas being of little importance. To value something in virtue of its internal properties is tovalue it for its own sake and, vice versa, to value something intrinsically is to value it forits own sake, and not, say, for the sake of its consequences.4

To understand why such an approach seemed perfectly legitimate for a long while,we need to say something more about G. E. Moore’s views on value. He went to quitesome pains in the Principia to determine the nature of intrinsic goodness, but arrivedsomewhat paradoxically at the position that the concept defies analysis. And by‘analysis’ he had in mind a quite particular activity. To analyse a concept was to

1 The distinction appears in the Timaeus, where Socrates asks what is good as such (pleasure). For morehistorical examples, see the introduction to Recent Work on Intrinsic Value (2005), by myself and MichaelZimmerman.

2 Although people generally have fairly strong hedonist instincts (an odd exception to this being one ofPlato’s nephews, who considered both pain and pleasure to be evils), it is easy to imagine examples that createproblems for this kind of intuition. Think of someone who finds pleasure in torturing another person. Again,what should we say about a case where someone finds it pleasurable when he falsely thinks he is makingsomeone suffer (unaware that his victim is in fact not suffering)?

3 The term ‘axiology’ is used here as a synonym of ‘value theory’; apparently it was given this sense by PaulLapie in his doctoral thesis Logique de la Volonte (1902). Two other early references are W. Urban, Valuation:Its Nature and Laws (1909) and E. von Hartmann, Grundriss der Axiologie oder Wertwägungslehre (1908).

4 Moore does not actually express himself in this way. He rarely applies the phrase ‘for its own sake’ togoodness. There are many works in which final and intrinsic value are treated as the same sort of value. See,for instance, Gregory Vlastos, ‘Human Worth, Merit and Equality’ (1969).

1.2 ISOLATING INTRINS IC VALUE 3

break it into simpler concepts, but, as he argued, not all concepts are capable of such atreatment. We cannot, he claimed, do it with a colour concept such as yellow, but wecan do it with complex notions such as, say, a mule or a mare. Goodness is like yellowin the sense, he thought, that it cannot be broken into parts. What we can say, hethought, about what is intrinsically valuable is that it is a value which depends on theinternal properties of its bearer, i.e. the thing which is valuable.5

In due course we shall return to the idea that value is a dependent, in fact, superve-nient, notion. Meanwhile, it must be stressed that Moore did provide his readers withguidelines to improve their grasp of intrinsic value, and, in particular, of what things carry(or possibly would carry) this kind of value. One of Moore’s more discussed ideas is themethod he proposed for finding out what things are intrinsically valuable, by which hewanted to ensure that we did not confuse what is valuable in itself with other kinds ofvalue (in particular, instrumental values). Thus, Moore maintained that:

it is necessary to consider what things are such that, if they existed by themselves, in absoluteisolation, we should yet judge their existence to be good.6

The instruction is quite clear and simple: think about the alleged value bearer as if itexisted all by itself. Doing this we should be able, he thought, to determine whetherthe existence of the thing is good or not. This so-called ‘isolation-test’ was supposed toenable us to sort what is valuable in itself from what is either valuable for somethingelse’s sake or has no value at all.7

But Moore’s thought-experiment has not been generally accepted. In part this is notthat difficult to understand. As Noah Lemos points out:

ontological isolationism is not very helpful since there are certain sorts of things that are intrinsicallygood but simply could not be the only things that exist. Consider the fact of Smith’s being happyand let’s suppose that it is intrinsically good. If there are certain abstract entities such as numbers orproperties or states of affairs that necessarily exist, it would be impossible for Smith’s being happy tobe the only thing that exists. More important, though, is the fact that Smith’s being happy couldnot exist without Smith’s existing, as well as, I suppose, Smith’s having certain pleasures and certaindesires satisfied and his having certain beliefs to the effect that he had those pleasures and that hisdesires were satisfied. Since it is necessarily false that Smith’s being happy could be the only existingthing, this sort of ontological isolationism is not very clear or very helpful.

(Lemos, 1994, p. 37; cf. Zimmerman, 2001a, p. 132)

5 Moore did not use the term ‘supervenience’ for the dependence relation between evaluative propertiesand non-evaluative ones. Johan Brännmark informs me (personal communication) that the term was usedamong British emergentists. See Samuel Alexander, Space, Time and Deity (1920), C. D. Broad, The Mind andIts Place in Nature (1925), and C. Lloyd Morgan, Emergent Evolution (1923). R. M Hare’s The Language ofMorals (1952) introduced the notion to many moral philosophers.

6 Principia Ethica (1903, p. 187).7 See Principia Ethica (1903, p. 187). As is pointed out by Lemos in Intrinsic Value (1994), Moore is not the

only one to propose such an isolation test. David Ross did so in The Right and the Good (1930, pp. 68–9).According to Zimmerman in The Nature of Intrinsic Value (2001a), Richard Price made a similar claim. See thelatter’s ‘A Review of the Principal Questions in Morals’ (1897, p. 148).

4 TYPES OF VALUE

The shortcomings of Moore’s isolation test led Lemos, and, later, Zimmerman topropose an amendment to it.8 The novel proposal is that we should reflect on thesupposedly valuable object, setting aside the consequences of it without paying atten-tion to the circumstances. The demand that we contemplate the object in this way isnot incoherent. It might be hard to comply with sometimes, but it is not asking theimpossible of us. This new proposed test tells us, then, to consider the thing withoutreference to context and what it brings in its train, and then to evaluate it. Lemos refersto it as the ‘intentional isolation test’ as opposed to Moore’s ontological one.9 If thiskind of contemplation ‘leads’ you to believe that the thing is valuable, or good, andgranted that you have in fact managed to keep circumstances and consequences out ofyour contemplation, you will have arrived at what is valuable, or good, in itself or for itsown sake. Or so the argument goes.

1.3 Isolating extrinsic final valueMoore’s isolation test has presumably played some role in the story of why, for a longtime, value theory treated the distinction between being valuable in itself and for itsown sake as one of little importance. By focusing so firmly on things in isolation, notadmitting external, relational properties to be considered as ‘valuable in itself ’-makingproperties, the perspective of value definitely came to be, for quite some time, toonarrow. Not everything that we value for its own sake carries a value that supervenesonly on the internal features of the value bearer.The intentional version is an improvement.10 However, once things are able to carry

extrinsic final value, i.e. be valuable for their own sake in virtue of at least somerelational property that is not internal to the value bearer, matters become much morecomplicated. Obviously, the foolproof air of Moore’s original test has evaporated; whatis left is a more modest mode of determining what has final value. The test continues tobe a catalyst that helps us trigger substantive evaluative reactions. However, we mustnow be on our guard; even the improved isolation test may be doing more harm thangood. For instance, we cannot conclude, just because we are not ready to ascribegoodness to something when we contemplate it in isolation, that the object is notvaluable for its own sake. We can only say that it is not a final intrinsic value.It might be retorted that we can always extend the intentional isolation test to

cases of extrinsic final value; after all, there is nothing particularly difficult about

8 Zimmerman shows that the ontological version is not only incoherent, but also misleading in certaincases. It can lead us to identify “certain states as parts of other states when they are in fact not parts of them atall” (2001a, p. 141).

9 Lemos introduces these expressions in (1994, pp. 10–11); cf. Roderick M. Chisholm, ‘DefiningIntrinsic Value’ (1981).

10 However, I think it is a mistake to suggest, as Chisholm appears to have done, that by saying thatsomething is intrinsically good we mean by this that it is what we should evaluate as good if we applied theintentional isolation test. It is one thing to regard the test as a method for learning what things are intrinsicallyvaluable; it is quite another to maintain that this test is part of the semantic analysis of value.

1.3 ISOLATING EXTRINS IC F INAL VALUE 5

contemplating a thing being related to something else. True, I do think the improvedisolation test can help us determine whether some object has or does not have extrinsicfinal value. But it is important to steer clear of some complications here. For one thing,the kind of contemplation that we are now considering cannot merely be the contem-plation of a state of affairs involving relata; focusing our attention on that would tell usat most whether this state of affairs has an intrinsic value. The relata would be internal tothe state, which is quite consistent with the state’s being an intrinsically valuable object.Rather, the contemplation must concern an object that is related to some other object.If we manage such contemplation, we should at least in principle be able to arrive atmore or less well-founded judgements about the extrinsic final value of objects. Again,this does not amount to the 100 per cent test of Moore’s original method. Nor,however, does it leave us empty-handed. Naturally, we cannot be certain that takingmore relational properties into the picture will not tilt our all-things-consideredjudgement of a particular object’s value. But this kind of tilting is just to be expected.11

Moore’s view that goodness is a simple, non-natural property has probably intriguedas many people as it has offended. The critics form a rather heterogeneous group.Without going into details that soon would take us far astray, we can at least identifythe following camps. Some object to what they deem mystical, namely that goodness isnon-natural. Others react to what they believe is only an expression of a false view onwhat we do when we evaluate. It is not difficult to find representatives of these viewstoday. Whether goodness and value are natural properties or not, is something that Iwill leave open in this book. An answer to this question would, eventually, emerge ifwe knew the answer to various questions on the semantic side of formal value theory.For example, can evaluations be true or false? Is their main function to express beliefs ordesires—to express attitudes that should fit how the world in fact is, or attitudes that aresuch that the world should change to fit the attitudes? However, today’s metaethicalconsensus leaves questions about the semantic nature of value judgements very muchopen, and so will I.12 (Much of interest can still be said about value and goodness thatdoes not depend on taking a decisive stand on these semantic issues.)

Some critics have gone a step further and argued that it does not make sense to speakof goodness in the way Moore apparently thought we are doing. This strain of thoughtgoes back to Peter Geach’s influential paper ‘Good and Evil’.13 Geach argued therethat every meaningful use (what he referred to as attributive use) of ‘good’ presupposesa certain domain of things to which the use applies. Things are never plain good. There

11 Moore is not alone in coming up with a test for what has value. Another towering figure in valuetheory, Immanuel Kant, suggested a different method. Instead of contemplating things in isolation, Kantwanted us to place the presumed bearers of value in a different context (my own suggestion resembles thisidea). For a discussion of Kant’s view, see Brännmark,Morality and the Pursuit of Happiness: A Study in KantianEthics (2002, sec. 2.2.).

12 In short, utterances of sentences containing value terms express the speaker’s value judgement. Thenature of these judgements throughout the present work, then, is regarded as being consistent with bothcognitivist and non-cognitivist metaethics.

13 Geach, ‘Good and Evil’ (1956).

6 TYPES OF VALUE

is no such thing as the predicative—as he called it—use of ‘good’. This discussion isimportant, and much can be learned from it. Despite this, I will sidestep furtherdiscussion of it. However, although I think this is an interesting approach, and clearlyworth pursuing, perhaps I should add that I also think it faces a couple of problems. Letme briefly state one. In maintaining that there are only attributive uses of ‘good’, weseem to assume the truth of a certain metaethical position—namely, one according towhich we can understand the goodness of, say, a car purely in terms of its non-evaluative properties. But, of course, this is not an obviously correct position. Wemight argue, for instance, that whatever the constitutive properties of the car are, in theend it takes a genuine evaluation to say which car is a good car. Now, since I believethat ‘goodness’ does have a genuine evaluative meaning (although it is not somethingthat I argue here), the attributive sense seems to me to be at most a non-interesting, butprobably common, way of expressing what is not a genuine evaluative judgement. The‘Geach camp’ could certainly deny this; they might argue that there are only attributivesenses. However, so far no one has managed to come up with an altogether convincingmetaethical view, and hence it is perhaps best to leave this issue open. (The analysis thatI will be proposing does just this.) Moreover, although Geach and others may have apoint as regards the use of “plain goodness/value”, it would be a mistake to confuse this‘goodness’ with a different sort of goodness that is used interchangeably with ‘intrinsic(final) value’. The phrase ‘object x is intrinsically (finally) valuable’ easily splits into ‘x isan object and x is intrinsically (finally) valuable’, and there seems to be nothingnecessarily wrong or odd about this.14

1.4 Competing notions of intrinsic valueSo far I have suggested that objects may exist that are valuable for their own sake butwhose value nonetheless does not depend exclusively on their internal features. It is, asmentioned, the task of substantive value theory to determine whether there is this kindof value. Still, we have to at least make the idea plausible. One reason the idea of anextrinsic final value has met with suspicion has been an assumption among someinfluential value theorists concerning the sorts of thing that can bear final value. Themain reason for limiting oneself to just one kind of value bearer has been simplicity. Thisis a very understandable motive, but it has not convinced me to give up a pluralisticposition when it comes to the nature of value bearers. I will return to this issue later on inChapter 10, where I present an argument for value pluralism. (Since a pluralist positionseems intuitively plausible, I prefer themore direct way to the analysis of personal value.)

14 This response to the Geach position comes originally from Zimmerman. In his (2001a) he rejects themain arguments put by Peter Geach, Bernard Williams, Philippa Foot, and Judith Jarvis Thomson. See JonasOlson, Axiological Investigations (2005, pp. 32–44), who also defends the notion of final (intrinsic) value.Recently Sven Danielsson has come to Geach’s defence in ‘On Geach on Good’ (2007). See also the recentarticle by Brännmark ‘Excellence and Means: On the Limits of Buck-Passing’ (2008).

1.4 COMPETING NOTIONS OF INTRINS IC VALUE 7

It is not difficult to find examples that plainly appear to be inconsistent with thetraditional Moorean view that all final values, without exception, are intrinsic. I willpresent some examples in a moment. Meanwhile, another possibility that has so far notreceived any attention here must be addressed. The fact that there is seemingly noagreement on what sort of things have, say, intrinsic value, may not necessarily mark adifference in substantive views. It might just be the case that there is more than oneconcept of value circulating. In fact, it has recently been argued by Ben Bradley15 thatthere is not a single concept of intrinsic value but two: Mooreans accept the threetenets that intrinsic value: (a) supervenes only on the value bearer’s intrinsic properties;(b) accrues to the value bearer by necessity; and (c) is such that something that hasit would continue to have this value “even if it were alone in the universe” (2006,p. 113). Kantians reject these principles. On their view, intrinsic value is “the sort ofvalue that requires respect”. Bradley also argues, in many ways convincingly, that oncertain points Mooreans and Kantians have been talking past each other. Whereas theMooreans regard intrinsic value as something that ought to be promoted, or somethingthat makes the world a better place, Kantians look upon it as something that should berespected or honoured.

The position defended in this work is obviously not Moorean. It does recognizesomething like Moore’s intrinsic value, though. However, it also claims that since it canbe subsumed under the notion of final value, this is not the fundamental value notion.Nor is the position Kantian, however, although it bears some resemblance to whatBradley characterizes as such. Thus, on my view, certain valuable objects ought to berespected or honoured; but equally there are a great number of different attitudes thatone ought to take with regard to objects that are valuable. Bradley’s idea that there arejust two value notions is not that intuitive from this perspective. If there are indeedvaluable things that we should cherish, protect, love, admire, and so on, it might wellbe suspected that there are not two value concepts but many—one each, say, forobjects that ought to be cherished, promoted, loved, desired, and so on.

This is, of course, a possibility. However, we might also try to look for a theory ofvalue that could bring all of these values under one and the same roof. Later on, inChapter 2, I will outline a pattern of value analysis which has received considerableattention lately and purports to do precisely this.16

Let us consider some examples that seem inconsistent with Moore’s views onintrinsic value. Shelly Kagan, for instance, discusses the value of a unique or rarestamp; John O’Neill refers to a wilderness that may be considered valuable preciselybecause it is untouched by humans,17 and Wlodek Rabinowicz and I have suggestedthe following:

15 Bradley, ‘Two Concepts of Intrinsic Value’ (2006).16 Bradley recognizes this. See his discussion of the fitting attitude analysis (2006, pp. 127–8).17 O’Neill, ‘The Varieties of Intrinsic Value’ (1992).

8 TYPES OF VALUE

Another class of examples involves cases in which a thing is valued for its own sake in virtue of itsspecial relationship to a particular object, event, or person. An original, say, an original work ofart, may be valued for its own sake precisely because it has the relational property of being anoriginal rather than a copy. Its final value supervenes, in part, on its special causal relation to theartist. Princess Diana’s dress may be another case in point. The dress is valuable just because it hasbelonged to Diana. This is what we value it for.

(Rabinowicz and Rønnow-Rasmussen, 2000, p. 41)

Being unique or rare, untouched by humans, original, or related to someone ofimportance are all external relational properties, so they would not be admitted in thesupervenience base of an intrinsic value.Examples of this kind can easily be multiplied, which makes one wonder why the

Moorean dogma has not been very widely abandoned once and for all.

1.5 The source of valueRecently a number of philosophers have argued that there can be final values that are notintrinsic. One argument in particular, by Christine Korsgaard, has been widely discussed.Korsgaard calls an object intrinsically valuable if the “source” of its goodness lies in theobject itself. But, in her view, not all things valuable as ends are intrinsically valuable.

Separating the two distinctions in goodness [intrinsic/extrinsic and final/instrumental], however,opens up another possibility: that of something which is extrinsically good yet valued as an end.An example of this would be something that was good as an end because of the interest thatsomeone took in it, or the desire that someone had for it, for its own sake.

(1983, p. 252)18

The source of some end values is located outside the value bearer. Thus, instead ofbeing internal the source may well lie, say, in our interests, or in desires directed ontothe objects themselves, in which case the source will be external (Korsgaard, 1996[1983], p. 252).Now, the notion of a value source is not straightforward. As Rabinowicz and I have

argued elsewhere, it is important to separate two possible interpretations of this notion;but when this is done it becomes clear that Korsgaard’s argument is misdirected. Toidentify the source of value may be either to detect the supervenience base of the value orto discover the constitutive grounds of an object’s final value.The distinction between supervenience and constitution is important if you want to

understand why value theorists understand value in so many different ways. But bothnotions are far from transparent, and a more detailed discussion of the distinction isrequired. Therefore, in the remainder of this chapter, I will outline how I think thesenotions should be understood. Let us begin with supervenience.

18 See her ‘Two Distinctions in Goodness’ (1983), which is also published in her Creating the Kingdom ofEnds (1996), from which the quotation comes.

1.5 THE SOURCE OF VALUE 9

1.6 SupervenienceSupervenient properties are ‘dependent properties’.19 To insist that value alwaysaccrues to an object because it has certain other, non-evaluative properties is to expressa belief, by implication, that there is a relation of dependence between value and thoseother, subjacent (i.e. underlying) properties. This does not seem obviously true in thecase of mental predicates. At least, it is not obviously implausible to suggest that we, aslanguage users, were mental ‘predicaters’ long before we became aware of the body/mind problem.

The second job the supervenience claim is customarily taken to do is that ofcapturing what we might call the consistency intuition. This is the conviction that ifyou assert that x is valuable, and if you agree that y is relevantly similar to x in its naturalproperties, you must be prepared to assert that y, too, is valuable. To deny this wouldbe to leave the listener in a kind of perplexity similar to that which follows thedetection of a contradiction. I take it that if we were to believe that values are, so tospeak, free-floating ontological entities that just land on objects (either haphazardly orbecause, say, they were bestowed in accordance with God’s will), we would not belogically committed to ascribing value to y here. It is at least not a requirement ofconsistency to draw the conclusion that y is valuable, for x and y may differ precisely inthat value has ‘landed’ on one but not the other.

It might be suggested that the fact that we can imagine such ‘value butterflies’ showsthe supervenience thesis to be, not a conceptual thesis, but an expression of our deeplyrooted beliefs about what and how things are. However, this reasoning is flawed. Fromthe fact that we seem to be able to imagine, without contradiction, ‘free-floatingentities’, it does not follow that we are conceptually able to imagine that values aresuch entities.

The discovery that this ‘butterfly’ reasoning seems to be fundamentally confusedshould not obscure the fact that it is hard to come up with a knock-down argumentshowing that the supervenience thesis is not, at bottom, based on ontological convic-tions. I am not sure what such an argument would look like. Whether this mistrustwarrants serious disbelief in supervenience is another matter. For my own part, I do notthink so. At any rate, I am certainly inclined to say that value terms are, for purelyconceptual reasons, supervenient terms. Hence, in what follows I will treat the super-venience thesis as a conceptual thesis. Here I merely wish to make it clear that I amunable to see how to settle beyond question whether or not my intuitions are purelylinguistic.

19 In other words, when, henceforth, I talk about value-makers, I have in mind those properties on whichvalue depends, leaving it for discussion how ‘depends’ should be read. Perhaps dependence here is oneinvolving “makers”, or “enablers”, or both. See Jonathan Dancy, Ethics Without Principles (2004). Cf. CajStrandberg, Moral Reality: A Defense of Moral Realism (2004, pp. 229–39).

10 TYPES OF VALUE

As mentioned earlier, there is, or so it has been argued, more to supervenience thanthe claim that if two things, x and y, are similar in their natural properties, they must—and ‘must’ here is conceptual—have similar value properties. In recognizing that thesupervenience thesis expresses a consistency intuition, we must not lose sight of afurther important relation that is said to hold between natural properties and valueproperties—namely, that the former necessitate the latter. This necessity relation mightbe moral or metaphysical, or it might be some other kind of necessity. It might mean, asSven Danielsson has expressed it, “something like ‘fact-independently’. If Socrates isgood because he is brave and wise, then, whatever the facts that may obtain, anyonebrave and wise would be good” (2001, p. 95).20

The remedy for butterfly properties seems rather obvious. If our aim is to dislodgebutterflies from value supervenience we need to formulate a relation between valueproperties and natural ones such that if the latter are present, then, in every possibleworld, the former are too. We need, in other words, to go from intra-world to inter-world supervenience. Taking ‘V’ to refer to value, ‘N’ to refer to natural properties,and ‘Nec’ as a necessity operator, we need the relation captured by the followingformula, due to Jaegwon Kim:21

Strong Supervenience: Nec 8x (Vx ! ∃N (Nx and Nec 8y (Ny ! Vy)))

If this strong relation is what the supervenience relation is all about, properties thatbehave on the analogy with the butterfly are blocked; such properties cannot besupervenient. There is a necessity relation (the second one above) of some sort(moral necessity, metaphysical necessity?) between N-properties and V-propertiesthat excludes this possibility.Strong Supervenience captures something essential about value supervenience.

However, the question is whether it catches all there is to this dependence relation.One problem with it needs to be resolved. As far as I can see, it leaves out an importantaspect, namely what we might call the ‘direction-feature’ of the ‘because of ’ or ‘invirtue of ’ relation. These relations suggest that moral and natural properties areconnected in a special sense—the latter give rise to (or support) the former, and notvice versa. Consider the relation between ‘x is coloured’ and ‘x is extended’. Thisrelation is not, in my view, a ‘because of’ or ‘in virtue of’ relation. But suppose wereformulate Strong Supervenience in terms of properties of colour and extension.Strong Supervenience would then seem to fit the relation that exists between colourand extension. In other words, if we do not want to say that colour and extensionexemplify the kind of relation found between moral properties and natural ones, weneed to qualify Strong Supervenience in some way. At present I do not know how todo this.

20 Danielsson, ‘The Supervenience of Intrinsic Value’ (2001).21 See Kim, Supervenience and Mind (1993).

1.6 SUPERVENIENCE 11

1.7 The Invariance ThesisThe idea that values are constituted by subjects is at heart of subjectivism. Now,subjectivists are vulnerable to an objection that we need to examine. The attitudeswe have appear to depend heavily on arbitrary factors. Hence, if we tie values to suchcapricious entities as attitudes, we make the question of what values there are highlycontingent; and this is not how we normally conceive of values. The suspicion, inother words, is that if we relate value to attitudes, value becomes a relative notion, andvalue relativism is hard to combine with the following:

The Invariance Thesis: The final value of an object is invariant over possible worlds.

It follows from this thesis that if an object is valuable, the object (and any object exactlysimilar to it) will carry precisely this value in every possible world in which it is present.Or, more accurately: if something is a supervenience base in the actual world, it will besuch a base in every possible world in which it is present. It is a standard objection tosubjectivist accounts that they have to reject the Invariance Thesis. If we fix value todesires and other kinds of attitude, the complaint runs, a valuable object x will not bevaluable (i.e. its value will not have been constituted) in possible worlds in which x isnot the object of a desire. If the world were different, with an alternative geography ofdesires, the set of valuable objects would be quite other than the actual one.

On this matter we may expect various subjectivist responses. I have discussed whichones elsewhere (Rønnow-Rasmussen, 2003), and I will therefore not repeat themhere. In this context it is paramount to appreciate that the idea that value is invariant iscontentious. Although a number of influential axiologists, including G. E. Moore,have defended the Invariance Thesis over the years, it has come under considerable firemore recently.22 Its plausibility depends on the nature of the value bearer x or, tobe more precise, on what kinds of property are included in the supervenience base.If x is valuable in virtue of its non-essential properties, the Invariance Thesis looks tooencompassing. If La Gioconda had been painted with, say, different colour properties, itwould not have been valuable (that is, assuming that the colours are not essentialproperties of La Gioconda). However, things are different if x is a value-bearing objectlike an abstract entity which is valuable in virtue of its essential properties. Such anentity will be the same in all possible worlds, and hence its value will remain the samein all worlds. If x is not the concrete La Gioconda itself but rather some abstract entitysuch as the fact (the obtaining of the state of affairs)23 that La Gioconda presents such andsuch features, then we cannot ‘tamper’with the features making this very fact what it is.To do so would be to create a distinct fact whose possible value or disvalue would haveno bearing on the issue under discussion.

22 For a recent defence of this thesis, see Zimmerman (2001a).23 See, here, Lemos, Intrinsic Value: Concept and Warrant (1994, pp. 25–6) and Zimmerman (2001a, p. 50).

12 TYPES OF VALUE

In an unqualified form, then, the Invariance Thesis is ruled out as a test ofdemarcation between objectivists and subjectivists. Advocates of either position canagree that if an object supervenes on non-essential properties, then, in a world in whichthe object lacks some of these properties, it will no longer be valuable. It should benoted, however, that the objectivist’s view may be quite a different one, for they (likesubjectivists) can adopt a certain kind of value-bearer monism. On this view it is onlyentities that have whatever features they have essentially (e.g. states of affairs) that carry(final) value. If objectivists and subjectivists share this view, their responses to theInvariance Thesis will inevitably differ.Subjectivists will naturally reject the Invariance Thesis, but they have quite a lot of

leeway as to how they go about discarding it. Some ways, for instance, will be more‘objectivistic’, or attitude-independent, than others. Suppose there is a state of affairsthat I want to see realized for its own sake—say, that people look at, and appreciate, LaGioconda. Putting aside the substantial question whether this is a good example of avalue bearer, the nature of my attitude here is taken to be universal. In other words,I have this attitude not only for the actual work of art, people, and world. Rather, thescope of my attitude ranges over all the possible worlds in which people appreciate apainting that is exactly similar to the actual La Giaconda in universal features; in any suchworld I prefer this sort of appreciation to be a fact.24 Given my actual preference, thevalue of the object x will, in other words, be invariant. One salient aspect of thisexample is that, in it, the constitutive ground of value does not coincide with thesupervenience base. Actually, there is nothing particularly odd about this; we need justremind ourselves that some objects of our attitudes (say, objects of preferences) do notcoincide in time with our preference. I may, for instance, now prefer that, when I amold, no one will seek to ensure that I go on living at any price. I may prefer thisalthough I have good evidence that most people do in fact want to go on living, even ifthe cost is very high, when they grow old. Again, some of my preferences areconditional in form. For instance, in the bath I can prefer that, were I on the train,I would have clothes on.

1.8 ConstitutionThe idea that we should differentiate between the constitutive grounds and super-venience base of value has appeared in several places in recent years.25 I first became

24 Here I sidestep a complication, namely whether value can supervene only on universal features. InChapter 6 I shall return to this matter.

25 However, it is also a distinction which, over the years, has been overlooked. Thus Henry D. Aiken(1950) objects to views that have defined intrinsic value in terms of “the object in which interest, satisfactionor pleasure is taken” (p. 11). On his view, it is rather satisfaction in itself that is valuable (a position which herefers to as subjectivism). In comparison with the former views, his own avoids the paradox of ascribingintrinsic value to something on the basis of a relational feature (e.g. that it is an object of someone’s desires).The separation of the supervenience base of the value and its constitutive grounds would clear away this‘paradox’, though.

1.8 CONSTITUTION 13

familiar with it through a work by Wlodek Rabinowicz and Jan Osterberg.26 Theirdistinction illuminates the division between subjectivists and objectivists, i.e. betweenthose who believe value is necessarily related to subjects and those who deny this.27

Rabinowicz further clarifies the distinction between supervenience and constitutionby looking at the role conventions play in chess (Rabinowicz and Osterberg, 1996,p. 21). Thus, in a chess game certain moves will be permissible and others will not. Thepermissibility of a particular move depends on how the board is set at the time of themove. For instance, a white rook at square a1 can take the black queen on a7 becausethe features of the game now make it possible to move the rook all the way up to a7(while before there was, say, another piece blocking that row of squares). Now, thepermissibility of the move a1xa7 does depend (supervene, if you like) on the (internal)features of that particular game, at that particular time. However, the constitutiveground for this move is to be found in the social convention that maintains the rules ofchess, and which, among other things, specify how the rook and other pieces move.

Rabinowicz’s analogy is illuminating. Of course it does raise important issues thatneed to be commented on. However, since I have done so in Rønnow-Rasmussen(2003), and since it would lead us too far astray from our main concern, I prefer toconsider a problem for this subjectivist approach.

1.9 Subjectivism and objectivismThere is, then, a possible objection to my approach here that I need to address. Thiswill lead to yet another way of illuminating the difference between supervenience andconstitution. The objection, which would most likely figure in an objectivist responseto the subjectivist approach, is this. What the subjectivist refers to as the constitution ofvalue is nothing but itself a supervenient feature. When the subjectivist claims that anobject x is valuable, and that x’s value is constituted by the subject’s attitude, theconstitutive attitude in question is really part of something else that is valuable—namely, the subject’s direction of a certain attitude on x. It is this obtaining of a stateof affairs rather than x that is valuable. Since the attitude is in a sense part of the valuebearer, it is more appropriate to regard it as a value-making, subjacent feature. Hencethere is no need to distinguish in the original way between supervenience andconstitution. Accordingly, subjectivism should be regarded as a position that ascribesvalue only to certain kinds of object: it ascribes value only to those facts that involve thesubject directing an attitude on some object.

26 A detailed elucidation of it can be found in Rabinowicz and Osterberg, ‘Value Based on Preferences’(1996). See also Rabinowicz and Rønnow-Rasmussen, ‘A Distinction in Value: Intrinsic and For Its OwnSake’ (2000) and Rønnow-Rasmussen, ‘Subjectivism and Objectivism: An Outline’ (2003b). I oweWlodekRabinowicz a great deal for having so generously discussed the distinction with me over the years. However,he is not to be blamed for anything that I say here.

27 David Gauthier’s Morals by Agreement (1986) is a good example of a subjectivist position. I owe Fritz-Anton Fritzson for pointing this out to me.

14 TYPES OF VALUE

There is some truth in this reply. Thus, it seems correct to say that if the obtaining ofthe state of affairs of a favouring x is valuable, then it is reasonable to regard the attitudeas an element of the supervenience base. The problem with the reply lies elsewhere.Why should a subjectivist confine herself to the analysis of value judgementsconcerning facts about the subject directing some attitude on something? Of course,we might have taken a stand on what things are, and what things are not, in factvaluable; and we might have made the restriction in accordance with our evaluativeviews. But if subjectivism and objectivism are positions designed to answer questionsabout value and valuing, there is something fundamentally wrong when either subjec-tivism or objectivism is depicted as a position that says, necessarily, that certain valuejudgements cannot be analysed according to their own pattern of analysis (i.e. in thecase of subjectivism, patterns of the kind ‘Object x is valuable’). This would be todescribe the two positions from a standpoint that is both evaluative and, in this case,biased in favour of objectivism. Of course, objectivism might be true. If this is shown,subjectivists will have to give up their idea that attitudes have constitutive powers. Butsubjectivism might also be true, and if it is objectivists will have to face up to the factthat there is more to value than supervenient features.The above brings me to another way of elucidating the difference between super-

venience and constitution. This will further support the idea that we should notinclude a subject’s attitude too readily in the supervenience base.Properties (or facts about these) belonging to the supervenience base are customarily

invoked as reasons for or against various actions taken for the sake of the object. Toappeal to the subject’s attitudes—say, those of his preferences or desires directed on theobject—as a reason for acting is not always plausible. At least, it is not if what you havein mind here is not what caused you to act, but rather what you take to be yournormative reasons for acting.28 In case we have in mind actions performed for theobject’s own sake (rather than for the sake of being desired by us) this possibility is ruledout. The attitudes should therefore remain in the background, not the foreground, ofour motivational set. They can be the constitutive grounds of final value, but this doesnot make them a part of its supervenience base.Let us pick up the threads and return to Korsgaard’s suggestion that we should

understand the difference between intrinsic and extrinsic final values in terms of thelocation of the source of value. Recall her point: if the source is found in the objectitself, we are dealing with intrinsic value; if it lies in the interests or desires that aredirected on the object, the value is extrinsic (1996 [1983], p. 252).29

28 We shall return in Chapter 5 to the difference between what is part of the intentional content of anattitude (say, some features that call for you to act) and what caused the subject to have the attitude.

29 Thus Korsgaard writes: “Separating the two distinctions in goodness [intrinsic/extrinsic and final/instrumental], however, opens up another possibility: that of something which is extrinsically good yet valuedas an end. An example of this would be something that was good as an end because of the interest thatsomeone took in it, or the desire that someone had for it, for its own sake” (1996, p. 252). As noted by BengtBrülde in The Human Good (1998, p. 390), in her discussion of final value Korsgaard does not clearly

1.9 SUBJECTIV ISM AND OBJECTIVISM 15

That this is not a compelling way of understanding the distinction between intrinsicand extrinsic values should be clear. ‘Source’ may refer either to the properties of anobject on which its final value supervenes or to the constitutive grounds of the object’s finalvalue. The latter may well lie outside the object itself—that is at least what subjectivistswant us to believe—even though the former are internal to the object. For instance,the value of an object is, on a preferentialist conception, constituted by the preferencesor desires directed on the object. But notice, where the object is desired only forfeatures that are internal to it, then, on this subjectivist position, this externally con-stituted value is nonetheless intrinsic. The object’s good-making properties are allinternal properties of the object: they are those for whose sake the object is preferred.30

To relate value to the desires of a subject under the heading ‘value source’may be toconflate the supervenience base with the constitutive grounds of value. To establishthat values can be externally constituted is in itself not an argument for extrinsic finalvalues. Final values may all still be intrinsic values if what makes them final is theirsupervening only on features internal to the value bearer.31

Before we proceed, there is something I would like to emphasize. An anonymousreferee for Oxford University Press pointed out that even if we grant that subjectivism,as I have presented it, is compatible with supervenience, it will remain a contingent factthat the values constituted by attitudes actually deliver that supervenience. This shouldcertainly be underlined. That is, subjectivists cannot guarantee supervenience when-ever an objectivist claims there is value supervenience. As I pointed out in Section 1.7,subjectivists cannot agree with all that objectivists want to say about value super-venience. However, they need not agree with everything objectivists say about value.Supervenience, according to the subjectivist, is a relation that holds between value andan object in virtue of the value bearer’s non-essential properties. Subjectivists recognizethe logical requirement that if you say that x is valuable, you must agree that somethingabout x makes it valuable. Since they do so, unless they change their minds about whatis valuable, they are therefore forced to apply their value predicates in a consistent way.Of course, objectivists will claim there is more to “force” than “consistency”.

distinguish between being valued and being valuable (as an end, or for its own sake). In what follows,however, I focus on the latter.

30 When, in ‘The Limits of Well-being”, Kagan maintains that on “a radically subjectivist conception ofvalue, [ . . . ] nothing would be valuable as an end in the absence of there being some creature who values it”(1992, p. 184), he (too) conflates the constitutive grounds of value with its supervenience base. Subjectivistswill locate the constitutive ground in the subject, and not the supervenience base; and it is the nature of thelatter which determines whether or not the value is intrinsic. See Rabinowicz and Rønnow-Rasmussen(2000).

31 See also Korsgaard’s later work, ‘Motivation, Metaphysics, and the Value of the Self: A Reply toGinsborg, Guyer and Schneewind’ (1998), in which she claims that our particular ends “have only extrinsicvalue, since their value depends on our own desires and interests in them and is conferred on them by ourown rational choices” (p. 63). Here, Korsgaard conflates supervenience with the constitutive grounds ofvalue.

16 TYPES OF VALUE

Obviously, subjectivists cannot constitute “objectivist” value—at least, not the sortof value that supervenes on the essential properties of the value bearer. But this is quiteconsistent with the idea that (subjectivist) value is constituted by a subject’s preferencesand other attitudes. Supervenience tells us something about “where” a value is located.The difference between objectivists and subjectivists is that whereas the former thinkthat if we know where a value is located we also know (or, chances are, we also know)what grounded (constituted) the value, the latter deny this, believing that it is one thingto identify a value location, but quite another to settle what constitutes the value.There are several serious problems with subjectivism, and these incline me to reject

it. The most important is the fact that the very notion of value constitution remainssomewhat obscure. However, objectivism is also problematic. But, my goal here hasnot been to settle the subjectivism/objectivism issue, but rather to outline how thedebate should be understood. My contention is that the disagreement does not, in thefirst place, concern the location of value—both parties can agree on what is valuable; itconcerns the basis (constitution) on which the value is located where it is located.G. E. Moore was a stout defender of the idea that intrinsic value is in a sense the

property of objectivists; only objectivists can speak of it coherently. I think Moore waswrong about this—just as I think those (mostly subjectivists) who argue that the notionof intrinsic or final value is incoherent are wrong.32 There is nothing to preventsubjectivists from recognizing the final or intrinsic value of something. There is animportant difference, of course, between subjectivism and objectivism, but it is not onethat prevents subjectivists from ascribing intrinsic or final value to things.The distinction between supervenience and constitution is not the only tool needed

to separate subjectivists from objectivists. Nonetheless, its being overlooked continuesto give rise to unnecessary confusion. It is also true that proper appreciation of itssignificance allows one to move beyond certain casual, one-dimensional assump-tions—for instance, the idea that the subjectivist only locates value in subjective states,or the idea that objectivists cannot share the evaluations of a subjectivist. On theapproach defended here, subjectivists and objectivists may well share each other’sevaluations, and they need not disagree about what objects are the appropriate bearersof value.33

32 Moore found at least two faults with subjectivism—it makes the objectivity of good disappear, and itmakes it impossible for things to be intrinsically valuable. The following passage is often taken to express, byimplication, the latter point: “To say that a kind of value is ‘intrinsic’means merely that the question whethera thing possesses it, and in what degree it possesses it, depends solely on the intrinsic nature of the thing inquestion” (‘The Conception of Intrinsic Value’ (1922)). The subjectivist, according to Moore, makes valuedependent on something external, the subject, and the subjectivist’s alleged value is therefore not an intrinsicvalue. For another early attempt to argue that subjectivists cannot coherently speak of intrinsic value, see theGerman value theorist Erich Heyde’s Wert: Eine philosophische Grundlegung (1926).

33 The distinction has a clear bearing, for instance, on the way we should understand a central claim inL.W. Sumner’sWelfare, Happiness & Ethics (1996) that only subjectivists make welfare logically dependent onthe attitudes of the welfare subject.

1.9 SUBJECTIVISM AND OBJECTIV ISM 17

Sometimes positions in this area (in particular subjectivism) are depicted as first-ordernormative claims. For example, it has been said that you are a subjectivist if you thinkthat each individual ought to do whatever she happens to think she should.34 Ofcourse, there is no real point in quibbling about names. But in my view, sincesubjectivism and objectivism are naturally seen as arising from questions about howto understand the nature of value (and ultimately evaluative and normative-judge-ments like the one mentioned above), a great deal would be gained if we could retain‘subjectivism’ and ‘objectivism’ for second-order views on values.

34 See, for example, J. L. Mackie, in Ethics: Inventing Right and Wrong (1977), who refers to this as a kind offirst-order subjectivism. Moreover, when Josiah Royce refers to the enquiry ‘Why should I obey the morallaw?’ as “the old question of subjectivism” he is obviously not thinking of subjectivism in the same way(‘Discussions: The Outlook of Ethics’ (1891, p.109).

18 TYPES OF VALUE

2

Fitting-attitude Analysis

To think that something possesses personal value is to evaluate it as having value forsomeone; in cases of positive or negative personal value we think the thing is good orbad for this someone. In my view, evaluations, whether they involve personal orimpersonal values, may be more or less well grounded. Perhaps they may even be trueor false. Butwhether or not they carry truth-value, the fact that it ismy evaluation does notin itself make these evaluations true or invulnerable to criticism. And this, I take it, holdsfor impersonal as well as personal evaluations. I am not going to argue for this here.Though it is tempting to do so, I will spare the reader such a digression frommymain task.The literature on well-being (philosophical, as well as non-philosophical) is inundated

with suggestions as to what is good, and what is bad, for you. But these more or less well-grounded pieces of advice and suggestions are seldom accompanied by analysis of the verynotion they employ. Of course, there are some important exceptions to this, and we shallconsider these in due course. Naturally, the literature is rich in suggestions about the kindsof thing that contribute to human welfare or well-being; and in all likelihood welfare orwell-being are bona fide examples of things that have personal value to us. Philosophicalworks on what makes us feel well, or what makes our lives go well, devote much spaceto the discovery of the constitutive parts—or the necessary and sufficient conditions—ofa good life or good mental state. For many years this has been, and still is, a stimulatingand vivid area of philosophical analysis. However, it is also a field which, perhaps tooreadily, has tended to monopolize the notions good-for and value-for. The presentmonograph tries to put up a bit of a fight and resist this trend. That is, if am right there ismore tovalue-for thanwell-beingorwelfare.The lattermaywell be typical cases ofwhat isgood for a person, but they do not exhaust the category of things that are good for us.A word on the very nature of this resistance: philosophical analysis is mainly

reductive—it typically splits concepts into smaller conceptual elements wherever itmakes sense to do so. This has proven to be a most fruitful intellectual enterprise,gradually affording us more and more fine-grained conceptual tools. I will not departfrom this approach. However, what I will initially be proposing is that the readershould take a step back and be open-minded enough to entertain, as at least a

possibility, the idea that value-for is a more complex notion than ‘welfarists’ haveassumed. Doing so, we shall eventually be led down another reductive avenue. Thekind of analysandum that I have in mind is one typically accruing to certain mentalstates, but also to things like poems, tombstones, and the like, and to the obtainings of(abstract) states of affairs such as, say, the fact that my favourite football team wins. Myway of arguing for this more complex notion of value-for is, then, this: I shall simply tryto outline an analysis that is sufficiently cogent and attractive, and that is such that itbecomes evident that the list of things just mentioned are all, at least potentially, moreor less plausible carriers of value-for, as my analysis understands this notion.

Now in recent years a certain pattern of value analysis—what I will refer to as thefitting attitude (FA) analysis—has had something of a renaissance. This pattern leads,not necessarily, but very naturally, to a position that admits many different kinds ofvalue bearer; and it is a promising pattern, since it shows what all values—thin ones(like goodness) as well as thick ones (like admirability)—have in common. Later on, inChapter 10, I will present an argument for why I think we should recognize differentkinds of value bearer. I shall argue that a well-known substantive view ofvalue, hedonism, is best understood as ascribing value to sensations with a particularquality—a hedonic property. Hedonism alone is a rather imperfect value theory;things other than mental states carry value—things indeed that do not belong to thesame metaphysical category as mental states. This initial contention leads quite naturallyto a search for a pluralist value theory. The argument requires some unravelling,though. It seems therefore wise not to let it get in the way of the analysis’s main steps.

In this and the next chapter I will outline some of the more important features of theFA analysis. I shall also highlight some of its more problematic elements. Unfortunately,the FA analysis does not only have much that speaks in favour of it; it also introduces aconsiderable number of more or less serious problems, some of which presently appearto lack any obvious solution. My aim is not to solve, or even to discuss, all of theseproblems, but to argue that if we fine-tune this kind of analysis, we may add yet anotherpositive feature to it—that is to say, it will become apparent that it is capable of giving usan account of personal value. Of course, the possibility that values are more heteroge-neous than traditional FA advocates have so far realized cannot be regarded as a positivefeature of the FA approach if, eventually, it is shown that wemust give up on the patternanyway (because, say, we cannot solve the problems mentioned above). However, myhope is that, by pointing at a fairly easy way of extending the analysis, we shall keep thefocus on what is a most intriguing, and for years too-much neglected, view of value.

2.1 The originsA detailed history of the FA analysis has yet to be written. Here I shall confine myself tosaying something about two major exponents of it,1 although in fact the list of

1 For discussion of some of the leading proponents of FA analysis, I refer the reader to Jonathan Dancy’sexcellent ‘Should we pass the buck?’ (2000a). See also Wlodek Rabinowicz and Toni Rønnow-Rasmussen(2004a). This section of the present text rests heavily on this work.

20 FITT ING-ATTITUDE ANALYS IS

prominent figures who, at one point or other, have advocated it is arguably quite long:it includes C. D. Broad, A. C. Ewing, and more recently Richard Brandt, SvenDanielsson, Alan Gibbard, Jonas Olson, Thomas Scanlon, John Skorupski, and Mi-chael J. Zimmerman.2

The Austrian philosopher Franz Brentano (1838–1917) appears to have been thefounding father of the analysis.3 However, his insights left no lasting, salient impact onhis contemporary value theorists in Austria and Germany. There is at least no clear traceof the FA approach in the mature work of prominent Brentanists like Alexius Meinongand Christian Von Ehrenfels; and again, if we consider works on the history of ethicsproduced during, and for many years after, Brentano’s life, we find that, strangely, theFA view is often neglected or misunderstood.The core idea of Brentano’s view can be expressed in a fewwords.Here is an example:

. . . the good is that which is worthy of love, that which can be loved with a love that is correct.4

Notice that he is saying not that the good is worthy of love, but rather that what isworthy of love is good. The attitude is, in other words, directed not on the goodness,but on something else. Brentano had a quite specific idea in mind when he talkedabout love. ‘Love’ refers to what he regards as a “higher mode” of taking pleasure insomething.5 Moreover, this feeling of pleasure has a further feature that distinguishes itfrom what he thought of as mere compulsive and instinctive feeling, namely that it is“experienced as correct [als richtig charakterisiert]”;6 and because the experience ofcorrectness is conceived of as a part of the very attitude itself, Brentano is able to arguethat such attitudes function as the basis of our knowledge of the good.Some further points about Brentano’s views ought to be made. He distinguished

between what was good in itself (that which is “pleasing in itself ”) and what might bereferred to as good as a means (that which is “pleasing in virtue of what it brings about orpreserves or makes probable”).7 But what is more interesting is that he also applied theanalysis to value comparisons. As a consequence, his analysis accommodated the idea that

2 C. D. Broad, Five Types of Ethical Theory (1930); and A. C. Ewing, The Definition of Good (1947). See alsoAlan Gibbard,Wise Choices, Apt Feelings (1990); Richard Brandt, ‘Moral Valuation’ (1946); Thomas Scanlon,What We Owe to Each Other (1998); Michael J. Zimmerman, The Nature of Intrinsic Value (2001a); JohnSkorupski, ‘Buck-Passing about Goodness’ (2007); and Sven Danielsson and Jonas Olson, ‘Brentano and theBuck-passers’ (2007).

3 Franz Brentano, The Origin of Our Knowledge of Right and Wrong (1969 [1889]). Other names sometimesmentioned in this context include Adam Smith (1759/1790) and Henry Sidgwick (1907 [1874]).

4 Brentano (1969 [1889], p. 18).5 Brentano conceived of this higher mode as an emotive state. But taking pleasure in something is not

necessarily being in an emotive state. In fact in recent years Fred Feldman has objected to what he refers to asthe “distinctive feeling view”: the idea that pleasures are phenomenologically uniform; see his Pleasure and theGood Life: Concerning the Nature, Varieties, and Plausibility of Hedonism (2004).

6 Brentano, (1969 [1889], pp. 21, 22). It should be mentioned that Roderick M. Chisholm’s translationdeviates from the German original but may still be faithful to it in spirit. On this point, see J. N. Findlay,Axiological Ethics (1970, pp. 21ff.). Cf. Rabinowicz and Rønnow-Rasmussen, ‘The Strike of the Demon: OnFitting Pro-Attitudes and Value’ (2004a).

7 Brentano (1969 [1889], p. 18).

2.1 THE ORIGINS 21

the sphere of value should be understood in terms of more than one kind of attitude oremotion. This is an importantmove. It suggests that valuemight need evenmore attitudesto be fully understood. Moreover, if these attitudes take different sorts of object, then thismanoeuvre can be seen to carry the seeds of a pluralistic position on value.

Brentano understood value comparisons of the kind reported when we say thatsomething is better than something else in terms of the former being “worthy of greaterlove”.8 He carefully stressed how this greater love should be understood. It was not aquestion of more intense love, or of a more strongly felt pleasure. Instead ‘greater’refers to:

. . . the peculiar phenomenon to be found within the sphere of emotions—namely, to thephenomenon of preferring . . .When we call one good ‘better’ than another, we mean that theone good is preferable to the other. In other words, it is correct to prefer the one good, for its ownsake, to the other.9

Evidently Brentano regarded preferences as emotive, rather than purely connative,states (the latter being a common view today).10 That apart, the twist, or turn, whichvalue analysis takes via this expansion is of great importance. “Worthy of greater love”can apparently be read as meaning fitting to be preferred. The value analyst’s toolbox isenriched by one more item besides love.

2.2 A. C. EwingIf Brentano founded the FA analysis, Ewing is its foremost cultivator. During theperiod in which Ewing presented his views on value in a series of works, formal as wellas substantive value theory was rapidly becoming marginalized in certain influentialEnglish-speaking philosophy circles. This explains in part why his views were some-what neglected for several decades.

Ewing objected to naturalism as well as to the idea, widespread at the time, thatgoodness was some indefinable quality. Actually, his dismissal of this Moorean idea(and the corollary that we somehow discern such a quality) becomes the springboardfrom which he launches his own definition (Ewing, 1959, p. 81). While he thoughtthat we are unable to obtain a clear idea of goodness, he maintained that this is not thecase with obligations. In his view it was:

. . . quite clear in fact, that we are aware of obligations and that the concept of obligation is quitedistinct from an empirical concept. (p. 82)11

8 Ibid., p. 25.9 Ibid., p. 26.

10 Brentano would not, in other words, regard preferences as attitudes that aim (at least, directly) at therealization of some state of affairs.

11 Just how much trust Ewing had in introspective knowledge is not clear to me. There are passages inSecond Thoughts in Moral Philosophy suggesting that he regarded introspective awareness (e.g. of obligations)with some suspicion, as something that does not necessarily disclose something in the real world (see 1959,


Originally, in the early work The Definition of Good (1947), Ewing distinguished betweentwo sorts of obligation and two corresponding oughts: he understood the first, andprimary, notion as “signifying ‘fittingness’” (1947, p. 151), which he took to be an“irreducible ethical concept”; and the second as the “‘ought’ of moral obligation” (see1959, p. 90). It is not obvious why he refers to ‘fittingness’ as an ethical concept. Thepassage would make more sense had he characterized it as a normative concept. He thendefined goodness, in terms of the first kind of ought, as a “fitting object of a proattitude”.12 However, later, in Second Thoughts in Moral Philosophy, he modified this view:

I still recognize these two meanings of ‘ought’, but now I wish to lay more stress on the moral‘ought’ and to reduce the ‘ought’ of fittingness to reasonableness. (1959, p. 90)

He still defines ‘good’ as “what ought to be the object of a pro-attitude”, but his newposition stresses that both the moral ‘ought’ and the ‘ought of reasonableness’ areunique in the sense that they cannot be reduced to other concepts. Certain things maybe morally good, and we morally ought to adopt a favourable attitude to these objects.Other things are good in a non-moral sense: they justify, or provide reasons for, theattitude. For example, my own pleasure is such that I ought to adopt a positive attitudeto it, but the ‘ought’ here is hardly the moral ‘ought’ but rather the ‘ought ofreasonableness’: pleasure provides a reason to adopt a pro-attitude towards it, but notnecessarily a moral reason.13

Brentano worked with a limited set of pro-attitudes. Ewing’s notion of a pro-attitude extends to a much wider range of attitudes.14 In fact, he takes it to cover“any favourable attitude to something”: “choice, desire, liking, pursuit, approval,admiration”.15 He also underlines the notion that different pro-attitudes, or evenconglomerates of such attitudes, fit different kinds of valuable object; and he takesthis to show that ‘good’ can have different senses.

p. 54). For a most interesting essay on Ewing’s metaethical views, see the forthcoming ‘A. C. Ewing’s Firstand Second Thoughts on Metaethics’, by Jonas Olson and Mark Timmons. At the end of his career Ewingdeveloped a peculiar metaethical hybrid theory combining elements of both non-cognitivism and cogniti-vism. Olson and Timmons argue convincingly that this theory is flawed in several respects.

12 Ewing (1947, p. 152).13 As mentioned earlier, Ewing’s metaethical views changed considerably over the years, and in ways that

are not always entirely transparent. Perhaps the most noteworthy change was that he gave up on the idea heonce had (in 1947) that evaluative language is primarily concerned with describing (i.e. ascribing properties),replacing this with a more prescriptivist (expressivist) account (in 1959). He then tried to combine the latterwith the idea that value judgements have truth-values.

14 His account of preferences is interesting and quite different from Brentano’s: “‘Prefer’ is such a generalword that I think it may be used to cover any case where a more favourable attitude is adopted to somethingA than to something else B, whatever the pro-attitude in question” (Ewing, 1959, p. 85). This seems tosuggest that, of two attitudes, x and y, if the former is more favourable than the latter, and I apply x to objectA, and y to object B, then I have a preference for A over B. Needless to say, such an account raises a numberof difficult issues. For instance, what makes us say that one attitude is more favourable than another? Isadmiration of A a more favourable attitude than the attitude of respecting B?

15 Ibid., p. 149.

2.2 A. C. EWING 23

2.3 The primacy of the normative over the evaluativeBrentano’s and Ewing’s decision to give primacy to normative concepts (suchas fittingness, correctness, ought, should, and reason) over evaluative concepts (princi-pally, the good) is, in my view, highly persuasive. Thus, the FA theory that will beproposed here reduces evaluative claims to deontic claims about attitudes, to thevaluable objects in question, that it is fitting to have, or that one ought to have, or thatwe have reason to adopt. The theory developed here departs from Ewing’s later view,however. Where he distinguished between two sorts of ought—the moral ought andthe ought of fittingness, interpreted as an “ought of reasonableness”—I will be morecautious. True, it is a central feature of values that they are primarily normative incharacter: we think there is a reason why a valuable object demands, requires, or callsfor, certain attitudes from us, and that this reason derives from the object. But whetheror not the reason is moral in character seems to me a secondary matter.16 At least, I havenot found any compelling reason to follow Ewing here. This is not to deny that thereare two oughts. It is merely to leave the matter open, and to assume that what isessential to an analysis of value is that it brings out the normative character of value.

Here it might be a good idea to point out, as Joshua Gert17 recently did, thatnormativity has two dimensions: There is the strong level of requirement—what weought to do—and there is the weaker level of permission—what we may do. AlthoughI will not pay any particular attention to these two levels, it is an important point tokeep in mind when I give examples of what there might be reasons to favour. Thesereasons may not all be about what we are required to do.

A noteworthy feature of the FA analysis is that, in it, it is properties other than thevalue property that provide reasons to respond to the valuable thing by taking up anattitude to it.18 The reason we ought to take a certain responsive stance is to be lookedfor among the subjacent properties and not in what supervenes on those properties.19 It

16 Zimmerman, in ‘UnderstandingWhat’s Good for Us’ (2009), does not seem to agree with this. The factthat I leave it unspecified whether the reason I have in mind is prudential or moral is a problem for my view,according to Zimmerman. Perhaps. However, the concept of personal value I have in mind does not implyanything other than that something is a reason—what this reason is will become clear, eventually, when wetake into consideration the supervenience base of the value. However, since there is no mention of anysubjacent properties in the analysis, it cannot be determined whether the reason is moral, prudential, or someother kind of reason. There is nothing peculiar about this; in my view we often identify something as a reasonwithout having a clue as to whether the reason is moral, prudential, or of some other kind.

17 See Gert, ‘Value and Parity’ (2004).18 Fittingness (worthiness, correctness) might be regarded as a primitive notion that cannot be subsumed

under standard generic deontic notions, including notions such as ought, must, and should. On this view,fittingness constitutes a special kind of deontic notion. Ewing took this approach in his early work TheDefinition of Good (1947). See also Danielsson and Olson (2007).

19 Cf. Dancy, ‘Should We Pass the Buck?’ (2000a, p. 161): “Ewing seems . . . to be in a position to say thatgoodness is not a distinctive evaluative and intrinsic property in objects, one whose presence we can discernand to which we do or at least should respond with approval and admiration. The goodness of the object justis the relational fact that we should respond to it with approval, admiration or other pro-attitude. Theevaluative ‘good’ has been defined in terms of the deontic ‘should’. And with this result, the intuitionists


is the set of value-making properties that provides us with a reason to take a pro-attitude to the valuable. Following Scanlon (1998), we could say that what carries thebuck is not value but what value supervenes on! In other words, to be valuable is tohave the property of having a value-making base that gives us reason to have a pro-response to the valuable object. A caveat is in place here, however: although I referreda moment ago to the supervenience base of the valuable object in order to explain whatthe analysis is all about, it must not be assumed that the analysis somehow directly refersto the relevant subjacent properties.20 The FA analysis is, in effect, a very simpleanalysis; it neither mentions nor refers to the properties in virtue of which the objectis valuable. It just says that to be valuable is to be an object there is reason to favour.The FA analysis defended in this work is pluralistic in the sense that it acknowledges

that final value accrues to different kinds of metaphysical entity, including abstract statesof affairs and concrete objects such as persons.21 Great importance is attached to theidea that different kinds of valuable object invite different kinds of pro-response. Forinstance, a brave person ought to be admired, a precious artefact calls for care andprotection, and a desirable state should be desired. In fact, in many cases we maydescribe the response that is fitting more accurately as a conglomerate of different kindsof attitude—or even as a conglomerate of various kinds of attitude-cum-behaviour.Some of these attitudes are especially thing- or person-oriented: we care for ourchildren; we cherish and preserve objects of historical importance like the MagnaCarta; we protect a part of the Brazilian rainforest that is as yet untouched by humans.Other attitudes are instead state of affairs-orientated. The most obvious cases are desiresand preferences, both of which typically take states of affairs (e.g. that Denmark willqualify for the World Cup, or that people do not say bad things about one behind one’sback) as their objects rather than (concrete) things and persons.

2.4 Different value idiomsThe FA analysis has typically been employed to analyse what is valuable ‘for its ownsake’ or ‘as an end’ or has value ‘in its own right’. However, none of these descriptors is

reversed Moore’s position . . .Moore defined the right, that which we ought to do or should do, in terms ofthe good. Ewing defined the good in terms of how we should respond.”

20 This is important when it comes to the question whether evaluative properties may themselves bereason-providing. As outlined here, FA analysis leaves this issue open, and as far as I can see there are goodreasons to allow that value may supervene on other kinds of value. Scanlon, for instance, was originally of adifferent opinion, believing that the subjacent properties were all natural properties. However, he changedhis mind in response to R. Jay Wallace, ‘Scanlon’s Contractualism’ (2002, pp. 447–9): see Scanlon, ‘Reasons,Responsibility, and Reliance: Replies to Wallace, Dworkin, and Deigh’ (2002, p. 513).

21 See, for instance, Elisabeth Anderson, Value in Ethics and Economics (1993); Christine Swanton, ‘Profilesof the Virtues’ (1995); Marcia W. Baron, ‘Kantian Ethics’ (1997); Wlodek Rabinowicz and Toni Rønnow-Rasmussen (2003a). For monist approaches see Noah M. Lemos (1994); Michael J. Zimmerman (2001a) andhis ‘Intrinsic Value and Individual Worth’ (2001b); and Jonas Olson, ‘Revisiting the Tropic of Value: Replyto Rabinowicz and Rønnow-Rasmussen’ (2003).

2.4 DIFFERENT VALUE IDIOMS 25

quite satisfactory. The notion that something is valuable as an end, for instance, appearsnot to be consistent with the possibility that the value bearer is a non-propositionalobject, such as a thing (a painting, a stamp, or the like). It seems right, at least, to agreewith W. D. Ross that ends are objectives and never things. In this sense, the other twoexpressions seem to represent an improvement. Both ‘in its own right’ and ‘for its ownsake’ imply something about whether or not the value is dependent on somethingother than the valuable object.

Of the two expressions ‘in its own right’ and ‘for its own sake’, I prefer the second.The notion that something is valuable in its own right might lead one to wonderwhether the object has value in virtue of a capacity of the object that has been realized.If that were so, this kind of value would also exclude, it seems, certain objects from thefield of value bearers. (Would a state of affairs, say, have value in its own right in thissense?) ‘In its own right’ also seems to suggest that the value accruing to an object doesso without the help of a subject. This suggests that the expression would lend itself tosome extent to an objectivist analysis of value.

Admittedly, the advantage of the phrase ‘for its own sake’ may not be immediatelyobvious, and certainly it has a somewhat counter-intuitive ring in expressions such as ‘xis valuable for its own sake’ (compare ‘x is desired for its own sake’). However, I prefer it.Following others, I will say that something that is valuable for its own sake has final value.My reason is that the value in question here is a natural counterpart, in a sense in whichthe value picked out by ‘in its own right’ is not, to two other familiar kinds of value:contributory value (or what is valuable for the sake of some whole) and instrumentalvalue (or what is valuable for something else’s sake rather than for its own). Moreover,‘valuable for its own sake’ does not impose bias on the value-bearer issue; it rules outneither abstract states nor concrete things, unlike the notion of something being valuableas an end. Therefore, in what follows I will speak of final values in terms of what isvaluable for its own sake (and, accordingly, of what is good/bad for its own sake anddesirable for its own sake). If the value is intrinsic (one kind of final value), the object isvaluable for its own sake in virtue of its internal features; if the object has extrinsic finalvalue, it is valuable for its own sake in virtue of at least some external relational feature.

Rae Langton has objected to this sort of characterization: “to talk of somethinghaving ‘final value’ that is ‘extrinsic’ is to talk of valuing for the thing’s own sake,something that has extrinsic value. We keep the talk of finality and ends to a distinction,not in the way things have value, but in the way we value things” (2007, p. 165).Although Langton’s proposal has its merits, ultimately I think we should resist it. Whatdetermines whether a value is intrinsic or extrinsic is, in the end, the nature and scopeof the supervenience base. Moreover, her suggestion is not obviously consistent withthe pattern of value analysis that I employ.

It should be stressed, though, that there is a sense in which my aim here concernsonly indirectly final values. My main concern in this book is to discuss the extent towhich some version of the FA analysis, if true, is helpful when it comes to understandingpersonal values.


The caveat ‘if true’ here is important, and it will occupy us in a moment. Mean-while, some further features of the FA analysis need to be highlighted.

2.5 The FA analysis of final positive valuesFor reasons of simplicity let ‘favour/disfavour’ be schematic place-holders for a range ofpositive and negative responses called for by various kinds of valuable object. Appliedto final positive values, the analysis then says the following:

The FA analysis of final positive values: The final value of an object x consists in theexistence of normative reasons for favouring x for its own sake.

This pattern of analysis can be extended to other kinds of value, too. For instance, anobject’s instrumental value might be said to consist in the existence of reasons to favourit for its effects’ sake. That is, given that instrumental value is in fact a kind of value.That this is not obviously the case is something that I will comment on in Section 4.2.The FA approach also applies, and equally well, to both thick and thin varieties of

final and instrumental value.22 Thus an admirable person is a person whom there isreason to admire, a respectable person should be respected, and so on. Of course, thematch between the value and the fitting attitude is not always as obvious as it is in thesecases. For instance, to be brave is also, for the FA analyst, to be the object of anormatively required attitude. The question, of course, is what kind.That the FA pattern of analysis might encounter a problem with certain values has

recently been pointed out by Roger Crisp:

A painting may be beautiful because of its sublimity, delicacy, profundity, boldness, imagination,vitality, grace, honesty . . . According to the version of BPA [i.e. buck-passing analysis23] we areconsidering, since these properties are evaluative, we might assume that buck-passing accounts ofthem also ought to be available. How will BPA distinguish them? Reference to the appropriateresponse may be appropriate in certain cases, especially those in which reference to a response isanalytic to the concept in question. ‘Being awesome’, then, may be said to be the higher-orderproperty of having lower-order, natural properties that provide a reason for awe. But the BPA isunlikely to be able, by reference to different responses, to distinguish properties with no suchanalytic reference to responses themselves. Consider, say, grace and delicacy. To be sure, they callfor certain responses, but the responses themselves are too similar to enable us to distinguish theproperties.24

22 For the distinction, see Bernard Williams Ethics and the Limits of Philosophy (1985).23 The term “buck-passing analysis” is due to Scanlon (1998, pp. 95–100).24 Crisp, ‘Value, Reasons and the Structure of Justification: How to Avoid Passing the Buck’ (2005). See

also Andrew Reisner, ‘Abandoning the Buck Passing Analysis of Final Value’ (2009). That it is often notpossible to pair up emotions and values in any neat way was also argued by Kevin Mulligan in ‘FromAppropriate Emotions to Values’ (1998, see pp. 174–5).

2.5 THE FA ANALYS IS OF F INAL POSIT IVE VALUES 27

Is this pattern of analysis too blunt an instrument, then, to deal with values like graceand delicacy? Could it discriminate between these kinds of value? Crisp offers tworeasons for thinking that the buck-passing, or FA, analysis fails. First, thick values do notalways connect with a unique attitudinal response. For instance, being delicate does nothave an attitude or emotional response that is appropriate to it. Second, the propertiesin virtue of which the object is delicate—what Crisp refers to as the lower-ordernatural properties—are too coarse-grained to enable us to distinguish between valuessuch as delicacy and gracefulness.

Crisp’s first observation is correct. As to his second point, I am not sure whether itreally is impossible to detect a difference between delicacy and gracefulness in terms oftheir supervenience bases. Crisp thinks that an account of these values in purelynaturalistic terms will not do. With this I fully agree, but that is because I am inclinedto deny that naturalism is a viable theory in the first place. However, this denial doesnot commit me to holding that we cannot differentiate between values in terms of theirlower-order properties, or what I would refer to as their supervenience bases. We can,even if doing so is often a difficult business; it is just that there might be more to valuethan their supervenience bases.

Thus, in response to Crisp, I would suggest that it is the combination of lower-orderproperties and their reason-making nature that enables us, in principle, to differentiatebetween these two values. In the case of a delicate object there will be a set ofproperties such that they provide us with a reason to feel, say, delight. In the case ofgrace, another set of properties will give us reason to have another attitude—say,aesthetic enjoyment (as has been suggested by Jonas Olson).25 In fact, since the twovalues are very similar, they might well provide us with a reason to adopt the sameattitude to the object. The difference would then be accounted for only in terms of thesupervenience bases. For instance, the properties that make a vase into a delicate vase donot seem to be the properties thatmake it into a graceful vase; among the latterwewouldfind, say, its flowing shape, whereas a delicate vase (in at least one sense of ‘delicate’)would have properties that have something to do with how easily it is damaged—say,that it is made of porcelain. Of course, things are often both delicate and graceful, and sothe supervenience base will contain common elements. Again, this is not in any way toassert, or imply, that these values can be captured in a purely naturalistic analysis. Arejoinder might be to say that, if these claims are correct, we will have to admit that weare really dealing not with two different values at all, but with one kind of value that werefer to differently because it has two distinct supervenience bases. But supposing thatthis were the case, would that be a serious problem? I do not think so.

Crisp also suggests that the buck-passing, or FA, analysis fails in another aspect. Theexperience of beauty is not merely the experience of the natural properties in virtue ofwhich the object is beautiful. The reason we admire the artefact is that it is beautiful:

25 Olson, ‘How to Pass the Buck’ (2004).


. . . the reason for admiration lies not in the natural properties—these could be understood bysomeone with no aesthetic sense—but in the beauty.26

The issue Crisp raises here is both complex and deserving of detailed discussion.However, I will confine myself to a short reply. Even a buck-passer might go alongwith the possibility that two persons do not reach an agreement onwhether something isbeautiful or not despite perceiving the same natural properties. But instead of invokingthe ability of one, and the disability of the other, to perceive non-natural properties, theexplanation would be that one grasped and the other failed to grasp that the naturalproperties are reason-providing properties. Moreover, I am not confident that my own‘value phenomenology’ rules out the latter explanation; I cannot be certain that forphenomenological reasons the first explanation is better (or worse) than the second.Given this, and given that the second explanation does not require us to posit a non-natural property, the FA pattern of analysis appears to be the more plausible one.27

There is another recent objection to the FA analysis that deserves a comment.28 Itwas first pointed out by Wlodek Rabinowicz in ‘Value Relations’,29 and later MarkAlfano has expressed a similar point.30 Alfano’s argument sets out from the idea that itis conceptually true that ‘good’ and ‘bad’ are polar predicates, i.e. they are contraries.And so an analysis of good and bad should keep this polarity in its analysans, but the FAanalysis runs into problems here since it understands value in terms of attitudes that arenot necessarily contraries. There is nothing conceptually problematic about being in anambivalent state of favouring and disfavouring one and the same object. So if wesuppose that it is actually fitting to be in such a state, then on FA, we need to say thatthe object is both good and bad. Now, a modified FA need not be endangered by this.Suppose we take the FA to say something to the following effect: x is good if and onlyif it is fitting to favour x and it is not fitting to disfavour x. This should handle this sortof challenge.31 Rabinowicz’s point (2008, p. 40) is more general, though; FA reducesconceptual truths concerning value notions to formal constraints on appropriateattitudes. For instance, consider the ‘betterness’ relation. As Rabinowicz suggests thisseems to be as a matter of conceptual truth, a transitive relation. However, it is notobviously a conceptual truth that preference in all permissible preference orderings istransitive (e.g. I may prefer x to y, and I may prefer y to z, but it is not permissible that I,at the same time, do not prefer x to z). So if the FA analysis forces us to treat apermissible preference ordering as transitive, (which is not unreasonable but nonethe-less not conceptually true), it obviously has non-tautological implications. Of course,

26 Crisp (2005, p. 82).27 Crisp raises other objections as well. For discussion of these, see Olson (2004).28 Recently Crisp has returned with a new objection to the buck-passing account. I will return to this

objection at the end of the next chapter.29 Rabinowicz, ‘Value Relations’ (2008).30 Alfano, ‘A Danger of Definition: Polar Predicates in Moral Theory’ (2009).31 The point is not mine. Unfortunately I cannot remember its source, or the person who originally

suggested this solution to Alfano’s challenge.

2.5 THE FA ANALYS IS OF F INAL POSIT IVE VALUES 29

an FA analyst should perhaps bite the bullet here. After all, as I have pointed out earlier,FA is to some extent a revisionary analysis. However, there is another, albeit moreradical, possibility, namely that we need not understand ‘better’ in terms of dyadicattitudes (such as preferring x to y). Perhaps we could understand ‘better’ in terms of itbeing fitting to have certain monadic attitudes rather than other monadic attitudes. Wecould then account for the ‘betterness’ of x over y in terms of the difference of strengthbetween the two monadic attitudes. For example x is better than y, if and only if, it isfitting to, say, like x to a degree n+1, and it is fitting to like y to a degree n. Needless tosay, whether such an analysis can be worked out in a plausible way is an interestingissue. However, I will not pursue this matter here.32 Instead, at this point it isconvenient to draw attention to one of the earliest attacks on FA analysis.

2.6 W. D. Ross’s objectionRoss objected to the whole FA project.33 Attempts to reduce statements such as ‘x isgood’ to ‘it is fitting to admire x’ appeared to him not to offer a genuine analysis:34 inRoss’s view, FA ‘analysis’ got things backwards. However, this was not his only worry;FA seemed to him to also involve a circle:

[A]dmiration is not a mere emotion; it is an emotion accompanied by the thought that thatwhich is admired is good. And if we ask on what ground a thing is worthy of being thought to begood, only one answer is possible, namely that it is good. It would be absurd to say that a thing isgood only in the sense that it is worthy of being thought to be good, for our definition of ‘good’would then include the very word ‘good’ which we were seeking to define.35

To analyse goodness in terms of the attitude of admiration would be circular, ifadmiration involves a judgement to the effect that the object admired is good. Thisis an important objection.36 Given his other claim, then—that goodness is the veryfeature in virtue of which the object is, to use Ewing’s original phrase, worthy ofadmiration, and cannot therefore be understood in terms of being worthy of admira-tion—Ross appears to be challenging advocates of FA analysis on two fronts.37

32 Recently Rabinowicz has developed what a modelling of value relations would look like from amonadic-centred perspective; see his ‘Values Compared’ (2009).

33 Ross, Foundations of Ethics (1939, pp. 276, 278). His direct target was Ewing’s view.34 At least, if what is valuable is an action. If it is instead an experience of pleasure, then, insofar as it is

valuable, it is a fitting object of satisfaction rather than admiration (ibid.).35 Ibid. p. 279.36 According to Mulligan, “Meinong [in (1968) Abhandlung zur Werttheorie, Gesamtausgabe, Vol. 3, Graz:

Akademische Druck- u. Verlagsanstalt] noted that analyses of value in terms of justified or correct emotionsavoid circularity only if justification and correctness are deontic properties and if deontic properties andaxiological properties are different types of property. He accepted both claims and took deontic propertiesand value-properties to be formal objects of desire and of emotions, respectively” (2009b, p. 490).

37 The circularity need not be vicious. It depends, I suppose, on what we hope to accomplish by theanalysis. For instance, David Wiggins accepts the circularity. He argues that, owing to its “detour through

30 FITTING-ATTITUDE ANALYS IS

Ewing replied to Ross’s criticism, arguing:

[T]he reason why it is proper to admire anything must be constituted by the qualities whichmake the object of admiration good, but it does not follow that the thought that it is good must,if the admiration is to be justifiable, intervene between the perception of the factual qualitiesadmired and the feeling of admiration.38

[T]he ground [for a pro attitude] lies not in . . . goodness, but in the concrete, factualcharacteristics of what we pronounce good. Certain characteristics are such that the fittingresponse to what possesses them is a pro attitude, and that is all there is to it.39

Thus (i) admiration is a response to the object’s ‘good-making’ qualities and not to itsgoodness. For this reason, (ii) the attitude of admiration need neither involve, norpresuppose, any judgement that the object admired is good. Do pro-attitudes have toinvolve any evaluations? In recent years Ross’s position has been questioned. JustinD’Arms and Daniel Jacobson,40 in particular, have mounted an attack on various formsof ‘judgementalism’ and ‘quasi-judgementalism’41 that purport to analyse emotions:the fact that evaluative language may be useful for describing our emotions should notlead us to believe that all emotions essentially involve some judgement of value. Andlikewise, it would be a mistake to think that to be in an emotive state the agent has tobe entertaining evaluative judgements or concepts.42 The evaluative concepts weemploy to depict emotions need not correspond to any ‘phenomenological’ realityin the emotion itself. D’Arms and Jacobson therefore join ranks with Ewing inrejecting the Ross-style circularity objection to the FA analysis.D’Arms and Jacobson have a point, as I see it. However, even if they were mistaken,

and emotions somehow did necessarily involve evaluations or were defined in terms ofcertain values, this would not in itself be an insuperable obstacle to the FA approach. Itwould probably present a problem for any attempt to provide a semantic reduction ofvalue claims to claims about the attitudes and emotions we have reason to have. But theFA analysis is not (at least, primarily) an exercise in semantic reduction. Suppose thatthe attitudes FA employs can be characterized in ways other than by their relation tovalue. If so, there is a point to explicating value in terms of a normative component andthese attitudes. At some level such an approach might well be circular, but the circlewould be informative.

sentiments”, the analysis remains informative to some extent; see David Wiggins, ‘A Sensible Subjectivism?’(1987, p. 189). Cf. D’Arms and Jacobson, ‘Sentiment and Value’ (2000b, pp. 732ff.).

38 Ewing, The Definition of Good (1947, p. 158).39 Ibid., p. 172.40 See D’Arms and Jacobson, ‘The Significance of Recalcitrant Emotion (or, Anti-Quasi judgmentalism)’

(2003). Also relevant is their ‘Sentiment and Value’ (2000b).41 Quasi-judgementalism is, according to D’Arms and Jacobson, a revision of judgementalism in that it

“still type-identifies the emotions by their defining propositions, and claims that certain thoughts are partiallyconstitutive of being in an emotional state, but loosens the requirement that these thoughts must be affirmedby the agent” (2003, p.130).

42 For an overview of different positions on this matter, see Mulligan’s ‘Emotion, Value, and Morality’(2009b).

2.6 W. D. ROSS ’S OBJECTION 31

Skorupski has recently addressed this issue. Supposing the morally wrong is thattowards which there is reason to feel blame, he considers the objection that this (FA)proposal:

cannot yield a definition of ‘morally wrong’, since ‘blame’ in the relevant sense should be definedas that sentiment which is appropriate—which there is reason to feel—towards the morally wrong.A good objection, but not the end of the story. Suppose that there is a distinctive sentiment ofblame whose object is the morally wrong; suppose also that we can be given an independentcharacterization of this sentiment, say in terms of the actions to which it disposes, or by having itexplained as the sentiment we feel when we consider some specific paradigm cases. Suppose,finally, that when we have been made familiar, in the first-person way, with the sentiment, wefind ourselves able to go on spontaneously, and reasonably convergently, making confident newjudgements about when there is reason to feel that specific sentiment. In that case we haveeverything we need to grasp the concept of moral wrongness. The morally wrong is that which isblame-sentiment-worthy in the absence of an excuse. The concept can in this sense beexhaustively captured in terms of the concept of a reason and the concept of a certain sentiment,even though ‘morally wrong’ cannot be defined in terms of ‘reason’. Furthermore if we canindividuate the sentiment by these methods we can also use them to introduce a term to refer tothe sentiment, say ‘BS’. The morally wrong is that which is BS-worthy (which there is sufficientreason to respond to in that way) in the absence of an excuse.43

Skorupski is right, I think. We can indeed say something about what it means to be inpossession of the concept moral wrongness even if we cannot semantically reducestatements about what is morally wrong to statements about what there is reason toblame. Once we grasp the concept of a reason and the concept of blame (particularly interms of a morally wrong independent characterization) we are in a position to graspmoral wrongness. Even if we allow, then, that emotions and attitudes necessarilyinvolve evaluations or somehow relate to values, it remains to be shown that the sortof circularity involved in FA analysis, given such an allowance, would be devastating tothe analysis. Some, but far from all, circles are vicious.44

43 Skorupski, ‘What is Normativity?’ (2007b, p. 249). That FA analysis need not involve a vicious circlewas considered as a possibility in Rabinowicz and Rønnow-Rasmussen (2004a, 2006).

44 See I. L. Humberstone’s seminal ‘Two Types of Circles’ (1997). See also J. A. Burgess, ‘When isCircularity in Definitions Benign?’ (2009).

32 FITTING-ATTITUDE ANALYS IS

3

The Wrong Kinds of Reason

The fitting attitude (FA) analysis of value has many attractions, but it also faces severalmore or less serious internal problems, and these will need to be acknowledged andexamined. Its more salient positive features will emerge, eventually. But at this stage itis only fair to the reader to present what has been thought of as a major problem for theanalysis. This is far from the only problem. Some of the problems concern the analysisin general;1 others relate to my specific application of it to personal values. The latterwill be examined in due course. In this chapter I focus mainly on what WlodekRabinowicz and I have called the ‘wrong kind of reason’ problem (henceforth, theWKR problem). At the end of the chapter I will comment briefly on another problem,one that concerns whether an FA analyst has to admit that goodness cannot be areason-giving property.

3.1 The wrong kind of reason problemThe FA analysis equates value with the existence of reasons to favour the objectcarrying the value. One obvious advantage of the FA analysis is that it satisfies arequirement that all formal axiological theories should meet: it is silent as to whatobjects carry value. Substantive axiological reasoning aims to tell us what kinds of thingare in fact valuable, but this is not the objective of formal axiology. This notwithstand-ing, if the analysis is too generous and implies, counter-intuitively, that certain kinds of

1 Although I do not share his conclusion, Krister Bykvist’ s recent article ‘No Good Fit: Why the FittingAttitude Analysis of Value Fails’ (2009), contains an interesting discussion concerning fitting attitude analysesthat take non-obtaining states of affairs to be valuable. Bykvist considers whether we can favour a (non-obtaining) state of affairs such as “there being happy egrets but no past, present or future agents (i.e. beingswho intentionally bring something about)” (p. 5). He considers several possible ways of understanding “tofavour” something, but none of them, he thinks are such that we can reasonably say that it is fitting to favoursuch a state. For example, it would not make sense to say that it is fitting to promote such a state or to desire it.This certainly seems true. However, it might be fitting to direct some optative attitude on this state of affairs.We can certainly wish that it obtains even if we know that it cannot be promoted.

apparently valueless objects are valuable, then there is obviously a reason to be scepticalabout it. The problem—perhaps, indeed, the biggest problem—facing advocates of FAanalysis is precisely this: the FA analysis seems to open the door to worthless objects,and to construe them as bearers of value. Equally, we might have reasons not to havepro-attitudes to objects that obviously are valuable.2 Discussions of the WKR problemhave focused mainly on the first kind of case. Most likely the explanation is that in casesof the second kind we do have reasons to have a pro-attitude, but these reasons areoutweighed by reasons not to have such an attitude. So, these are not counter-examples to FA analysis. Here are two examples in which there seem to be perfectlygood reasons to take pro-attitudes to objects of no value at all:3

(i) A demon threatens you with punishment if you do not admire a saucer of mud.4

(ii) A demon threatens you with punishment if you do not admire him.5

These examples appear to take the FA analyst straight into a cul-de-sac. Here is why:the saucer of mud has no value (or so we assume); yet, if the analysis is correct, we areforced to say that it does have value, since there are reasons to take up a positive attitudeto it. The same goes for the demon in (ii). What is significant about not only (i) and (ii)but WKR cases in general is that it is seemingly the pro-attitudes themselves, ratherthan the objects, that are valuable in these cases: it is the admiring of the saucer ordemon, not the saucer or demon in themselves, that carries value. However, in thecontext of the FA analysis, the fact that we have reason to admire the object is justanother way of saying that these objects are in fact valuable. We are apparentlycornered. Objects which obviously are not valuable are depicted in the analysis asbeing of value.

2 Justin D’Arms and Daniel Jacobson have identified and discussed this difficulty, which they refer to asthe “the Conflation Problem”, in their ‘Sentiment and Value,’ (2000b). See also D’Arms and Jacobson(2000a) and their ‘The Significance of Recalcitrant Emotion (or, Anti-Quasi judgmenta-lism)’ (2003).

3 As Sven Danielsson and Jonas Olson have pointed out in their recent ‘Brentano and the Buck-passers’(2007), G. E. Moore raised a kind of WKR objection that would be an example of the other type of case. Inhis review of the English translation of Brentano’s Vom Ursprung Sittlicher Erkenntnis, Moore argued thatbeautiful objects are such that we should favour them. However, this does not mean, he thought, that valueaccrues to the beautiful object; it is rather our appreciation of them that is valuable. Moore’s contention is notuncontroversial, though. It is not obvious that it is only our appreciation that is valuable, but not the beautifulobject itself.

4 This example is from Roger Crisp’s review of ‘Value . . . and What Follows’ by Joel Kupperman(2000a). If you prefer a more realistic example, recall the scene from the movie Catch 22 in which the twocorrupted colonels, Cathcart and Korn, offer to send Captain Yossarian home from the war (they want tohush up an ugly affair they are involved in), something the captain has wanted for a long time. There is acatch, though. Yossarian hates the colonels, and their offer comes with a price. Yossarian has to, as thecolonels put it “like us”. The scene is at the end of the movie.

5 This example is discussed in Wlodek Rabinowicz and Toni Rønnow-Rasmussen (2004a); it wasoriginally suggested by Folke Tersman (personal communication).

34 THE WRONG KINDS OF REASON

Several philosophers have tried to solve this problem. To do justice to all theattempts would require too much space, but I will say something about what is perhapsthe most popular general approach. It sets out from Derek Parfit’s distinction betweenstate-given and object-given reasons for attitudes:6

Of our reasons to have some desire, some are provided by facts about this desire’s object.These reasons we can call object-given. We can have such reasons to want some thing either forits own sake, or for the sake of its effects . . . Other reasons to want some thing are providedby facts, not about what we want, but about having this desire. These reasons we can callstate-given. Such reasons can also be either intrinsic or instrumental.7

In other words, facts provided by the very attitude, and not the object of the attitude,introduce what Parfit calls state-given reasons; and facts that have to do with the objectof the attitude introduce object-given reasons.How does this distinction offer a solution to the WKR problem? Suppose we

have identified a reason for favouring an object. Given this, we can ask: Is ourreason for favouring the object an object- or a state-given reason? The straight-forward solution on offer at this point is to claim that the right kind of reason willalways be an object-given reason, and the wrong kind of reason always a state-givenreason.8

Recall the example of the saucer of mud. This is an object without value, as (we aresupposing) is the demon. In cases (i) and (ii) we seem to have a reason to admire theobject (the saucer or the demon), but the properties that provide these reasons are ofthe wrong kind: they belong to the attitude rather than to the object. If I do not admirethe saucer or the demon, I will be punished. This is a fact about the attitudes involved,not about the saucer of mud or the demon.But the success of this approach is not quite as clear-cut as it seems to be. There is a

complication, as Rabinowicz and I pointed out:

It is easy to see that for any property P of the attitude there is a corresponding property P’ of theobject: If a pro-attitude towards an object a would have a property P, then, ipso facto, a has (orwould have, if it existed) the property P’ of being such that a pro-attitude towards it would havethe property P. Consequently, to the attitude-given reason, provided by P, corresponds theobject-given reason, which is provided by P’. In exactly the same way, of course, for anyproperty P of the object of the attitude there is a corresponding property P’ of the attitude itself:the property of being such that its object has (or would have) property P. Thus, to each object-given reason corresponds an attitude-given reason, and vice versa.9

6 Parfit, ‘Rationality and Reasons’ (2001).7 Ibid., pp. 21f. Parfit returns to this distinction in his forthcoming ‘OnWhat Matters’. See its Appendix B.8 Parfit himself denies in ‘Climbing the Mountain’ (circulating ms) and in ‘OnWhat Matters’ (circulating

ms) that state-given reasons are reasons for favouring, and so he would deny that there is a WKR problem inthe first place. I will return to this approach in a moment.

9 Rabinowicz and Rønnow-Rasmussen (2004a, p. 406).

3.1 THE WRONG KIND OF REASON PROBLEM 35

To differentiate between these two sorts of property—between “object” propertiesand “attitude” properties—will, then, not obviously serve to remove the presentdifficulty.10

Now, there might in fact be a rather easy way out of this problem, namely toexclude all relational properties from the supervenience base of value. At least, it mightseem we can do this for final value. If it is only among an object’s internal, non-relational features that we find value-making properties, and if, therefore, final value isnothing but intrinsic value, which the object has in virtue of its internal features alone,then the WKR issue seemingly will never arise. And so it can be argued, then, that notevery kind of FA analyst has this problem to contend with—only those, like me, whoconsider final value to be extrinsic in some cases.11

There is a rejoinder, however, that questions the idea that the WKR problem onlyarises for final extrinsic value. All we need to do is to find an object whose internalfeatures are such that they provide us with a reason to favour the object despite the factthat it is obviously not valuable. If there were such an object we would have to be able,as FA analysts, to distinguish the sort of reason involved in such a case from the kindemployed in an FA value analysis. Perhaps there are such objects. Rabinowicz recentlysuggested to me (personal communication) that a fact of the following kind might be anexample: I will be punished if I do not favour this fact (that I will be punished . . . ). This is apeculiar fact—a self-reflexive one. It does seem to give me a reason to favour the threat.So if we are ready to accept this particular kind of fact, then it does seem as the nature ofthis sort of reflexive fact is such we have a reason to favour a valueless object in virtue of

10 In ‘How to Deal with Evil Demons: Comment on Rabinowicz and Rønnow-Rasmussen’ (2005),Phillip Stratton-Lake is worried that if we tolerate such properties as that of being an object whose favouringwould have property P, we will end up with an objectionable ontology. This might be true if all such ‘mirrorproperties’ had an ‘imaginary’ touch to them, but this is not the case. For examples of more robust objectproperties that reflect the properties of the attitudes, see Rabinowicz and Rønnow-Rasmussen (2004a, p.407). For example, take the case of the demon who demands that we admire him rather than the saucer ofmud, and who makes it clear that he will strike us if we do not. We seem to have a reason to admire thedemon which is of the wrong kind. That the demon is disposed to strike is not something that makes himadmirable or valuable in a positive sense. Nonetheless, his disposition to hurt us provides us with a reason foradmiring him. But this feature of the demon certainly is a most robust feature.

11 Even if we were to allow relational properties among the value-makers, could we not get rid of theWKR problem by qualifying the supervenience base in some way? For instance, I discussed the followingexample with Heath W. White (personal communication): consider a thick value such as praiseworthiness.We might want to say that if, and only if, a property contributes to the overall purpose for which we have thegeneral practice of praising can it be a praiseworthy-making property. So if we were to find some object thatdoes not contribute, or even runs counter, to this overall purpose, we could be sure that it is not valuable. Butwhether or not we believe that there is such an overall purpose, this is not the way to handleWKR problems.Suppose we did argue that a certain alleged valuable object cannot be valuable, because the properties invirtue of which it is claimed to be valuable are not, or are not consistent with, the properties belonging to‘real’ valuable objects. But is it possible to determine the purpose of the practice of praising without taking astand on what are the right and wrong reasons for praising? I suspect it will be quite hard to do so. In otherwords, if what we do here is merely set out from a substantive view of the right kinds of reason, we have notachieved very much. In this particular case, I think it would be a mistake to restrict the properties on whichpraiseworthiness depends in the manner suggested above. Why should we be excluded from praisingsomeone just because he or she, say, tries to change the practice of praising?


its internal features alone. Rabinowicz’s suggestion therefore indicates that the WKRissue has a more general relevance than what Rabinowicz and I originally believed(Rabinowicz and Rønnow-Rasmussen, 2004a). At least certain objects give rise to theWKR issue in virtue of only their internal features.

3.2 Dual-role attitudesSeveral attempts to deal with this sort of problem have been made. The followingsuggestion is one that Rabinowicz and I have considered:

On the FA-analysis, the value of an object consists in there being reasons to favour the object inquestion. But favouring is not supposed to be a non-discerning attitude. Rather, the idea is thatwe are to favour the object on account of some of its properties. These properties, then, appear inthe intentional content of the pro-attitude. At the same time, they are supposed to make theobject valuable. Consequently, they also provide reasons for favouring the object. Thus, theyhave a dual role. On the one hand, (i) they appear in the intentional content of favouring as thefeatures on account of which the object is favoured. On the other hand, (ii) they justify favouringthe object in that way; i.e., provide reasons for the pro-attitude in question.12

In cases where the right kinds of reason are in play, properties of the valuable objectjustify our favouring of the object, but these properties also appear in the intentionalcontent of the attitude as those for the sake of which we favour that object. Forinstance, someone I am fond of, whose friendship I value very much, will call forvarious pro-attitudes, and these attitudes will have in their intentional content suchproperties of the friend as his kindness, wit, generosity, and so on. These are thequalities for which I like him. They are part of the intentional content of the attitude.But they also provide reasons for my liking him.Consider now the saucer of mud example again. Here the idea is that we should

desire it for its own sake, since doing so would protect us from the fury of the demon.In this case the property that justifies our favouring the saucer of mud for its own sake(i.e. the property of being such that favouring will protect us from the fury of thedemon) cannot appear in the intentional content of that favouring as the property onaccount of which we favour it in this way.The requirement that reasons play a dual role appears to explain why certain cases

involve the wrong kind of reason. Unfortunately, however, it cannot handle everykind of case. There are examples in which the reason for favouring an object does playa dual role but is still the wrong kind of reason. To see this, consider the demon case,(ii), in more detail: the demon wants us to admire him for his own sake precisely on account of hisdetermination to punish us if we do not do so (i.e. he wants to be admired as an individualmaking that threat). Under these circumstances, the dual-role requirement seems to besatisfied: the demon’s determination to punish us if we do not comply with his wish

12 Rabinowicz and Rønnow-Rasmussen (2004a, p. 414).

3.2 DUAL-ROLE ATTITUDES 37

provides the reason for our admiration and at the same time appears in the intentionalcontent of that attitude as the feature for which its object is being admired.

Certainly, the demon’s determination to inflict punishment if we do not admire himprovides us with a reason. But it is the wrong kind of reason, since it relates to a featurethat makes him worse, not better.

3.3 Solving the WKR problem?Should we then consider the WKR problem unsolved? Perhaps not. Let us return tothe saucer of mud example, and consider what might provide us with a way of tacklingWKR cases.

Let us call the attitude involved in admiring a saucer of mud W. As we have seen,two sorts of reasons seem to be involved here. That is, certain features of the mud giveus a reason not to admire it. (These are, in Parfit’s terms, the object-given reasons forthe attitude.) On the other hand, in cases involving a threatening demon, intuitively,we also seem to have reason to admire the saucer of mud, and this reason has to do withthe positive effects of having this attitude. Admiring the mud would prevent us frombeing punished.

Now, Parfit argues that the way I have just described this case is not ideal.13 It is amistake to think that the property of the attitude (of seeing to it that the demon doesnot carry out his threat) gives us a reason for the attitudeW. Rather, what we are given,according to Parfit, is a reason for a second-order attitude: a pro-attitude favouringpossession of the first-order attitude of admiring the mud or bringing it about thatwe admire the saucer of mud.14

John Skorupski has taken a similar approach to the WKR problem:15

Suppose the violin performance is not good, but the evil demon will punish me with eternaltorture if I fail to admire it. Is that not sufficient reason for me to admire it, even though it is notgood?16

Skorupski then suggests that:

. . . we can apply the distinction between reasons to believe or feel on the one hand and reasonsto bring it about that one believes or feels on the other. Thus: there is no sufficient reason for meto admire the performance, though there certainly is sufficient reason for me to bring it about

13 Personal communication.14 Personal communication. For clarification of Parfit’s case for saying that we should describe facts about

attitudes as giving us a reason to have a second-order attitude towards the attitude, see his ‘Climbing theMountain’, Chapter X.

15 See also John Skorupski, ‘Buck-Passing about Goodness’ (2007a), and Ingmar Persson, ‘Primary andSecondary Reasons’ (2007). Though the distinctions presented in these works are not identical to Parfit’s, allthree works take the same broad approach.

16 Skorupski, ‘What is Normativity?’ (2007b, p. 258).


that I admire the performance, if I can. In other words, in this case there is reason for me to bringit about that I admire something which there is no reason for me to admire. (p. 258)

Skorupski’s solution is then quite straightforward: the wrong kinds of reasons arereasons to cause or bring it about that we have attitudes (say, that you intend to dosome action); these should not be confused with the right kind of reasons to favour(say, admire the demon).But Parfit and Skorupski’s contention is surely somewhat contentious. It rests on the

far from obviously correct idea that having a reason to adopt an attitude can never giveus a reason to actually favour the object of the latter; it is always a reason to bring it aboutthat we favour some object (while not bringing about our actually favouring it). But itis hard to understand on what ground we make this categorical exclusion. Whatdifference between first- and second-order attitudes ensures that only the latter, notthe former, can be called for by certain kinds of reason-giving property? (Moreover,it might be suspected that we have smuggled in a substantive conception of reasons inorder to shut out the conceptual possibility that a reason to have an attitude providesthe subject, in effect, with a fully-fledged reason to favour the object rather than givingher merely a reason to bring about our favouring the object. However, since I am notsure whether this is in fact what is going on here, I will not press this point.17)Suppose we go along with the idea that facts about an attitude do not give us reasons

to have the attitude; they only call for us to favour having it. If we are ready to ignorethe ad hoc character of this move, we have a way of tackling WKR cases. However,this still does not clear the way for the FA analysis, for, as Rabinowicz and I havepreviously pointed out,18 a problem remains: we need next to distinguish what makessomething a reason for admiring, wanting, or desiring an object from what gives us areason to a have a second-order attitude to these ‘direct’ attitudes. The challenge is todo this without involving our notion of what is valuable. At the moment I am not surethis can be achieved. It might be replied that we could just say that something X is areason for an attitude F if and only if X makes F fitting or correct. This would, in effect,be to follow in Franz Brentano’s and A. C. Ewing’s footsteps and regard notions such as‘correctness’ or ‘fittingness’ as primitive notions in the analysis. Of course, it is neverideal to rest one’s analysis on an unexplained notion. For one thing, it seems quitepointless to replace one obscure notion (value) with another one (correctness, fitting-ness). If we do not know how to distinguish between considerations that make first-order attitudes fitting or correct and considerations that make second-order attitudesfitting or correct, nothing seems to be gained by taking such a step.

17 See Andrew Reisner, ‘Abandoning the Buck Passing Analysis of Final Value’ (2009), who proposes acounter-example to Skorupski’s solution: “One difficulty with Skorupski’s view is what I have calledelsewhere blocked ascent. In blocked ascent, the evil demon will cause trouble for you if you cause yourselfto have the relevant pro-attitude. The only way to get the prize, so to speak, is just to have the pro-attitude. Ifyou are lucky enough to already have it, then necessarily you can have it. So, there can be at least some casesin which the Skorupski solution will fail to provide any reason at all” (p. 388).

18 Rabinowicz and Rønnow-Rasmussen (2004a).

3.3 SOLVING THE WKR PROBLEM? 39

I would like to consider yet another approach, recently suggested by Sven Daniels-son and Jonas Olson (2007). What they are proposing is not, according to themselves,identical to the Skorupski (and, we might add, the Parfit) solution. This might, strictlytaken, be true. Nonetheless, as I shall be arguing, it is quite difficult to spot any realdifference.

3.4 Danielsson and Olson on the WKR problemDanielsson and Olson’s approach is strongly influenced by Brentano. They distinguishto begin with between two sorts of reason for attitudes: a ‘holding-reason’, which theyunderstand as a reason for having the attitude, and a ‘content-reason’, which is a reasonfor the correctness (or fittingness) of the attitude. To give a holding-reason for anattitude is, then, to provide arguments supporting the idea that we ought to have theattitude. On the other hand, to give a content-reason for the attitude is to argue thatthe attitude is correct. (We shall return to this notion of ‘correctness’ in a moment.)

Moreover, Danielsson and Olson believe that:

a content-reason for some attitude, a, implies a defeasible holding-reason. But crucially, thatthere is a holding-reason for some attitude does not entail that there is a content-reason for thatattitude. In other words, that there is a reason to have an attitude does not entail that the attitudeis correct. (2007, p. 515)19

The distinction between these two kinds of reason furnishes Danielsson and Olsonwith a response to theWKR problem. Consider again, the demon case. Danielsson andOlson can acknowledge that the demon does give us a reason to favour the saucer ofmud or the evil demon (when the demon insists, on pain of a severe punishment, thatwe admire him rather than the saucer of mud) but insist that it is the wrong kind ofreason. We ought indeed to favour the saucer of mud (or admire the demon), but thatis not the correct attitude to take towards the mud (or demon). The authors’ proposal,then, is that “‘x is good’ means that ‘x has properties that provide content-reasons tofavour x’” (2007, p. 520).

This is a neat suggestion. However, how to distinguish between holding- andcontent-reasons remains a problem. Danielsson and Olson suggest that we followBrentano and take correctness as a primitive. As I mentioned earlier, this is obviouslynot an optimal way of dealing with these issues, but perhaps it is in the nature of thingsthat we are forced to take such a step. So I will not repeat my reservations, but ratheraddress the more interesting issue, namely whether their approach adds somethingsubstantially different to the Parfit–Skorupski solution. Danielsson and Olson are quite

19 Later the authors claim that “to say that there is a holding reason to favour the evil demon is to say thatthere is a content-reason to favour favourable attitudes towards the evil demon, or possibly a content-reasonto disfavour the non-occurrence of such attitudes. This implies that favouring the demon is good, which isthe right implication; favouring the demon is instrumentally good since it shields us from punishment” (2007,p. 519).


explicit about this matter: their distinction between holding- and content-reasons doesnot coincide with the one between reasons for attitudes and reasons to bring it aboutthat one has an attitude (nor with Parfit’s distinction between state- and object-reasons). Notwithstanding this, I think it is hard to see that their approach is substan-tially different from the Parfit–Skorupski solution. To see this we need to understand inmore detail how Danielsson and Olson conceive of the relation between holding- andcontent-reasons. Consider the following passage:

. . . the notion of a holding-reason should be analysed in terms of the notion of a content-reason.To say that there is a holding-reason to have some attitude is to say that there is a content-reasonto favour the occurrence of this attitude, or possibly that there is a content-reason to disfavour thenon-occurrence of this attitude. (2007, p. 518)

Apparently, the talk about there being two kinds of reason is not entirely accurate.What Danielsson and Olson in effect are saying here is that holding-reasons are nothingbut a certain kind of content-reason, namely content-reasons to have second-orderattitudes.But if Danielsson and Olson’s approach is to reduce holding-reasons to content-

reasons, what difference is there between their solution and the one provided by Parfitand Skorupski? Parfit, for instance, recognizes only object-given reasons for attitudes.20

On his account, there is no reason to admire the demon but a reason to favouradmiring the demon. Danielsson and Olson would concur: there is no content-reasonto admire the demon, but there is a content-reason for the second-order attitude(admiring the demon).Danielsson and Olson’s talk of “holding-reasons” becomes therefore empty if

holding-reasons are nothing but content-reasons for second-order attitudes. Thus,the difference between the two positions seems purely verbal if holding-reasons areanalysable in terms of content-reasons.21 Danielsson and Olson also maintain that whatwe ought to do and what we have the right kind of reason to do do not necessarilycoincide. Thus, it might not be correct to admire the demon (in the sense that there is noright reason to do so, and hence no content-reason to do so) even though that is whatwe ought to do, where this ‘ought’ should be understood in terms of holding-reasons.Intuitively, we do seem to admit that there is a difference between what we ought to

20 Parfit’s ‘On What Matters’, Appendix B (circulating ms), ends in a way that I find puzzling. Aftermaking it clear that “we do not, I suggest, have state-given reasons to have beliefs or desires”, he adds thefollowing paragraph: “Wemay have state-given reasons to be in some other kinds of state. I might truly claim,for example, that I have a reason to be in Paris next April. But as I have argued, such reasons would have noimportance. It would be enough to claim that I have reasons towant to be in Paris next April, and to go there, ifI can” (p. 426, with reservation for changes). If there can be state-given reasons of this kind, why cannot therebe state-given reasons for attitudes? I owe Rabinowicz for drawing my attention to the paragraph.

21 Cf. Gerald Lang, ‘The Right Kind of Solution to the Wrong Kind of Reason Problem’ (2008), whoargues (and I believe for different reasons) that Danielsson and Olson’s solution is “uncomfortably close toJohn Skorupski’s favoured solution to the WKR problem”. That Lang’s view is based in part on amisunderstanding of their views is argued by Olson in ‘The Wrong Kind of Solution to the Wrong Kindof Reason Problem’ (2009).

3.4 DANIELSSON AND OLSON ON THE WKR PROBLEM 41

favour and what it is correct to favour. That I ought to favour the demon seems to saysomething different from that it is correct to favour the favouring, say, of the demon.However, since holding-reasons are just content-reasons for second-order attitudes (ofthe kind exemplified in the claim that we should admire the demon), then Danielssonand Olson might in effect be said to offer a solution that blurs this distinction. So unlessDanielsson and Olson can explain in some other way why we should divide content-reasons in the way they suggest, little will have been gained.

Resort to the claim that a certain attitude is correct (as Brentano would say) or fitting(as Ewing would say), and to the further claim that correctness, or fittingness, areprimitive notions, is not quite the optimal solution to the WKR problem (cf. Reisner,2009). The ‘primitive’way out of the problems with the FA analysis might nonethelessbe quite tempting. Let me therefore finally express yet another worry about theDanielsson and Olsson approach. The notion that something is correct is problematicin that there seem to be ways of understanding it that are of limited present interest. Forexample, it is incorrect in some circles to clean your dinner plate by licking it, but thisdoes not show that we are dealing with a fully-fledged normative notion of correctnesshere. Nor does the fact that it is correct to drive on the left side of the road in Englandbut not in Sweden express something fully prescriptive. This suggests that we need todistinguish between at least two kinds of correctness: one that is somehow conditional,and one which is not. If we can achieve this without employing or presupposing anyvalue judgement, the Danielsson and Olsson solution will be unaffected. However, ifwe need to separate the fully normative notion of correctness from the conditional oneby invoking value, it will turn out that we are not well served by their approach. At anyrate, little can be gained if we are obliged to understand value in terms of a notion ofcorrectness that itself needs to be illuminated by invoking value.

3.5 The biconditional buck-passing accountI want to close this chapter with a discussion of another feature of the FA analysis. Tointroduce this idea it will be convenient to consider what Crisp recently has said are thetwo key elements of a buck-passing account.22 Crisp thinks we should reject whathe refers to as the biconditional buck-passing account on which “whenever somethinghas lower-order properties such that there is reason to respond to it” by favouringit in some way, “that thing is valuable”. One reason this account should be abandonedis that there are what Crisp calls deontological reasons that do not ground value.Here is one example of such a reason: suppose I sincerely utter “I promise to ç”;the fact that I have said these words gives me, according to Crisp, a reason to chooseto ç.23 On a biconditional account, “because ç-ing has the property that I have

22 Crisp, ‘Goodness and Reasons: Accentuating the Negative’ (2008).23 Notice that Crisp formulates the buck-passing account in terms of responses. However, I will for the

sake of simplicity and discussion simply assume that choosing to ç is in fact a form of favouring ç� If this is tostrain the meaning of ‘favouring’, just think of an example in which my promise to ç gives me a reason tofavour to ç�


promised to do it”, it follows that ç-ing is therefore valuable. But is it? Crisp quiterightly questions this; ç-ing does not seem to be valuable. Noteworthy here isthe discovery that we are stuck, apparently, with a WKR problem whether or not weactually think there is a distinction between ‘object-given’ and ‘state-given’ reasons. AsCrisp himself notes, this distinction “has no purchase” when it comes to this kind ofexample.Crisp goes on to argue that biconditional accounts are eliminativist theories of value;

and this, he thinks, is too big a step to take. He formulates what he takes to be the twokey elements in a buck-passing account:

Negative component: Being good is not itself a reason-providing propertyPositive component: Being good is merely the higher-order property of having lower-orderproperties that provide reasons to respond in particular ways. (2008, pp. 263–4)

Of these, he thinks it is the positive component that we need to throw overboard. Weshould retain the idea that goodness, or value, is not itself a reason-providing property.But is this what we should do? Is it really clear that goodness, or value, is not a

reason-provider on its own? At this point one might be tempted to interject that weshould give up on the negative component but that buck-passers have no choice: theyhave to endorse the negative component, qua Buck-passers. This is a mistake, howev-er. Let us therefore consider the matter in more detail, and let us do so from the perspectiveof the FA analysis.Suppose I am told that an object x is good: someone tells me that I should give

money to a certain organization; I believe the organization is good; and this is prettymuch all I know about the organization. On the FA approach, what I believe is that xhas the property of providing reasons to favour x. Let us focus on this very property.Could it not be argued that it is itself reason-providing? Might not the very property ofbeing a reason-provider provide us with a reason to favour O? In other words, is it notpossible that the fact that something is such that it gives me a reason to (say) desire itfurnishes me with a new reason to desire it?Now, although I am not quite sure what kind of reason we are talking about here, I

am nonetheless inclined to think that we are dealing with some sort of reason. But if theproperty of providing-reasons-to-favour indeed provides us with a reason to favour,

is in itself reason-providing

PR propertythe property of

providing reasonsto favour

Figure 3.1

3.5 THE BICONDIT IONAL BUCK-PASS ING ACCOUNT 43

then, obviously, we should give up what Crisp refers to as the negative component ofthe buck-passing account: goodness (the positive component) in the FA analysis is itselfa reason-providing property.

In passing, we should recall that the goodness of x provides us with a reason to favourx and not (at least not necessarily) with a reason to favour the goodness of x. Thisdistinction is important not to overlook. We might otherwise end up believing thatvalue accrues to goodness. It does not. It accrues to the bearer of goodness.

Goodness plays a role, then: it tells us that certain properties are reason-providers. Toshed some light on the issue of whether goodness also is in itself reason-providing, weneed next to say a bit more about what a reason is, and in particular about what anormative (i.e. justifying) reason is. Letting ‘R’ refer to a fact, we might follow JohnBroome and suggest the following:

(R1) R is a normative reason to ç iff R explains why we ought to ç24

Suppose this is correct. To believe there is a reason for me to ç (favour something) is tobelieve, by implication, that there is a fact that explains why I ought to ç� But now, ifthis fact signals that there is an explanation of why we ought to ç we appear to have areply to the question above. Neither goodness nor value is a reason-provider. This iseasily shown on Broome’s account: the fact that there is a reason to ç is not itself areason to ç since this fact does not itself explain why we ought to ç. Hence, value andgoodness are not reason-providers.

(R1) needs to be qualified. Explanation is a success notion; if a fact fails to explainsomething it is obviously not a reason according to (R1). However, as I do not thinkthe notion of a normative reason is a success notion, I find it hard to understand reasonsin terms of explanations. But suppose we could find a way of dealing with thiscomplication.25 There is still, I think, a drawback with the Broomean response.Although Broome’s notion catches something important about normative reasons, itseems to fail to capture all there is to our notion of a normative reason. There is, at least,a different notion of normative reason in play, namely:

(R2) R is a reason to ç iff R bears positively on the question of whether oneought to ç

Pamela Hieronymi has argued that the fundamental relation in which a considerationbecomes a reason is something like that adumbrated in (R2).26 (By ‘consideration’ she

24 See Broome’s ‘Reasons’ (2004). I am here simplifying matters. Broome summarizes his view in thefollowing way: “A reason is either a perfect reason or a pro tanto reason. A perfect reason for you to ç is a factthat explains why you ought to ç. A pro tanto reason for you to ç is a fact that plays a characteristic role in apotential or actual weighing explanation of why you ought to ç or of why you ought not to ç or of why it isnot the case that you ought to ç and not the case that you ought not to ç” (p. 55). Broome treats ‘explain’ as aprimitive.

25 I am not quite sure what Broome would say about this.26 See Hieronymi, ‘The Wrong Kind of Reason’ (2005).


has in mind a broad notion that includes, among other things, facts and propositions;this is how I, too, would regard this notion.) By differentiating the questions, shemanages to separate different kinds of reason. For instance, that it is raining bearspositively on the question of whether we ought to believe that it is raining. It should bepointed out that Hieronymi wants to deal with the WKR problem in cases of value, asshe deals with reasons for belief. The fact that you would get beaten by the demonunless you believe that London is the capital of France obviously does not bear onthe question of whether London is the capital of France. Hieronymi, then, goes on toargue that we can reason in an analogous way when it comes to cases of value. But,from the perspective of the FA analysis, this conclusion does not follow. The fact thatthe demon wants to hurt me unless, say, I favour a saucer of mud, clearly bears on thequestion of whether we have a reason to favour the saucer of mud (i.e. whether it isvaluable).Now, depending on what kind of reason we have in mind (R1 or R2), the question

‘Is the buck-passer’s goodness reason-providing?’ will receive rather different answers.As we saw earlier, on Broome’s account the answer would be a clear ‘no’. OnHieromymi’s account, by contrast, the answer is ‘yes’: the fact that there is a reasonto ç is itself a reason to ç, since this fact bears on the question of whether I ought to ç�Thus, if you are a buck-passer who does not want to give up on the idea that

goodness is a reason-provider, endorsing (R2) rather than (R1) will be of great help;(R2) explains why we should, as FA analysts, throw the negative component of thebuck-passing account overboard. Embracing (R1) (with the proviso mentioned) onthe other hand, paves the way for the negative component. FA analysts have, in otherwords, a choice: they may or may not throw the negative component of the buck-passing account overboard.27

27 Cf. also Mark Schroeder’s recent ‘Buck-Passers’ Negative Thesis’ (2009). Schroeder contests what herefers to as the “buck-passing inference” (p. 342)—that the negative thesis follows from the positive one.

3.5 THE BICONDIT IONAL BUCK-PASS ING ACCOUNT 45

4

Mistaken Value Analyses

The general aim of this work is to develop a novel way of understanding values that weregard as personal, i.e. as being in some way good or bad for us. Sometimes what haspersonal value will also have an impersonal value. This fact complicates matters in moreways than I care to think about. What if these values conflict with each other? How dowe compare them?Which should we give priority to? These and many other questionsare all important; but although I will have more to say about some of the issues later on,I have had to leave out the general question about the relation between personal valuesand other values.

My own welfare and well-being are examples of something having both kinds ofvalue.1 Certain of mymental states are good for me, but such states, regardless of whetherthey have value-for me, might also carry impersonal value. The link between good-forand welfare or well-being is tight, though. In fact, numerous writers treat this connectionas analytic. This is not obviously the best way to regard the relation between thesenotions. Depending onwhat one takes it to be, well-beingmight be just one of the thingsthat exemplifies value-for. Other ‘things’ might carry value-for as well. In whatfollows I will examine various possibilities of this sort. I will seldom discuss well-being.This does not reflect my view of its importance. Most examples I discuss will, most likely,not be especially important values—not by comparison with the value of well-beingat any rate. But the less important values need to be understood if we want to obtaina complete picture of value. Moreover, in my experience it is often easier to discusscases that appear to be rather trivial. It is easier in the sense that one can focus on what isreally relevant.

Here is an example of a putative personal value which I am inclined to say carries noimpersonal value. My father made a bookshelf for me when I was a child. It was,perhaps, not an aesthetic revelation—more a piece of robust and heavy furniture thatserved its function well. As an adult I took it with me as I moved from place to place

1 I do not take well-being and welfare to be the same. However, in what follows I will not pay anyattention to these differences, since they are not of crucial importance to what is discussed.

(a bookshelf always comes in handy). Each time it looked more and more shabby.At one point, it was in such a bad state that I decided to get rid of it, and finally I gotaround to throwing it away. Or so I thought. Now, many years later, bits of it still popup in the most unexpected places. I find this quite amusing, and I have begun towonder whether, in my own life, I shall ever see the back of this item of furniture.These screws, shells, wood pieces, ornaments, and so on, hardly have impersonal value,but they do seem to carry some sort of value for me.However, it would be a gross simplification to conclude that personal values, and in

particular those that are not also of impersonal value, must all be found in objects relatingto one’s past. I mention these examples since they appear to me as more obvious.All examples of personal value have this much in common: they are receptive to the

FA pattern of value analysis. What we need to do is fine-tune the relevant analysans sothat it brings into the analysis a reference to the person whose personal value it is. Buthere a caveat is in place. This requirement on the analysis (that it should somehowbring in the person whose value it is) is ambiguous. It might refer to what is perhaps bestdescribed as an owner claim relation between the value and the person whose value it is.Like G. E. Moore, to whom I will return in a moment, I have a problem with this kindof view. The suggestion I endorse sets out from another idea, namely that personalvalues are just like other kinds of value but for the fact that they stand in a certainrelation to some person (or perhaps group of persons). But this relation has nothing todo with the value being ‘his’ or ‘her’ private value.

Here, then, is how I want to modify the standard analysis, where FAP stands for theFitting Attitude analysis of Personal value:

FAP: An object x’s value for a person a (i.e. x’s personal value), consists in theexistence of normative reasons for favouring/disfavouring x for a’s sake

FAP is not as original as I once thought when I proposed it in a series of talks in 2004;there are points of contact between it and the view Stephen Darwall takes on welfare.There are also some important differences, though; but more about this later, inChapter 8. Also, as we shall see in Chapter 7, there is a chance that Henry Sidgwickunderstood good-for along the lines of FAP.The analysis is stated in terms of reasons and pro- and contra-attitudes; as may be

recalled, “favour” is a notational alternative to “having a pro-attitude”. In the chapterthat follows I will say something more about how I understand the reason componentof FAP. The notion of a pro-attitude (and with the proper adjustments, a contra-attitude) is here understood to cover quite a large set of different pro-responses that arecalled for by the value bearer. I will not try to specify what makes a certain mental stateinto a pro-attitude. I trust that this notion is sufficiently clear to render FAP intelligible.Some cases will clearly fall outside FAP. To feel thirsty or hungry would not, in myview, count as having an attitude. Other cases will be more difficult. What should wesay, for instance, about a specific kind of thirst—say, feeling thirsty for orange juice?

4.1 MISTAKEN VALUE ANALYSES 47

There are all sorts of problems involved here, but I will not pursue them.2 (FAP willeventually be modified somewhat in Chapter 5, where I seek to say something moredetailed about one feature of the attitudinal element.)

As to the variable x, it should be understood as gathering a wide variety of things,including living beings (persons and animals that carry the personal value to x). Oneadvantage with this pattern is its wide range: very many different kinds of value areanalysable in terms of the same general formula. But perhaps more important, bysituating the distinguishing quality in the attitude, rather than reason, part, the analysisallows that personal value is recognizable as a value not only by the person for whomit has personal value but everyone else too. In this way we avoid introducing twocompletely different notions of value, one impersonal and the other personal.

Of course, it is not immediately obvious why not having two completely differentnotions of value is a good thing. Why, for instance, should we not locate thedistinguishing mark in the much-discussed agent-relative/neutral dichotomy? Just aspersonal values are well suited to be understood in terms of agent-relative reasons, soimpersonal ones appear connected to agent-neutral ones. One can imagine that thismight indeed come in handy. However, as will emerge in Chapter 9, there arepowerful reasons to be wary of such a move. However, for the time being I will notdiscuss in any more detail just how we should understand the normative component ofthe analysis.

Moore saw that there is something intuitively odd about the idea that value isrelative. His scepticism is shared by many philosophers, though. A more recentopponent of good-for is Donald H. Regan, whose views I will return to in Chapter7. Regan expresses well what is strange about this notion:

The good for Abel must be peculiarly Abel’s—the goodness or the value must be peculiarlyAbel’s—in a way that the mere occurrence of universal good in Abel’s life does not necessarilysatisfy. But if the good for Abel is peculiarly Abel’s—if its value is somehow essentially a value forAbel—then why indeed should Cain care? We think there is a deep connection between valueand reasons. That suggests precisely that if the good for Abel is a matter only of value for Abel,then it creates reasons for Abel, and for no one else.3

However, notice that even if you think that value-for is a coherent notion which doesnot depend on value period, it does not follow that it must be analysed in terms ofagent-relative reason. Connie Rosati has argued in ‘Objectivism and Relational Good’(2008) that these reasons may be agent-neutral. She suggests, contrary to Regan, that

2 See Thomas Scanlon (1998), where the issues are delineated in the following way: “The class of attitudesfor which reasons . . . can sensibly be asked for or offered can be characterized, with apparent but I thinkinnocent circularity, as the class of ‘judgment-sensitive attitudes.’ These are attitudes that an ideally rationalperson would come to have whenever that person judged there to be sufficient reasons for them and thatwould, in an ideally rational person, ‘extinguish’ when that person judged them not to be supported byreasons of the appropriate kind” (p. 20).

3 See Regan, ‘Why am I My Brother’s Keeper?’ (2004, p. 211).

48 MISTAKEN VALUE ANALYSES

the notion good-for does not involve a relativization of normativity or goodness. WhenI started writing on personal values in 2004 (later published as ‘Analysing PersonalValues’ (2007a)), I had the same thoughts as Rosati. The way to avoid the problemsthat Moore pointed to was to understand these values in terms of agent-neutral reasons.I now believe that the real distinction between good and good-for is not to be foundin the reason element, but rather in the kind of attitude that is called for. More aboutthis in the coming chapters, though.Moore’s sound scepticism about ‘private’ values led him astray, to the mistaken

conclusion that we should give up the notions of good-for and bad-for altogether(or at the very least be aware that these expressions are elliptical). His arguments arenot convincing, as I shall argue in the next section. We need not give up these notions.We just have to find a more appealing account of good/bad-for than the one Mooreobjected to.

4.1 Moore’s objection to good-forIn his much-discussed attack on egoism in Principia Ethica,4 Moore expresses greatscepticism about certain ways of understanding the expressions ‘my own good’ and‘good for me’:5

What then is meant by ‘my own good’? In what sense can a thing be good for me? It is obvious,if we reflect, that the only thing which can belong to me, which can be mine, is somethingwhich is good, and not the fact that it is good.6

The passage is bewildering (perhaps especially as it comes fromMoore). First, Moore is,of course, right that the fact that something is good is not something that can belong tome. Facts are not the kind of object that can be in someone’s possession. But this doesnot force us to say that ‘my own’ in ‘my own good’ cannot refer to anything but theobject that is valuable and hence cannot refer to the value itself. It is quite surprisingthat Moore in the first place takes our relation to facts to be an issue. It is hard to believethat any of Moore’s opponents would actually disagree with Moore on this matter.7

It is important to underline the fact that Moore cannot simply assume there are nofacts about ‘my own good’. As yet he has given us no reason to suppose that there are

4 For a classic and quite devastating objection to Moore’s attempt to show that the doctrine of egoism isself-contradictory, see C. D. Broad’s ‘Moore’s Ethical Doctrines’ (1942).

5 As Rosati (2008) has recently put it: “Moore appears to have found the expression ‘good for me’ akin tothe peculiar expression ‘true for me’” (p. 324). Rosati’s interesting paper contains detailed discussion of notonly Moore’s, but also some modern objections to the notion good-for. Her idea that critics of this notion“need not be understood as advocating that we purge ordinary discourse of good for talk (as if we could dothat), or even (more realistically) that we theorists refrain from such talk” is, I believe, correct. Of course, itremains possible that Moore was trying to make us stop using the expression ‘good for’.

6 Moore, Principia Ethica (1993 [1903a], p. 150).7 It may be that, in concluding the passage quoted with the words “and not the fact that it is good”Moore

intended no more than to assert that goodness is not something that can belong to a person. However,whether or not this is true, it does not change our assessment of what Moore accomplishes argumentatively.

4.1 MOORE’S OBJECTION TO GOOD-FOR 49

not two different kinds of value fact: one, that something is good, period (foreveryone), and another, that something is good for person a. The states of affairsinvolved in these two cases may well, when they obtain, constitute two differentkinds of fact.

Here is Moore’s own conclusion as to what we can mean by ‘my own good’:

When I talk of a thing as ‘my own good’ all that I can mean is that something which will beexclusively mine . . . is also good absolutely; or rather that my possession of it is good absolutely.The good of it can in no possible sense be ‘private’ or belong to me; any more than a thing canexist privately.8

Again, if it is not merely that it is mine or that it exists that is valuable, but the thingitself, then Moore has not shown that the value of such an entity cannot, in aninteresting sense, be private. If there is something like a peculiar ‘good-for fact’, thenit will be a fact for everyone, since facts are not the kind of thing that can be insomeone’s possession. But this should not prevent us from saying that the fact containsa ‘private’ element that makes it distinguishable from facts involving good, period.

Moore’s objection to understanding ‘my own good’ as meaning anything but thatwhich is exclusively mine carries an absolute good must be understood in the light of hiscriticism of egoism—a view he believes he can refute. Moore is not easy to follow here,but a core idea concerning his objection to ‘my own good’ seems to be a claim to thefollowing effect:

(1) To be a good at all, the good must be (a) good in itself, i.e. an intrinsic form ofgoodness, and (b) a ‘universal good’ (cf. pp. 150–1)9

Moore’s famous attack on egoism and the idea of something being ‘good for me’,which runs over several paragraphs, boils down to this very idea that to be a good at allthe putative good must be a universal intrinsic good. However, he does not support theidea—at least, not in a persuasive way. Therefore, the attack is in effect not so much anargument as an assumption that Moore is asking us to join him in making, or workwith. He has not given us any convincing reason why the notion good-for should berejected.

As mentioned, Moore is not alone in finding the notion of good-for objectionable.Thomas Hurka (1987) argues:

As currently used, ‘good for’ is multiply ambiguous, and for none of its meanings is ‘good for’ theappropriate term. (p. 72)

For this reason Hurka thinks it should be “banished from moral philosophy” (p. 72).The senses he lists are: “‘good for’ means satisfies the desires of, or further the interests

8 Moore, Principia Ethica, p. 150.9 The notion of a ‘universal good’ is left undefined. A natural interpretation (especially since he is

discussing Sidgwick’s view on the matter) is that Moore has in mind everyone’s good.


of”; it may also mean “good from the point of view of ”; what is ‘good for’ a personmay be what “she believes good”; and finally it can “mean that portion of the good(from whatever point of view) that falls within a person’s own life” (pp. 72–3). Hurkahas misgivings about all of these suggestions. I share these, and do so for the reasonshe gives. However, as I hope to make plausible by developing an analysis of the notion,this does not mean that ‘good for’ cannot be given, or associated with, a single senseof its own.10

4.2 Instrumental valueFA analyses of final impersonal value say that an object has final value if there is reasonto favour it for its own sake. Applied to final personal values the FA approach adds thatthe attitude must be of a certain kind: it must be a favouring of x for a person’s sake.(This kind of attitude has not received much attention in the literature, so we shallneed to examine it below in Chapter 5.) An object of impersonal value is then saidto be instrumentally valuable if there is reason to favour it for the sake of its effects.In the case of instrumental personal value we have, on the analysis, a reason to favour,for a person’s sake, something for its effects’ sake. Here it is important to pause fora moment, though. For this analysans is, of course, only plausible to begin with ifinstrumental value is a genuine kind of value in the first place. However, this is notobvious.Standard suggestions—such as ‘x has instrumental value’ means x is conducive to

something that has final value—are not very helpful if the aim is to find out justwhat the speaker means by ‘instrumental value’. Such suggestions tend to leave usin the dark with regard to whether x is a bearer of something that belongs to thecategory of value or whether x is merely related in some way to something belongingto this category. In other words, to say of an x that it is an instrumental y does notyet determine whether or not x is a kind of y. What happens here is somethingthat seems to occur with many expressions we employ to qualify something. Justas ‘quicksilver’ does not refer to a kind of silver, ‘instrumental value’ may not refer toa kind of value.In the literature it is possible to detect at least two distinct usages of ‘instrumental

value’. To begin with, there is what I will refer to as the Strong evaluative sense:

S: ‘x has instrumental value’means ‘x bears a (certain particular) value, and it does soonly if x is conducive to (the existence of something with) final value.’

10 Regan (2004) suggests that instead of speaking about good-for, we would do better to employ“goodness occurring in a person’s life”. Rosati has convincingly shown, I think, that this notion of a valueoccurring in a person’s life faces a number of difficulties of its own. More importantly, she shows, too, thatwhat is “picked up” by good for is not covered by the notion of “goodness occurring in the life of ”. I willreturn to Regan’s views in Chapter 7.

4.2 INSTRUMENTAL VALUE 51

The second conjunct here is open to more than one interpretation; but since I haveconsidered these elsewhere, I will keep things simple and side-step the complicationsthat ensue.11

For instance, George Dickie’s way of expressing the traditional distinction betweenintrinsic and instrumental goodness suggests that he takes the latter to be a kind ofvalue: ‘‘Philosophers distinguish between intrinsic goodness, which is good all by itself(independent of its relation to anything else), and instrumental goodness, which is thegoodness that something has because it is a means to something else which is good” (1979,p. 157; my emphasis). Here Dickie appears to assume that there are at least two kinds ofgoodness: intrinsic and instrumental.12

The point I wish to emphasize is that the value described in S should not be confusedwith instrumental value in the Weak evaluative sense:

W: ‘x has instrumental value’ means ‘x is conducive to (the existence of somethingwith) final value’.13

In contrast with their stronger cousins, weak instrumental values do not belong to thecategory of value. But, it might be asked, is not that which leads to something of finalvalue a good example of something with a kind of value—namely, the kind of valueaccruing to objects that lead to what is of final value? The suggestion implied hereseems to set out from the idea that ‘being valuable’ is somehow part of the meaning of‘being conducive to value’. But that is surely a mistake. Notice that the definiens says,merely, that x is conducive to what is of value. There is no reason why ‘being valuable’should somehow logically follow from the meaning of ‘being conducive to value’.To conclude that what is conducive to value must therefore be of value is to commitwhat we might dub the value-by-association mistake.Since the expression ‘instrumental value’ is equivocal and does not, in itself, clearly

exclude either S or W, we will have to turn to metaethical theories of what constitutesvalue to settle what an instrumental value is. (That is to say, we need here somethingmore than value’s supervenience base; we need some kind of account of how it is thatthe subjacent properties ground or constitute the value.) Meta-theories of value mightimply, of course, that there is only one way of understanding ‘instrumental value’.In that case, we should expect a reason for this exclusion. For my own part, I believethat there are S-values. Moreover, I think that FA analysis gives us a good account of

11 Rønnow-Rasmussen (2002b).12 See also R. T. Allen, The Structure of Value (1993, p. 59), in which the author seems to assert that the

instrumental value of objects is ‘realized’ when we use these objects. Cf. Robert Nozick’s claim “[ . . . ] thereis something’s originative value which is a function of the value it newly introduces into the world, the newinstrumental or intrinsic value it introduces that was not presaged by or already fully counted in previousinstrumental value’’ (1981, p. 311).

13 Johan Brännmark has pointed out that there might be a structurally similar but nonetheless differentsense of instrumental value—one relating the instrument to a goal rather than to a value (personalcommunication). I think he is probably right, but I will not discuss the suggestion further.


them as objects that we have a reason to favour for their effects’ sake.14 The importantthing is to be aware of the two senses, S and W. I will not argue for this contention.Doing so would lead me astray. The point I wish to make clear, however, is that theanalysis of instrumental value mentioned earlier is an analysis of an evaluative notion ofinstrumental value.

4.3 A bad approach to personal valueI now want to open a discussion of the two key elements of FAP—the attitude andnormative parts—by considering an approach which, in my view, is mistaken. Thisapproach quite naturally suggests itself. It sets out from the notion of extrinsic finalvalues. What it suggests is that the road to an understanding of personal value is clearonce we accept the idea of extrinsic final value. Thus, according to what I will refer toas the De Facto Attitude analysis of Personal Value (or DEFA), personal values make up asubgroup of extrinsic values. As I say, DEFA is not a good approach to personal value.However, it might prove useful to glance at this erroneous alternative if that helps us toappreciate the merits of the more serious suggestion presented in the next section.DEFA is arrived at by the following path. If final values can be extrinsic—i.e. accrue

(at least, in part) to objects in virtue of their non-internal relational properties—we canexpect some object of extrinsic value to carry this kind of value in virtue of its relationto a particular person. Once value is relativized to persons in this way, the stage is setfor an analysis of personal values: we need, next, to specify the relata to which personalvalue accrues more carefully. This must be done in order to distinguish personalvalue from other kinds of extrinsic value accruing to objects merely in virtue ofbeing related to a person. Intuitively, it seems reasonable to assume that a personalvalue is something more than a value that is related to a person. We need, that is,to analytically separate from each other cases like the following: Napoleon’s hat anda drawing made by my five-year-old daughter.It can certainly be argued that Napoleon’s hat has some kind of final, impersonal

extrinsic value.15 Even if Napoleon never endorsed any evaluative judgement whatso-ever about the hat, it might still be of value because it belonged to, and is famouslyassociated with, a historically important person. My daughter’s drawing, on the otherhand, is unlikely to have impersonal value. It appears rather to be a carrier of personalvalue, if it has value at all.Suppose we accept these examples in the terms they are described. The question

then is: How should we account for the difference? What makes the first example acase of impersonal value and the other a case of personal value?

14 This will particularly be the case if we take into account the fact that “favour” refers not only to variousdifferent attitudes but also to attitudes-cum-behaviour and perhaps even to choices. That the latter possibilityshould be taken into account was pointed out to me by Brännmark (personal communication).

15 Cf. Shelly Kagan, ‘Rethinking Intrinsic Value’ (1998) and Rabinowicz and Rønnow-Rasmussen(2000).

4.3 A BAD APPROACH TO PERSONAL VALUE 53

The role that Napoleon played in the past, and my, and my daughter’s, historicalinsignificance, may explain why the hat has impersonal value while the drawing doesnot. But these facts do not account for the fact that the drawing, and not the hat, haspersonal value. Here we need to look for something else.

A natural strategy would be to search for something special about the two salientobjects involved in the drawing example, i.e. the drawing x and the person a forwhom, ex hypothesi, the drawing has personal value. On the face of it, it is likely that a asa matter of fact favours the drawing. DEFA could, but need not, be cast in some kind offitting analysis mould. That is to say, it could say something to the following effect: xhas personal value for a if and only if we ought to favour x in virtue of the extrinsicproperty that x is de facto favoured by a.Thus, my daughter’s drawing would have personal value only if the value accrued

to it in virtue of being an object that is favoured (say, cherished or esteemed) by me, a.Whatever intuitive appeal it has, however, DEFA establishes at most that the value

accruing to the drawing is relative; and that is not specific enough if it is personalvalue we are looking for. It might be objected that DEFA need only make analternative, and quite reasonable, subjectivist claim, i.e. that we should regard theattitude of a as having value-constitutive powers. This might well be true. However,while the issue between subjectivism and objectivism is not settled in a convincingway, I will allow, as might be recalled, the analysis to reflect this by formulatinga neutral analysans.

Personal value is in a sense relative: since it is value-for, we need to mention, inanalysing it, the person for whom it is valuable—but relative values are certainly notnecessarily personal values. In other words, although the inference from “x is a personalvalue” to “x is a relative value” is valid, the reverse inference is invalid.There are two more reasons why we should regard DEFA with scepticism. First, it

places personal values exclusively among extrinsic (final) values. But this seems to be anarbitrary restriction. An account including intrinsic values would be more appealing.Of course, it might be that all personal values are in fact extrinsic final values. Asubstantive evaluative argument might force us to take such a view. But the issue nowis whether we could coherently think that some object x carries value-for in virtue ofx’s internal properties and nothing else. I am inclined to think that examples ofphenomena of this sort will be fairly controversial. (I will discuss some later on.) Butin principle I do not see why the relation picked out by the expression ‘good-for x’cannot supervene on the internal features of an object alone.16

Second, it is not obviously a necessary condition of something being a personal valuethat it is valued by the person for whom it is a value. Surely a convincing case could bemade, albeit on substantive grounds, for the idea that there can be objects of personalvalue that are not valued (esteemed, cherished, and so on) by the relevant person.

16 I will return to certain problems with this idea in Chapter 6.


5

For Someone’s Sake

If one wants to understand values in terms of reasons, there seems to be a perfectdichotomy to use—namely, the dichotomy of agent-neutral and agent-relative rea-sons. After all, value-for is a relative notion, and what would be more logical than itsconnection with agent-relative reasons? However, as I shall argue in Chapter 9, thereare good reasons to be cautious about this distinction; we should therefore be equallycautious about understanding the distinction between personal and impersonal valuesin terms of the dichotomy between agent-relative and agent-neutral reasons. Butwhether or not my scepticism concerning the reason dichotomy is sound, the sort ofanalysis that I am proposing stresses that the defining feature of personal value is not tobe found exclusively in its normative element, but is rather located in the analysans’sattitudinal element. Objects of personal value are such that there is reason to direct acertain type of attitude—i.e. what we might call a ‘for-someone’s-sake’ attitude—onthem. I believe that these reasons are all agent-relative. Be that as it may, what is typicalabout objects that are good for, or bad for, persons is that a for-someone’s-sake attitudeis called for. This is what distinguishes such values from impersonal ones.1

In this and the next chapter I attempt to add flesh to the bones of these for-someone’s-sake attitudes (henceforth FSS attitudes). They have not yet been properlyanalysed in the philosophical literature.How we should delineate attitudes in general, and how, in particular, we should

isolate FSS attitudes in what follows from, on the one hand, pure non-propositionalcravings such as hunger and thirst, and, on the other, purely cognitive beliefs, arematters I will bypass here. As might be recalled, what I have in mind by ‘attitudes’ arenot only desires and preferences, but also ‘thicker’ attitudes such as admiration, respect,and love, to mention only a few examples. Moreover, I will assume that favouring (i.e.

1 Given the great variety of values, and given the ‘monotony’, to borrow Kevin Mulligan’s description(personal communication) of the normative, if we are to understand the former in terms of the latter, we hadbetter come up with something more than a reduction of the valuable to the normative; this is precisely whatmy account purports to do here.

wanting, admiring, and so on) x is an attitude that differs from favouring y, and thatthey differ because their intentional contents are different: the former contains apropositional element involving x; the latter, one involving y. I am, then, setting outwith the tentative assumption that there might be a special kind of attitude, namely theFFS attitude, which, intuitively, is the kind of attitude we have when we favour x withan eye to someone (else); and that this too is reflected in the intentional content of theattitude, which we describe as favouring x for someone’s sake.Here a caveat is in place. The idiom ‘to do something with an eye to’ is somewhat

unclear. Often it means that we are doing something with regard to, in view of, or forthe purpose of something. My way of using the expression here and throughout thisbook is not quite idiomatic, though. Aswill emerge later on, since Iwish to stress that it isthe person rather than the effects on or the implications of my favouring for that personthat we have inmindwhenwe ascribe personal value to something, I will be as bold as tostrain the English and say that we can favour an object with an eye to a person.

Unfortunately, there is no simple way to characterize ‘favour x for-a’s-sake’-attitudes.In fact, my own view is that the prospects of specifying the characteristic features of thissort of attitude in a fully satisfactory way are bleak. The fact that they seem resistant toanalysis might suggest that they are not genuine attitudes. Perhaps this is so. However,since analysis of these attitudes has been largely ignored by philosophers, to brush themaside at this stage would be premature. Of course, although they remain something of apuzzle, this does not mean that we cannot say anything about them. FFS attitudes arediscerning attitudes: like most attitudes, they are directed on objects on account of aparticular property, or some properties, of the objects on which they are directed.However, as I shall be arguing, not all attitudes are discerning in the same way. Somediscerning FSS attitudes are, as I shall put it, ‘identity-involving attitudes’ (for the sakeof brevity I shall refer to these as Identity attitudes), while others are non-identity-involving.2 To support this categorization, I will later consider two contrasting sorts ofdiscerning attitude, namely love and admiration. Love, I shall claim, falls in the firstcategory; admiration in the second. Different conceptual barriers constrain the kinds ofproperty that can be invoked to justify love and admiration. The kinds of property thattypically justify love (on certain conceptions of it, at any rate) are not the kinds ofproperty that justify admiration. In fact there is a good reason to believe that theproperties justifying love could not justify admiration.

In Rønnow-Rasmussen (2008b), I introduced the distinction between justifiers andidentifiers, which refers to two kinds of property differing from each other in the rolethey play in the intentional content of the attitude (justifiers justify, and identifiersidentify). I argued that when it comes to admiration, properties of the admired objectplay the role of justifiers rather than identifiers, but that in the case of love, properties ofthe beloved are identifiers. This distinction needs to be further clarified, to be sure.

2 I owe thanks to Christian Piller for these expressions.

56 FOR SOMEONE’S SAKE

However, once that is done, the distinction will help us to separate two kinds of FSSattitude, both of which are pertinent to the analysans of personal value. It will takesome unravelling, in this and part of the next chapter, to make this identifier/justifierdistinction comprehensible; and in the course of this unravelling I shall need to touchupon certain issues that have been much debated by contemporary philosophers. ThusI will say something about the intentional content of the sort of attitude that I will bediscussing, and as I do so I will run into various problems concerning the sense andfunction of proper names in an attitude’s propositional content.Since I do not know how to settle these issues in a satisfying way, and since I am not

familiar with any theory that provides all the right answers to the intricate problemsthat arise in this area, I have two options: to omit speculation in this area, or to enterinto it regardless of the fact that it is bound to involve claims that are problematic oncertain views on reference. With some qualms I have decided to take the latter option.Moreover, I have taken the decision to keep the discussion as simple as possible, and toignore a number of views that have a bearing on much that I say even if they would nothave settled the issues. Of course, I would very much have wanted a general theoryabout how to understand the propositional content of attitudes to back up some of thepoints I make in this chapter, but I do not have such a theory. Had I one, I wouldprobably have written a book about this kind of theory in the first place rather than oneabout personal values.In this and the next chapter I will also say something about the importance of

people’s identity. From experience, I know there is a chance that some of my claimswill suggest that I take a firm stance against certain views on what matters when itcomes to ‘identity over time’, i.e. diachronic identity. But, as I hope will become clearin the next chapter, what I say here is, in my view, consistent with a number ofpositions regarding who we ought to care for in different possible scenarios.

5.1 Types of attitudeLet us begin to fill out the attitude required in the analysis in a more substantive way.To begin with, consider the following two attitudes, where a, as before, refers to someperson, and x to some object (understood in a wide sense):

(1) a favours x for its own sake(2) a favours x for some other object’s sake

Here (1) describes what I shall call a final attitude—the attitude I have, for instance,when I desire pleasure for its own sake. If we take the attitude in (2) to be about x’seffects on the object, then Claim (2) describes an instrumental attitude. I have this kindof attitude when I desire to drink a glass of water for the effects of drinking; drinkinggives rise to, say, pleasure.Recall next the content of FAP: an object x’s value for a person b consists in the

existence of normative reasons for favouring x for b’s sake.

5.1 TYPES OF ATTITUDE 57

What should we say about the kind of attitude referred to in FAP?

(3) a favours x for b’s sake

Although it may be hard to specify the details of this attitude, it is quite easy to think ofexamples in which it figures. Here is one: one treats a stranger amicably for a friend’ssake—the stranger being the friend of a friend, say. Interestingly, emotions also seemcapable of this sort of construction. For instance, on learning that someone for whomwe care has had good fortune, we are glad for his or her sake. Here is another example:I sometimes want to go to the theatre for my wife’s sake. She enjoys a good play, andalthough I frequently fail to see what is so good about these plays, I want nonetheless togo. Not for my sake, but for hers.3

To avoid yet another possible misunderstanding, it should be clear that FFS attitudes,and attitudes-cum-actions (and perhaps emotions) need not involve other people andfrequently do not do so. I take it to be uncontroversial and familiar to most of us thatwe often favour things for our own sake. For instance, I may desire to know for myown sake whether or not I have a fatal illness, and I may have this desire whether or notI have the further desire to know this for my family’s sake. It is not especially hard tofind other examples.

Here is a different example which suggests that we can believe that something haspersonal value without us knowing whose values they are. As a tourist I have visited anumber of churches and I have often overheard comments from other tourists that oneshould not behave in certain ways for the sake of the religious believer. People aresupposed to treat, say, the statues of Madonna as special objects for the sake of peopleattending worship.

Two kinds of FFS attitude in particular are relevant to my analysis of personal values.One is needed to understand final personal values, and the other to understandinstrumental personal values—as is the case with impersonal values, personal valuescan be final (as when something is finally good4 for us) or instrumental (as whensomething is good for us as a means). There may be other sorts of personal value too,such as personal contributory value. However, I will mostly be discussing final personalvalue, which is the most interesting sort. I propose to explain final personal value interms of what I shall call final FSS attitudes (FFSS attitudes for short):

(4) a favours x for its own sake for b’s sake

3 So what is carrying the personal value in this particular case? The play? Perhaps, but it would besomewhat strained to say the play gives me a reason to want to accompany her to the theatre. Morereasonably, what is valuable for her is rather seeing a good play together with someone whose companyshe enjoys. This is more likely to give me a reason to accompany her. I will address objections to the analysisin Chapter 8.

4 ‘Finally good’ should here and in what follows be understood as referring to what is good for its ownsake, and not as what is good, at last, or something to that effect.


In (4), x ranges over abstract as well as concrete objects; and, among the concreteobjects, x may refer to things as well as to people. Things as well as persons have valuefor us.The idea I am advancing here, then, is that one way of understanding the attitude in (3)

is to read it as a special case of (4).Of course, the very idea thatwe can favour something forits own sake for someone’s sake is counter-intuitive and linguistically awkward. Let ustherefore consider in more detail what this sort of qualification might mean.Admittedly, the inclusion of two for-the-sake-of qualifications appears at first sight

confusing. We might therefore be tempted to understand such descriptions as referringnot to a single kind of attitude but rather to a complex consisting of at least twoattitudes. This suggestion should not be rejected out of hand. There are, then, at leasttwo salient ways to split an alleged FFSS attitude (such as the one we think is presentwhen a favours x for its own sake for y’s sake) into more basic attitudes. The first is totreat the example as one involving these attitudes:

A1: a favours x for its own sakeA2: a favours x for b’s sake

This is a likely scenario. That is, a person we describe as having an FFSS attitude can,quite naturally, be taken to have attitudes like those in A1 and A2. Certainly, if afavours x for its own sake for b’s sake, then it is true that a favours x for b’s sake in somesense not yet specified and it is true that a favours x for its own sake in some sense not yetspecified. However, from this we cannot simply conclude that an FFSS attitude is merelythe conjunction of the attitudes in A1 and A2. This becomes evident once we realizethat these attitudes are less specific than the FFSS attitude. As we shall see in a moment,they can be made more precise in two principal ways. Meanwhile, consider thefollowing second suggestion as to how FFSS attitudes can be regarded as multipartattitudes. In connection with the example above, the idea is that we should take theFFSS attitude to be a combination of these attitudes:

A1: a favours x for its own sakeA3: a favours A1 (i.e. a favours favouring x for its own sake) for b’s sake

Notice that A3 is a normal second-order attitude.There are at least two problems with this deployment of A1 and A3. I shall confine

myself to the more obvious of these.5 A1 and A3 cannot be the whole story behindFFSS. (The points I shall now make apply mutatis mutandis to the combined deploy-ment of A1 and A2.) This becomes clear once we realize that A3 can be made moreprecise in two main ways: we can favour A1 either as a means (i.e. we can favour it for

5 The second problem (whose details would require considerable unravelling) has to do with the fact thatif the only thing we favour when we are having an FSS attitude is another attitude, it would appear to follow,given the pattern of value analysis employed in this work, that the only kind of object that could havepersonal value would be other attitudes.

5.1 TYPES OF ATTITUDE 59

the effects it has on something else) or non-instrumentally (i.e. for its own sake).Suppose it is the latter we have in mind. In that case we are back where we started,namely with an FFSS attitude. For the attitude described in A3 has the same form as thetypical FFSS attitude: a favours some object x (e.g. the attitude referred to in A1) forits own sake for b’s sake.

5.2 Sake, end, and objectiveFFSS attitudes seem to involve an ability to harbour attitudes for someone else’s sake—anability, by the way, that appears to be an important part of empathy. To recognize thatsomething is of personal value to a, on this line of reasoning, is in some way to echo, for areason, a favouring that a has or ought to have vis-a-vis the object of value for a’s sake.Onemight object that if we take this route we shall be left stranded with two notions ofpersonal value: one employable only by the person for whom something is a personalvalue and another for those who recognize that something carries this person’s personalvalue.Wewould have to face, in other words, a first-person value-for, and a second- andthird-person value-for. However, this need not follow. The attitude that I am ready totake to, say, the stranger who is a friend of a friend of mine has something to do with myfriend’s attitudes to the stranger. But it is hardly my friend’s de facto favouring that I will‘take over’. Rather it is the kind of attitude that a has reason to take in those circum-stances—a reason that has met with my approval. Given this, I cannot see why a’s claimthat something has personal value should not be analysed in the same way.

These speculations apart, the following response—with the focus on final rather thaninstrumental pro-responses—can be anticipated: make up your mind! Either youfavour x for its own sake or you favour it for your own sake. You cannot have itboth ways. Actually, I am convinced that we can have our cake and eat it here. At anyrate, I have failed to identify a single, precise content of ‘a’s sake’ that cannot becombined with ‘favouring x for its own sake’. This seems to suggest that the word‘sake’ is ambiguous: it does not necessarily have the same content in sentences contain-ing ‘for its own sake’ and those containing ‘for a’s sake’.

Two things in particular complicate the evaluation of these FFSS attitudes. First, weshould keep in mind that ‘favours’ is used here as a technical term that covers a widerange of attitudes. So what is possible with regard to one kind of attitude may beimpossible with regard to some other kind of response. However, I have not found aclear example of an attitude that could not take the form of ‘a favours x for its own sakefor b’s sake’ (or the instrumental attitude ‘a favours x as a means for b’s sake’). Of course,when we look at more specific attitudes various issues will arise. Still, since I obviouslywill not be examining all of the different attitudes covered by ‘favour’, we shouldwatch for questionable generalizations on this matter. There might well be attitudesthat cannot be transformed into the ‘for someone’s sake’ pattern.

Evaluation of the FFSS attitude is problematic, secondly, simply because it involvesawkward, unnatural-sounding descriptions. In general, it is surely a good policy to be


suspicious about descriptions of this sort. But, of course, the fact that we need anunnatural-sounding formulation to capture a complex mental state is in itself no reasonto dismiss the associated attitudes on logical grounds. (Moreover, in this particular casethe FFSS description is actually an attempt to specify an attitude rather than to give amore general description.) I am inclined to think that the perplexity created by the twofor-the-sake-of qualifications derives to a great extent from a particular way of reading‘sake’. Claims about a’s sake and x’s sake might suggest that I am ascribing, at least to theformer, a sake or an end, in the sense of an objective (cf. the German ‘Sache’; seeRabinowicz and Rønnow-Rasmussen, 2000). However, generally, an analysis ofattitudes does not essentially require (as a description of a psychological profile mightdo) reference to an objective in the analysans. This needs to be kept in mind.Moreover, FFSS attitudes may be directed on a large number of different kinds ofthing, only some of which can be understood, in a reasonable sense, as havingobjectives. Furthermore, the kind of final personal value in which I am interestedalso accrues to objects that often cannot be understood as having objectives(cf. Rabinowicz and Rønnow-Rasmussen, 2000, p. 48). A poem that my daughterwrote to me, which plausibly carries personal value for me, is a case in point.However, as I hinted above, there is a simpler way of understanding ‘sake’. Suppose

all attitudes are either final or instrumental. Obviously, this simplifies matters. I might,for instance, favour something for its part or contribution in a whole (organic unity).However, I am quite sure that the inclusion of a discussion of these attitudes would notchange the general picture—other than by making the presentation more complicated.So in what follows I will set these attitudes aside.Supposing, then, that there are only final and instrumental attitudes, I can mean one

of only two things by saying that I favour x. First, I might favour x as a means. Wemight explain this further by saying that we favour x for its instrumental properties,these being the properties that we believe make x causally efficient in some way. Thatthe attitude is directed on a means will then be reflected in its intentional content; thisoccurs when the instrumental properties are part of the intentional content of theattitude, i.e. when a favours x only for its instrumental properties.

As to the second possibility, sometimeswe favour x not as ameans. In that case x’s beingcausally efficient is not that forwhich x is favoured.The features forwhich x is favoured donot, or at least do not only, include x’s instrumental properties. I propose that we take ‘forits own sake’ to indicate that we are favouring the object in a final, non-instrumental way,as specified above. There need not, therefore, be anything perplexing about having twofor-the-sake-of qualifications in our description of an FFSS attitude.There is further complication here. As has been argued in recent years, sometimes

we value objects for their own sake in virtue of their instrumental properties (e.g. seeKorsgaard, 1983; Brännmark, 2001; Rønnow-Rasmussen, 2002b). The attitudesinvolved in such cases would have to be distinguished from the two attitudes I amdiscussing here. I am not sure about the details, but I suspect the intentional content ofan attitude might include an object’s instrumental properties, though it would not do

5.2 SAKE, END, AND OBJECTIVE 61

so only for the sake of their instrumentality or causal effectiveness. Whether this ispossible remains to be shown. However, if it is we should be able to differentiate suchattitudes from the two I discuss above.

To advance the process of clarification, then, I shall take the first and second ‘sake’ in‘a favours x for its own sake for b’s sake’ to indicate what sorts of property of x and b arepart of the intentional content of a’s attitude. If x as well as b are not viewed merely asmeans, the properties for which we favour x for b’s sake will not be their instrumentalproperties.6 As to the values understood in terms of these attitudes, the first ‘sake’ in‘favouring x for its own sake for b’s sake’ specifies that the value accruing to x need notbe conducive to something else that is valuable. Final personal value is best described,then, as an end-point value or an ultimate value; this sort of value is not determined bysomething else being valuable. Similarly, the second ‘sake’ indicates that the favouringattitude is not instrumental. Our attitude is to b, and it is final in the sense that it doesnot depend on some other favouring. An attitude with the attainment of someobjective as its object will only be relevant to the analysis of personal final value ifthe attainment of the objective is favoured for the person’s sake.

It might well be incoherent to interpret the attitude ‘favouring x for its own sake fora’s sake’ in terms of the ‘end/objective’ sense of ‘sake’. Certainly it would be reasonableto request more details about the ‘objective’ a person might be said to have. However,in the ‘end-point value’ sense, there is nothing strange about ‘sake’ referring to one setof properties in the case of x and another in b. Why a particular set of properties suppliesus with a reason to finally favour something for a person’s sake is ultimately anevaluative issue; attempts to address this issue will require substantive justification.(Think of the hedonist who only acknowledges that certain kinds of experience arefinally valuable; such a person will reject alleged examples of non-derivative finalpersonal values that are said to accrue to things other than experiences of pleasure.)

The interpretation of ‘sake’ in terms of b (rather than in terms of b’s objective) isperhaps, therefore, most plausibly understood as being about b’s (discerned) properties.However, at this point it is important to see that the idea that attitudes take as theirintentional object properties can be understood in various ways. As I have recentlyargued (2008b), when it comes to at least one kind of love, and one kind of admiration,the former but not the latter is a bona fide example of an attitude that can, in a certainsense, be said not to take properties as its object. Here, this claim needs to be amplified,so in a moment I shall say a little more about the way love and admiration differ fromeach other in an important respect. Meanwhile, in order to avoid a possible misunder-standing, let me say something about how I do not understand personal values.

Suppose that it is the case that:

6 One more possibility ought to be mentioned. I might have the non-instrumental properties of x and theinstrumental properties of b in mind. This would be a complicated (but, I believe, genuine) attitude involvingyet another object or person.


(1) x has final personal value for b.

And that the following therefore is true:

(2) There are normative reasons to favour x in a final way for the sake of b.

That we can have normative reasons to favour something because we favour somethingelse is plausible. But in that case, the question is whether we should understand (2) andfinal personal value along the lines of (3)?

(3) There are normative reasons to favour x in a final way because b is favoured.

As one referee from Oxford University Press suggested, this would seem to be confused:“If one has a normative reason to favor x because one favors b, then how can that be anormative reason to favor x in a finalway? To have a reason to favor x in a final way is tohave a reason to favor x that is not dependent on one’s favoring some other thing.”This is a fair worry. However, my analysis is not assuming the truth of (3). I also want

to emphasize that I am not taking (2) to mean (3), and hence, do not understand finalpersonal value along the lines of (3). What follows from (1) is that we should favour x,not as a means but as an end for the sake of b. This is not confused, at least notnecessarily, if we understand ‘sake’ as I suggested above. It depends on what role theproperties of x and b play in the intentional content of a’s attitude. If your attitude canbe directed to two objects, x and b, then there is the possibility that the properties ofthese objects can play the role of “instrumental properties” or “end-properties” in theintentional content of such an attitude. Moreover, if there are no conceptual obstaclesfor thinking that the properties of x and b are not instrumental properties in theintentional content (i.e. they are not discerned with an eye to what they can lead to)then I do not see why it cannot be the case that there is facts that call for us to have suchattitudes. In other words, if we can make sense of such an attitude, we should also beable to make sense of the claim that there is a reason to have such an attitude. Whenthere is a reason for such an attitude there is an object that carries final personal value.

5.3 Discerning attitudesAlthough we often invoke FSS (i.e. for someone’s sake) attitudes, it would obviouslybe desirable if we could describe them in a more precise way. Ideally, we do not wantto rest our case on linguistic reports which, in my view, rarely if ever capture the fullpropositional content of our attitudes. For instance, it is not hugely clarificatory to saythat we are dealing with an attitude the propositional content of which is ‘thatsomething is favoured for someone’s sake’. I think the attitudes at issue here displayseveral quite different kinds of propositional content, and I suspect that these contentsare seldom of the above-mentioned type. Still, although I am unable to give a fullaccount here, I will at least separate two kinds of for-someone’s-sake attitudes, both ofwhich are pertinent to the analysans of personal value; and I will do so by consideringthe intentional content of these attitudes.

5.3 DISCERNING ATTITUDES 63

I suggested above that itmight be incoherent to understand FSS attitudes in terms of the‘end/objective’ sense. I also said that there is another approach—one pressing ‘end-pointvalue’ into service—and that this approach looks plausible. On that latter view therewould be nothing odd in allowing the two instances of ‘sake’ in FAP refer to different setsof properties. In one sense (the precise content of which I will come to shortly) this talkabout understanding ‘sake’ in terms of properties is quite understandable. However, thisfact should not hide another fact, namely that not all attitudes take properties as theirintentional content. Love, for instance, is a bona fide example of an attitude that does notin any obvious way have properties as its object. Moreover, I believe that the mostinteresting group of FSS attitudes have a good deal in common with love, and so, Ithink, there is good reason to clarify what is so particular about love. At least, on oneconception of it. The notion of love is highly complex. For instance, is it, as differentauthors have argued, an apprehension of something in the beloved, or an emotional responseor perhaps a volitional state that has the beloved as its object? The latter suggestion is HarryFrankfurt’s, and my view, which I briefly touch on in this chapter, is in several respectssimilar to his, and in particular his account of what he calls “active love”.7 People areinclined to ascribe value to persons they love. However, the relation between love andvalue is far from straightforward. This is particularly evident if love is depicted, as I think itshould be, as an attitude that takes non-replaceable persons as its intentional objects.

The extent to which people do in fact display the kind of love depicted here is amatter I leave open. The romantic literature is full of references to this sort of love,though. For instance, consider how W. B Yeats ends his poem ‘For Anne Gregory’:

But I can get a hair-dye/And set such colour there,/Brown or black, or carrot,/That young men in despair/May love me for myself alone/And not my yellow hair.

Love is an attitude typically directed on some person. I will not discuss here whetherthis happens in a more direct or indirect way. However, this much can be said rightaway: many of our attitudes require for their explanation mention of the fact that welove someone. This lends some support to the ‘indirect’ idea that love plays a ‘master’role among our attitudes.

We love persons, that is, and not, say, the properties of the persons or states of affairsinvolving the beloved.8 Of course, people can love things other than persons (animals,and perhaps even things, including their cars or football teams). But that apart, we

7 See Frankfurt, ‘Autonomy, Necessity, and Love’ (1999a). See also his description of love in ‘On Caring’(1999b, p. 165).

8 For example, Max Scheler, as Mulligan pointed out to me (personal communication) rejects theproperty view of love. See Mulligan, ‘Scheler: Die Anatomie des Herzens oder was man alles fühlen kann’(2008).


typically love persons but not properties. Now this claim has been contested. In fact, itis not uncommon to find it argued that what we love, when we love a person, is theperson’s properties. This idea can be found in Blaise Pascal’s Pensees.9 He refutes theview that what we love is a person with the observation that when one’s belovedchanges in some more or less substantive way one’s love also does. If the change is greatenough, the love will eventually fade away. This shows, he concludes, that what oneloves cannot be a person. Instead, love is love of qualities:

A man who sits at the window to watch the passers-by; can I say that he sat there to see me ifI pass by? No; for he is not thinking of me in particular. But someone who loves a person becauseof her beauty, does he love her? No; because smallpox, which will destroy beauty withoutdestroying or killing the person, will ensure that he no longer loves her.And if someone loves me for my judgment, for my memory, is it me they love? No,

because I can lose these qualities without losing myself. Where is the self, then, if it is neitherin the body nor in the soul? And how can you love the body or the soul except for itsqualities, which do not make up the self, since they are perishable? For would we love thesubstance of a person’s soul in the abstract, whatever qualities it contained? That is impossible,and would be unjust. Therefore we never love a person, only qualities. (Pensees and OtherWritings (1995 [1897], pp. 130–1)

It is not obvious why anyone has taken this argument seriously. If the target is the ideathat the qualities of a person have absolutely nothing to do with love, he has a point.However, his observation is not sufficient to establish something that he also con-cludes, namely that the love of persons is a love of qualities. So why the popularity?For sure, the position has a distinctly unromantic ring to it, which I am sure has someattraction to at least certain people. But this cannot be the whole story. Perhaps theargument benefits from the confusion of this non sequitur with the plausible claimthat to love someone is, minimally, to stand in a causal relation to some of his or herqualities.There is some truth to Pascal’s sinister claim, though. Nevertheless it is the expres-

sion of either a quite confused idea or, at best, an incomplete description of what itmeans to love someone. The kernel of truth is, of course, that we love persons in virtueof their properties (I will get back to the complex copula ‘in virtue of’ in a moment).But it is a fallacy to go from this to the conclusion that we love the beloved’s properties.To make this move is to overlook the familiar difference between the cause of anattitude and its intentional object. The object of love is a certain person; and we shouldnot confuse an attitude’s causes with its intentional object. Perhaps these might, insome cases, coincide. The fact that Romeo loves Juliet in virtue of her black hair doesnot mean that what he loves is the blackness of her hair. Her hair might well have

9 For a critical survey of different positions on love, see Christopher Grau, The Irreplaceability of Persons(2002). This dissertation drew my attention to the fact (among many others) that we owe to Pascal the viewthat it is the qualities we love, not the person.


something to do with the fact that he loves her, but it is Juliet that he loves, not a featureof her hair.

Let us dwell a moment on this idea that what causes us to have an attitude towards anobject need not be found in the intentional content of the attitude. It plays animportant role in what follows. Unfortunately, and surprisingly, there seems to beno suitable word or simple expression that indicates decisively whether we have thecontent or the causal source in mind. All the candidate expressions are ambiguous.Thus, to say that a favours y ‘on account of ’, ‘in virtue of ’, or even ‘for’—to mentionsome readily imagined possibilities here—is to say something that can be understoodboth as being about what caused a to have the attitude and as being about the contentof a’s attitude. This is a peculiarity of the languages that I am familiar with. To avoidmisunderstandings, it must therefore be kept in mind that I will be using theseexpressions, henceforth, with the attitude’s content (not cause) in mind, leaving itopen whether what caused the agent to have the attitude is also what appears in thecontent.

Although English leaves us in the lurch here, there is a further reason not to forgetthis distinction between attitudinal cause and content. That it is not a case of philo-sophical pedantry is clear, I think. Here is a down-to-earth case. (I will touch on ananalogous matter in Chapter 9, Section 9.4, when discussing the person who believedGod had told her to jump into the water.) Suppose you admire a person for herhonesty. If asked why you do so, you might truthfully say something to the followingeffect: ‘because she is honest’. Now, it is quite likely that her honesty caused you toadmire her, but it would also be possible, in a different case, for you to admire her forher honesty even though she is in fact a dishonest person. In short, you may be mistakenabout her character, but this is consistent with your admiring her for her honesty.Honesty, here, appears in the intentional content of the attitude, but since the person isnot honest, it cannot be what caused you to have the attitude. Again, suppose I like myboss because he or she is generous. It is, of course, possible that his or her generositycaused me to like him or her. But it is also possible that I am deluding myself, and thatwhat makes me like the boss is something quite different—namely, that he or she is myboss, or that my positive attitude will, in effect, make it more likely that I will befavoured by the boss.

Clearly the distinction between attitudinal content and cause illuminates theseexamples. It allows us to understand how people can be mistaken about what theyfavour, and this is important, since, as a consequence of favouring things for what wetake them to be, we sometimes favour them for what they are not. Plainly, propertieswe wrongly take people to have cannot cause us to have corresponding attitudes.However, this is quite consistent with the fact that these properties appear in theintentional contents of the attitudes.

Notice, too, that in distinguishing between cause and content I am not implying thatnon-veridical attitudinal content is somehow without cause. I am merely saying that,whatever the actual causal story happens to be, what caused the content to be what it is


need not be what caused the person to have the attitude in the first place (whichcertainly is one of life’s mysteries).It is an interesting metaphysical question, albeit not one I will address here, just how

to more precisely understand what it means to say that a property is contained,exemplified, manifested, or more generally figures in a propositional attitude. Theseexpressions, which are the ones I employ in this work, are, as far as I can see, consistentwith classical metaphysical views on property instantiation.Let us now return to the attitude of love.10 I think it is clear that, though it is in many

ways a special case, love is not the only attitude that takes persons rather than, say, theproperties of the person as its intentional object. Attitudes in general seem to be capableof taking this shape. Suppose, therefore, that the identity of a person can in some way bea feature of the intentional content of the attitude. This would at least be one way ofunderstanding the somewhat cryptic claim that love ‘takes a person as its object’. Thesupposition can be made, I think, in connection with the sort of love that I aminterested in here. Naturally, this ‘essentialist’ claim is not easy to spell out in anymore detail; but I would like to provide a further indication of what I am driving at.Love appears to consist in a complex set of attitudinal dispositions. Sometimes these

dispositions are activated, and accordingly the person then has the attitude. Love, in sucha case, is describable as an occurrent attitude. These dispositions often play no active partin our motivational set-up. When this is the case love can be said to be in a dormantphase, being ready to move out of the motivational background when, for instance, onethinks of one’s beloved, or when one actually sees him or her. Suppose we consider acase of occurrent love. One person is, say, directing an occurrent love attitude on a, theperson he or she loves. How should we then express the fact that a’s identity is a part ofthe attitude? The gist of a suggestion I have previously made (in Rønnow-Rasmussen,2008b) is this: the identity of the beloved plays an indispensable part in love, in the sensethat the persistence of the attitude in question is conditional on the lover’s belief that theperson on whom the attitude is directed is identical to the beloved. Unfortunately thisidea is not as clear as it needs to be, so let me try to develop it further.What is characteristic of the conception of love I have in mind is that it pictures love

as taking a non-fungible (i.e. non-replaceable) person as its object.11 Love, thusunderstood, is nonetheless quite consistent with what are generally taken to be factsabout people’s psychology—namely, that a person can love more than one person at atime, or that one can fall in love with someone b despite already loving a. Love, likeother attitudes, can be individuated. Hence, in principle, nothing prevents us fromhaving many such attitudes.

10 In his discussion of love’s constancy, ‘A Conceptual Investigation of Love’ (1973), W. Newton-Smithsays much that I am in agreement with. However, what I miss in this otherwise insightful work is an awarenessof the distinction between what causes us to love someone and the intentional content of love (e.g. see p. 122,where Newton-Smith uses the expression ‘on account of’ without specifying what he has in mind).

11 This conception of love raises philosophically interesting issues that I have had to set aside here. I discusssome of these in (2008b).


Once love is thought of as an attitude taking a non-fungible (i.e. non-replaceable)person as its object, some interesting but rather counter-intuitive features come to thesurface. Perhaps the most vital of these is the following: How should we harmonize theidea that the identity of the beloved is a part of the relevant attitude with the idea thatattitudes are in general discerning? By the latter I have in mind, as may be recalled, theidea that our attitudes are directed on objects on account of the properties of thoseobjects. That is, at least some of the properties of the object are exemplified in theintentional content of the attitude. To be more precise, what makes an attitude moreor less discerning is that the intentional content of the attitude contains these (dis-cerned) properties to a greater or lesser extent.12

Earlier, in Section 3.2, the dual role of properties in attitudes was discussed. This roleadds an extra dimension to this matter. If we take into account the fact that, in a valuecontext and according to a fitting attitude account of value, attitudes are called for, theaforementioned properties will also be what makes the object valuable. That is, theseproperties will make up the supervenience base of the object (they are subjacentproperties). Thus in the case of a unique painting, uniqueness will be what (in part)makes the artwork valuable, or at least more valuable. However, facts about thisproperty will also provide a reason for favouring the object (e.g. for caring for it).We should favour the painting on account of its uniqueness.

Can there be entirely non-discerning attitudes?13 Could love be characterized assuch? Could properties of a beloved individual be wholly absent from the intentionalcontent of the love borne to that individual? Is this what might be meant by the claimthat love ‘takes a person rather than properties as its intentional object’? The old refrainthat love is blind might suggest as much. But though we cannot just rule out thispossibility, the idea of a non-discerning attitude is somewhat puzzling. Attitudes arediscerning in degrees, but it is hard to imagine a wholly non-discerning attitude to anobject x, if by this we mean an attitude the intentional content of which does notcontain, or represent, any of a’s properties. However, there is in fact another approachto be considered, which looks more viable. This will also improve our understandingof the cryptic idea that FSS attitudes take non-fungible persons as their objects.

5.4 Identifiers and justifiersWe need, to begin with, to introduce a general distinction that seems applicable toattitudes generally. This is the distinction between what I shall call justifiers and

12 En passant, recall here that by ‘on account of ’ I have in mind what appears in the intentional content ofthe attitude; whether that is also what caused the attitude is something I leave open.

13 On certain views on reference, when the content of an attitude only contains, say, a proper name, theattitude is not discerning: such an attitude contains an element that has reference but no sense. What I say inthis work about discerning attitudes might, therefore, if these views are correct, concern the special case ofattitudes that do contain more than a proper name—that is, they contain some of the properties of the objectof the attitude.


identifiers. We can separate two kinds of property which differ from each other in therole they play in the intentional content of the attitude: justifiers justify, identifiersidentify. Later on, this division will help us to understand the two kinds of FSSattitudes.I will be claiming that the identity of the person for whose sake we have an attitude

may be a part of the attitude’s intentional content, and that this is consistent with thefact that the attitude is discerning. The explanation is that properties can figure in theintentional content in more than one way. Various conceptual barriers limit the kindsof property we can invoke to justify attitudes. For instance, the properties that typicallyjustify love are not the kind that justify admiration. In fact there is good reason tobelieve that the kind of property that justifies love cannot be what justifies admiration.This suggests that properties may have different functions, or play different roles, in anattitude. This is true of normal attitudes as well as being true of the more complex FSSattitudes. Thus, I shall argue that, when it comes to admiration, properties of theadmired object play the role of justifiers rather than identifiers; but that, by contrast, inthe case of love properties of the beloved are identifiers. Of course, this needs to befurther elaborated. That elaboration will make it quite clear that there is nothing tosuggest that FSS attitudes are not capable of being similarly distinguished.Let me therefore return to the case of admiration, which is in many ways a more

typical discerning attitude. We admire a person a for his or her wit, or courage, say.Properties such as these are what make a admirable. They form the supervenience baseof admirability, i.e. the set of admirability-makers. Now, these subjacent propertiesmay well play some sort of value-making or attitude-justificatory role in associatedintentional contents. Actually both possibilities seem likely. Thus, sometimes when weadmire a for his or her wit and courage, the content of the attitude is such that theseproperties of a appear in it as value-makers, as what makes a admirable. At other timeswe are not quite sure what it is about a certain object that makes us admire it: we justdo. When this happens we are often ready to argue our case, or to take measures to findout what it is about the object that we admire. In such cases the role of the properties islikely to appear in the content as that which justifies the attitude rather than as makersof value. There will also be, I imagine, times when attitudes combine both kinds ofcontent. However, for simplicity’s sake let us just make a mental note of this possibility.Let us again consider love. Here things are different; what causes one to love a

person a will to some degree appear in the intentional content, but not obviously aswhat makes a valuable or justifies one’s love.14 We might try to express the difference asfollows: in cases of admiration we admire a, and a is made admirable by his courage andwit. In cases of love we love a, and a possesses such-and-such properties. That we love aperson in the sense I have in mind here is quite compatible with the fact that we may

14 Sometimes we might think that our love is justified. I am just not assuming that to love someone is toregard one’s love as justified. There is ample evidence that people sometimes love persons they do not thinkdeserving of their love.

5.4 IDENTIF IERS AND JUSTIF IERS 69

have a number of other attitudes to the beloved. For instance, Elly, my wife, has anumber of qualities; she is clever, unselfish, and considerate. Her having these proper-ties makes her a good person; her qualities are such that I have a reason to favour her invarious ways. These and many other features she has make her desirable, someone thatI find worth caring for, and so on. If the qualities were somehow lost, I would stopdesiring her for the sake of those properties. I would not necessarily stop loving her.

In the admiration case the properties figuring in the intentional content are whatmake a carry value in the first place (ceteris paribus, they are together what justifies us inadmiring a). In the love case, the properties are rather identifiers of a; they do not playany value-making or attitude-justifying role.15 This latter role is instead played by theidentity of the person. The properties discerned are what identify the person, albeit notnecessarily successfully; and it is the identity of the person that justifies our directing ofthe attitude on the person. It is important to realize that the properties that function inthis way as identifiers are unlikely all to be part of the intentional content. This impliesthat in order for a property to qualify as an identifier it is sufficient that it contributes toidentification—whether veridical or not. (The notion of an identifier is, in otherwords, not a success notion.) That the content of an attitude does not always encirclethe identifiers (of a person) is a complication, and one to which I will return in Chapter6. For the time being we can set it (and the question why it is important) aside.

It might be objected that the distinction between identifiers and justifiers is improp-erly artificial; there is no real argument for ascribing different functions to properties.Admittedly, there is an artificial flavour to this distinction. Despite this, however, thecriticism can be deflected, as I shall show shortly. The matter is complicated, though.So let me in the meantime briefly restate what I so far have claimed.Discerning attitudes are attitudes directed on some object on account of some of the

object’s properties. In some cases, some of these properties provide a (right kind of)reason for directing the attitude on the object. Suppose I admire a person a for beingW. We can then construct the following schema, where the subscript shows that thearrow signifies the justification relation:

W !j admiration of a

In other cases, our attitude may be of a quite different kind. Love (at least, a certain kindof love) is perhaps the most obvious example. Since it too is a discerning attitude, it willcontain the properties of the beloved in its intentional content. However, theseproperties will not fit into the above schema. It is not my beloved’s possession of theproperty W that justifies my love for him or her. Rather, it is the identity of the personthat appears to have that function. Despite being present, in the intentional content,the properties do not play the same role here as they do in the admiration case. It is

15 Cf. Robert C. Roberts’s point, made in Emotions: An Essay in Aid of Moral Psychology (2003), that “love,in the dispositional sense of attachment... is a construal that identifies the object of concern” (p. 288).


therefore necessary to interpret the arrow in a different way when it comes to love. Tomark this need let us use the subscript ‘ij’ to signify ‘identity-justification’:

W !ij love of a16

Now, it is not obvious that ‘justifies’ is the right term for what is going on in the case oflove. Certainly, one reaction at this point would be to say that the person who loves ahas no justification for his love (on this picture of love). He or she just loves a, period.‘I-justifies’ would then mark the fact that something other than justification isinvolved. For the time being, however, I will regard it as an open issue whether weshould consider such attitudes properly justified attitudes, or whether we need to lookat what is going on here in entirely different terms.When it comes to property-justification, with one important exception all or some

of the properties of the object on which the attitude is directed are invoked asjustification. The exception has to do with the fact, which attentive readers will haveconsidered, that identity is, of course, a property too. But property-justification doesnot cover the identity of the object. As to identity-justification, only the identity of theobject on which the attitude is (either directly or indirectly) directed is invoked.Having stressed this difference—and it is perhaps important to emphasize, in particular,that we should not merely assume that there are two sorts of justification—I believe it issafe to continue without noting on each occasion whether we are discussing property-or identity-justification. Henceforth, then, I will stick with the term ‘justification’. Wewill return later on to the question whether I-justification is really a variety ofjustification or merely a mark of something else. I am inclined to think the former isthe case.For the time being I am more interested in establishing that properties can play

various roles in attitudes. Thus, having said that the identity of the beloved is whatjustifies one in loving this particular person, we still need to allow a role to be played bysome of a’s other properties. We need to do this because we have agreed that love is adiscerning attitude. Where love is claimed to fit into the above schema—and we canleave it open for the moment whether love is held to be like this for conceptual reasonsor merely on occasion—a’s properties function as identifiers of a.Let us therefore return to the complaint that the distinction between the two roles is

artificial and of no substance. There is more than one reason to contest this reply. For astart, the distinction is backed up by the observation that we have agreed that certain ofour attitudes cannot always be properly justified although they are directed on objects.That is, the relevant attitudes are not justified by invoking the subjacent properties ofthe object (the identity of the object not included). However, a caveat is in place. Theproblem is not epistemological. Sometimes we might be more or less unsure what therelevant justification is, but this is not what I am discussing now. The point is rather that

16 I owe David Alm thanks for pointing out (in personal communication) that my first attempt to presentthe scheme was more confusing than helpful.

5.4 IDENTIF IERS AND JUSTIF IERS 71

in some cases we acknowledge that an attitude of ours is such that we find ourselvesmerely having it; we have sided neither with nor against it. The properties of the objectare ‘there’ all right, in the intentional content—but they are not doing any job at all asjustifiers. We have, as it were, been struck by the attitude.17 In fact, at times we do notmerely experience the attitude as something planted in us; we also believe that there aregood reasons (either moral or prudential) for us not to have the attitude in question.

A similar phenomenon has been discussed at great length by theorists analysingemotions. Take a textbook example. Many people suffer from fear of flying althoughthey know very well that it is much safer to fly than it is to drive a car. Their fearremains unaffected by their belief. Another example is a person who has a certainpreference for, say, a particular sexual activity, but who believes that there are goodreasons for him not to live out his sexuality.18 Justin D’Arms and Daniel Jacobson referto these kinds of emotion as “recalcitrant emotions”, and this label catches very wellwhat these affective states are all about.19 As far as I can see, emotions are not the onlyrecalcitrant mental states we have. There are many attitudes that we somehow regardourselves as having, but which we look upon as alien to the kind of person we wish wewere; these attitudes are like uninvited guests that we try to show the door but oftenjust have to learn to cope with.

Notice that I am here describing a general phenomenon. I am not saying, for instance,that love should be regarded as an alien, uninvited attitude. It might be like that, but itneed not be. I am merely pointing out that sometimes we do not intend to, and cannot,justify a discerning attitude in terms of properties of the object on which the attitude isdirected. This strongly suggests that the properties of objects can play various roles inattitudes. The distinction is therefore not a pure invention. There is a sound basis formaking it. In fact, a further observation—besides the purely phenomenological one thatwe sometimes experience ourselves as having an attitude we recognize ourselves as justhaving—a fortiori strengthens the idea that properties which figure in intentional contentmay be playing various roles. Thus sometimes we are ready to accept that it is logicallylegitimate for a person not to provide a justification, cast in terms of the properties of theobject, for having an attitude. Of course, it is far from obvious exactly how logicallegitimacy should be determined. Much could be said here. I shall merely hint at what Ihave in mind. In some cases, we seem to accept that people have attitudes that theysimply recognize as theirs—we allow them, if you like, and crudely put, to be struck bythe attitude. Moreover, we recognize that there is no linguistic ground on which todemand justification for the attitude. Certain attitudes are merely had.

17 This observation has been made by some authors: e.g. see Frankfurt (1999a).18 A case of this is the Swedish priest who declared in the newspapers in the summer of 2007 that he was

homosexual. While he thought it was not a sin to be a homosexual, he did think it was a sin to performhomosexual acts, and he had therefore decided to live in celibacy.

19 See D’Arms and Jacobson, ‘The Significance of Recalcitrant Emotions (or anti- quasi-judgementalism)’(2003): “Wewill say that an emotion is recalcitrantwhen it exists despite the agent’s making a judgement that isin tension with it” (p. 129).


A typical example is having a desire for something. Consider a person who wants herhouse painted red. She may have no other ‘justification’ than the fact that she desires tohave her house painted red. After a period of indecision she just realizes that it is a redhouse she wants. By refusing to come up with an answer to the question ‘why do youwant it red?’ (other than something along the lines of ‘I just do’) this person is notcommitting a logical mistake. Compare this case with one in which someone a admiresanother person. Here, a may not be able to provide a justification for her attitude, but,for logical (i.e. conceptual) reasons, she cannot deny that a justification is called for. Tosay ‘I admire her for no reason at all’ is to commit a conceptual error, to misuse the term‘admiration’. ‘I desire a red house for no reason’ is not in the same way a logicallystrange thing to say.Whether the ‘desire’ case should be handled in the way I propose for love is

doubtful, but I will not pursue this issue here. Another tricky issue that I will not gointo requires us to say to what extent we can genuinely admire something that weacknowledge not to be admirable. However, notice that whether or not we admit thispossibility, it remains the case that, for conceptual reasons, if you admire something,you can always legitimately be asked to justify your attitude in terms of the admiredobject’s properties.We do not so much love a person for his or her properties as we love him or her

regardless of his and her properties.20 As a result, we cannot (nor do we recognize theneed to) supply a justification for our love cast in terms of the beloved’s discernedproperties. This, in its turn, explains why we sometimes feel that we are not in a senseresponsible for having the attitude. However, on this view of love we are stranded witha problem. For supposing that attitudes are never wholly non-discerning, we need tounderstand what role properties do play in the attitude. I suggest that the distinctionbetween identifiers and justifiers accounts for this.

5.5 Two kinds of FSS attitudeLet us now return to the FSS attitudes mentioned in the analysis of personal value.I repeat (3) above from Section 5.1, then:

(3) a favours x for b’s sake

Now it might seem tempting to ask what this kind of attitude resembles more, love oradmiration. However, since ‘favours’ is but a notational alternative to ‘has a pro/contra-attitude’, this question would appear to be too general to admit of any definiteanswer. Still, we might confine ourselves to the kind of FSS attitudes that we believewill, as a matter of fact, be expressed when we think something has value for someone.We might then ask: Should personal values be understood in terms of love-like or

20 Recall that I have set the psychological realism of these attitudes aside. I am interested in a certainnotion of love as a logical possibility.

5.5 TWO KINDS OF FSS ATTITUDE 73

admiration-like FSS attitudes? Or to put the question differently: Is favouring x for b’ssake an attitude whose persistence is conditional on the identity of b? Is it, in otherwords, one, like love, which accommodates identity-justification? Or is it an attitudewhose persistence is independent of the identities involved—one we might even becompelled logically to adopt, on pain of conceptual error, vis-a-vis a person, or thing,that we agreed did not relevantly differ from b?Perhaps the question should be rephrased. We must keep in mind that love is an

Identity attitude, i.e. an attitude capable of identity-justification, while admiration is aNon-Identity attitude. The issue then, I believe, is not whether there are FSS attitudesthat are dependent for their existence on the identity of a person, for clearly there are.What makes things less straightforward is that it also seems clear that FSS attitudes donot all have to be like love (in this respect, that we are not conceptually compelled tolove someone who shares the same features as someone we love). That is, there can alsobe Non-Identity FSS attitudes—attitudes which, in common with admiration, aresuch that the persistence of the attitude is conditional not on the identity of a person a,but rather on universal properties of a. If these observations are correct—and in amoment I will argue that they are—the real crux of the matter appears to be whetherpersonal values should be understood in terms of only one type, or both types, of FSSattitude.

Consider the following scenario. I am at a dinner. The people at the table are allunknown to me. Suppose person a in front of me politely asks me to pass the salt to b,the person sitting to my left. This person, too, is a complete stranger. If I pass it, I willhave acted on what might be described as a desire to pass the salt to b for a’s sake. Is thisdesire conditional for its persistence on the identity of a? This seems rather unlikely.Still, I can correctly say that I passed the salt to b for a’s sake.Consider next a different scenario. The woman sitting beside you at dinner presents

herself in the following way: ‘I am John’s mother’. When you realize that you have themother of your friend, John, beside you, you decide to take especially good care of her.You soon find her quite boring, but you do your best to be entertaining—and you doso for your friend’s sake. When the evening is over, you realize that she was John’smother all right, but not ‘your’ John’s mother. Here your attitude of being entertainingfor your friend’s sake does seem to be conditional for its persistence on the identity ofJohn. The ‘John’ that figures in your intentional content is not one that corresponds toyour real friend. This conclusion may not seem obviously correct, but the followingobservation supports it. The John of your attitudinal content is obviously a person whois related to the woman sitting next to you in a mother–son relationship; it is when yourealize that this ‘content-John’ does not correspond with the real John that yourattitude changes.

Let us now return to the question I posed earlier: Should personal values beunderstood in terms of Identity FSS as well as Non-Identity FSS attitudes? Here Iam afraid that we have gone as far as formal analysis can take us.


Certainly, it might be argued that in the case where I pass b the salt for a’s sake it israther obvious that, in some sense, I am not really doing it for a’s sake. Instead I amdoing it for the sake of anyone who exemplifies the universal properties of a that makeme pass the salt.21 In this case these properties pretty much amount to being seated infront of me, being a stranger, and asking me to pass the salt to another stranger sittingbeside me. However, this does not really answer the question; it only spells out againthat there is a difference between two kinds of final attitude. Cogent argument thatpersonal value should never be understood in terms of Non-Identity FSS attitudeswould require evaluative reasoning. I think, therefore, that the best approach to thisissue (at least, in a work of a formal character) is to acknowledge that personal valuescan be analysed in terms of both Identity and Non-Identity FSS attitudes. The moreinteresting personal values—those expected to have a substantial impact on a person’slife—will, in my view, be those that require for their analysis Identity attitudes; and itwill be these with which I concern myself as we continue.On the other hand, in my view we should allow even Non-Identity-final attitudes a

role in our evaluative reasoning. Consider again the example of the tourists in thechurch, which indicated that it is possible for us to take something to have personalvalue even if we do not know whose value it is. The tourist believed that he shouldbehave in certain ways for the sake of the religious believer. People (he thinks) ought torespect certain religious objects for the sake of people attending worship. In cases likethis, at least, people appear to recognize that they have reasons to direct Non-IdentityFSS attitudes on the relevant objects.22

5.6 Summing upI fear that this chapter covers several somewhat elusive issues. A brief summary mighttherefore be in place. After introducing some examples of FSS attitudes, I consideredthe possibility that these attitudes are, in effect, combinations of at least two or morestandard attitudes. I am not closing the door on this possibility. However, since the twomost salient examples that I can think of do not seem entirely satisfactory, I willcontinue to regard them as cases of a single attitude. Unfortunately I have had toemploy a linguistically awkward formulation to persist in this regard. But, as I argued,this should not deter us. It is not necessarily incoherent to think people have suchattitudes. It is possible, I argue, to make reasonable sense of these for-someone’s-sakedescriptions. In Section 5.3, I moved on to explain what I have in mind by ‘identity-

21 The example is simplified. In point of fact two personal values could be in play here: it might be the casethat you ascribe personal value to the satisfaction of b’s request or desire to make me pass the salt to a, or itmight be the case that you think a’s getting the salt is what carries a’s personal value.

22 The example is complicated, though. Some of us will also think that, fundamentally, the worshipperdoes not have any reason to favour the Madonna, which suggests that he or she might not after all considerthe Madonna to be carrying someone’s genuine personal value; our value judgement expresses rather a kindof inverted commas use of ‘value for’.

5.6 SUMMING UP 75

involving attitudes’. To this end, I discussed a typical example of such an attitude,namely love. I went on to contrast love with admiration, which I employed as anexample of a Non-Identity attitude. Having prepared the ground in this way, thedistinction was then, in Section 5.4, applied to FSS attitudes. They, too, which wassuggested in Section 5.5, can easily be divided into Identity- and Non-Identity-involving cases.

In Chapter 6, I consider some foreseeable objections to my notion of an Identity-involving attitude. For example, it would seem, on my account, that the dropping ofsuch an attitude is difficult to imagine. However, I will show why there is no need toworry; the objection overlooks something that plays an important role throughout thiswork, namely the distinction between an attitude’s cause and its intentional object.Chapter 6 also considers some other issues, the toughest of which arises when we try

to analyse value in terms of these FSS attitudes. In their nature, the Identity FSSattitudes (which I think the really interesting personal values require for their analysis)appear not to be well suited to value analysis. However, as I shall argue, although this iscertainly true when it comes to impersonal values, it need not be true about personalvalues.


6

Examining the Analysis

The preceding chapter was intended to add more flesh to the bones of the ratheropaque ‘for its own sake’ idiom, and to the thought that personal values accrueto objects that we value with a particular individual in mind. To accomplish this,I provided intuitive support for the idea that there is conceptual space to be filled insuch constructions as ‘O ought to be favoured for a’s sake’. If the suggestion I havemade about FSS attitudes (i.e. for-someone’s-sake attitudes) holds water, we now havethe materials we need to analyse personal values in a novel way.Before continuing any further, I should point out that in what follows I will, unless

clarification is necessary, speak of ‘favouring’ without specifying the kind of favouringinvolved, i.e. whether or not it is final or instrumental. This simplification allows me toavoid the cumbersome locution ‘favour x for its own sake/for its effect’s sake/forsomething else’s sake’.Now, FSS attitudes, like many other attitudes, are such that the properties in their

intentional content play various roles. In fact in the philosophically interesting casesthey appear to be particularly striking examples of discerning attitudes that displayintentional content-properties with more than one function or role. Their peculiardirectedness on something for someone’s sake strongly suggests that identifiers mayplay a role in their intentional content. Moreover, this approach implies a quite naturalway of separating the kind of attitude called for by objects carrying personal value andthe kind of attitude called for by objects with impersonal value. That is, the identity ofthe person for whose sake we have an attitude plays a particular, and crucial, role. Forinstance, when I care for a person b for my friend a’s sake, the properties of a will, inpart at least, be identity-makers in the caring attitude. The ‘justification’ of the attitudeis intimately related to the identity of a. If I somehow come to believe that I ammistaken—that the properties identify the wrong person—my caring conduct willemerge in a different light. I will have been caring for bwith an eye to a person I take tohave properties identifying a. But the person in question is not a, but a*. Of course,

I might care for b nonetheless—it might, for instance, be the moral thing to do. But if bis not somehow related to a, but rather to a*, and if I have no relation to a* (who is notmy friend), and I initially but mistakenly cared for b for a’s sake, it will be, if notinexplicable, then at least highly unexpected that I am caring for b for a’s sake. Myoperative reason for caring for b will be mistaken.

This is an important point, so let me dwell on it a little more. What I am trying tomake clear is not that it would be impossible for someone to care for something for aperfect stranger’s sake. It would be absurd to deny that possibility. This was what the‘pass the salt’ example was designed to show. The point is rather that, on manyoccasions, realizing that you have mistaken the identity of a person is sufficient foryou to drop your FSS attitude. I take this to be a fact about us as persons. In order toexplain this phenomenon, given that properties do play a role in the intentionalcontent of the relevant attitudes, we need therefore to come up with some sort ofstory. The suggestion is that in some cases the identity of the person plays no vital role,and in other cases it does. Personal values, then—at least those that interest me—shouldbe analysed in terms of these latter attitudes. Accordingly, we can now spell out FAP ina bit more detail:

FAP*: An object x’s value-for a person a (i.e. x’s personal value), consists in theexistence of a normative reason for favouring/disfavouring x for a’s sake—where thisfavouring/disfavouring is an Identity FSS attitude.

6.1 Dropping an FSS attitudeIt might seem that the view of FSS attitudes outlined here faces, in effect, a reductio adabsurdum. If these attitudes (at least, those of major relevance for FAP, the so calledIdentity FSS attitudes) characteristically have non-fungible persons as their objects, itbecomes hard to explain how people might come to stop having these attitudes,especially if the people for whose sakes we have the attitudes are qualitatively thesame. If FSS attitudes are genuine attitudes, they will be expected to behave like otherkinds of attitude, and this will mean, among other things, that they can be dropped andreplaced by other attitudes (though not necessarily in any voluntary direct way). Butsupposing this is true, and given that identity plays the peculiar role it does when wehave love in mind, as outlined above, we seem to face something of a mystery. In thecase of regular attitudes we often explain the loss of an attitude by saying that we madea mistake about the properties of the object of our attitude: for instance, the car youonce wanted turned out not to have excellent brakes as you originally thought, and soyou do not want this car any more. There is nothing strange about this. But when itcomes to Identity FSS attitudes, since no universal properties of the object (i.e.properties capable of being described without reference to any particular individual)justify, or explain, your direction of the FSS attitude on the object, you cannot ever, it

78 EXAMINING THE ANALYS IS

seems, avail yourself of a similar explanation. Therefore, it might appear that the viewoutlined here leads in effect to an absurd position. Does it not oblige us to ascribe to theperson who has lost the attitude some sort of amnesia concerning the identity of theindividual for whose sake she originally had the attitude? But that sounds odd; peopledo stop having attitudes, and they do so without having any doubts whatsoever aboutthe identity of the person they once, say, loved or cared for. Someone might be quitepositive that they are not suffering from amnesia, and that the person they do not loveis the very same person they used to love. So there must be something wrong after allwith the way love and other attitudes are described here.This objection can be dealt with quite easily. It is paramount not to blur the

distinction between what figures in the intentional content of one’s attitude andwhat caused one to have the attitude. Causes need not appear in content. Once we acceptthis, it should come as no surprise at all that what causes an attitude to disappearneed not be part of its intentional content. So if the explanatory reason (i.e. theexplanation in causal terms of ) why I favour something for a’s sake is some universalproperty, then a’s loss of this property might well result in my dropping the attitude.What might have seemed mysterious then boils down to the fact that what causesone to have, and not have, attitudes is often beyond one’s epistemic reach. But thisis nothing particularly troublesome for FSS attitudes, and indeed it is a feature ofmost of our attitudes.But then, it might be replied, we face another problem, namely that we are

unwarrantedly mystifying the typical person’s psychological set-up. That is, does notthe above response imply that when we lose an attitude, we must be at a loss as to whywe did so—why we stopped, say, being in love, or are no longer ready to care for b fora’s sake? Are we really epistemically thwarted in this way by our own psychologies?The implication does not follow, and besides I think we are sometimes surprised by

the attitudes we have or do not have. Just as we in some cases are struck by an attitude,we can be struck by its absence. Literature is full of examples of men and women whowake up one morning realizing that they no longer love the person next to them, orwho find that they have changed, overnight, into people who no longer need alcoholor drugs. Such flashes of realization are, at least, reported. However veridical thesereports are, we must of course grant that they are seldom followed by a genuine senseof perplexity; but this shows only that people often have beliefs about the causesinvolved.It is far from obvious that a certain kind of value should be analysed in terms of

attitudes that take non-fungible persons as their objects. For one thing, attitudes likethese are not expressible in universalizable value judgements. Value and universaliz-ability are, in many philosophers’ eyes, intimately related. Though universalizabilitymay in fact be an obstacle to FAP*, I will postpone discussion of it at this point. InsteadI want to briefly comment on an issue which, I am sure, has been lurking in thebackground for some time now.

6.1 DROPPING AN FSS ATTITUDE 79

6.2 Non-fungible persons and identityEarlier I stated that I take it to be a psychological fact that the persistence of some of ourattitudes is conditional on the identity of the person on whom they are directed.1 Arich, and in many ways intriguing, literature suggests that we should be careful whenemploying the notion of identity. This warning is based on the fact that it is fairly easyto imagine cases in which, for logical reasons, it seems impossible to say whether or notan individual at time t retains her identity at a later time t*.

Given this background, I can imagine the following sort of response to what I havebeen doing so far. It might be pointed out that whether or not an attitude is, as amatter offact, conditional on the identity of the person, some philosophers concerned with so-called diachronic identity (identity over time) have argued that attitudes ought not to beconditional in this way. So whether or not there happen to be such attitudes, we shouldgive up on them. There are good grounds for not distinguishing between objects only invirtue of their identities, and if two objects are qualitatively identical, it is irrational not todirect the same attitude on one object you have directed on the other.

The position I set out from is not inconsistent with this response—at least, so long aswe clarify it by making some very plausible amendments to it. Let us therefore returnto the example of love, which makes for an interesting test case.

To begin with, recollect what has so far been said about this attitude. The love onehas for a particular person at time t1 is not dependent, for its persistence over time, onone’s beloved’s possession of precisely those properties that he or she has at t1. Thisindependence feature explains why love does not necessarily disappear when theproperties of the person loved change. To love a particular person is to love him, orher, whatever his, or her, universal properties are—where these universal features aredescribable without reference to any particular individual. Love might therefore becharacterized as unconditional from one perspective. However, if we turn to whatcauses this attitude, love may well be dependent on the fact that the beloved does notexhibit certain properties (e.g. being unfaithful or a child molester). But acknowledge-ment of this does not commit one to saying that one loves one’s beloved for not being,say, a child molester. All it means is that if she were one you would not love her.2

This picture constitutes a truly romantic view of love. Needless to say, it is anincomplete account. It says nothing about what it is to feel or experience love: itconcerns the object of love. Furthermore, I hold it to be a clear possibility that love, asdepicted here, might be an element in different kinds of love. A description of, say,parental love would not coincide completely with a description of love between adults,and the love I have for my wife is not the same as the one I have for my brother and

1 I am inclined to think that in certain attitudes—love would be an example—this conditionality isanalytically built in. However, this is not a claim I shall argue for here.

2 See here Jonathan Dancy (1993, p. 77).

80 EXAMINING THE ANALYSIS

sister. Still, love as characterized here may well be a core element in different kinds oflove. In some cases it is more obviously present than it is in others.It might be objected that we have no reason to believe that anyone actually has this

sort of love for anyone. We might acknowledge that there is a conception of love suchas the one depicted, and agree also that this conception is pressed into service inromantic novels and the like. But in real life people do not have this attitude to eachother. Perhaps this is the case. It is at least a reasonable view. I am just not sure how onewould go about settling whether or not it is accurate in a convincing way. The fact thatpeople divorce, break up or fall out of love is not conclusive evidence. As we have seenabove, falling out of this (demanding) kind of love is quite possible. However, let me atleast bring up some evidence that love of this kind might not after all be a merelyromantic fantasy.Suppose your beloved dies and is replaced by an exact replica with the same physical

appearance andmental set-up. The two persons here are so similar to each other that youwould be unable to point to any difference between them in terms of appearance,behaviour, and thinking. If you think that you have a reason to feel sorrow, or at leastbelieve that you would feel sorrow, over the loss of your beloved, the likely explanationis that (you believe) your love is of the romantically resistant kind. The person you lovein virtue of whatever properties he or she has is no longer with you. So you feel sad.3

To feel sorrow in a case like this is quite consistent with the idea that you can cometo love the replicant; after all, this is quite likely to happen. (As might be recalled, it isquite consistent with the view outlined here that one could replace one love attitudewith another one.) We may expect that many of the properties that once caused you to(fall in) love (with) your beloved are present in the new person too. Whether or notthis is the case, however, the point here is that lovers will in fact say, about the scenarioabove, that they would be saddened by the news that their beloved has ceased to exist:the fact that a perfectly similar (or even better) replacement would be provided doesnot in itself cancel this reaction.One might try to resist this by arguing as follows. Suppose, without your knowing it,

we replaced your beloved with an identical copy. Would you not go on—in fact,would you not have to go on loving this person as much, and in the same way, as youdid with your beloved? This kind of challenge is well known, but hardly impressive.A person who experiences the love that I have in mind will go on loving the personwhom he falsely believes to be his beloved. How could it be otherwise? But from this itdoes not follow that on discovering that his love is directed on a copy, and that the

3 Cf. Michael J. Zimmerman’s (2001b) description of romantic love: “Suppose Kath has an identical twin,Kay, identical not only in terms of appearance but also in terms of personality, and so on. If Kath’s wit andfigure give me equally good reason to love her for her own sake, then Kay’s wit and figure give me equallygood reason to love her for her own sake. But this is a disturbing thought. It’s disturbing because it seemsincompatible with the attachment that I have to Kath in particular. The fact is that neither Kath, nor Kay, norI believe that reason somehow requires me to love Kay as I do Kath. Romantic love is simply not subject toduplication in this way. In this sense, it is not a rational attitude” (p. 129).

6.2 NON-FUNGIBLE PERSONS AND IDENTITY 81

original is somewhere else or no longer alive, he will not go and look for her or feel sadthat he has lost the person he loved.4

This, at any rate, is my sense of the matter. Not everyone will share my intuitions.Some will say that the realization that their beloved has been replaced by a qualitativelyidentical copy will annul any feelings of sadness: they would not expect to experiencethe slightest negative feeling. To back this up, they will stress that there is somethingtruly fishy about attaching weight to particularity in the way I am doing. They willcontend that, as far as love goes, nothing of importance happens when we replace thebeloved individual with an exact copy. Again, how realistic these people are is a matterof intuition. I am certainly inclined to think that any person in the circumstancesdescribed would feel some sadness, and that there can be cases of the sort of ‘resistant’love I have in mind. But suppose, as was once suggested to me, that by somemetaphysical oddity my beloved disappears in thin air, and a split second later thereappears on the same spot a being identical in all universal features with my beloved.Surely I am rationally required to love this new person in the same way as I loved theone who disappeared?

Again, let us set aside the fact that I would probably love the ‘new’ person—I woulddo so for the simple reason that whatever caused me to love the original individualwould cause me to love his or her replica. But this is one thing. The question I am nowaddressing is different. I am asking: Am I somehow irrational if I do not love thisperson—or, at least, do not love her in the same way?

Intuitively, we distinguish between properties that are internally and externallyrelated to an object. Suppose we consider our beloved only in terms of his or herinternal properties.5 Perhaps this will place too much strain on our imaginative capa-cities, but suppose it can be done. In that case, I venture to say that the whole idea ofdirecting an attitude on a that differs from the attitude you direct on b when a and bshare all internal properties becomes incomprehensible. If a and b are exactly alike intheir internal properties, and if you are not allowed to consider external properties atall, how could you feel differently about a and b?

Things look different if we include externally related features. Which ones tocontemplate is an open question. However, the most obvious are, I think, thoseconcerning the history of the person. I am not ready to replace my beloved with acopy, however similar in internal properties, for the simple reason that the copy willnot be the person with whom I have a shared past, to whom I owe obligations, havemade promises, and so forth.

The idea that our beliefs about people’s identity are in part based on externalproperties gives us occasion to further illuminate the notion of an identifier. So farthe identifiers that we have been speaking about have been assumed to be parts of the

4 Cf. Christopher Grau (2002).5 There is insufficient space in this work to discuss the complex nature of this distinction. Its fruitfulness in

application will, I hope, be displayed, though.


intentional content of the attitude. This picture needs now to be slightly modified.Although some relational properties seem discernable, obviously, this cannot be thecase with all features. In cases in which something was related to an object in the past,we are forced to say that the relevant property cannot be what we discern. This seemsincontestable. The implication, then, is that in some cases (I would even go as far as tosay, in most cases) the set of identifiers is not entirely made up of properties belongingto the content of the attitude.6

Now, to return to the earlier issue, one might be morally required to love thereplica, and it certainly would not make any sense to deny that one would love him orher if one were unaware of the switch. However, since the replica is not the person oneshared one’s past with, this suggests a way of explaining the fact that one would besaddened by the news of a switchover: the replica is not the one who, say, gave birth tomy daughters (she believes so, but she is mistaken).In fairness, by referring to the ‘new’ person as a replica I might be accused of evading

the issue. Why not imagine a possible world in which this person has all of thebeloved’s properties, internal and external? Then what?At this point it becomes hard to follow the example. But notice that the reason for

this is not that the notion of love outlined here rests on some metaphysical idea of a‘personal substratum’ (e.g. that to which a person’s features attach). Nothing I have saidso far saddles me with this commitment. (Keeping our metaphysics to a minimum, weshould allow for the possibility that we love persons even if we happen to believe thatthey are nothing but, say, bundles of properties or, if you prefer, tropes.) Of course, Ican imagine my beloved in a possible world, but I am not sure it makes sense first tothink of someone, and then to imagine this someone having his or her own properties(this we must assume) replaced by all of my beloved’s properties in some possibleworld. So I will not pursue questions about what is going on in this strange kind ofexample.

6.3 Good-for and welfareFAP* captures, I believe, something important about value-for. Does it also exhaustthis notion? Most probably, it does not. I have already noted in passing that we shouldnot expect there to be a perfect match between value and normativity. Some values inparticular do not perfectly fit the idea that values are nothing but reasons for attitudes.Beauty is perhaps the most arresting example. For instance, people who are beautiful(or ugly, for that matter) have been said to strike their observers with their looks.7

6 Should we allow these relational properties to be part of the attitude indirectly? Perhaps this makes sense.However, it does not affect the point being made here, i.e. that certain properties cannot be said to bediscerned.

7 Recently Kevin Mulligan has voiced this sort of objection to the buck-passing account; see his‘Emotion, Value, and Morality’ (2009b). For a similar point, see also Johan Brännmark, ‘Goodness, Values,Reasons’ (2009).

6.3 GOOD-FOR AND WELFARE 83

A particular kind of experience is involved in such cases, and it is not obvious that wehave experiences of reasons—at least, not in the sense that we are, for instance, takenaback by the beauty of a certain face. This discrepancy is worth examining, but to takesuch cases at face value and conclude that we actually experience sui generis valueproperties, and that a fitting-attitude pattern of value analysis therefore cannot be thewhole truth, would be unwarranted. There is ample evidence that we should notautomatically trust phenomenology of this sort to be veridical. Ideally, then, a buck-passer will provide a suggestion as to why we should not rely on phenomenology incases of ‘value experiences’. This is not something I will be engaged in here, though.Rather, I prefer to explore another possibility, and one that seems to me moreimportant—namely, that there might be more than one notion of value-for, andthat FAP accounts only for one of these.

This possibility certainly looks to be worth pursuing, especially in connection withpositive or negative expressions of value-for—that is to say, in connection with‘good-/bad-for’. Let me therefore say something about this matter here. I shall focuson the relation between good-for and welfare, although the approach I shall take tothis matter could equally well be applied to the many attempts to link good-for tointerest or benefit.8 I will follow up on this discussion in Chapter 7, where I will considersome examples of good-for analysis that appear to involve a notion of good-fordiffering from the one FAP* aims to clarify.

One way of conceiving of the relation between good-for and welfare is quiteconsistent with FAP*. I shall refer to this as the thin-thin synonym relation. Supposesome x is constitutive of, or is, the welfare-making base of a person a’s welfare. For thetime being we can leave out the details of x. It seems perfectly natural to say then that xis good for a (where ‘good for’ thus expresses a positive value-for). This seemsinoffensive because many of us would like, at least, to say that whatever is good for aperson is welfare-constitutive, and that whatever is welfare-constitutive is good for theperson. Effectively, we regard the connection as holding between two value concep-tions (or, if you prefer, between two evaluative notions). It is merely a verbal matter,on this view, whether we should use ‘good for’ or ‘welfare’ (or indeed ‘of benefit to’)in order to express these two value conceptions. So whatever we take to be good-formaking properties are also, analytically, welfare-making properties, and vice versa.But notice that in acknowledging this we have agreed neither that the notion of good-for is analytically related to a particular non-evaluative term, nor that welfare is soconnected.

I doubt that welfarists (who can be treated here as anyone who ascribes value-for towelfare) would analyse the relationship between good-for and welfare in the way justdescribed. More likely they would take welfare to be either a natural (descriptive)

8 E.g. see Zimmerman in ‘Understanding What’s Good for Us’ (2009), who takes “a benefit account ofgoodness-for to be correct; what is good for someone (in the relevant sense) is what benefits him and, all elsebeing equal, his life is good to the extent that it benefits him”.


notion, rather than a value notion, or a thick value notion. Later on I will have more tosay about especially the latter possibility. Here I want to introduce the thin-thinconception, and to outline yet another possibility, which is the one I favour myself.Welfarists (whether or not they are in a position to explain what they mean by

welfare in terms of natural, non-evaluative properties) need not be treating good-for/welfare in accordance with any of the approaches mentioned. They might be thinking,say, that welfare, understood as a certain mental state, is just an example of what is goodfor the agent. ‘Good-for’ and ‘welfare’ are not regarded as synonyms, since for onething the former can be applied to things other than welfare. But welfarists aregenerally not quite that permissive when it comes to what can be good for an agent.They might acknowledge that something other than mental states of the kind theyfocus upon can be good for the agent. However, such a good-for claim would then beregarded by this sort of welfarist as being about a different kind of good-for value. Itwould be an employment of a notion of good-for of a different kind to the one theyemploy when they think welfare is good for a person. Their good-for notion becomesnecessarily (i.e. analytically) connected with, for example, some mental state, such asfeeling pleasure, or happiness, that is regarded as being what welfare/good-for is allabout. Welfarists of this kind tend, in other words, to think that there is sufficientdescriptive content in the notion of welfare to allow one, on purely conceptualgrounds, to rule out a number of candidates as good-for/welfare constituents. So, forinstance, when Connie Rosati,9 as we shall see later on in Chapter 7, makes a sharpdistinction between ‘good for’ as a term referring to personal welfare and other sensesof this expression (e.g. in the assertion ‘water is good for the plant’), she has in mind anarrower notion than the one I am seeking to capture in FAP.A third notion also seems to be intimately related to welfare and good-for, namely

‘better off ’. Thus, we might say that something is good for a person if and only if itmakes him better off; and that something makes a person better off if and only if itincreases his welfare. Again, I am strongly inclined to think that if we are not merelyregarding it as a synonym of ‘good for’, ‘better off ’ will at some point or other have torefer to either an impersonal or personal value, in which case what I have said so far willalso apply to ‘better off’ approaches.I am not assuming that there are things which, in virtue of the meaning of ‘value for’

or ‘good for’, cannot exemplify these values. In my view, it is an evaluative (or, if youprefer, a normative) issue what things are good-for/welfare-constitutive. If we regardthe connection between ‘good for’ and ‘welfare’ as analytic in the thin-thin sense, noobstacles to FAP* arise. However, I suspect that for many observers this will be to strainthe meaning of ‘welfare’ too much. Be that as it may, the view that welfare must beunderstood in terms of natural properties, such as the enjoyment of certain mentalstates, and that good-for accrues only to those properties, squares badly with a fitting

9 Rosati, ‘Objectivism and Relational Good’ (2008).

6.3 GOOD-FOR AND WELFARE 85

attitude analysis which, like FAP*, involves a normative element. I remain uncon-vinced that there is a notion of good-for that FAP fails to capture. This notwithstand-ing, the important thing is that what FAP does denote is at least as interesting a value-kind as the other notions of good-for outlined here.

6.4 Two sorts of personal valueIn my view FAP* has a strong intuitive appeal as an analysis of personal and not merelyrelative value. The object x does not only have value relative to some person. It oughtto be favoured in a special way; it should be favoured for some person’s sake. To dosomething (in a general sense of ‘do’) for someone’s sake is to act with an eye to thisparticular person. A value understood in these terms is more appropriately described asa personal, rather than impersonal, value. Presently, I will point to some other positivefeatures of FAP, but in the meantime let me continue to comment a little more onwhat we have just discussed.

Suppose we take ‘welfare’ to have some kind of descriptive meaning. For example,we assume that a person is in a certain state of welfare if and only if she is in a certainmental state. A great many writers discussing welfare tend to agree that the presence ofsuch a state is at least a condition of welfare. Of course, they also disagree not only overthe precise nature of the mental state, but also over whether this condition is sufficient.Now, I have no problem with the claim that welfare is to be understood in terms ofmental states. However, if this is how we look upon welfare, it will be an entirelyevaluative issue whether we should ascribe value-for to these states. This is consistentwith the idea that many of the things we take to be welfare states (and, for that matter,states of well-being)10 are typical examples of something that carries personal values. If Iam right about this substantial claim, we might wish, for purely classificatory reasons, togroup together a range of what are typically considered bona fide welfare-makingproperties of a person a in virtue of which we can reasonably favour x for a’s sake(whether or not the pro-response is final or instrumental).The thought that a single, large group of ‘welfare’ states is part of the set of objects

carrying personal values seems intuitively plausible. It is, at any rate, very likely that thesupervenience bases to which we normally ascribe ‘welfare value’ are such that theysupply (or, at least, are believed to supply) us with reason for favouring them for theperson’s sake.

Personal values, I shall say, consist of two fundamental categories of value, one ofwhich contains those values-for whose realization is welfare-enhancing or welfare-preserving (i.e. such as to enhance or preserve the kinds of state that are typicallyregarded as constituting welfare). For the sake of illustration, I assume here that‘welfare’ is not a synonym of ‘good for’, but merely something that might provide

10 Recall that I do not take well-being and welfare to be the same.


one kind of supervenience base for good-for. We need not go into the precisenature of this something, but what I have in mind are objects that contribute tothe existence of certain feelings, beliefs, and attitudes of the person whose welfare isbeing affected.The second category of personal values contains a related, but nonetheless different,

kind of value-for. Objects carrying values of this other kind are what we have reason tofavour for a’s sake without it being necessarily the case that this favouring willsomehow contribute to a’s welfare, even if welfare is conceived in quite a broadsense. Despite the fact that a favouring of x would not have any welfare-contributingproperties, values belonging to this category would nonetheless exemplify somethingthat is good for a.We need to dwell a little longer on this last matter, since it might appear somewhat

counter-intuitive. Setting aside any substantial reservations we might have, the ques-tion we are addressing is this: Is it really possible to favour something for a’s sake that isnot in some minimal way welfare-promoting (in this narrower mental-state sense ofwelfare)? I think it is possible, and I shall argue as much in a moment. However, itmight be protested right away that favouring something for someone’s sake simply hasto be understood in terms of what makes the person better off (especially if we takefavouring to be equivalent to preferring) or what is in the person’s interest. Perhaps thisis correct. Again, it depends how we understand these terms. If we take ‘better off’ or‘interest’ here to be synonyms of ‘value-for’/‘personal value’ understood in terms ofFAP*, then, obviously, we should go along with the protest. But otherwise I think it isbest to be sceptical. It would surely strain the words ‘better off’ a little too much to saythat everything which we have a reason to favour for a person’s sake will make himbetter off.Indeed a moment of reflection should show that it is possible to favour something

for a’s sake that is not in the slightest bit welfare-promoting (again, in this narrowermental-state sense of welfare). First, we need to recall that not all attitudes have thesame kind of intentional object. Consider the attitude respect. Showing a person respectis not necessarily something one does in order to contribute to this person’s welfare. Itmight even have welfare-impairing effects. Respect has rather to do with what theperson is or represents, and not with what you hope to achieve for the sake of thisperson’s welfare.Once upon a time it was not uncommon for soldiers to respect their enemies. This

did not prevent them from wishing that the welfare of their enemies should come to anabrupt and most definitive end. Of course, many attitudes like this one are old-fashioned, and some are quite ridiculous. But this fact does not controvert the notionthat, conceptually speaking, we can do something for someone’s sake that does notnecessarily involve enhancing her welfare. Doing something for a’s sake must, howev-er, in some minimal sense, involve acting in favour of a—in the sense (to return to theexample) that being respected is regarded as positive, while being ignored or disre-spected counts against a person.

6.4 TWO SORTS OF PERSONAL VALUE 87

It might be argued that we are not respecting a for a’s sake at all. We are ratherrespecting him for his dignity’s sake (e.g. see Stephen Darwall, Welfare and RationalCare, 2002). Now, it is quite acceptable to say that we respect a for his dignity’s sake aslong as this is not taken to imply that the proper object of our attitude is dignity. Surely,what we respect is persons and not properties. It would be odd to argue the latter.

Respect belongs to a group of attitudes presenting the starkest counter-examples tothe idea that doing something for someone’s sake is necessarily doing it to preserve orpromote that person’s welfare. Other examples are: to admire, to honour, to esteem, tolike, and even to desire that something is the case. Even a response like desire, whichhas traditionally been intimately associated with welfare, need not always be under-stood as welfare-related. For instance, a person may live an immoral (or inauthentic, orirrational) life, and we might well desire that certain states of affairs obtain that wouldmake her life moral (or authentic, or rational) despite any negative impact this mighthave on her welfare. We would still be favouring the relevant states of affairs for thisperson’s sake.11 It would be possible to try to counter this point by analytically buildinginto the very notion of welfare something that blocks this eventuality, but this wouldjust land us with a rather implausible and uninteresting notion of welfare.

We still lack a full understanding of what the ‘for a’s sake’ idiom comes to. ButChapter 5 showed, I think, that we now have enough to go on to conclude that thereis at least a conceptual gap that the complex construction ‘O ought to be favoured fora’s sake’ fills—even when we are talking about attitudes such as respect.

6.5 Personal intrinsic valueLet me next bring out another advantage of FAP, namely that it does not necessarilymake personal value conditional on some de facto favouring. (We saw in Chapter 4that this was an implication of DEFA, i.e. the idea that some x has personal value forperson a if and only if we ought to favour x in virtue of the extrinsic property that x isde facto favoured by a.) An object of personal value is one that we ought to favour for aperson’s sake, and this might well be the case whether or not anyone actually favoursthe object for this person’s sake. This point is less counter-intuitive than it may appearto be. Suppose a did not know what her parents looked like; a knew them only byname. Furthermore, assume that there exists a photo of her parents, unknown to her.Such a photo could plausibly be said to carry a personal value for her whether or not

11 Here is another example. A person whom I once knew well lived a very self-destructive life. Weeventually lost contact with each other as time passed. Later on, when I learned she had passed away from theafter-effects of her destructive lifestyle, I immediately felt sadness. However, I also realized that my mourningof this person’s death had a different character from what I experienced when, for instance, my father died. Inthe latter case, I was sad for my own sake at having lost someone I cared about greatly. In the former, I cannotsincerely say that I actually cared much for this person. My sadness was rather of the following kind: I was sadfor her sake. Her disastrous life called forth a special feeling, sadness, which I experienced for her sake.


she knew about it. The point is not, however, that the value of the photo would beonly hypothetical—say, conditional on her cherishing the photo.Here is a different argument that suggests that we take it to be possible, at least, for

something to have value for a person despite the fact that he or she rejects the idea.Suppose you believe that a certain object x has personal value for your friend a.However, when you show a the item, she rejects the idea that it has this sort ofvalue: x has no value for her, she says. Suppose, further, that a cannot give you anexplanation that you are satisfied with as to why x is different from other objects of thesame kind that do have value for her. In such a case you may draw the conclusion thatyou were somehow mistaken about what has value for your friend. Alternatively, youmay think your friend is mistaken. Both possibilities are open. You may take the latteroption because you believe your friend is being inconsistent. But this is not the onlypossibility. Surely you may also think that she fails to see something in the item thatyou think you see—something that is such that you believe there is a reason for you tocare for the object for her sake. You expect that at some point awill see what you see inthe object, and then come to realize that x has in fact personal value for her.

Notice, too, that nothing of relevance changes if we add that the attitudes arehypothetical. For example, whether or not the photo has personal value is, on theanalysis outlined here, a question of whether there is reason to favour the photo for a’ssake, and the truth or validity of this normative statement is not dependent on the truthof a hypothetical statement. However, suppose a declares, when we draw her attentionto the photo, that she could not care less. Suppose further that this is actually true.There is no reason to deny this possibility: a might be absolutely indifferent to thephoto. Does it follow that the photo now has no value—of a personal or some otherkind? It might of course follow if we adhere to a meta-valuational theory, according towhich, say, values are constituted by attitudes (e.g. classical subjectivism, or a moresophisticated non-cognitivism). But theories apart, does it follow that the photo has novalue if a is indifferent to it? I do not think so!12

FAP does not assume the truth or falsity of value subjectivism or value objectivism.The deontic statement in the analysans might well be given a subjectivist or anobjectivist analysis. I fear that I have repeated this point one time too many, but thefact that FAP accommodates both perspectives is crucial for understanding that, on it, itis possible for personal value to be something other than extrinsic. Of course, to whatextent there are intrinsic final personal values is, as I have mentioned earlier, in part asubstantive issue. The really interesting question is whether there is a notion of intrinsicpersonal value. Is that notion really coherent? Given the way FAP is formulated it is notobvious that we can rule it out on purely formal grounds. To see why, consider tobegin with a new version of FAP:

12 Of course, this is a substantive question (which it would be a pity to rule out on formal grounds). Toanswer it, we would need to engage in substantive reasoning—something there is no room for here.

6.5 PERSONAL INTRINS IC VALUE 89

FAP-intrinsic: An object x’s value for a person a (i.e. x’s personal value) consists in theexistence of normative reasons for favouring/disfavouring x in virtue of its internalfeatures alone for a’s sake.

As can be seen, there are no straightforward obstacles here. That is, it seems to bepsychologically quite possible for a person to take into account only the internalfeatures of an object and favour it for someone’s sake. Suppose, for example, thatyou have a friend who collects small bronze sculptures. You find one of these at anauction. Your friend is unaware of this sculpture’s existence, and he cannot thereforebuy it himself. It seems that I could ascribe value-for to such a sculpture in virtue of itsinternal features alone. That is, I could take myself to have a reason to bid for it for myfriend’s sake.

My point is not, however, that this is necessarily the most plausible interpretation ofthe case. It might be objected that we cannot avoid somehow relating the object to thefriend. Perhaps this is true, but it is not enough to establish that these relation featuresturn the object into an extrinsically valuable object. The relational properties mightsimply not figure in the supervenience base of the object.

Though the example is psychologically plausible, it does not, however, establish thatthere are intrinsic personal values.13 The bronze sculpture case is obviously aboutsomeone who takes certain features to be reason-making, but it establishes thatsomething has been considered a normative reason, not that something is a normativereason. Hence the case does not supply a knock-down proof of the existence of a realnormative value. Notwithstanding this, the fact that we seem to recognize suchintrinsic personal values is not irrelevant. After all it underlines our openness to theidea of something being intrinsically valuable for some agent. That FAP can be seen asconsistent with it is therefore a great plus, I think.

But what if we take seriously the suggestion that all reasons are agent-relative? Surelywe must then conclude that there are no intrinsic values on the fitting attitude analysis?Perhaps. Certainly, I am inclined to think that this conclusion is correct. The fact thatI still can see some small openings for a notion of intrinsic value-for, even if it isaccepted that all reasons are agent-relative, is not something that I will go into here;currently my thoughts on this matter are not sufficiently worked out to be presentable.The important thing to bear in mind—and I would readily agree that it is easy to forgetthis—is that fitting attitude analyses do not, in general, analyse value in terms of specificsets of reason-making properties. The analysis—and this goes for FAP, too—says onlythis much: to be valuable is to be an object that we have (the right kind of ) reason tofavour. In other words, it is not an analysis of the reason-making properties of valuable objects.It is important to keep this in mind, especially in this context. Since the analysis leaves

13 Here I leave out other complications. It might be argued that the example is only about instrumental,rather than intrinsic, values. I would resist this, but to make this resistance really plausible I would need to(and here cannot) present a fuller description of the case. I think this could be done, though.

90 EXAMINING THE ANALYSIS

questions about the normative driving forces, and the precise nature (objective orsubjective) of this normativity, open, it also leaves the question whether there can beintrinsic final personal values unanswered. If normative reasons are facts involving theagent for whom they are reasons, it is hard, as I shall argue in Chapter 9, to understandthe claim that there can be intrinsic values. Reasons qua facts all appear to be agent-relative.I am inclined to think that reasons are indeed facts of the kind mentioned above.

Given this inclination, I think it is difficult to make room for the notion that there canbe personal intrinsic values. But I do not think that fitting-attitude analyses in generalanalyse value in terms of a specific ‘reason notion’. Rather, they understand value interms of a general notion of reason or normativity. I am sure many will see this as aweakness. For my own part, I think this is very much its strength.

6.6 Agent-relativity and private valuesFAP has inherited some serious problems from the fitting attitude pattern of analysis.Most notably, the WKR problem discussed in Chapter 3 is an issue that it would havebeen nice to have buried once and for all. But FAP faces several problems of its own.I am bound to overlook some of the difficulties, but in the remaining part of thischapter and the whole of Chapters 7 and 8, I shall attempt to improve our understand-ing of the analysis by considering a number of more or less tough responses to it.The first concerns a foreseeable reaction to FAP raising some difficult issues. The

idea is that FAP runs the risk of making personal values private. Two features of theanalysis appear to trigger this kind of Moorean reaction.14 One has to do with what wehave just discussed, namely the great likelihood that FAP is formulated in terms ofagent-relative reasons. Agent-relativity makes personal value too private: just as therecannot be private truths—propositions that are true for me, but not for anyone else—there cannot be private values. What is more, it makes even so-called impersonal valuestoo private, since they too are most likely to be understood in terms of agent-relativereasons.The second trigger has to do with the fact that personal value is being analysed in

terms of so-called FSS attitudes some of which take non-fungible persons as intentionalobjects (i.e. the persistence of these attitudes is dependent on the identity of the objecton which they are directed). This last feature seems to run into a problem witha generally acknowledged characteristic of value, namely universalizability. This sec-ond reaction is the more serious of the two. However, I will begin by consideringthe first.Personal values would indeed be strange if value-for/good-for was similar to ‘true

for’. We have difficulty making sense of ‘true for’, because we read it as referring to

14 See the discussion of G. E. Moore’s objection to ‘good for’ in Section 4.1.

6.6 AGENT-RELATIVITY AND PRIVATE VALUES 91

something true which, in addition, is only true for a particular person; and that justseems confused. But the value-for expression is not like this. It does not refer to value,period, and require us to understand that to be a value for a particular person. Value-foris a particular variety of value: a relational variety. The fact that this relational value isunderstood in terms of agent-relative reasons does not alter this picture; even if wewere to understand FAP* in terms of agent-relative reasons, this would not turnpersonal values into items which, by their nature, are inaccessible to persons otherthan the one whose values they are. As we have seen, it is to be anticipated that anobject may be good for person a despite the fact that a has no knowledge of that object.Personal values do not somehow become someone’s private property. The relativiza-tion of the reason (in the analysis) does not give the reason-holder ownership of it; nor,for conceptual reasons, is it impossible for anyone else to have a corresponding agent-relative reason. Naturally, this is often quite likely to be the case, as a matter of fact.

Still, these comments might not be seen as going to the heart of the issue. Recall theexample from earlier about the bronze sculpture that was for sale. Suppose I am at theauction, but that I do not know that my friend collects these artefacts. Here, since I amunaware of my friend’s interest in the sculptures, it seems that I cannot reasonably besaid to have a reason for, say, wanting to buy one for my friend’s sake. Must I nottherefore conclude that the personal value accruing to the bronze figure is ‘not therefor me’? Must not the conclusion be, therefore, that FAP makes value too private?

This is a strange line of reasoning. The scenario described does not show that,because I am unaware of the facts of the case, the reason does not apply to me orsomeone else. A fact can constitute an agent-relative reason regardless of whether youare aware of it—or so I will assume. Moreover, we should not forget that agent-relativereasons are probably universalizable (see the next section, however). In the case justdescribed, the facts prevent me from knowing that I have a reason in this case. Theconclusion to draw, then, is therefore that the facts are such that I am unable toappreciate that a value for the friend accrues to the bronze statuette. There is nothingpeculiar about this, nothing that makes values into private entities.

6.7 Personal value and universalizabilityLet us next return to the second, and tangibly more complex, matter. This concerns theextent to which so-called FSS attitudes can figure in an analysis of value. A possibleresponse to such a view would set out, I believe, from the following commonly madeclaim: if two objects differ in value, theymust differ as to some universal feature—where auniversal feature is one the description of which does not necessitate mention of anindividual. This idea is, then, bolstered by the claim that value judgements are universaliz-able: if we say that an object a is valuable, we are committed to saying, of any other objectwith the same universal features as ahas, that it too is valuable. If, then, universalizability is arequirement on all judgements of value, as Richard Hare has so painstakingly argued over


the years,15 then an analysis like FAP might appear to run into problems here. However,notice to begin with that these alleged problems have nothing to do with the fact that thenotion of reason employed in FAP is likely to be agent-relative. As I have just pointed out,it is generally recognized that such agent-relative reasons are as universalizable as theallegedly neutral ones. The problem has a different origin. It originates in the peculiarnature of the FSS attitude. Specifically, in the present context, it originates in the peculiarnature of IdentityFSS attitudes. Asmaybe recalled, these takenon-fungible persons in theirintentional content, and this does seem to prevent them from being employed in analysisof a universalizable notion of value.16 That is, suppose x carries person a’s personal value.On FAP*, this means there is a reason to favour x for a’s sake, i.e. there is a reason to directan FSS attitude on x. Suppose, too, that the value is in fact an extrinsic personal value andis such that there is an a-relational property among the reason-making properties of x.This implies that another object (if there is one)which is exactly similar to x in its universalfeatures might still not carry a value-for. The fact that it did not stand in this relation to awas, so to speak, a sufficient reason for not ascribing value-for to it.But obviously, to reject the implication that if x has value, any exact copy as regards

universal featuresmust also have valuewill not sit well with an advocate of universalizability.A quite obvious reply waits around the corner at this point.17 Since value-for is a

relative notion the comparison base will have to include, albeit in universalizable terms,not only the valuable object, but also the person for whom it is a value-for. Thus,suppose I ascribe value-for to an object x with an eye to a. I am now committed, itseems, to ascribing value-for to any other object y with an eye to a* if x and y, and aand a*, share exactly the same universal features. And this does seem to be somethingthat many would agree with. If y and a* stand in the same relation, and are identical, inuniversal respects, with x and a, then it would not make sense, it seems, to deny that yhas value for a*. That is, it would not make sense if you were a believer in universaliz-ability. And perhaps someone with an FSS attitude need not reject this possibility. Butsuppose she did? She might not even deny that y has value for a*; she might confineherself to claiming that she does not take a stand on this issue, and that no logic in theworld can force her to say that y has value for a*.Now, universalizability is a very strong requirement, and it might therefore be

argued that it would be far from devastating to discover that FAP is inconsistent withit. It is simply too strong a demand; value analysis can manage with something lessdemanding: supervenience.

15 For some problems with Hare’s views, see my Logic, Facts, and Representation: An examination of R. M.Hare’s Moral Philosophy (1993).

16 The problem might seem to arise only for those FSS attitudes that have identifiers in their intentionalcontent. However, unfortunately this is not the case. The invocation of identity in any sort of justification(whether the identifiers are internal or external to the attitude) will run into problems with universalizability.

17 This reply is not available to so-called Mooreans, like Donald H. Regan (2004) and Thomas Hurka(1987), who reject the idea of a relativized notion ‘value for’ that can be defined independently of value(period).

6.7 PERSONAL VALUE AND UNIVERSALIZABIL ITY 93

Although universalizability and supervenience are intimately related, they must not beconfused with each other. The latter, in contrast with universalizability, is compatiblewith unicity, i.e. there being only one object exemplifying a given set of properties.Supervenience requires us to ascribe value to any object y that has the relevant propertiesof x, and the relevant properties in question may well pick out a single, unique individualsuch as the being who shared the past with me, gave birth to my children, and so on. So ifI ascribe to x value-for having in mind the person who shared the past with me (and soon), value-for is uniquely determined by these external relational properties.

One option, at this juncture, is therefore to throw universalizability overboard, and tostick with the idea that values are supervenient features. I do see this as a viable option.My problem is that I am not altogether sure whether universalizability is too strong arequirement. At least, it mightwell appear that a rather low price attaches to its acceptanceonce supervenience is agreed to. Surely it cannot be all that difficult to acknowledgethat any other object y which has the relevant subjacent features of x is also valuable, i.e.ought, also, to be an object of an FSS attitude directed on a*? True, even a firm believerin non-fungibility attitudes, if you will pardon the description, ought to agree withthis. But the real crux of the matter then becomes whether the next step follows too:does it follow that the person who directed an FSS attitude on x having a in mind mustalso direct one ony with an eye to a*? Here I feel I have no good argument; that is, I donot think this step is motivated by the logical features of our value notions (alone). Nor doI think rationality (consistency) obliges one to take this step. I am therefore unableto provide an argument to show that the rejection of universalizability is a mistake.

The above issue also has a clear bearing on something discussed in Chapter 5. I amthinking here of the distinction, made there, between two sorts of attitude-justification:identity- and property-justification. I suggested that identity-justification might not, afterall, really be a sort of justification. If you have strong intuitions in favour of universaliz-ability, quite obviously, youwill notwant to allow the identity of a person to be among anobject’s value-making properties, and so you will eventually reject the idea that identity-justification is a genuine variety of justification. However, if you think invocation of thesupervenience base justifies the claim that an object is valuable or explains whether thereis reason to favour the object, you will probably agree that identity-justification is, afterall, a bona fide form of justification. I am inclined to subscribe to the latter view.

For-someone’s-sake favourings constitute a loose-knit set of very different kinds ofattitude. In this chapter I have mainly focused on non-universal varieties, i.e. those thattake non-fungible persons as their intentional objects. These are particularly interestingfor two reasons. First, they are philosophically more problematic inasmuch as it is notobvious how to make room for such attitudes in a value analysis. However, as I argued,although it may be difficult to analyse value simpliciter in terms of these attitudes, we caninclude them in an analysis of value-for. Second, non-universal attitudes deserve ourattention because they appear to have a great impact on our lives. It is reasonable tolook upon them as the kind of attitude that matters more to us: most of us would divertconsiderable resources to avoiding their frustration or accomplishing their satisfaction.


7

Mo(o)re Objections

So far my main concern has been to pave the way for what I believe is an interestingapproach to understanding value-for. The time has now come to have a look at somemore or less obvious objections to, and enquiries about, my approach. There are anumber of reasons why one might be sceptical about this sort of value, some sounderthan others. I will begin with a readily entertained suspicion. Given the long traditionof value theorists who have dedicated their work exclusively to the notion thatsomething is good, period, it seems quite natural to suspect that we do not need toadd any ‘personal values’ to our value typology; we can take care of alleged cases ofpersonal value within extant views. Obviously, such a line of reasoning is bound tocome in an array of forms. I will confine myself to commenting on only two of these.They have in common that they do not purport to show that we have to throw ‘goodfor’ overboard: this expression can remain in the lexicon of value. However, they arguethat it would be fundamentally confused to think that ‘good for’ is an independentevaluative notion—that is to say, a notion capable of being analysed independently of‘good’. To believe this is to fall into serious confusion.But before we proceed, it should be stressed that the distinction between good,

period, and good-for (and value, period, and value-for) raises questions that I have hadto set aside. Perhaps the most important of these centres on an issue for those who claimthere is only one kind of fundamental positive value notion, i.e. only one kind ofgoodness. The issue is then whether this value is goodness simpliciter, or goodness-for?Historically important as this discussion is, I have nonetheless chosen to disregard it. Inmy view it is basically a misconstrued debate between objectivists and subjectivistsabout value;1 since good-for is not in any really interesting sense a more subjectivist

1 The idea that there is not one but two independent notions of goodness has been a major topic in moralphilosophy. Henry Sidgwick, in hisMethods of Ethics (1907 [1874], pp. 120–1, 403–5), famously claimed thatthe mark of modern ethics, as opposed to ancient Greek philosophy, was precisely the realization thatpractical reasoning faces a choice; it can help us determine what we should do to secure either maximumperson-relative goodness or maximum impersonal goodness. Much of this historical discussion is, in my view,closely tied to the question ‘What is the precise relation between value and people and their attitudes?’

notion than good period, I find this discussion less interesting. A more interestingposition rejects both these “monist” views in favour of a view embracing goodsimpliciter and good-for; both are perfectly meaningful value notions, and neithertherefore should be dismissed. Since I do not think we should give up on the notionof good period, I will be concerned chiefly with arguments against good-for.

7.1 Hurka’s views on good forVarious philosophers have, over the years, joined ranks with G. E. Moore in attackingthe view that good-for is a normative notion on its own, one that is not eventuallyunderstandable in terms of the notion that something is good, period. In particular,Thomas Hurka (1987), and more recently Donald H. Regan (2004), have launchedobjections to certain ways of understanding good-for to which I suspect Moore wouldhave consented. Their criticisms are in many ways quite understandable, and in factI share many of their misgivings. But, like Moore, they have shown only that there areways of understanding good-for that are confused. An analysis like the one I propose is,as far as I can see, untouched by their attacks. Obviously, one reason for this is that thereis chance, as mentioned earlier, that what I am analysing has little in common with theconcept that Hurka and Regan attack (and Connie Rosati defends: see below). Thismight be the case. I shall return to this matter in due course. However, since theexamples these authors discuss all appear to involve what are personal values in mysense, it will be wise to see whether their objections have any real bearing on what I amproposing.2

Hurka identifies four meanings of ‘good for’, three of which he thinks we should getrid of for the sake of clear thinking. These are rehearsed below; only (iii), whichcaptures what might be properly referred to as the Moorean sense of ‘good for’, meetswith his approval:

(i) “ . . . ‘good for’ means satisfies the desires of, or furthers the interests of . . . [or]answers (somehow) to the subjective states of ” (p. 72);

(ii) ‘Good for’ can sometimes be used to mean what a person believes good;(iii) “‘Good for’ can also be used to mean good from the point of view of ”;(iv) “‘Good for’ can also mean that portion of the good (from whatever point of

view) that falls within a person’s own life”.3

Hurka’s objections to (i) and (ii) seem sound to me. Meaning (i) fails, in his view,because the defining claims are descriptive, and we cannot account for the normativity

2 In other words, the assessment of alternative views on good-for from my perspective is not asstraightforward as it might initially seem. Recall what I said at the outset of Chapter 6: that there are waysof understanding good-for which, in comparison with the idea I am analysing, involve a narrower notion.

3 Hurka, ‘“Good” and “Good for”’ (1987). Hurka’s main concern is to attack the position that “Good[ . . . ] is always good for someone” (p. 71). This is not a position that I defend.

96 MO(O)RE OBJECTIONS

of ‘good’ with these senses. Of course, if you want to be picky, the strength of Hurka’spoint is not that clear-cut. Why should an advocate of (i) feel vulnerable because thedefiniens in (i) does not live up to the normativity of another notion that it does not tryto define, namely ‘good’? It is not ‘good’ that is at issue here, but ‘good for’. It is notobvious how Hurka would get around this, since he does not want to acknowledgeany independent good-for notion. However, once we accept that not only ‘good’ but‘good for’ is normative—which I think we should—(i) fails to account for thisnormativity, since it reduces good-for to a mere psychological state. It should bestressed that Hurka’s own (1987) account of impersonal good places him amongbuck-passers: ‘good’ is reduced to what we ought to morally desire and pursue.4 Sohis claim that the senses given in (i) are insufficiently normative is quite understandable.The meaning described in (ii) is, I believe, fairly common: what is good for a is what

a believes to be good. Whether or not we find this sort of relativism appealing or plainappalling, it clearly has nothing to do with what FAP is intended to disclose, so it neednot detain us. It is a sense that acknowledges only one kind of goodness, namely ‘good’,not the relational ‘good for’. That is, it understands the expression ‘good for’ in termsof what is good, period, for some person.As to (iii), if we are equating ‘good for’ with ‘good from the point of view of ’, we

should probably, for the sake of clarity, be speaking about ‘good from a point of view’to begin with. Hurka reduces this latter expression to what we ought to desire orpursue. In other words, sometimes when we are speaking about good-for we haveimpersonal goodness in mind.Finally, Hurka has a similar reply to (iv)—the notion that ‘good for’ means that

“portion of the good (from whatever point of view) that falls within a person’s ownlife”. If this is what we mean, there is no need to allow that we have an independentnotion of good-for. It would make for greater clarity if we were to speak instead ofwhat is “good in her life”, where ‘good’, here, is understood as impersonal good.Hurka does not have much to say about this latter sense; but he is not alone in

thinking along these lines. Regan has more recently argued that this latter sense is theonly sensible thing we can have in mind when we speak about something being goodfor a person. Michael J. Zimmerman (2009) calls this kind of sense “an ownership accountof goodness-for”. He thinks there are two versions of it—“‘x is good for P’ = def. x isgood and belongs to P” and “‘x is good for P’ = def. x’s belonging to P is good”—neither of which he thinks corresponds to what people mean today by good-for.Regan defends the former version, so here I will only be concerned with this sense.I will return to Zimmerman’s own favoured sense shortly. However, let us firstconsider Regan’s approach.

4 Hurka recently expressed during a visit to Lund (personal communication) doubts about the buck-passing account, and so I want to emphasize that what I discuss here are his views from (1987).

7.1 HURKA’S V IEWS ON GOOD FOR 97

7.2 Regan’s views on good forRegan (2004) contrasts good-for with good simpliciter. That something is good simpli-citer gives us an agent-neutral reason to promote it, according to Regan. Regan alsoassumes that good-for must be understood as a source of agent-relative reasons. (Noticeby the way that this is not necessarily a fitting-attitude analysis; a number of theoriesmight accord reason-making status to moral features.) In his view good-for combinestwo inconsistent ideas, one about goodness and another about some inexplicable ‘for’;and the first of these does not capture anything of interest over and above Moore’s ideaof ‘good possessed by a person’. As might be recalled (see Chapter 4, note 10), Reganrefers to this Moorean notion as a “good occurring in a person’s life” (cf. Hurka, 1987).He also proposes an argument against the usefulness of good-for—if it is taken to beindependent of this Moorean good. By lack of ‘usefulness’ he seems to have in mindthe fact that such an independent notion can make no claim on persons other than theindividual whose good it is. Once this worry was my own,5 and so I am sympathetic toit. An analysis of good-for must avoid turning this kind of value into something entirelyprivate—a goodness that is only available to (or ‘there’ for) the agent whose good it is.6

Why cannot good-for be a genuinely independent relational goodness? Regan’sidea is that at least one of the incompatible claims made by the good-for advocate has tobe jettisoned: suppose, for instance, that the good-for theorist agreed with the Moor-eans that ‘good for’ only denotes ‘good in the life of’. The advantage derived fromaccepting this, according to Regan, would be that the good-for theorist could nowclaim that what is good for a person gives us an agent-neutral reason. If some x is goodfor a in the sense that it is a “good in the life of a” anyone has a reason to care about andpromote x for a.

Of course, to accept this would be to acknowledge that ‘being good for’ is no longera distinct normative property. However, on the other horn of the dilemma, the good-for theorist could be obliged to give up the claim that she is, in effect, talking about a‘good occurring in the life’ of the person whose goodness it is. But then the problem,according to Regan, is that good-for is no longer a normative property for everyone.

Regan’s latter worry is exaggerated, though. It would indeed be problematic if‘good for’ came out as a property which, in its nature, only carried normative force fora single individual; and certainly if good-for was somehow rigged to a particular personnot in virtue of, say, the circumstances obtaining, but because the relevant goodnesswas in some way uniquely someone’s value property, I would join Regan in thinkingthat such an entity would be difficult to comprehend. But FAP does not give rise tosuch worries. As a fitting attitude analysis it understands goodness in terms of norma-tivity. This fact also soothes the privacy worry. The fact that something that is good for

5 See Rønnow-Rasmussen (2007a; 2008a).6 But notice: the fact that something that is good for me requires me to be mentioned in its supervenience

base does not necessarily make the value private; since FAP is formulated in terms of reasons that apply tomore than the person whose good-for is in question, the privacy argument has no bearing here.


me probably requires me to be mentioned in its supervenience base does not necessar-ily make the value private; since FAP is formulated in terms of the existence of a reasonwhich, in the circumstances, may, and probably will, apply to people other than theperson whose good-for it is (most reasons being, like this one, such as to apply to manypersons), the privacy argument has no bearing.7

In fairness, Regan, whose position renders problematic a number of views onwelfare/well-being, does not consider the kind of view outlined in this work. Hedoes consider Michael Smith’s theory on relational good, and he gives reasons forsetting it aside that would probably also apply to my approach. I will return to thesereasons in a moment. Meanwhile, I wish to draw attention to what Regan thinksought to replace the ‘good for’ idiom (at least if good-for is understood in terms ofwelfare). This is relevant to my analysis regardless of how closely the good-for Regan isattacking coincides with my analysandum.From my point of view, the really interesting feature of Regan’s account is actually

something different—namely, whether the job done by ‘good occurring in the life of aperson’ is what we take ‘good for’ to be doing. According to Regan, the former,Moorean notion is preferable in that it suggests a way of understanding the (in his view)problematic ‘for’ in ‘good for’: “If ‘for’ means ‘occurring in the life of’ this gives us anempirical relativization to the agent, but the normativity involved is still the universalnormativity of ‘good’”. But, again, since normativity is built into the very notion ofgood-for, Regan’s worry here need not be shared by the fitting-attitude analyst.8

7.3 Rosati and good occurring in a lifeRosati has recently questioned Regan’s suggestion that we should replace ‘good for’with ‘good occurring in a life’, or GOL. I support her conclusions, albeit, I suspect, forsomewhat different reasons. She is, I think, quite right to maintain that GOL isproblematic in its own terms, and that none of the more reasonable ways of spellingout what ‘occurring in a life’ refers to satisfactorily replaces the notion good-for.Rosati begins by asking whether GOL should be understood as stating the following

(where ‘TP’ refers to time and place):

GOLTP: “Good occurs in the life of P when good occurs in the time and place inwhich P lives.” (2008, p. 334)

7 See also Zimmerman (2009), who suggests we can dispose of the dilemma posed by Regan by drawing adistinction between prudential reasons and moral reasons.

8 Regan seems to be saying that if the normativity of good-for is not relativized, good-for coincides withgood. Rosati questions this, claiming that the normativity of good-for need not be relativized. It is stillpossible to argue that good-for is distinguishable from good simpliciter. Rosati’s point is that there are manythings that are universally normative (e.g. rights, rationality) but which nonetheless differ from good (see2008, p. 329). In Rønnow-Rasmussen (2007a) I maintain something to this effect. However, if normativereasons (at least, in an interesting sense) are all agent-relative, as I shall argue in Chapter 9, Regan has no pointin the first place.

7.3 ROSATI AND GOOD OCCURRING IN A LIFE 99

The occurrence of goodness is, in other words, understood as a spatio-temporalperson-related event. I suspect Mooreans would not be happy with this proposal.However, since I am uncertain how it might be improved, I will treat it as a candidate.

According to Rosati, GOLTP may well lead to some rather counter-intuitivesuggestions about what we ought to do:

If good occurring in a life required only spatiotemporal coincidence, then individuals could sleepthrough lives in which a good deal of good occurs and this would have peculiar results for whatwe ought to do. We ought, other things equal, to redecorate meticulously the rooms of thepermanently comatose, pipe beautiful music into their rooms, send in the clowns. To be sure, wewill promote more good by expending our energies elsewhere. But the suggestion that we couldhave any reason to promote good occurring in the lives of the permanently comatose, at least inthis sense, is dubious at best. If this were all the Moorean had in mind, then I think we couldreasonably doubt that good occurring in a life makes a claim on just as much as we can doubt thatthe self flagellator’s Sumnerian good for makes a claim on us. Good occurring in the life of P, inthe bare sense sketched by TP, does not capture what the good-for theorist has in mind. (2008,pp. 334, 335)

Rosati might find the idea of a good occurring in a comatose person’s life dubious forreasons other than my own; I am not quite sure. Her idea seems to be that no point canbe found here—nothing can come from making goodness occur in the life of someonewho cannot appreciate it. Perhaps the idea is rather that the alleged values mentioned,like beauty, are such that they cannot play any role in the life of a comatose person, andhence that adding such value cannot plausibly be said to be good for the person. If thecomatose individual cannot use her senses, what is the point in adding value objectsthat require to be sensed for their appreciation.

This reasoning makes sense on many plausible substantive accounts of value, but Icannot be certain Rosati subscribes to it.9 I am also inclined to think that a defender ofGOLTP need not be that worried about it; but that is because I think the notion ofgood-for is consistent with the idea that the promotion of what is good for a personneed not have any impact on the person. Needless to say, this is a contestable claim,but, as I have mentioned, fundamentally the issue is an evaluative one, and not a matterthat can be settled with the flick of a conceptual lever. This notwithstanding, I am notimpressed by GOLTP, and I share Rosati’s conclusion that it cannot replace good-for.Let me explain.

Imagine two separate, beautiful rooms filled with the same beautiful music, eachcontaining a comatose patient. (I am afraid many readers will think this example is toostrained. How can a silent room have any value for a comatose person? This is certainlya reasonable question. Perhaps the correct reply is that it cannot, the reason being thatcomatose people cannot have personal values—or cannot, at least, have personal values

9 Is she perhaps thinking that nothing can be good for a comatose individual? Rosati might be read asclaiming, for instance, that only conscious people can have a good-for; but it is clear, I think, as we shall see ina moment, that this is not what she believes.


of the kind accruing to objects detectable through the senses. However, it is alsopossible that none of the objects that carry value are (in some way) capable of sensorydetection. Respecting a dead or comatose person’s wish might be like this. So althoughit is somewhat strained, I think the example serves a purpose.)It seems to me fairly obvious that we need a notion of good-for that allows us

sensibly to claim that we should leave patient a in his room, but let b move to asilent room, since she has all her life hated music. Silence has a value for patient b.It might be contended that Mooreans like Regan can acknowledge this. They cansay that we should move b because she does not appreciate certain values. For sure,this is certainly a possibility. Here we must proceed carefully, though. We must notoverlook something fairly obvious. We might justify moving b out of the room byappealing to the fact the b had certain desires in the past. However, whether or notwe believe that this involves, at some level, taking an evaluative stand, I think weshould (also) be able to evaluatively motivate our decision to move b. To do this,however, at least as far I can see, we require an agent-relative conception ofgoodness/badness. That is, we need to be able to appeal to something more thanthe non-evaluative fact I take ‘b as a matter of fact did not appreciate certain values’to express. We must appeal to what is good for or bad for the patient; not what isgood or bad according to the agent. Again, I think we should be able to say thatremaining in her room is not good for b, or that the sound of music in the roomhas a negative value for b. I would move her or stop the music, and I would do sofor her sake.Notice, too, that nothing here hinges on the substantive position we take on

what is of final value. In the example it is beauty, but we could pick more or lessanything that is good, period. It still makes sense that we should be able todifferentiate the two agents’ presence in these rooms, and to do so with an eyeto what we should do for b’s sake (rather than a’s). That is, we need to be able toevaluatively express the relationship between an agent and what is good, period.Good-for, as handled in FAP, is in my view a serious shot at this. The invocationof merely impersonal values (i.e. of good simpliciter) would not explain why weshould either find b a new room or turn off the music in the room b is currently in.How could it? After all, non-relational values are the same for everyone, so theycannot be employed in a case like this.A possible rejoinder at this point is to suggest there is negative value in b’s room that

is not present in a’s room, and that is why we should move b but not a. But suppose weimagine both patients in the same room. Would we not be obliged to say, in such acase, and given GOLTP, that this negative value is present in a’s life as well as in b’s? Itseems so. But in that case nothing is really gained by introducing a negative value. Wewould still need, I believe, to find a value that is particularly related to b in order tojustify treating b differently from a.Ultimately I am not quite sure what to say about the coma case. It seems quite clear

that Mooreans cannot reasonably have GOLTP in mind when they think of a good


occurring in someone’s life; it simply does not sound like a convincing account of whatmight be understood by that phrase. Before moving on, let me therefore voice apossibility that should not be readily dismissed, namely that we understand ‘goodoccurring in a life’ in terms of FAP*.

Perhaps ‘good’ and ‘good occurring in the life of someone’ are two quite differentnotions of value (such that ‘good occurring in a life’ is a variant of Moorean goodsimpliciter). ‘Good’ would refer to a monadic property, while ‘good-occurring in thelife of p’ would refer to a dyadic property. So far we have merely assumed that thenotion of goodness picked out by the expression in ‘good occurring in a life’ is the samenotion as Moorean good. But if they are different we need a better way of under-standing the former notion than that provided by GOLTP. The sort of analysis I amoutlining in this book might, in effect, be an interesting way of understanding ‘goodoccurring in a life’. An adjusted FAP* might do the necessary work here. The ideawould be to look for reasons for favouring something for the person’s sake—facts thathave something to do with the person’s (whole) life. It would take some time to set outsuch an approach in full, and I will not try my hand at this here. In fact the case wouldclearly be a special case, subsumable under FAP*, and I prefer to focus on the moregeneral pattern. The idea sounds worth pursuing, though.

Interestingly, Sidgwick in The Methods of Ethics (1907 [1874]) put forward a fitting-attitude (FA) analysis of “my own good” and good-for that Moore might well haveaccepted. Hurka points this out in “Moore in the Middle”:10

By “my own good” and sometimes “good for me” Sidgwick meant what I ought to desire“assuming my own existence alone to be considered”. This should not have been objectionabletoMoore: it defines my own good as that portion of the good, however that is understood, that islocated in me. (2003, p. 611)

Sidgwick’s proposal is interesting. The problem is that it is not obvious how one shouldunderstand “assuming my own existence alone to be considered”. But perhaps wecould take it as implying that something is good for me, if and only if it is what I oughtto desire for my own sake, in which case FAP might not have been objectionable toSidgwick.

Rosati considers another interpretation of GOL. We can label this GOLTPC:

Perhaps we should say . . . that good occurs in the life of P when good occurs in the time andplace in which P lives and P is conscious of the occurrence. Call this characterization of goodoccurring in the life of the time-place-consciousness characterization, or ‘TPC’. (2008, p. 336)

Rosati also dismisses this suggestion. Her reason is interesting:

If all that good-for talk were about was good occurring in the life of as prescribed by TPC, good-for theorists who do not accept a consciousness requirement would not be talking about

10 Hurka, ‘Moore in the Middle’ (2003).


good for, despite what they may think, These good-for theorists could be mistaken, but I do notthink their position should be ruled out prior to our efforts to advance and assess analyses of goodfor P. (2008, pp. 336, 337)

Rosati has a point here. The discussion between mental-state theorists of good-for andtheir opponents is, I would say, a substantive discussion; it concerns what ‘good for’judgements we ought (and ought not) to make.Before moving on to problems concerning Rosati’s own position, I want briefly to

comment on Regan’s reason for setting Smith’s views11 on relational good aside in hisdiscussion of good-for theorists.12 Regan, who thinks good is the pleasurable experi-ence of objectively appropriate activities, considers different ways of understandinggood-for that purport to retain its normativity. Like Hurka, he dismisses variousnaturalist positions, all of which, he thinks, fail to secure the necessary normativity.However, considering the example ‘good for Brown’, he believes another approach ispossible:

. . . we might take it to mean just ‘what Brown ought to promote’. The most interestingdevelopment of this idea is Michael Smith’s, although Smith sensibly prefers not to refer to hisconcept as a version of ‘good for’ but rather as a subscripted ‘goodagent’. Interesting though thispossibility is, it is clearly remote from any concept of welfare. The goodBrown in this sense mayinclude any number of states of affairs outside Brown’s life; and furthermore there is noplausibility at all in suggesting that the goodBrown as such makes any claim on Jones. It may be,of course, that the goodBrown and the goodJones overlap, or are even identical in extension, but itwill not be the goodnessBrown, the fact that Brown should promote it, that accounts foranything’s being goodJones. So we can ignore this subscripted good. ‘Good for’, as used in thisessay, is a normative concept, and a synonym for the other normative concepts of ‘welfare’,‘well-being’, and ‘good of’. (2004, p. 208)

It is a pity Regan does not explain in greater detail why Smith’s subscripted goodnessshould be ignored in discussion of good-for and welfare. First, it does not strike me asnecessarily sensible to disassociate this sense from good-for. Surely it is possible that thissubscripted sense of ‘good’ is just what welfare theorists apply to welfare, or, to be morespecific, what they apply to whatever it is they take to be constitutive of personalwelfare; and quite why they do not apply it to something else is in part a substantive

11 Smith, ‘Neutral and Relative Value after Moore’ (2003).12 As Rosati points out, in a footnote Regan seeks to clarify his view: “It may seem that by restricting the

good to pleasurable experiences, the Moorean reveals that his real concern, like the ‘good for’ theorist’s, iswith value for the subjects of these experiences. Why else should an experience of a beautiful sunset, or a greatmathematical theorem, or a Bach cantata, need to be enjoyable in order to be valuable? I cannot answer thepoint fully here. The short answer is that what is really valuable (non-relatively) is the appreciativeengagement of the subject with a worthy object.” A few lines later Regan adds: “To my mind, whenthere is the right sort of engagement, we could as well say that the value created is value ‘for’ the sunset, or thetheorem, or the cantata, as insist that the value is ‘for’ the subject” (2004, p. 221). Now, this is an interestingpoint, but it is not necessarily problematic for the advocate of FAP. Fundamentally there is nothing in thekind of attitude required by FAP that necessarily prevents us from having such attitudes (mutatis mutandis) toobjects, such as sunsets and theorems, as opposed to people.


issue. The fact that goodagent may cover a “number of states of affairs outside Brown’slife” (and, I would add, concrete things) is not a reason for saying that it is a differentconcept from the one good-for theorists talk about. It might be—I have admitted this.However, it just might be that people are (merely) evaluatively disagreeing about whatcarries goodness-for. (Consider a slightly different case, one involving value, period.The fact that some people want to say that this kind of value only attaches to pleasuredoes not necessarily mean they cannot disagree with a preferentialist about what isvaluable.)13

7.4 Personal and impersonal valueLet me now say something about Rosati’s views. She has in recent years producedseveral perceptive papers on good-for. As she notices herself, my notion of personalvalue differs from her view of personal good or welfare.14 In fact I suspect ourperspective on good-for is quite different. I suspect, for instance, that what I take tobe the desiderata of an analysis of personal good/value are not what she has in mind,and vice versa. But I will not pursue this comparison here; instead I refer the reader toRosati’s own illuminating work. A couple of differences should be underlined, though,since this will help to highlight certain features of FAP.

Here is how Rosati describes certain key features of good-for:

Good-for value is relational, not relativist value. It is objective in that it gives anyone a reason foraction, though the good for might . . . be otherwise subjective; perhaps the normative relationgood for itself involves a mental-state requirement, or perhaps the relational property being goodfor P is instantiated only by items that produce particular mental states. Finally, good for is a formof extrinsic rather than intrinsic value. (2008, pp. 344–5)

Despite certain terminological discrepancies, I find much to agree with here. Perhapsmost importantly, I think Rosati is right in keeping a mental state requirement out ofthe analysis. I have some queries too, though. As I suggest in Chapter 9, I am not at allsure that it really makes sense to say that good-for gives anyone a reason for action.I have also voiced my suspicion that good-for may be an intrinsic value. Now, it mustbe said that by ‘extrinsic’ Rosati has something in mind other than the meaning I havewhen I use the term. She relies on Christine Korsgaard’s distinction between two wayssomething can have value—a distinction we considered in Sections 1.5 and 1.9.‘Extrinsic’, then, is to be understood in terms of where the source of the value islocated. Perhaps it is with regard to this matter that our views more obviously deviate.On Rosati’s account, good-for is a derivative value. Thus something’s being good-forderives from “its relation to a being with value” (2008, p. 343), and more specifically

13 Regan’s worry, expressed at the end of the passage quoted, about goodagent not being sufficientlynormative need not concern an advocate of FAP (not even if reasons are agent-relative).

14 See Rosati (2008, p. 318).


the being whose good-for it is. This means that if a being has no value, nothing canbe good for it (or, alternatively, nothing can be considered good for it). Rosati alsothinks this has a bearing on the sorts of value judgement we can make:

The property of being good for a person . . . well matches the normative role of judgmentsabout what is good for a person. Whether from the first-person or third-person standpoint,our focus in making these judgments is on the person herself and what preserves and advancesher, presupposing her value and attending to the particulars of her nature and circumstances.(2008, p. 346)

If I read this correctly, the idea is that someone who asserts, say, ‘x is good for a’cannot at the same time deny that a has a value, period; to do so would be either tomisuse ‘good for’ or to mean something by ‘good for’ other than what Rosati has inmind. But I fail to see why good-for has to derive in this sense from impersonalgoodness. It is my belief that, often, when we ascribe personal values to someone, wealso regard this person as valuable. But even if it were true that we always regardpeople as valuable, we would still need to provide an argument for the claim thatgood-for necessarily derives from this impersonal value. Rosati suggests the follow-ing approximation of what we mean when we say that something is good forsomeone. When we say:

. . . X is good for P, we indicate that there is a reason to promote X with P as the beneficiary ofthe action in light of or out of regard for the value of P. (2008, p. 344)

Now it strikes me that Rosati here points out just one possibility; surely quite anumber of reasons, or considerations “in light of or out of regard for”, might beinvoked when we say that something is good for P, especially if we have final good-for in mind. For instance, why must the value be impersonal? Since I suspect Rosatithinks that we all have this impersonal value qua persons, such value cannot explainwhy we “should promote X with P as the beneficiary” rather than Q as thebeneficiary. The real issue is a different one, however. Why must the promotionof X be out of a regard for a value? To put the question in terms of judgement, whydo we have to evaluate P in the first place? Is it not sufficient, for example, that welike the person, love him or her, or some such thing? Of course, we might make ananalytic connection, as certain naturalists would want, between the propositions ‘Ihave a pro-attitude towards P’ and ‘I evaluate P’. However, I doubt that Rosatiwould agree that her own analysis is, even in effect, naturalistic in this sense. Be thatas it may, I think it is a mistake to build into the analysis a certain kind of metaethicalposition. Of course, I might be wrong about this.

7.5 Thin-thin and thin-thick conceptionsNow that we have seen some examples of good-for analyses, let me address an issuethat has concerned me while reading not only Hurka, Regan, and Rosati, but a

7.5 THIN-THIN AND THIN-THICK CONCEPTIONS 105

number of other philosophers who have expressed views on what is good for us (orhow we should understand welfare or well-being). The issue can be put thus: Towhat extent, if at all, am I and these other writers discussing the same notion ofgood-for? As I pointed out when we touched upon this issue earlier, there is achance that we are dealing with different notions. However, since I also believethere is a chance that we are concerned with one and the same notion, let meexplain why I think this just might be the case.

One way of regarding the relation between good-for and welfare is to say that good-for can accrue to great many different things, including welfare. The kind of thing itwill accrue to is not, in the end, something the analysis of good-for should determine(anymore than an analysis of good should determine which things are in fact good); andsince the analysis does not specify the supervenience base of good-for, it may accrue toa wide variety of objects. As an upshot of this, people may disagree about whetherwelfare is good for a or not—and when they do so they are disagreeing about whetherwelfare is a supervenience base of good-for.

In other words this approach does not treat ‘welfare’ as a synonym of ‘good for’.Perhaps this is too counter-intuitive to imagine. Suppose therefore that there is a tightrelation of synonymy between the two notions. But such a relation is not straightfor-ward (and not merely because of familiar problems concerning synonymy and analy-ticity). On one account, the relation corresponds to what I earlier referred to as thethin-thin relation. Just as ‘good for’ expresses a thin notion, so does ‘welfare’.15

However, one might agree that there is this tight relation between ‘good-for’ and‘welfare’, but nonetheless deny that the synonymy relation holds between two thinnotions. It is probably welfare, then, that will be regarded as a thick notion—I mean, anotion with a fixed descriptive (non-evaluative) content.I suspect that at least some of the philosophers I have been considering here conceive

of the relation in this way. But whether or not this is the case, I think there is reason toquestion such a relation. (I suppose one might deny that welfare is a thick evaluativenotion on the grounds that it is not an evaluative notion at all. Welfare would then be apurely descriptive notion and open to a naturalistic analysis. I shall set this positionaside; I doubt it would be endorsed by any of the philosophers I am engaging with.)We would then have two possibilities, each of which seems to be salient in the presentcontext:Now, I think (1) is more plausible than (2), and that neither is as good as the approachI am taking—which is to regard the welfare as just one candidate as what is valuablefor someone, and hence not to treat ‘welfare’ as a synonym of ‘good-for’.

Before I explain why we should be cautious about (1) and (2), I want to enter thefollowing proviso. If welfare were indeed a thin value notion, I would have little

15 Zimmerman (2009) is ready to accept such a synonymy relation, and he is not particularly troubled byits being trivial or circular. Whether he conceives of this relation as thin-thin or thin-thick is not clear.


objection to the thin-thin approach. For one thing, it would have the advantage thatRosati, Regan, and I might after all be discussing the same notion. The approachwould not allow for much in the way of analysis, of course, but my real problem with itis that I am not convinced that welfare is such a thin notion. The notion of welfareseems, at least, to be thicker than good-for. On the other hand, (2), the thin-thicksynonymy relation, strikes me as implausible, because I do not see how a thin anda thick notion could be synonymous.Rosati and Zimmerman, and many others, seem to treat ‘good for’ as a synonym of

‘welfare’. It is merely a verbal matter whether we use one or the other term. Sowhatever we take to be the good-for making properties in a specific case are also,analytically, welfare-making properties in that case, and vice versa. But notice that, inacknowledging this, we neither agree that the notion of good-for is analytically relatedto what we express by a certain non-evaluative term nor accept that welfare is soconnected. We might simply have the thin-thin relation in mind.As mentioned above, Rosati makes a sharp distinction between ‘good for’ when it is

used about a person’s welfare and the other senses this expression has in statements like‘water is good for the plant’. This strongly suggests that she has in mind a narrowernotion than the one I am trying to capture with FAP.If Rosati treats welfare and good-for in accordance with (1), the differences between

her position and mine are substantive. We disagree over the good-for/welfare makingproperties. However, we are probably concerned with the same notion. However, ifshe treats welfare as a thick notion, or at least as a notion thicker than good-for, whichshe in turn regards as a thin notion, we do not have the same notion in mind.Could she not regard the relation between good-for and welfare as one obtaining

between two thick notions? Perhaps. However, this strikes me as no less implausiblethan the idea that the synonymy relation obtains between a thin and a thick notion.Good, as well as good-for, are in my view bona fide thin notions. However, I confessthat I have no compelling argument for this view—other than to say that the idea thatgood-for has more in common with good than with various kinds of thick notion, suchas admirability, strikes me as the more credible view.

(1) Thin-thinrelation?

(2) Thin-thickrelation?

Whatever is good forperson x isconstitutive of x’swelfare, and whateveris constitutive of x’swelfare is good for x

Figure 7.1

7.5 THIN-THIN AND THIN-THICK CONCEPTIONS 107

What I have said here applies, I think, mutatis mutandis, to Zimmerman’s (2009)suggestion that good-for should be understood in terms of what benefits the person.Is the relation between these notions of the thin-thin kind, though? If it is, the analysislooks acceptable; we should be able to accept it. However, the notion that somethingbenefits someone seems to me to be a thick notion, and if this is right, we have animplausible synonymy relation.


8

Problems and Possibilities

Connie Rosati’s view that good-for is a derivative value is shared by many authors,but it is not an element of FAP. In FAP the core idea is that the personal value of anobject x for a person a consists in there being normative reasons to favour x for a’s sake.It is to be expected that a will often (perhaps always) carry, in the eyes ofthe ‘evaluator’, some impersonal value. However, in my view this arises from thefact that most people, most of the time, are ready to ascribe some value to most otherpeople. Whether or not I am right to make this empirical claim about what peoplein fact evaluate, I wish to make it clear that I am not assuming that personal values arenecessarily derivative values.I want to turn, in the section below, to a feature of FAP that appears to be

worth expanding. Later, in Section 8.2, I will respond to some objections that havebeen raised to my account. FAP seems vulnerable to various kinds of counter-example.For instance, if values are understood in terms of the existence of agent-relativereasons, FAP seems to imply interpersonal value incompatibility (i.e. that personsa and b et al. will have values that are incompatible). Another kind of case concentrateson the apparent fact that FAP obliges us to care for people that we dislike. The generalidea goes something like this: it is plausible to assume that people we dislike havepersonal values; according to FAP we therefore have reason to favour things forsomeone’s sake that we in effect do not care at all about, which is counter-intuitive.In this connection we shall have occasion to look at Stephen Darwall’s insightfulapproach to welfare in Section 8.3. His analysis appears to handle these more or lessproblematic cases better than my own. There are also some more detailed objections tomy account that need to be addressed. I do this in Sections 8.4 and 8.5. Finally, I willsuggest in Section 8.6 what I would like a second volume on personal values to beabout. Pace the Tractarian Wittgenstein, there is always unfinished business to attend toin philosophy.

8.1 Janus valuesSo far the idea has been that objects of personal value should be understood in terms ofthere being a reason to take up some attitude vis-a-vis those objects. But the questionremains whether some personal values might not have more than one side—one facingthe person a for whom the value is ‘personal’ and another facing the rest of us (at anyrate, those of us who have a reason to favour the object for a’s sake). This issue seems tobe pointed up by the fact that the attitude that it is fitting to take to an object such as adrawing can vary depending on whether that drawing is, say, your daughter’s or yourown. Perhaps the attitude can also vary with regard to the strength of the reason. Somereasons appear to have a stronger normative force than others, and it is perhapsconceivable that one and the same object is Janus-like, also, in this regard. This strikesme as less likely, though. The first possibility seems more worthy of examination.

I find the bare idea that personal values might (in the first sense) be Janus-valuesintellectually inspiring. Moreover, once one starts thinking along these lines, thequestion arises whether there might be attitudes that everyone except a (the personwhose personal value we are considering) ought to take towards some object. Theremight be attitudes/emotions that we may direct onto, or have in relation to, others butnot direct onto, or have in relation to, ourselves. Perhaps there are even attitudes/emotions for which only the attitude holder can be the proper object of the attitude.The following may be an example of this: it might be (morally) unbecoming, but it isnonetheless the case that many people sometimes feel pride about something they havedone. However, although people often say that they are proud of what others, such astheir children, have achieved, or what their country or favourite soccer team hasaccomplished, it is not obvious in what sense one can be proud of something thatsomeone else has done. It is like taking credit for another’s accomplishments, andalthough we often seem to try to do it, the notion that we can do so can be queried onconceptual grounds. As to the former possibility, promises are often presented asexamples of attitudes that we cannot direct on ourselves. Certainly, one can say tooneself ‘I promise to stop smoking’, but such utterances are often said to express anavowal of some kind (say, stating your intention) rather than something that creates apromissory obligation. To promise is to promise someone other than oneself; other-wise it is not a real promise.

These two examples show, perhaps, that personal values can be ‘double-sided’. Onthe other hand, if some personal values are Janus-like in this way, I strongly suspect thatthey are quite rare. The examples are hardly beyond question, and since I cannot comeup with any more convincing cases, it is perhaps best to continue to regard this double-sided feature as a mere possibility. FA versions formulated in terms of, not reasons, butfittingness may be better suited to account for this Janus-like character of value. Thus itmight be fitting to feel pride for something one has done, but quite unfitting to feelpride for something someone else has done (especially if you are unrelated to theperson of whose actions you are proud).

110 PROBLEMS AND POSS IB IL IT IES

It might be objected that I am overlooking a problem here, and that this problemjeopardizes the whole analysis. Suppose therefore that there are attitudes (not neces-sarily of feeling pride in another’s accomplishments or promising oneself something)that we can have vis-a-vis an object x for a’s sake but which a conceptually cannot havefor her sake. Does this show that x cannot be a personal value? Does it even show,perhaps, that something can be a personal value for everyone except the person forwhose sake we are favouring it? Both of these worries can safely be ignored. An analysisin terms of reasons would be incredible if it did not leave room for the possibility thatreasons are at least agent-relative in the sense that only people who can favoursomething have a reason to do so. So if I cannot, on conceptual grounds, promisemyself something for my own sake, then this is to be looked upon as any other case inwhich someone does not have a reason to favour something because she cannot favourthe object. That the analysis is sensitive to this possibility is part of its very force.Accordingly, the reason FAP does not contain any clause to the effect that person a’s

personal value must be understood in terms of at least a’s having an agent-relativereason has much to do with the fact—as I take it to be—that we should accommodatethe possibility of personal Janus values.

8.2 Some counter-examplesHere is another worry. FAP might fail to account for everything we consider to be ofpersonal value. Think of a teenager a who wants to be ‘different’, and who values aband because it appeals only to him. He would stop valuing the band if he realized thathis parents liked the group or its music.1 It seems, in other words, that the band mighthave personal value for a only if certain others do not favour it or even considerthemselves as having a reason to favour it.Although teenagers’ attitudes do sometimes display this kind of conditionality, this

does not afford a counter-example to FAP. It may nonetheless prove illuminating todiscuss why it misses its target. To offer a counter-example, the teenager case wouldrequire us, first, to establish on intuitive grounds that the object of the conditionalizedattitude can carry personal value. It would then be necessary to say that FAP impliesthat the parents have a reason to favour the band (it has, after all, a personal value fortheir son). The final move would be to insist that this is incompatible with the value-making base (the subjacent properties of the band): the fact that the parents seethemselves as having no reason to favour the band.It is, I suppose, rather obvious why this example fails to present a problem to

advocates of FAP. Surely, whether or not the parents have a reason to favour theband—or, to be more specific, that which a favours in the conditionalized manner—for a’s sake is not obviously settled by the fact that the band is favoured by a on

1 I owe this example to Michael Brady (personal communication).

8.2 SOME COUNTER-EXAMPLES 111

condition that the parents do not think they have a reason to favour the band. The latterdoes not exclude the former. Moreover, suppose that a’s favouring is of the followingsort: a desires to listen to the band at a very high volume, on condition that the parentsdo not desire, or take themselves to have a reason to desire, the same thing. Again, thisis not itself a reason to rule out the possibility that the parents have reason to favour theeventuality that the states of affairs a wants to obtain obtains for a’s sake. This is quiteconsistent with the fact that they do not have to desire to listen to the band.

It is important to underline that FAP is formulated in terms of reasons that have an‘objective’ reading in the following sense: quite what reasons there are to favour anobject is not thought to be dependent on our beliefs. We may not know what featuresof an object are reason-providing, but the relevant features are nonetheless reason-providing. In fact, objectivity in this sense requires not only ignorance, but also error.For instance, to take an extreme case, though we might all believe that something is areason to favour an object, it is nonetheless possible that it is not a reason, since we allhold a false belief.

There are yet other, more or less problematic, sides to FAP. For example, it impliesthat all that it takes for x to carry a’s personal value is that there be some reason forsomeone to favour x for a’s sake; but that seems too permissive. Moreover, it does nottake much effort to imagine prima facie counter-examples. Here is such an example. Itsuggests that FAP forces us to ascribe conflicting personal values to a person. Supposeyour father is a convinced theist, and that your mother an equally firm atheist. Yourfather wants you to believe that God exists, and he maintains that he wants this (amongother things) for your sake. Your mother claims, on the other hand, that she wants you,for your sake, to believe that God does not exist. Surely we want to say both that yourmother has an agent-relative reason to favour your adoption of atheism and that yourfather has an agent-relevant reason to favour your adoption of theism. But now, sincewhat FAP requires for something to be a’s personal value is that there exists an agent-relative reason to favour this something for a’s sake, we have to conclude that you havetwo incompatible personal values; and this is hard to swallow. Atheism and (simulta-neous) theism cannot both be states carrying your personal value.

The conflicting-parents example has a number of facets. It opens up some interestingissues that must certainly be addressed. I will begin with the main issue: whethersomething can carry conflicting or incompatible personal values. To discuss this matterlet us accept pro tem that your parents do, in the case described, have agent-relativereasons to favour your being and acting in certain ways. This seems plausible. It is lessobvious, though, that they have a reason to favour your being an atheist, or a theist, foryour sake. As might be recalled, it is one thing to favour x (instrumentally, or for its ownsake) and another to favour x (instrumentally, or for its own sake) for someone’s sake.We need therefore to keep in mind that, on the most plausible interpretation of theconflicting-parents example, the kind of attitude involved is perhaps not an FFS (i.e.for-someone’s-sake) attitude at all—and that, if it is one, it may not be of the right kind.By ‘plausible’ I have in mind how easily, or to what extent, we can make sense of the


example (rather than the frequency of such cases); and by ‘right kind of attitude’ I meanthat we cannot exclude the possibility that each of the parents are in fact favouring x fortheir own sake rather than for their child’s. Where this is the case, we are dealing withthe wrong kind of attitude with regard to the personal value of the child.So does the conflicting-parents example show, or at least render it plausible, that

people can have conflicting personal values? I do not think so. In fact, if we arewilling to make an assumption about reasons that appears quite persuasive, we canstraightforwardly reject the intended conclusion of the example. Moreover, since thereare, as I have just mentioned, other apparently reasonable and realistic ways ofinterpreting the example, our rejection of the example need not come with a costlyprice tag.The assumption I would be willing to make about reasons can be stated in a few

words: one cannot have a reason to favour x and a reason to favour not-x. Certainly, thisneed not be true of every sort of reason. Thus, perhaps you can have a pro tanto reasonto favour x and a pro tanto reason to favour not-x. But not all reasons are, I think, protanto reasons. So, setting these (and the corresponding pro tanto personal values) aside, Ithink there is something fundamentally counter-intuitive about the very idea that oneand the same maximal set of relevant facts could give you a complete (or perhaps all-things-considered) reason to favour both x and not-x (whether that reason is agent-relative or not). In some of its features or parts, the world can, for example, be vague orindeterminate; hence we can anticipate that there will be ‘vague’ or ‘indeterminate’facts (and reasons and personal values). It remains hard to believe, however, that theworld could incorporate incompatible facts. Without doubt, I am here giving expres-sion to a basic metaphysical view, and I am not sure that I can come up with someheavy artillery in its defence. Be that as it may, the view seems inoffensive when itcomes, at least, to certain kinds of reason.Now, the scenario in the conflicting-parents example is different. It involves two

persons; and since we are, I assume, dealing with agent-relative reasons, this opens upthe possibility that these persons have different reasons—one has an agent-relativereason to favour x and the other has a similar reason to favour not-x. This makes sense.However, if we accept that facts cannot give one and the same person an agent-relativereason to favour both x and not-x, this implies that there must be some relevantdifference between the father and the mother which gives rise to these two differentreasons. Facts about x cannot on their own constitute an agent-relative reason to favourx and not-x. This is consistent with the claim that facts about x and the mother (and herrelation to x) and facts about x and the father (and his relation to x) can generate twoagent-relative reasons in this situation.Assuming this observation is correct, two connected points can be made. The first

concerns the fact that we need to turn our attention to the mother and father in orderto determine whether the example involves two kinds of reason. We need, that is, tofigure out whether there are relevant differences between these two persons, and thenask ourselves whether these differences are, together with other facts, such as to

8.2 SOME COUNTER-EXAMPLES 113

constitute two different reasons. My guess is that, often, once the parties agree aboutthe facts of the case, they will also agree that they have the same kind of reason, or thatonly one of them will have a reason (this reason not being shared by the other person).People are most certainly different. However, at the risk of oversimplifying matters,I would add that it also seems clear that we are more alike than we are unlike. I amtherefore inclined to think that most of the time facts will give people the same reasons(even if they are agent-relative such reasons); and that, depending on the circum-stances, the facts will sometimes generate a reason for one but not another person.Much less commonly the facts will be such that two persons have different agent-relative reasons.

As to the second point, suppose I am wrong to take this optimistic view of people’ssimilarities. So let us stick with the original assumption and suppose that the father doeshave an agent-relative reason to favour xwhile the mother has one to favour not-x. Thisdoes not imply that being an atheist and a theist is something that has personal value tothe child. Since no one has an agent-reason for favouring both x and not-x for his or herown sake, we can safely conclude that in this very case, where x and not-x are features ofthe child (being an atheist and a theist) that the child could not have a reason to favour hisown sake, it is clear that the issue of incompatible personal values does not arise at all.

The more plausible interpretation of the conflicting-parent example is, I suppose,the following. Given the content of the mother’s conviction, she has a reason to see toit that she convinces the son to become an atheist (for his own sake); and given thecontent of the father’s belief, he has a reason to see to it that he convinces the son tobecome a theist (for his own sake). But from this it does not follow that the son has areason to be a theist and an atheist for his own sake.

Other examples appear to show that FAP is counter-intuitive. These set out from thevery plausible idea that we do not confine ourselves to speaking about the welfare of (orwhat is good for) people we care about, since we also express judgements about thewelfare of people we are indifferent to, or even dislike. For example, although wedislike, or would have disliked, Hitler (to put it mildly), many people would neverthe-less be inclined to agree, if asked, that Eva Braun was good for Hitler. FAP appears tohave counter-intuitive results here. Why ought we to favour Braun for Hitler’s sake?2

As I shall explain in a moment, the advocate of FAP can deal with this kind ofobjection in various ways. Meanwhile, I would like to comment on Darwall’s analysisof welfare, which seems to handle cases like the Hitler example without any counter-intuitive results.3

2 If you think the example has excessively strong ‘welfare’ connotations, you might prefer a differentexample: instead of letting Eva Braun carry Hitler’s personal value, we might think of one of his paintings orhis summer residence.

3 For a review (in Swedish) of this book, see my ‘Recension av Stephen Darwalls Welfare and RationalCare’ (2004).


8.3 Darwall on welfareThe central tenet of Darwall’s proposal, which he calls a rational care theory of welfare, isthe following:

a person’s good is constituted . . . by what one (perhaps she) should want insofar as one cares abouther. (2002, p. 4)

The care-proviso is, of course, what stands out here. For Darwall it is crucial; it makes itpossible, according to him, to account for the normative feature of welfare:

To understand the normativity of welfare . . . we must see it in relation to care . . .What is aconceptual truth is that to care for someone is to be in a relation to him such that considerationsof that person’s welfare are normative for one’s desires and actions with respect to him.What is forsomeone’s good or welfare is what one ought to desire and promote insofar as one cares for him. (pp. 6–7)The normativity is not, however, the agent-relative kind of rational preference. It is rather

an agent-neutral normativity grasped from the perspective of someone who cares for the person.(p. 45)4

Darwall’s approach seems promising. For instance, thinking that something is good forHitler does not require that we have some actual attitude; it only requires that webelieve that it is something that there is reason to want insofar as we care about Hitler.On the other hand, this approach also raises a number of questions. I discuss theseproblems in (2004). For instance, one problem concerns the fact that the analysis seemsto give us something more than an analysis, namely a criterion with which to single outwhat is in fact beneficial to individuals. See, for instance, page 31, where Darwall says“Something is for someone’s good if it is what that person would want for herself, asshe actually is, insofar as she is fully knowledgeable and experienced and unreservedlyconcerned for herself ”. This is presented on the assumption that “any informed-desirestandard can serve” as a “plausible criterion of welfare”. Moreover, the criterionresembles his analysis of welfare quoted above. Admittedly, it is not always easy todetermine the distinction between analysing x and suggesting a criterion for whatcounts as x. But by adding the care-proviso Darwall appears to be trying to ensure thatthe analysis will not lead us to call things good-for that do not make the person better.If this is what Darwall in fact suggests (I might misunderstand him here, as mentioned),we should not, I believe, follow him.Here I will confine myself mainly to just one feature that I find problematic.

According to Darwall’s analysis, a person a’s good accrues to an object x if and onlyif x is what one should want “insofar as one cares about” a. My own analysis resemblesthis claim, but there are also a couple of important differences between my account andDarwall’s. The more obvious one is, of course, that Darwall’s analysandum is muchnarrower than mine. He is interested in welfare. Second, this might account for the fact

4 Darwall assumes that care, or sympathetic concern, as he also refers to it, is like a psychological naturalkind; see his (2002, p. 50).

8.3 DARWALL ON WELFARE 115

that he formulates his analysans in terms of only one sort of attitude, whereas FAPadmits (and even encourages us to consider) a whole spectrum (perhaps even spectra) ofattitudinal kinds. Objects carrying what I consider to be personal values typicallyprovide us with reasons to be considerate. In fact, at one point I thought of thedistinction between personal values and impersonal values in terms of what I called‘(person) considerative’ and ‘non-(person) considerative’ values. To believe there isreason for you to be considerate in your dealings with something is merely to evaluateit as a personal value, as something that is good or bad for someone. However, this doesnot necessarily mean that you care for the person. Often you will, but caring is onlyone sort of attitude that objects of personal value call for. A number of attitudes can bedirected at someone, or something, for someone’s sake, and it is not always the casehere that we care for the person. This notwithstanding, we may still think there isreason to behave in a certain way, in the light of some fact or with regard to someobject, and we might come to think this merely out of regard for the person. Again,our doing so reveals that we are treating something as a personal value. But once again,to be a considerate person is not necessarily to be someone who cares for the personinvolved. In this sense, being considerate is a much less demanding attitude than caring;to care for someone is to show among other things that you are considerate. Thereverse is not necessarily the case. Admittedly, ‘caring’ might be given a broad sense soas to cover a variety of different attitudes. However, even granting this, it would, in myview, strain the sense of this term too much to include under it a number of theattitudes (including respect and desire) which, on FAP, we have reason to have.

Third, Darwall’s analysans is not a straightforward FA analysis; it is formulated interms of what we want insofar as we have some other attitude, namely one of caring.Why this insofar-condition, and how precisely should it be interpreted?5 As we haveseen, Darwall is quite explicit about this matter. Were it not for the fact that we careabout this person there would be no way of saying what we have reason to want forthis person’s sake. Thus ‘caring for’ guarantees the normativity of ‘good for’; if we didnot care, we would not have a reason to want anything for a’s sake.I am puzzled by this line of reasoning. It would be quite plausible to see in it an

underlying and well-known view of practical reason. My problem is that I am unsurewhether Darwall actually endorses this view. The core idea of the view is that what wehave normative reason to do is all ultimately desire- or at least attitude-based—or (asone referee suggested), if we focus on welfare, based on certain kinds of attitude.6

5 Michael Zimmerman (2009) shares this worry: “It’s hard to know just what ‘insofar as’ should be takento mean here. It cannot be equivalent to the ‘if ’ of the material conditional, since then the failure to care for Pwould suffice for everything’s being good for P. Perhaps what is intended is the ‘if ’ of the strict conditional”(p. 437, n. 24).

6 In his earlier work Impartial Reason (1983), Darwall put forward the idea that “the content of thejudgment that there is reason for one to do A is simply that were one rationally to consider facts relevant todoing A, then one would be moved to prefer doing A” (p. 128). Plainly, this notion makes the relationbetween what we mean by ‘reason for acting’ and our rational preferences analytic.


Exegesis apart, suppose we did in fact endorse such a view. Would it be a problem?Not necessarily; but it would surely be preferable not to make the analysis dependenton a particular way of understanding reasons that is acceptable only to defenders ofthis view. Whether or not attitude-based views of either sort are plausible—I aminclined to think they are not—it would be a considerable drawback to reserve theanalysis for advocates of one just theory of reasons. As far as I can see, an analysis ofgood-for is capable of being illuminating even if the deontic notion is regarded asa primitive. It is one thing to understand value (be it impersonal or personal)in terms of a deontic expression like ‘ought’ or ‘reason’. It is quite another to requirethat only insofar as certain kinds of attitude are present does something, such as acertain fact, supply a reason. Darwall’s insofar-condition is, in my view, an interesting,but nonetheless superfluous, element in an analysis of good-for, and it serves onlyto ensure that the analysis is less interesting to those unwilling to accept therelevant, special view of reasons. I share Darwall’s conviction that good-for must beunderstood in terms of reasons. I also share his interest in interpreting whatthese reasons are. However, this is not something I include in the very notion ofreason I employ in FAP.Let me finally raise a minor point. The project of accounting for normativity in

terms of Darwall’s care-proviso raises an interesting issue: should we regard welfare asnormative when we need to formulate its analysis in terms of a condition, or should wesay that it merely has conditional normativity? It might be suggested that this is aterminological quibble. I suspect it is not. Consider the distinction between ‘a beliefon a condition’ (e.g. my belief that there is tiger in my living room, given that I seethe animal with my own eyes) and a ‘conditional belief ’ (e.g. that if there is a tiger inmy living room, I will believe I am in danger). Whereas the former is not a beliefthat I have unless the condition is met, the latter is a belief—it is just not a categoricalone.This suggests, it might be replied, that we should rather formulate the analysans by

analogy with a conditional belief. In the Eva Braun example we would thereby obtainsomething along the following lines: if you care for Hitler, you ought to favour Eva Braun( for Hitler’s sake). However, as was mentioned before, I think it is a mistake if weintroduce the ‘attitude (care)-clause’ in order to secure the normativity of the conse-quent. I would rule this out for the reasons mentioned earlier: I side, that is, with thosewho think that reasons must be attitude-given. If this is not what lies behind the‘conditional form’, I suspect the disagreement between an advocate of FAP andsomeone offering this kind of analysis will boil down to a substantive disagreementabout what reasons there are. I do not want to exclude the possibility that there aresuch ‘conditional values’; but in refusing to impose such an exclusion I am not obligedto deny that there are values in a stronger sense requiring a categorical analysans—which is what I try to deliver.

8.3 DARWALL ON WELFARE 117

8.4 Different strategiesLet us return to the defensive strategies that are open to an FAP-theorist confronted byproblems like the one set by the Eva Braun case. Recall the key objection: most of uscould not care less about what was done for Hitler’s sake, but it is nevertheless plausibleto say that Braun had personal value for Hitler. This clearly is imaginable whether ornot we adhere to FAP. However, if an analysis of personal value would imply that weshould favour things for Hitler’s sake just because we recognized that Hitler hadpersonal values, this would, the objection goes, talk against the analysis. Now, incontrast to Darwall’s account, FAP implies that if Eva Braun has value for Hitler,then we have a reason to favour her for his sake. This suggests then that FAP hascounter-intuitive implications.

There are in fact two quite familiar strategies open to advocates of FAP at this point:one explains why these examples appear counter-intuitive in such a way that theanalysis does not require that we hold actual pro-responses with an eye to Hitler. Onthe second approach we are forced to swallow the bitter pill; if Braun has personal valuefor Hitler then we should have pro-responses vis-a-vis Braun for Hitler’s sake.

Considering the first strategy, a combination of some well-known replies will behelpful. First, we must not forget that it is quite possible that Braun had no personalvalue for Hitler. But notice that to say this is not to deny that Hitler did in fact favourBraun. Apparently he did. But, again, from the fact that he loved and cherished Braunit does not follow that she was a carrier of personal value. It means only that she was theobject of some of Hitler’s pro-responses. It might be argued that this fact does notexplain our intuition that Braun was not merely liked by Hitler but had value for him.Perhaps we seem ready to hold that Braun had value for Hitler because we are soaccustomed to ignoring the meta-value issues that arise when we speak about personalvalues—the reason for this inattention being that we take the following as obvious: ifsomeone cherishes, or desires, an object, it has value for that person. But the fact that peopleoften reason in this way is no reason to say that the reasoning expresses a correctposition. For my own part, I fail to see why personal value should be exempted fromthe standard issues raised when we talk about (impersonal) value.

There is one possibility that is closely related to the above point. Suppose we admitthat “welfare” has more than one sense. Besides the one FAP is trying to capture, there isa purely descriptive non-normative notion. What if welfare constitutive in this sensemight also be considered as what makes something good for someone? Perhaps “goodfor” likewise is ambiguous. This ambiguity would help to explain why we believe it isfine to think that Eva Braun is good for Hitler despite our detesting Hitler to such anextent that we are not willing to favour anything for his sake. When this happens, theexplanation is that we have a sense of ‘good for’ inmindwhich, plainly, is not normative.

Of course, the reason we want to say that Braun carries value-for need not be thatwe are metaethically naive or that we have a notion of good for in mind that FAP is nottrying to analyse. Perhaps the judgement ‘Braun is good for Hitler’ expresses what


R. M. Hare once called an ‘inverted-commas judgement’. What we say is that Braunhas value for Hitler according toHitler. In other words, the ‘good for’ in the judgement isparasitic on the genuine evaluation of Hitler. These replies seem plausible. It is theirgenerality that worries me. Is it really always the case that we are either metaethicallynaive or expressing some kind of inverted-commas judgement when we judge thatsomething is good for a person we dislike?The second strategy acknowledges that the judgement ‘Braun is good for Hitler’ is a

genuine evaluative judgement; it handles the counter-intuitiveness of FAP in this sort ofcase by underlining that the reasons involved here are pro tanto reasons, i.e. reasons thatpersist even in the presence of weightier counter-reasons. To think that Braun is good forHitler is to think that we ought, pro tanto, to have some pro-response regarding Braunfor Hitler’s sake. In admitting this we are not, of course, admitting that Hitler is worthy,all-things-considered, of our pro-responses. Nor does it mean that he should be favour-ably treated until there is a stronger counter-reason to not do so. All it means is thefollowing: whatever it is that we think are the good-for making properties of Braun, it isonly in regard to these properties that we should favour Braun for Hitler’s sake.The approaches can in fact be combined. The fact that they can strengthens our

defence of the idea that FAP accounts for people’s personal values—whether or notthey involve malicious people.Still, patchwork solutions are rarely ideal—in particular, if they have to be backed up

by a complex, primitive notion such as the notion of a ‘genuine evaluation’. It istherefore wise to look for more plausible ways of dealing with these alleged counter-examples. One appealing approach, suggested by a referee, deals with the Braun–Hitlercase by distinguishing between moral and non-moral reasons to favour Braun forHitler’s sake. The atrocities committed by Hitler are perhaps such that we have nomoral reason to favour anything for his sake. However, this is consistent, or so it mightseem, with there being non-moral reasons to favour Braun for Hitler’s sake. Thesuggestion that the reasons to favour Braun for Hitler’s sake would be non-moral hasthe obvious merit that it kills two troublesome birds with one stone: the counter-intuitiveness of the patchwork explanation is gone; Braun has value for Hitler.However, recognition of this does not necessarily commit us to favouring Braun forHitler’s sake, although such a commitment would follow if we were non-responsive tomoral reasons, or if moral reasons did not override other sorts of reason.Consider again the pro tanto solution. It was argued that Braun’s being good for

Hitler supplies us with a pro tanto reason to have a pro-response to Braun for Hitler’ssake, but that it does not follow from this that Hitler is in toto worthy of our pro-responses. The referee’s suggestion might in fact strengthen this claim given thefollowing premise: in cases where a moral reason is a pro tanto reason and is in conflictwith another non-moral pro tanto reason, the non-moral reason is overridden.Just how passable a road this is remains to be shown. A serious obstacle certainly

remains, since we must explain what, exactly, makes a reason a moral one—a task I willnot undertake here.

8.4 DIFFERENT STRATEGIES 119

Let us move on to yet another worry about FAP. Like the Hitler example, thisworry encourages us to regard FAP as too demanding. However, this time the issuelooks more psychological than moral in nature. Consider the following example, then.Suppose winning the New York marathon has personal value for Charlie. Further-more, assume that you want to win this race, too. Is it really the case that you shouldthen favour his victory? Is that not plain counter-intuitive?7 Christian Piller (personalcommunication) has suggested another example: if your headache is good for youropponent in a game of chess because it increases his chances of winning, you surely donot thereby have reason to favour your headache for his sake as FAP requires. But sinceit does seem intuitively plausible to suppose that your headache is good for youropponent, there seems to be something wrong with FAP. Of course, in a case likethis we might not be in agreement over what is good for the person. However, supposeyou do believe that your headache is genuinely good for your opponent, and that youdo not take this to be what is sometimes called a ‘natural normative fact’. Theinteresting thing to ask then is whether FAP provides a way of explaining why theheadache is good for the opponent. When we take into consideration, first, thatfavouring your headache for his sake might involve a number of very different kindsof attitude (say, liking or preferring the headache for his sake), and second, that thereason you have to favour your own headache for another’s sake may well be out-weighed by other reasons (reasons having to do, perhaps, with the fact that you dislikethe headache for your own sake), cases like that of the headache do not seem to present aserious problem for FAP.

More generally, before surrendering to this sort of objection—the sort of objectiongenerated by the chess and marathon examples, I mean—we should investigate to whatextent ‘O has personal value for a’ expresses a genuine evaluation. I think that much ofthe resistance to the normativity of personal value derives from insufficient recognitionthat the examples concern inverted-commas uses of ‘value for’, or from the assumptionthat ‘O has personal value for a’ is equivalent to ‘O is de facto favoured by a’. Still,suppose someone claims to recognize that x has personal value for a, but denies that thisgives him any reason to favour x for a’s sake. Pointing out that reasons here mighthave pro tanto status will hopefully clear away much of the misunderstanding, but whatshould we say to someone who continues to deny that there is any reason at all to favourx for a’s sake?We could try reminding him that ‘favouring’ is a technical term covering awide range of attitudes. In other words, a variety of favourings might be relevant here,such as to be glad for a’s sake, or to respect the winner a’s sake (and not, say, spit on it) atthe end of the marathon. Many more examples could be given. Of course, some mightinsist that nothing whatever follows, regarding our attitudinal stance to x, from thefact that x has personal value for C. But now we have arrived at a point where I, at least,find it hard to understand what ‘value’ in ‘value-for’ stands for in this person’s view.

7 The examples are inspired by objections raised by Jonas Olson and Michael Brady.


An alternative approach to this issue would be to ask the persistent sceptic thefollowing question: Would you not accept, at least, that if you want to do everythingthat can be done for a’s sake, you ought to favour O? I owe this informal suggestion toMargaret Gilbert (personal communication). These suggestions are worth considering.However, they regard (personal) value as normatively conditional on a person’sattitudes (wants and carings-for), whereas my suggestion sets out from the idea thatwhat we have reasons to do is not necessarily a matter of what attitudes we have.

8.5 Heathwood’s objectionThere are further grounds for hesitating before seeking to interpret what is good foryou in terms of FAP. For instance, Chris Heathwood (2008) has recently argued in‘Fitting attitudes and Welfare’ that welfare, or what is good for a person, is notsusceptible to a fitting-attitude analysis. However, I think his conclusion can beavoided. His argument sets out from the idea that welfare is an intrinsic kind ofvalue. But it is also, he thinks:

. . . a relational kind of value. We express the idea when we say that something would be good forsomeone.Welfare value is intrinsic because we are saying the thing is good in itself for the person,and not merely good for what it leads to for the person. It is relational because it is a relationbetween the thing that’s good and the person for whom it’s good. (2008, p. 52)

Since welfare is an intrinsic value, its value will, he thinks, be unchanging; so ifsomething is good for me yesterday, it will be good for me today and in the future inthe sense that it will make “my life go better”. He then argues that “while an event’svalue for a person is unchanging, the attitudes he has reason to have towards such anevent can change over time” (p. 52).Suppose a person experienced a great pain yesterday, and that today, as the result of

taking a drug, he does not recall the pain he felt yesterday. Does he now have “reasonto be intrinsically averse to that ordeal for his own sake?” (p. 56). Heathwood thinksthe answer is negative, and that therefore welfare, and what is good-for a person,should not be understood in terms of a fitting attitude analysis.This reasoning is based on some assumptions that I do not share. However, even if

we focus on the case he describes, I am not convinced that the example has quite theimplications he seems to think it has for an FA analysis.8 He is correct that welfaremight have relational value. However, it is unclear why he simply assumes that welfareis an intrinsic value. He overlooks the possibility that welfare is relational in more waysthan one. Notice that Heathwood understands the “intrinsic value” of welfare as

8 Heathwood’s argument against an FA analysis of welfare is part of his general argument against fitting-attitude analyses. He argues that since it “would be theoretically unsatisfying to be forced to say that these twokinds of value [good simpliciter and good-for] have radically different natures”, we should give up the fitting-attitude analysis for good simpliciter too. As an illustration he uses the case ‘My Past and Future Operations’presented by Derek Parfit in (1984, p. 165).

8.5 HEATHWOOD’S OBJECTION 121

welfare being good, in itself, for the person. However, as we have seen, final value aswell as final value for might be extrinsic as well as intrinsic. Heathwood never reallyconsiders the idea that the pain I had yesterday might have an extrinsic final value—extrinsic in the sense that the time of the experience might well be one of the bad-formaking features of the event. If this were the case, the good-for accruing to the eventwould perhaps need to be time-indexed. But whether or not we want to accept thisidea, if the final value of the pain is an extrinsic final value for, there is nothing peculiarabout the idea that our reasons for favouring this event alter over time. In fact reasonsare not the only thing that might vary. The sort of favouring may also change. Forinstance, I might have reason to dislike now the pain I have now, but to wishtomorrow that the pain I had yesterday is not to be repeated.

Admittedly, this reply will not convince anyone who still thinks that the badnessaccruing to the pain yesterday is unchangingly bad, so Heathwood might still have apoint. However, I am inclined to think that the reason we regard the pain as somethingwith unchanging intrinsic value is that we take it to have final intrinsic value, period; itis at least reasonable to regard pain as having final value simpliciter. But if that is the case,we have, on an FA analysis, a timeless reason to disfavour the pain. This also seems to betrue if what is bad about the event in the past is rather that is makes our life worse. Insuch a case we might regard the value of the pain event as having a contributory value;what has the non-contributory value for me is, say, my life. In this case, given that mylife has final intrinsic value for me, it might be reasonable to say that the pain event is acontributory bad-for that gives me an unchanging reason to disfavour it.

8.6 The good life and the argument from fetishismLike most people I am interested in the truth. Some philosophers are more directlyconcerned with it than others; they are really very keen to tell us what the facts are. Notall philosophers are quite this straightforward, though. Some are concerned chiefly toprobe and explore possibilities and plausibilities rather than to disclose truths. I believe Ibelong, by temperament, in the exploratory camp. At the risk of sounding pompous,truth is something to hope for; it is interesting potentials that one should keep watchfor. In this work I have tried to explore the possibility of expanding our taxonomy witha new set of values.9 In doing so, as was made clear at an early stage, I have had to leavea number of interesting issues untouched. It was tempting to pursue at least some ofthese issues en route. Doing so would have considerably lengthened and delayed the

9 I have a twofold reason for being so bold as to speak of expanding classical value taxonomy. First, good-for has generally been understood as being dependent on the notion of good, period (or good simpliciter).However, I am treating good-for as a value notion independent of good, period. Second, the sorts of valuethat I am trying to understand in terms of FAP have been regarded by certain authors, including Rosati, asvalues that do not really correspond to what she (and others) consider to be good for people. My notion of apersonal value seems, therefore, to encompass much more than what is often considered to be good forpeople.


preparation of the book, however, and I would prefer to receive reactions to it at thisstage, before taking on some of these other issues. One such matter concerns therelationship between personal and impersonal values. Given that conflicts betweenthese kinds of value abound, this is a matter of serious theoretical concern that needs tobe addressed. Often we can only promote an impersonal value at a cost to somebody’spersonal value. To take a fairly simple example, suppose justice demands that a personshould return an object with personal value to her to its rightful owner. Perhaps sheacquired the object from someone who had stolen it and therefore did not have title toit, and therefore could not give her title to it. Can a situation be imagined in which itwould be the personal value rather than justice that we should secure? I think so (andI sincerely hope I am not alone in thinking so). I also suspect that most so-calledNIMBY (not in my back yard) cases concern conflicts between impersonal andpersonal values. If I am right about this, the issue is of not only theoretical but alsogreat practical concern. Or suppose by securing a certain personal value, I do some-thing that has (instrumentally, or is in itself a) negative impersonal value. ShouldI refrain from acting? In comparison with impersonal values, many—perhaps,most—personal values appear to carry little weight when we have to choose betweenthem and impersonal values. There are no ready-made answers or guidelines on howwe should resolve these conflicts. There are no clear-cut values about value, i.e. weneed at least to be able to compare these different values to each other in evaluativeterms. Of course, it would be marvellous if there were a rich set of values that accruedspecifically to value comparisons (and whose values were not classifiable as personal orimpersonal values), but this is still an underexplored area.10

Another issue to which I would like to return in the future concerns the place ofpersonal values in a good life. Obviously such values play a number of different roles ina person’s life. Some function as incitements, others as deterrents, to action; certainpersonal values require promotion, and others, as we have seen, call for quite differentkinds of response. But quite what roles, precisely, personal values play in the good life isas yet unclear to me. I shall not at this point try to address the difficult issue of whatmakes a life good. However, I would like to forestall a couple of potential misunder-standings about the contributory value of personal values in the good life.To facilitate discussion, let us confine attention to positive personal values. Should we

then conclude that a life full of personal values is a good life, and that a life less wellendowed with personal values, all other things being equal, is a worse life? Is therelation here really that straightforward? It seems hard to believe. When we considerthe nature of some of the examples discussed in this work, such as the bits and pieces ofa bookshelf that I mentioned at the outset of Chapter 4, it becomes absurd to say thatmy life would be better just because it contained more rather than less parts of thesebookshelves. Living a life rich in personal values is no guarantee that you are living

10 An exception to this is Wlodek Rabinowicz’s ‘Value Relations’ (2008).

8.6 THE GOOD LIFE AND THE ARGUMENT FROM FETISHISM 123

a good life (or, in one sense of these terms, that your level of welfare or well-being is high).Indeed there is an obvious problem with a life with too many personal values,particularly if they are of a certain kind. According to the argument from fetishism a lifefull of at least certain personal values is, in effect, a bad life. The reason for this varies,depending on the way the argument is advocated. The life might be bad because it issuperficial—a shallow sort of existence, one in which the person has the wrongpriorities. Alternatively, its being bad may have to do with some notion of authenticityor false consciousness: if you surround yourself with material things that have value foryou, you are not really living a good life. You may think you are, but you are deluded.In fact, to give weight to concrete things which you consider to have value for you is tolook at life with a blindfold on. ‘Real’ values—those capable of being elements in thegood life—will forever escape you.

Of course, the fetishist accusation is a sweeping attack, and perhaps therefore difficultto take wholly seriously. For one thing, it overlooks the distinction between my treatingsomething as a personal value and something’s actually beingmy personal value. I mightsurround myself with things that I claim have personal value, but whether these objectsreally carry value is, in the end, not a matter to be settled by personal fiat. Since I havealready addressed this matter enough, I will spare the reader yet another repetition.However, the fetishist argument does highlight an interesting issue concerning acertain group of personal values—namely, those that accrue to material things (e.g.the sheets of paper with poems inscribed upon them and bits and pieces of furniturethat I have considered in the course of discussion). Since one desideratum of the goodlife is surely that such a life will display good-making/contributive properties that areworth striving for,11 we seem obliged to argue that we should desire to have suchobjects if we want a good life. But it would be ridiculous to maintain this, so thingscarrying personal values are not part of a good life. Moreover, if these personal valuesplay no part in a good life, we might even question whether they are real values in thefirst place. Given the close, intuitive connection between the notion of personal valueand the notion of a good life, we should at least expect there to be some interestingrelationship between them. If FAP cannot guarantee this, it becomes questionablewhether it is about personal values in the first place.This line of reasoning does display certain insights. What is more, I suspect that the

assumption underlying it is that concrete things lack any value at all—or, that if theyhave value it is instrumental. If the latter is the case, then the thought is that we areprobably mistaking instrumental values for final ones. This is a classic mistake. Humewarned us about it. Cases of such conflation are probably real. (It is at least imaginable

11 Joseph Butler’s lesson to the hedonist, that it might be self-defeating merely to seek pleasure, must beborne in mind here. That is, it is probably a valuable piece of advice to anyone who thinks that there arethings that make a life good, and so should be pursued. Butler’s further point, that in order to maximizepleasure the hedonist should pursue things that are not valuable, may not be as easily followed by thoseespousing other views of value. See Butler, Fifteen Sermons Preached at the Rolls Chapel (1900 [1729], Preface,Section 31, and Sermon xi, ß1.9); see also Henry Sidgwick, The Methods of Ethics (1907 [1874], pp. 136, 403).


that, in the eyes of a miser, money has final value, whereas the reasonable position isthat it has at most instrumental value.) Still, the challenge is nonetheless flawed. It setsout from an even more fundamental idea, namely that values are things to bepromoted; and since promoting certain personal values (such as the value accruing toan artefact from one’s past) makes no difference to whether one has a good life or not, itis readily concluded that there is something fishy about these personal values. But thisreasoning is based on the false assumption that all valuable objects ought to bepromoted. Some should, but only some! In particular, concrete objects (e.g. peopleand things) of personal value are such that they should be, for example, respected,admired, or loved for some person’s sake. They are not such that we should seek tomultiply them. In most cases that would be absurd.From a buck-passing perspective, the above is suggestive of at least two lines of

further enquiry. If we are interested in what makes a life good, we might start bycontemplating what we suspect are the good-making properties of a good life. Thesemay be natural properties; equally they might be different kinds of personal as well asimpersonal value. If we are interested, in addition, in what makes a person considerate,I suggest we start looking at how the considerate individual deals with people’spersonal values (including her own). To be considerate is, in effect, to be someonewho favours something for someone’s sake. Being moral is not the same as beingconsiderate, but it is a good start.

8.6 THE GOOD LIFE AND THE ARGUMENT FROM FETISHISM 125

9

One Reason Dichotomy Less?

Since its introduction by Thomas Nagel (1970) and Derek Parfit (1984), the dichoto-my of the agent-relative and the agent-neutral has become widely popular.1 Forinstance, there is at present a lively discussion going on between philosophers interestedin taxonomizing moral theories who detect a potential line of demarcation in agent-relative restrictions.2 Certain theories (particularly utilitarianism) have been criticizedfor being unable to incorporate such restrictions. On the other hand, so-calledcommon-sense morality is said to be full of agent-relative restrictions. So why notmake use of this much-employed distinction? Why not simply draw the line betweenimpersonal and personal value with the help of these two kinds of reason?

It would be a tidy solution, certainly, but I think there are strong arguments forbeing cautious about this approach. My arguments have recently undergone a change,though. When, many years ago, I started thinking about the notion of value-for itwas very much because I was not satisfied with the idea of values being agent-relative.To look for an analysans in another dimension of relativity did not, therefore, seem agood idea. This was not the way to go. Actually, I had yet another motive, besides thatof avoiding making personal values private. I found the very notion of an agent-relativereason unclear. I could get a grip on a fairly unambitious way of understanding agent-relativity relatively easily, particularly when it was associated with aims, principles,statements, or judgements (propositions). But things get much more complicatedwhen we turn to reasons—or, at least, the normative kinds of reason FAP requires.My feeling was, and still is, that people try to squeeze out much more than a modest,unambitious sense of this notion. And to analyse a notion like personal value with thehelp of an imprecise concept did not seem like a good idea.

Recently my views on these issues have changed radically. I am still scepticalabout the dichotomy, but my scepticism now concerns to a greater extent its other

1 Nagel, The Possibility of Altruism (1970), and Parfit, Reasons and Persons (1984).2 See e.g. Samuel Scheffler (ed.), Consequentialism and Its Critics (1986) and David McNaughton and Piers

Rawling, ‘Deontology and Agency’ (1993).

component: it is the notion of an agent-neutral reason that I now find obscure. Thiswould be worrying were it not for the fact that I now also believe that there is a way toavoid ending up with private values even if we employ an agent-relative reason. In theremaining sections of this chapter I will explain my change of mind; and I will arguethat, given certain plausible assumptions, we should be sceptical about the distinctionbetween agent-relative and agent-neutral normative reasons. Just what conclusion weshould draw from this scepticism is still a somewhat open matter, to which I will returnat the end of the chapter. It might just be that there is a way of meeting the kind ofchallenge I shall pose to the coherence of the notion of an agent-neutral reason.Jonathan Dancy’s recent idea that reason considerations play different roles—theymay, in short, be favourers or enablers—might help us clarify and retain a notion ofagent-neutral reasons. This would be most welcome. I have some qualms, though.Dancy has argued that all reasons are contributory (pro tanto) reasons. I am not quiteconvinced by Dancy’s arguments, and although I do not take the matter to be settled, Ifind the notion of a complete reason to be quite plausible. If I am right about this—andI am not alone in thinking it to be so—then Dancy’s favourer/enabler distinction willstill be of use to us, although this use will be slightly more limited. It will help us savethe dichotomy when it comes to contributory reasons.However, if there is something genuinely fishy about this distinction, it would

certainly be bad advice to locate the boundary between personal and impersonal valuesin the fact that the former, but not the latter, are agent-relative reasons. We shouldneed to find another boundary. It was this insight, then, which prompted me tounderstand personal values in terms of for someone’s sake attitudes.

9.1 Introductory notesThe importance of the distinction extends to most areas in practical philosophy.Despite its popularity, it remains difficult to grasp. In part this has to do with the factthat there is no consensus concerning the sort of object to which it should be applied.Nagel distinguishes between agent-neutral and agent-relative values, reasons, andprinciples. Parfit focuses on normative theories (and the aims they provide to agents).In an early paper, McNaughton and Rawling (1991) are interested in rules, and in alater work (2002) they go on to discuss reasons. John Skorupski formulates thedistinction in The Domain of Reasons (2010) in terms of predicates, and there areother suggestions too. Some writers suspect that we fundamentally talk about oneand the same distinction (e.g. Dancy, 1993; Ridge, 2005).3 This latter issue will not beaddressed here, though. What is important to keep in mind, is that my focus continuesto be on (practical) normative reasons for action rather than theoretical reasons for belief.Moreover, I will understand ‘action’ in a broad sense of the word which includes states

3 See Dancy, Moral Reasons (1993) and Ridge, ‘Agent-Neutral vs. Agent-Relative Reason’ (2005).

9.1 INTRODUCTORY NOTES 127

such as desiring, wanting, intending, and (certain kinds of ) feeling. Skorupski (2010)makes a more fine-grained division, distinguishing between epistemic (what I calltheoretical), practical (reasons for desires), and evaluative reasons (reasons for senti-ments, emotions, and affective responses). However, for my purposes it suffices tooperate with the more general distinction between reasons for beliefs and reasons fordesires and emotions.

What is often overlooked is that there are at least two related, but nonethelessdifferent, ways of understanding the agent-relative/neutral reason distinction. One,outlined in Section 9.2, centres on whether reasons do or do not essentially involve, orrefer to, particular agents; this is the philosophically more interesting notion. Theother, briefly discussed in Section 9.3, takes into consideration the number of peoplewho as a matter of fact have a reason to do something. The remainder of the chapterasks in what way the notion of a normative agent-neutral reason is problematic.Needless to say, the fact that there are problems associated with the distinction neednot mean that it is suspect. The philosophical literature is full of (some would even saythrives on there being) intuitive distinctions that have not as yet received properanalysis. However, because the prospect of getting the distinction right looks some-what gloomy, as I shall try to show below, I think things are more serious in this case.Whether there are sufficiently strong reasons for giving up on the more philosophicallyrelevant distinction as a conceptual confusion is a matter I leave to the reader’s ownjudgement. I confine myself to setting out a diagnosis. The cure will depend onone’s attitude to a number of issues the discussion of which falls outside the scope ofthis work.

9.2 The essentialist senseThe philosophically more interesting sense of the distinction was formulated by Nagel(1986) in The View from Nowhere:

If a reason can be given a general form which does not include an essential referenceto the person who has it, it is an agent-neutral reason . . . If on the other hand the generalform of a reason does include an essential reference to the person who has it, it is an agent-relativereason.4

Nagel draws the distinction in terms of what does and what does not essentially referto a particular person, the owner of the reason. Various other writers have employedsimilar accounts. They have also tried to be more specific about what is required if areason is to qualify as agent-relative or neutral. Philip Pettit, for instance, describes anagent-relative reason as “one that cannot be fully specified without pronominal back-reference to the person for whom it is a reason”. He describes an agent-neutral one

4 Nagel, The View from Nowhere (1986, pp. 152–3).

128 ONE REASON DICHOTOMY LESS?

“as one that can be fully specified without such an indexical device” (1987, p. 75).5

Several other writers have suggested similar accounts, stressing that the distinctionconcerns reasons that do or do not contain ineliminable back-reference to the agent forwhom it is a reason (cf. McNaughton and Rawling, 1995; Portmore, 2001; Ridge,2005).6

We cannot, of course, be sure that Pettit’s and similar approaches to the dichotomyfully coincide with Nagel’s own approach. It is, for instance, a matter of interpretationwhen a statement is “fully specified”. But this notwithstanding, the concern aboutindexicals and ineliminable back-references surely reflects the authors’ efforts to expresswhat Nagel was driving at—namely, that the distinction is about reasons that do and donot essentially involve particular agents.7 How we present this in language is aninteresting but nonetheless secondary matter; the real issue is still what makes reason-statements true.Another influential view denies that reasons are facts on the grounds that normative

reason-statements express non-cognitive states (e.g. desires). Discussion of this alterna-tive here would complicate the presentation of what I take to be the problem with thedichotomy considerably, and I have therefore not entered into it.The ‘essentialist’ approach to the distinction is the interesting one, in that it tries to

unravel a conceptual difference (one which, among other things, is thought to help usdistinguish between different sorts of moral theory). It focuses on one particularfeature: whether or not a reason-statement necessarily refers to the agent for whomit is a reason. The important thing to realize, then, is that the distinction is silent aboutthe range of agents for whom something is a reason. It leaves this matter quite open.This is not true about the second approach, where the issue of quantification ispositioned centrally. But before commenting on this second, as I will call it, ‘numberapproach’ to the dichotomy, it will be convenient to make the first interpretation atleast intuitively accessible.Despite the fact that there is no general agreement8 as to how to formulate the

‘essentialist dichotomy’, it is at least possible to express in a few words what seems to beits core idea. Let P refer to a ‘reason-statement’ (by which I mean no more than a

5 Pettit, ‘Universality without Utilitarianism’ (1987). In ‘The Paradox of Loyalty’ (1988), Pettit suggeststhat “A reason for action [ . . . ] is the sort of proposition which may appear in the major premises of a practicalsyllogism”.

6 D. McNaughton and P. Rawling, ‘Value and Agent-Relative Reasons (1995), Douglas Portmore,‘McNaughton and Rawling on the Agent-Relative/Agent-Neutral Distinction’ (2001), and Michael Ridge,‘Agent-Neutral vs. Agent-Relative Reason’ (2005).

7 Nagel, originally in The Possibility of Altruism (1970), referred to the distinction as one between subjectiveand objective reasons. The terminology ‘agent-neutral and agent-relative’ was introduced by Parfit in‘Prudence, Morality and the Prisoner’s Dilemma’ (1979).

8 For some attempts to clarify the distinction, see McNaughton and Rawling, ‘Agent-Relativity and theDoing/Happening Distinction’ (1991) and ‘Conditional and Conditioned Reasons’ (2002), and Skorupski,‘Agent-neutrality, Consequentialism, Utilitarianism . . . : a Terminological Note’ (1995). For a review of thislast, see John Broome, ‘Skorupski on Agent-neutrality’ (1995). Objections to McNaughton and Rawling(1991) are given in Portmore (2001). For a reply, see McNaughton and Rawling (2002).

9.2 THE ESSENTIAL IST SENSE 129

statement expressing a reason). Nagel, Pettit, and others make this move from reasons tostatements, or what they call forms, expressing reasons. It will simplify the presentation ifwe sometimes discuss reason-statements rather than reasons.

The expression ‘reason-statement’ is ambiguous. Typically it will be a statementexpressing some fact—say, ‘that person a needs help’ or ‘that Anna is my daughter’.However, it may also refer to statements, which consist of two parts: one stating that (i)�-ing is normatively called for, and the other listing (ii) the reason-making grounds for�-ing. Here are some different examples: ‘Charlie ought to help Mary because she isdrowning’, and ‘There is a reason for Tom to help Mary, because she is his daughter’.The left-hand side of the ‘because’ expresses (i), whereas the right-hand side expresses(ii). Moreover, statements of the first kind are elliptical in the following sense: fullyspelled out, they will appear on the right-hand side of a statement containing parts (i)and (ii). Since nothing important hinges on this matter, I will, for simplicity’s sake,mainly use examples of the first kind.

Returning to the core idea behind the essentialist dichotomy, it can be expressed inthe following way—where a (and later b) refers to an agent, and � to some act:

If P states a reason for a to �, then:P states an agent-relative reason for a if and only ifP contains an essential reference to a,Otherwise, P states an agent-neutral reason for a to �.

Moreover:

P contains an essential reference to a if and only ifP is not logically equivalent to any other statement Q thatdoes not refer to x.

The following statement expresses, on the essentialist approach, an agent-neutralreason for a (in case a 6¼ b): (Pn) ‘Person b is drowning’. The fact that b is drowningconstitutes a reason to save b that is not agent-relative but agent-neutral. The statement(Pr) ‘My daughter b is drowning’, expresses in its turn a typical example of an agent-relative reason; it gives b’s parents a reason to save b that is different from the reasonthat I, not being her father, have to save her.

The above picture is oversimplified, however. The ‘relative/neutral’ status of thesereason-statements, Pr and Pn, seems to depend on the person to whom they areaddressed. For instance, if Pn is directed to person b, it should be seen as stating anagent-relative reason for b. And while Pr may state an agent-relative reason for b’sparents, it would, in many people’s minds, also state a reason for people other than b’sparents, since it seems to be an agent-neutral reason that we should save people fromdrowning. These dependencies can be quite confusing. They are one of the sources ofthe slipperiness of the distinction. That is, given that we have in mind a certain kind of


reason—which I will come back to later on—it is clear that whether a reason-statement expresses something agent-relative or agent-neutral depends, in part, onthe person to whom it is addressed. On the other hand, if the truth of a reason-statement is not relativized to the way the person to whom it is addressed understandsthe statement, the above confusion concerns something else, namely what people taketo be reasons. This distinction between, as I will refer to them, motivating and realreasons, will occupy us shortly. In the meantime the second important fleshing out ofthe dichotomy must be considered.

9.3 The number approachThe essentialist notion is not the only one in circulation. To say that P states an agent-neutral reason is, on the number approach, just another way of saying that P states thesame reason for all agents; and where this is the case P is perhaps best understood asstating the same reason for each agent. A neutral reason is therefore understood in termsof being a reason that everyone has. An agent-neutral reason is a reason that calls for thesame action from every agent (to which it applies), whereas an agent-relative one callsfor different actions from each agent (to which it applies). The number approachshould perhaps be labelled ‘Parfit’s approach’, because it seems to be the one Parfit hasin mind. In Reasons and Persons (1984), he explains it in the following passage, where‘C’ refers to a moral theory:

Since C gives to all agents common moral aims, I shall call C agent-neutral. Many moral theoriesdo not take this form. These theories are agent-relative, giving to different agents, different aims.

Further on, he adds:

. . .When I call some reason agent-relative, I am not claiming that this reason cannot be a reasonfor other agents. All that I am claiming is that it may not be. (1984, p. 143)

Two things particularly merit comment here. First, notice that there is no mention ofthe notion that some reasons are essentially an agent’s reason, so we can reasonablysuspect that we are here not dealing with an essentialist way of understanding thedichotomy. Second, it is not obvious how Parfit’s insistence (in the latter quote) thatwhat is an agent-relative reason may also be an agent-neutral reason should beunderstood. The natural conclusion here would, I think, be that whether somethingcan or cannot be a reason for other agents has do with the sort of people who happen toexist, or be present, at the time the act is normatively called for. If persons with a certainproperty S have a reason to do something, then it is an agent-relative reason if as amatter of fact not everyone has the property S. On the other hand, if people were tocome to have property S, the reason would be agent-neutral. In other words, if areason applies as a matter of fact only to a particular agent (or group of agents), it is hasagent-relative status. Otherwise it is agent-neutral. I confess I am not entirely certain

9.3 THE NUMBER APPROACH 131

how much of a conjecture this interpretative point is, so in what follows I will use theexpression ‘number approach’ rather than ‘Parfit’s approach’.9

The relationship between these two approaches—essentialist and number-based—needs clarifying. While the essentialist picture explains (to a degree) why certain reasonsprescribe different actions for different agents and others prescribe the same action toeveryone, the opposite does not hold. That is, a reason’s being neutral or relative on theessentialist definition tells us why it is neutral or relative on the number definition—butnot vice versa. On the number approach the dichotomy is dependent, not on aconceptual matter, but on how the world happens to look. In contrast with theessentialist notion, this approach does not purport to say (at least, not necessarily)anything about what a reason is. The essentialist notion is, therefore, the moreinteresting one; and accordingly my focus will be on this view of the dichotomy.

9.4 Reason-forI want next to draw attention to a particular feature of reasons—what I refer to asthe personalizability feature (PEF for short). PEF consists in the fact that all reasons arealways reasons for an agent to �. It was realizing the implication that this property hasfor the distinction between agent-relative/neutral reasons that recently made mechange my mind about this dichotomy. That reasons are always reasons for someoneseems to have clear consequences for the distinction between reasons that are agent-relative and those that are agent-neutral. In fact, setting out from this feature, we canchallenge the advocate of agent-neutral reasons to give us an account of it (the feature)and, in particular, ask why the following line of reasoning is erroneous (where PEFIrefers to ‘personalizability feature implication’):

(PEFI) Since all reasons are apparently reasons for someone to F, and a reason to F is only areason for someone if it somehow involves or refers to this someone, it follows that all reasons toF are in their very form reasons that refer to the person who has the reason to F. This, in itsturn, is just another way of saying that all reasons to F are, on entirely formal grounds, agent-relative reasons.

Whether PEFI poses a serious challenge or not depends on various things, however.First, we must be clear about the sense in which it is true that a reason is only a reason forsomeone, if it somehow involves or refers to this someone. We must also specify what kind ofreason we have in mind. As I have already said, I will confine myself to the discussionof normative reasons rather than theoretical ones. Moreover, I will be concerned witha certain popular view of normative reasons. What then becomes problematic is the

9 Cf. Mark Schroeder in ‘Reasons and Agent-neutrality’ (2007). He discusses what he refers to as the“Quantification strategy”, according to which “agent-neutral reasons arise when something is an agent-relational reason for everyone” (p. 280). Schroeder’s article is excellent, and I am just sorry that I read it toolate to include a more substantial comment.


combination of this widespread view and the essentialist way of understandingagent-neutral reasons.Before continuing, a caveat is in place: some reasons are not normative but explana-

tory. It might be that the latter, but not the former, kind of reason is consistent with thedistinction. Although the neutral/relative dichotomy is usually applied only to norma-tive reasons, in principle it can be applied to explanatory reasons.So-called explanatory reasons are psychological states that cause us to act (albeit not

necessarily in their own right). Could such reasons be propositional entities? This is asomewhat awkward idea, as we shall see in a moment. But for the sake of illustration,suppose for a moment that we regard reasons as propositions (of a certain kind). Wouldthat not pave the way for the dichotomy? This line or reasoning might then amount tothe following: many different propositions may, as a matter of fact, motivate a person.Moreover, there are no compelling grounds to be found in the literature for supposingthat it is only propositions that refer to the agent which can motivate her to act. Forinstance, believing that there is a stranger drowning may cause you to act.10 It mighttherefore seem that the challenge can be brushed aside if we employ a notion of reasonunderstood in terms of the propositional content of the mental states that cause the agentto act. What causes an agent to act may be the entertaining of a proposition that refersto the agent, but it is an open issue whether it has to be this sort of proposition thatmotivates the agent. Beliefs expressible with agent-neutral as well as agent-relativestatements may be motivating the agent.Is the above, then, the way to understand explanatory reasons? I think not! In fact,

we should disregard the above interpretation. It is not really propositional content thatconstitutes an agent’s explanatory reasons but rather her mental states. The descriptionpresented above is therefore, I suggest, a more accurate depiction of what we mightrefer to as motivating reason.That there is a point in keeping motivating (Dancy; Parfit) or as Thomas Scanlon

calls them, operative, reasons apart from explanatory reasons is, I think, obvious.11

Thus, it is the agent’s belief that p, rather than the fact that p alone, that explains why sheacted in the relevant way. Consider the following example. An agent a is asked why shejumped into the water and says something like “The reason is that God told me to savethe person from drowning”. In this case it would seem the explanation of why a actedas she did is hardly that God told her something; rather the explanation must be that

10 What precisely causes an agent to act is a complicated matter, and one that I do not want to go into. Mypoint here is merely that the belief must not necessarily involve, say, the idea that the stranger is in front of me,or the idea that I can help the stranger. It is (at least, in part) an empirical question which beliefs move us to act(alone or in company with other mental states).

11 See Scanlon (1998). Cf. also the notion of an apparent reason in Ingmar Persson’s The Retreat of Reason:A Dilemma in the Philosophy of Life (2005). Cf. Philip Pettit and Michael Smith (1990), and their distinctionbetween a first-person and a third-person perspective on reasons. That what motivates us is what we believeand not that we believe it is defended forcefully by Dancy in his Practical Reality (2000b, Ch. 6). Cf. Parfit(2001). Jonas Olson and Frans Svensson in ‘Regimenting Reasons’ (2005) also suggest a tripartite distinctionbetween what they refer to as motivating, deliberating, and normative reasons.

9.4 REASON-FOR 133

the agent believed God spoke to her. Nor does it seem correct to conclude that whatmotivated her to act was that she believed that God spoke to her; what motivatedher was something that was the case—namely, God’s telling her to do something—and not her merely believing it to be the case. As I argued earlier, we should, in otherwords, avoid confusing the content of the agent’s mental state with what caused herto act. It is beliefs and desires that make the agent act, not some proposition-like entity.

However, if we were to speak not about the content of the agent’s mental states,but rather about the agent’s mental states, we would in fact be referring to the agent, andthen explanatory reasons would all become agent-relative in character.12 It would bethe agent’s belief that God spoke that explained the acting. Moreover, if theseobservations are correct, the notion of a reason appears to decompose into at least threevarieties: (i) normative reasons, (ii) motivating reasons, and (iii) explanatory reasons.Whatmotivates the agent is hardly ever her belief that p but rather p. However, given thefalsehood of her belief we need to invoke her belief to explain why she did what she did.

The division between motivating and explanatory reasons is, of course, onlysketched here. Again, the general idea is that we sometimes have that which motivatesthe agent in mind when we talk about her reasons. Thus, what motivated DonQuixote to take up arms was that monsters stood in his way, but what explains whyhe spent time chasing windmills was his confused belief that these buildings wereactually living monsters (and his desire to kill them).

There is another reason why we should not mix up the notion of a normative reasonwith the related notion of a motivating reason. The expression “motivating reason”refers to that state of affairs which as a matter of fact is believed to be a (normative)reason by the agent; it is that which the agent takes to count in favour of some attitudeor act. Few would be ready to say, though, that just because we believe something tobe a reason, it is a reason. Only on a highly controversial metaethical view (a crude sortof relativism) are motivating reasons constitutive of normative reasons.13

The distinction between (normative) reasons and motivating reasons is important.The motivating kind appear to be quite consistent with the neutral/relative dichoto-my—that is, there seems to be no basis for doubting that the beliefs which the agenthas about what are her reasons need not always take a form that essentially involvesthe agent. Nothing necessarily prevents an agent from regarding the allegedly agent-neutral fact that b is drowning as a reason for her to try to save b. But again, this does notestablish that there are agent-neutral normative reasons—it indicates at most that theagent believes there to be (in some sense other than the essentialist one) agent-neutralreasons.

12 Michael Zimmerman reminded me of the latter interpretation (personal communication).13 The notion that a statement expresses an agent-relative or neutral reason does not show that the

relevant statement expresses a truth about a reason; it shows merely that it expresses a motivating reason. Theissue, discussed at the end of Section 9.2, of whether a statement expresses an agent-relative or agent-neutralreason depending on the person to whom it is addressed is an issue concerning operative, not real, reasons.


Now, since the fitting-attitude analysis understands impersonal as well as personalvalue in terms of normative (or justificatory, or good) reasons, my real interest concernsthese reasons, so I will not take the opportunity to enter into detailed discussion ofexplanatory and motivating reasons. I will leave it open, then, whether in particular so-called motivating reasons can be understood in terms of the dichotomy. Explanatoryreasons will run into most of the problems that normative reasons do with regard to thechallenge. However, this is not something that I propose to pursue here.Consider, next, normative reasons. There is much less agreement over the way these

should be analysed, although most observers agree that they are reasons that justify,rather than simply explain, an agent’s performance of an act. There have been all sortsof attempt to close the gap between normative and explanatory reasons. Here I willmake one assumption that rules out some of these attempts. Thus, in keeping with theway many influential writers regard reasons these days, I have supposed in this workthat normative reasons are facts.14 The nature of these facts can be left unspecified—they may be natural or, say, evaluative/normative.Some observations now need to be made about the above assumption. First, it is

important that facts should be understood as entities independent of what peoplebelieve to be the case (facts about people’s beliefs are no exception; if such facts arereasons, they are so independently of whether they are believed to be the case). Thisis also in line with much recent work on reasons qua facts. Second, ‘fact’ and(i) ‘obtaining of a state of affairs’ are considered synonyms.15 There are other options,though. For instance, we might think of a fact as (ii) an aspect of an obtaining state ofaffairs; or we could follow the common practice of using ‘fact’ to refer to (iii) thecontent of a true possible judgement, i.e. a ‘true proposition’.I remain unsure whether anything of genuine importance, for my purposes, hinges

on my preference for (i), and in what follows I shall have the more robust sense of ‘fact’indicated in (i) in mind—a sense allowing us to recognize that facts contain propertiesand objects. This does not square well with the idea that facts are true propositions.Propositions are generally thought to contain only concepts, whereas obtainings aregenerally taken to involve all sorts of objects.Again, how crucial this preference for ‘robust facts’ is for the discussion in this

chapter is not entirely apparent to me. As I made clear earlier, I have no problem withthe idea that we can formulate sentences that are agent-relative in the sense of beingspeaker-relative. Equally, I do not have any problem with propositions being either

14 See, for instance, Broome (2004) and Parfit’s ‘OnWhat Matters’ (forthcoming). Skorupski, in his turn,has suggested (in correspondence) that reasons might be understood as “facts plus modes of presentationthereof”.

15 The idea that it makes sense to establish a division like this between the non-obtainings and theobtainings of states of affairs raises difficult metaphysical questions that I will largely set aside. The notion thatwe need somehow to draw a line between these ‘entities’ seems obviously correct, as does the claim that theobtainings differ metaphysically from chairs, persons, and other concrete objects. Quite how we shouldaccount for the obtainings is another matter.

9.4 REASON-FOR 135

agent-relative or agent-neutral (i.e. not agent-relative); it is at least imaginable that weshould formulate propositions in two ways; those containing cross-references to theperson who expresses them, and those that do not. However, I do not think that thefact that we can draw these distinctions settles the question of whether there are anyreasons at all; nor do I think it settles the issue of whether there are any agent-neutral/relative reasons. One (but not the only) thing we need to do in order to settle theseissues is consider what makes such propositions true. My assumption is that this willhave to be a thicker entity, like an obtaining state of affairs, rather than a thin aspect(singled out by the proposition) of such an obtaining state. One reason I have forpreferring (i) to (ii) and (iii) is that it opens up a number of interesting questions aboutreasons. There is not space here to go into details. However, let me say this much. If weare suspicious about the idea that a statement expressing an alleged reason will in itselfsettle the question whether there are any agent-neutral/relative reasons, then consid-eration of the proposition expressed by this statement will hardly offer further assis-tance. And if the true proposition does not settle the question—“does the statementexpress a true normative reason?”—I do not see how the aspect, i.e. the truth-maker ofthe proposition, can do so. Something more is needed.

I have a further reason for choosing (i) over (iii). The notion that what makes a trueproposition agent-relative or agent-neutral is whether it contains a cross-reference tothe person for whom the true proposition is a reason seems to trivialize the distinction itis designed to illuminate. For instance, an agent-relative proposition that contains anindexical expression such as “my daughter” (e.g. “my daughter is drowning”) caneasily be transformed into what appears to be an agent-neutral proposition with somerigid designators. To save the distinction from this kind of trivialization, I suppose wecould say that even if the agent-relative proposition were transformed into an agent-neutral one, or vice versa, by employing or replacing rigid designators, it would still bepossible that only one of these propositions is a reason for the agent, so the distinction isnot trivialized. However, I must confess that I am not convinced by this reply.

My preference for a thick way of understanding facts has, then, to dowith the fact that Ifollow Joseph Raz in thinking that reasons have “vague and incomplete criteria ofidentity” (2006, p. 109). As an effect of this, reason-statements are more often than notvery incomplete descriptions (expressing incomplete propositions) that capture only somefeatures of the fact that constitutes the reason. I shall therefore assume that the examples ofreasons I give (or, more accurately, the reason-statements) are very probably incompletedescriptions. In a given situation there are almost certainly a number of facts that togethercall for a certain action. However, in discussing reasons we tend to single out only thesalient features of a certain situation as the reason. Amore detailed accountwould disclose amuchmore complex picture. Of course, it need not be like this, but in what follows I willset out from the idea that the full ‘identity of the reason’16 need not be disclosed by the

16 I use here ‘identity’ in Raz’s (2006) sense.


examples offered here. At the end of the chapter, when we discuss Dancy’s objections tothe notion of a complete reason, we shall have occasion to return to this matter.I now turn to the question more directly relevant to the view I am canvassing in this

work, namely: how readily normative reasons, qua facts, can be understood in terms ofthe agent-relative/neutral dichotomy. The idea that all reasons are reasons for an agent(see PEF above) does appear to create a problem for a view according to which reasonsare facts, i.e. obtainings of states of affairs.To examine this issue, recall first what it would be for a fact to be an agent-relative

reason in the essentialist view; it would be for some obtaining of a state of affairs to involvethe agent for whom the fact is a reason. Suppose next that a reason-statement Pr expressesthis state of affairs, and thatPr is true, so thatwhatPr says obtains does obtain.Moreover, thestate of affairs that Pr expresses is in some sense about the agent for whom it operates as areason—say, that the person who is drowning is my daughter. Here it is quite evidentwhich feature makes this fact into a reason for me: that it is my daughter who is in need ofhelp. An analogous reply does not workwhen it comes to agent-neutral reasons qua facts.On the essentialist approach, given the truth of an agent-neutral reason-statement, Pn,there is an obtaining of a state of affairs such that it is a reason for x.Moreover, it is a reasonfor x despite the fact that the state of affairs is not in anyway about x.17 The question ‘Whyis this fact a reason for x?’ cannot therefore be answered by pointing to some feature ofthe fact that even in a minimal sense concerns x. This is quite remarkable. The conclusionwe appear to be obliged to draw is that there is nothing about the fact thatmakes it a reasonfor x, but that the fact is nonetheless a reason for x.

9.5 Meeting the challengeThere is something troubling about this picture of normative agent-neutral reasons quafacts. But it might be insisted that this worry should not be exaggerated. Can we notsimply say that, in certain cases, the fact is that everyone has a reason to do the act, andthat since I fall under the thin descriptor ‘everyone’, I have a reason to do the act? Thisis an important question. In discussions of PEF and the challenge (PEFI) it poses tonormative agent-neutral reasons, it is often the concern people most readily press. If it isa fact that everyone has a reason to do some act, then I, like everyone else, have areason to do this act. But this reply is nonetheless question-begging; it will not meet thechallenge—at least, not if we are interested in the essentialist definition of reasons quafacts. If the fact that a person y is drowning gives me and everyone else a reason to savey, the reason I should save y is not that everyone has a reason to save y but rather that itis a fact that y is drowning. The crucial thing to realize here is that the number ofpeople who have this reason is not a reason-making feature. The fact that a person is in

17 Again, the reason it cannot contain a reference to the agent is that a reason-statement P expressing a factthat involves an agent is a true statement containing an essential reference to the agent; the statement is notlogically equivalent to any other statement that does not contain a reference to the very same agent.

9.5 MEETING THE CHALLENGE 137

need of help may well be such that it gives everyone a reason to perform the act. This isprecisely what is under discussion here. But if it is a reason for me, there must be somefeature of the fact that in a minimal sense concerns me—for its being a reason for me tosave y. Otherwise it would not be a reason for me. (I might believe that it is a reason forme, but we are not discussing motivating reasons or explanatory ones.)

I should add, though, that there might be other ways in which an agent could relateto a fact (or, say, a true proposition) that would explain why it (the fact or theproposition) is a reason for this agent. Suppose we agree that all reasons of the kindI am interested in here are reasons-for. It might be denied that a normative reason is areason for someone only if it involves, or refers to, someone. Michael Zimmerman(personal communication) has suggested that it seems sufficient that the reason shouldapply to the person for whom the fact is a reason. The agent does not need to be a partof the fact or in some way involved in the fact. It is enough that the fact applies to theperson for whom it is a reason. That the reason (i.e. the fact) applies to the personmeans, then, that the fact bears some kind of appropriate relation to the person. It doesnot have to be the case that the fact is about, or involves, the agent. On this suggestionwe would not find a reference to the person in the content of a reason-statementexpressing this fact. But it would still be the case that the reason was a reason for theperson, in virtue of the alleged relation between the person and the fact.

Zimmerman’s suggestion is appealing. In fact, I am ready to embrace it; there arecases in which a fact is a reason for us without our being involved in the fact. Recall thecase we discussed earlier in which it was a fact that everyone has a reason to perform someact. Certainly this fact can give you a reason, and so there is a sense in which the reasonapplies to you without it being the case that you are involved in the fact (i.e. a truedescription of the fact would not mention you). However, this is a peculiar kind ofcase. Such an assertion would naturally lead to the question: ‘What is this reason thateveryone has?’ And this suggests that we are dealing with what is best described as aderivative reason. There is some other reason on which the derivative reason depends.If the fact that everyone has a reason is a reason, it is so because of some other fact thatconstitutes a reason for everyone. But although Zimmerman has a point when it comesto derivative reasons I am not sure we can extrapolate and claim that this holds for non-derivative reasons as well. I, at least, am not sure what it means to be related to thismore basic sort of fact without it being the case that I am somehow related to theconstituents of the fact. Notice, too, that the suggestion cannot accommodate thenotion that the agent is somehow related to these constituents, since if there were sucha relation, a reason-statement expressing this fact would be agent-relative (i.e. areference to the person would be a part of the content of the reason-statement).

I could be wrong about all this. The matter is complicated. Perhaps Zimmerman’ssuggestion works for these non-derivative reasons, too. The explanation of why such areasonwould be a reason formewould then be that I am somehow appropriately related tothe fact. Suppose we set aside the question of what, exactly, is the nature of this relation;we charitably accept that there is such a relation. Will we then have met the challenge?


Perhaps. We can at least give some explanation as to why the reason is a reason for theagent; he or she is appropriately related to the reason. Whether this is a fully satisfactoryanswer is debatable, though. It depends, I suppose, on how one answers the furtherquestion: What is the underlying fact in virtue of which there is such an appropriaterelation between me and the fact that constitutes the reason? If this fact does not involveme as an agent, I suspect that there is something unsatisfactory about this sort of reply.Of course, one might regard things in a slightly different way. Reasons could be

taken to have two components. On such a view, my reason to � would consist of (P1)the fact that I am related in a certain way to fact (P2)—where (P2) is some other fact thatdoes not necessarily involve me. It seems reasonable to regard reasons in this way. Sucha view has the advantage that the reason now includes, or refers to, the agent. Apossible drawback is that this way of regarding reasons might easily be conceived of asmaking all reasons agent-relative. If so, we would not have met the challenge.These initial attempts to defend agent-neutral normative reasons did not in any

obvious way meet the challenge. However, there is a series of responses that we needto examine; and it is too early to make up our minds about the status of this challenge.Here, to begin with, are two replies that question the line of reasoning:

(i) Normative reasons need not be reasons for agents.(ii) An obtaining of a state of affairs may be about x and still be an agent-neutral

reason.

Now, (i) is a somewhat far-fetched response to the problem. It simply refuses to takeup the challenge presented by PEF. This is not an unexpected response. After all, I havenot offered any argument in support of the observation captured in PEF on which thechallenge sets out—at least, none other than that there is an obvious ring of plausibilityto PEF. (As is well known, one person’s modus ponens is another’s modus tollens.) Butperhaps there is, in effect, a suggestion that would help break this stalemate—and one,moreover, which would leave the distinction intact. It can be suggested that thequestion ‘Why is this fact a reason for x?’ is not always legitimate (as is it would be ifall reasons are reasons-for). Agent-neutral reasons are not reasons-for, but ‘merely’reasons. Accordingly, it would be to misunderstand these reasons to ask: ‘What is itabout this fact that makes it into a reason for x?’ By analogy with the core distinction inthis work between good, period and good-for, and value, period and value-for, reasons mightjust be reasons, period, but not reasons-for.As I have mentioned, there is something tempting about this response. However, it

would be a mistake to pursue it. For further thought reveals, I think, that this ‘analogy-reply’ is quite unconvincing. Reasons are not like values in the suggested sense. Quafacts, they do not just ‘call for’ an action, or make it eligible, but call for an action bythose who are able to perform the action. The idea that reasons can be altogetheragent-unrelated is mystical. If a fact is a reason, it is a reason for at least one agent. Thatsomething might merely be a reason, period, but not for any present, past, or futureagent, is, to say the least, an awkward idea.

9.5 MEETING THE CHALLENGE 139

Although the analogy objection is unconvincing, there might be another way ofsaving the first kind of response to the challenge. I will come to this saviour in amoment, but first, a word on this contrast between values and reasons.

It might be argued that if there is this dissimilarity between values and reasons, afitting attitude analysis of value cannot be correct. In reducing values to reasonssomething is lost—something, that is to say, that would account for this unlikeness.This is very probably true. But in my view this value-reason gap constitutes no realobstacle to the analysis. It would be naive, and quite futile, to expect there to be aperfect match between the associations we obtain from our value notions and those weobtain from our normative notions. The important thing, from the point of theanalysis, is that there is match between the significant features. That they are in somesense mystical entities is not, as far as I can see, an important feature of values.

9.6 An overcrowded boatThere may be another way of saving the first kind of response to the challenge. It mightbe that reasons after all need not be reasons for particular agents. The question iswhether there is not a rather obvious objection to my reply above (that there cannot bereasons that are not somehow reasons for certain persons). Wlodek Rabinowicz, forinstance, has suggested (in personal communication) that there might be examples thatunderwrite the claim that an action ought to be done by someone without under-writing the claim that it ought to be done by anyone in particular. For instance, if thereare too many people in a boat, someone ought to leave it. This seems to be an examplein which there is a reason for someone to act that is not a reason for any particularperson to act. This does not mean that everyone has a reason to leave the boat; it meansprecisely that someone but not everyone has a reason to leave the boat.

The example is indeed interesting. But what it shows is quite an open matter. Forinstance, even if we were to accept that it highlights a reason for someone that is not areason for any particular person, this would not automatically qualify the reason inquestion as an agent-neutral reason—at any rate, on the essentialist approach. Here iswhy: the statement expressing the fact that in a boat with persons x, y, and z, someonehas to leave, will not be logically equivalent with any other statement that does notmention x, y, and z. What we seem to have here, then, is a peculiar case of the agent-relative reason: one which is a reason for someone, but not for any particular person.

Advocates of normative agent-neutral reasons should not draw encouragement fromthe existence of peculiar agent-relative reasons for action. In fact, even these reasonsmight be questioned. For it appears to be quite counter-intuitive, as Rabinowicz hassuggested (personal communication), to regard the boat example as being, in the firstplace, a scenario in which there is reason to act. Consider again the details of the case.That the boat is overcrowded is a reason for there being someone who leaves the boat.However, this is hardly describable as a reason for an act. A reason for there being someoneis not about action, even if leaving the boat is. It is therefore unclear just what this


example shows. If the fact about the boat is a reason, it seemsmost plausible to regard it asa reason that is different in kind from reasons for action. Accordingly, if there are suchreasons, we cannot exclude the possibility that there are non-agent-relative reasons,even if we conclude, eventually, that all reasons for action are agent-relative.

9.7 More replies to the challengeIt is always logically legitimate to ask what it is about some fact that makes it intosomeone’s reason for action. And in the case of a so-called agent-neutral reason, the testis how does statement P, which allegedly contains no reference of any sort to x, expressa reason for a to�? What is it about the expressed fact that explains why it is a reason forx? What is it that the ‘for’ in ‘This fact is a reason for x to �’ signals, with regard to theagent, when it comes to agent-neutral reasons? Response (i) should be rejected.Acceptance of it is tantamount to an admission that reasons for action can, quiteliterally, be nobody’s reason, which is incomprehensible.The next response, (ii), is quite different from the first. It acknowledges the PEF

feature and does not deny that reasons are always reasons for an agent. However, itsuggests that the step from ‘this fact is (among other things) about x’ to ‘this fact cannotbe an agent-neutral reason’ is invalid. Facts can be about a certain agent, and nonethe-less be an agent-neutral reason for this very agent.There might be something to this reply. However, as far as the essentialist approach

goes, (ii) appears to be a non-starter. As might be recalled, it follows from ourinterpretation of ‘statement that essentially refers to an agent’ that if a fact is a reason,and if it is also, among other things, about person x (whether or not it is also aboutevery other agent y, z, . . . ), then the statement expressing the state of affairs P will notbe logically equivalent to some other statementQ that does not refer to x. Hence Pwillcontain an essential reference to x. So P will express an agent-relative reason for x. Inorder words, if reply (ii) is to be successful, it must provide a different interpretation ofthe dichotomy than the essentialist one.I would like to emphasize that I am certainly open to the idea that within what I

have referred to as agent-relative reasons, we should be able to distinguish betweendifferent sorts of reason, some of which might be referred to, in a sense, as agent-neutralreasons. For instance, Portmore has recently suggested (personal communication) away of understanding the dichotomy that preserves, he thinks, the philosophicalimportance. I am inclined to agree. Consider the following two examples:

AR: Person x doing � in c would ensure that Jones saves x’s child.AN: Person x doing � in c would ensure that Jones saves Jones’s child.

These counter-factuals involve the agent for whom they are a reason, and so, asPortmore pointed out, the personalizability feature has been accounted for; AR is still ina sense agent-relative in a way that AN is not. I think Portmore is right about this.

9.7 MORE REPL IES TO THE CHALLENGE 141

However, I also think his distinction certainly falls within what I have referred to asagent-relative reasons. These subjunctive facts involve the agent for whom it is a reason.

However, let us not rush things. It may be pointed out that not all features of a fact areequally relevant. So although a fact may to some degree be about the person for whom itis a reason, this agent-relative feature is not a reason-making property; and accordinglythe statement expressing the fact need not mention this agent-relative feature.18

There is something in this reply. In a while I will also consider a version of it thatperhaps constitutes the most promising way of meeting the challenge set by PEF. Butspeaking more generally this reply is far from compelling; it tries to solve a problem ofhow to clarify a conceptual issue by placing a substantive requirement on, in this case,what features of facts may or may not be relevant reason-providers. At best it will onlyconvince people who share the same normative outlook, i.e. those who agree to thesame normative statements. As far as possible, answers to formal questions should notdepend on substantive views.

If all reasons are reasons for someone, the PEF challenge is, in other words, to findwhat it is about the fact that explains why it is a reason for the agent if this fact does notinvolve the agent. If there is no such feature, it will be impossible to explain why thefact is a reason for the agent. Of course, we might not know, or we might be unsure,what the feature is. We might also question the notion that the fact is, in effect, a reasonin the first place. But this is not what is at issue here. The question is: Why is this fact areason for x? To maintain that a fact which does not involve person x might still bereason for x is to admit that there is no explanation of why the reason is a reason for x.That is, if x is not involved in the fact, it becomes inexplicable that the fact is a reasonfor x.

Perhaps we should accept that certain facts are, entirely inexplicably, reasons forcertain agents. Even if there is nothing about the fact that in the slightest way involvesthe agent, it might still be a reason for her to do something. That would be a case of aninexplicable reason. It could then be said that the set of inexplicable reasons coincideswith the set of all non-agent-relative reasons. This is a tricky manoeuvre to deal with. Ifyou are ready to accept the existence of inexplicable reasons, the PEF challenge canperhaps be met. But the price of this is high. Another possibility—as I will argue inSection 9.8—is to maintain that all reasons are agent-relative reasons.

But there are other replies to the challenge. Could it, perhaps, rest on an implausibleinterpretation of the expression ‘reason for x to�’? Surely something could be a reasonfor x to �, even if the reason failed to ‘get a grip on’ x? This reply implies thatsomething could be an agent-neutral reason for x to� despite the fact that x would not

18 As I understand Portmore’s suggestion, this is not something he would subscribe to. That the agentensures something is essential to his view of what makes something into a reason. In my view, Portmore’sinterpretation involves the wrong sort of reason-makers. For instance, in his account, the fact that mydaughter is drowning is not a reason for me to jump into the water. Nor is the fact that she is drowning and that Ican save her a reason for me to jump into the water. Rather, the reason I should jump into the water is thesubjunctive fact that my jumping into the water would ensure that she is or will be saved.


� even if she were aware of that reason. On such occasions the reason would still be areason for x.This reply leaves the challenge intact, however, since it is based on a misunderstand-

ing. That something is a reason for x does not, as I pointed out above, require x to besomehow aware of x. In fact, that some fact F is a normative reason for x to � is notconditional on its being the case that x would � if she believed that F. This might betrue of some facts, but it would be implausible to maintain that it is always the case.Sometimes two persons will have the same beliefs about what the facts are and still nottake the same normative views. More than that: even if I know a certain fact, and knowthis fact is a reason for me to �, I might still not �. Moreover, it is generally acceptedthat many facts are unknown to all of us. In the light of this it is reasonable to expectthere to be many normative reasons that are unknown to all of us. That reasons arealways reasons for x should not, in other words, be taken to imply that x mustsomehow be aware of, or even respond to, the reason. The expression ‘reason for’establishes that there must be something that relates the fact to the person; but therelation in question need not be epistemic. Once there is no relation at all between theagent and the fact, the suggestion that the fact is a reason for x becomes incomprehen-sible. If this contention is correct, there is one sense of ‘reason for x’ that appears to bedifficult to understand in terms of a normative agent-neutral reasons.But why not go straight to the point and say that some facts essentially involve

certain individuals, while others only contingently involve people? This looks like astraightforward solution. But it remains to be shown what sense can be given to thelocution ‘this obtaining of state of affairs contingently involves x’. Metaphysicians tendto agree that propositional (abstract) entities like states of affairs have all their elementsnecessarily. The same holds, I would say, about obtainings of states of affairs; they toohave their constituents necessarily.There is the possibility that the challenge sets out from a narrow view of acts.

Suppose (a) ‘helping x’ is one kind of act, and (b) ‘helping x if you can’ is a differentkind of act. The reason for performing (a) must somehow be a fact that among otherthings involves one’s being able to help x. Thus, the fact that x is in need of help is anincomplete reason for performing the act in (a). The fact that x is in need, together withthe fact that you can help x, is a reason for you to perform the act mentioned in (a).However, the former fact is not obviously incomplete when it comes to the act in (b).That is, it might be argued that the fact that x is in need is sufficient to provide a reasonfor performing the act in (b). Hence, there seem to be facts that are reasons which donot necessarily involve the person who has a reason to act.This is a quite strained reply. First, it sets out from a highly implausible view of what

acts are. In what sense does (b) refer to an act? There is no obvious answer to thisquestion. Certainly, it is one thing to act on a condition, but so acting is not necessarilydoing something different from what one does when one acts, but not on a condition.Sowhat this reply in effect shows is that reasonsmight perhaps apply to things other thanacts and persons, and that there might in fact be agent-neutral reasons as long as they are

9.7 MORE REPL IES TO THE CHALLENGE 143

not reasons for acts, but for other things (such as acting on a condition). The discussionabove of the overcrowded boat has already suggested something along these lines.

9.8 A positive argumentFor all its simplicity, the challenge turns out to be a hard nut to crack. It sets off fromtwo ideas: first, that claims such as ‘This is a reason’ and ‘The reason is . . . ’ are, at leastin the case of normative reasons, elliptical—they are all shorthand for ‘This is a reasonfor a to �’ or ‘The reason for a to F is . . . ’. Second, if normative reasons are facts, thosefacts that are reasons must somehow reflect or explain this ‘for-a’ feature of reasons,since otherwise it would be inexplicable why those facts constitute reasons for theagent. The challenge for adherents of agent-neutral reasons is, then, to find a fact thatdoes this without involving the person for whom it is a reason.

Our consideration of the challenge to this point has not established that it cannotbe met. In fact, a further reply quite correctly points out that since all that hasbeen shown so far is that certain replies do not work, caution is called for. This is, ofcourse, a sensible warning. However, there is also a strong positive argument (alreadyinvoked more than once in this work) for the claim that all normative reasons are agent-relative. It centres on a generally accepted feature of practical reasons: a feature that canbe captured, in a rather rough and ready way, in the ‘Reason-Implies-Ability’ principle:

(RIA) There is reason for a to � ! it is possible for a to �.

Suppose, then, it is true that to have a reason to � you must be able (physically andperhaps psychologically) to �. Furthermore, suppose that this ‘ability constraint’ onreasons applies to all practical reasons. This restricts in its turn the kinds of fact thatconstitute reasons: fact F is a reason for a to � only if F involves in part a’s ability to �.A full description of the fact (that is a reason) would, in other words, mention that theagent is able to act.

So if RIA does indeed identify a formal feature of reasons (or more accurately ofreason-statements), it seems a fortiori to strengthen the claim that there are no agent-neutral normative reasons. (The cautious ‘it seems’ claim is warranted, as we shall see ina moment.) All normative reasons are agent-relative.

It might be argued, perhaps, that the ‘ability constraint’ is an uninteresting, trivial, orin some sense disregardable feature of reasons; and that we may, or should, dividereasons into different groups without paying any attention to this shared feature. Thismay be right, but it does not change the fact that these groups would consist exclusivelyof agent-relative reasons; you cannot uphold the reason dichotomy discussed here ifyou believe in the doctrine that practical reasons imply the ability to act. Or so it seems.

Dancy’s intriguing idea, set out in Ethics without Principles (2004), that we shoulddistinguish between reason considerations that are “favourers/disfavourers” and thosethat are “enablers/disablers”19 suggests how the positive argument might be refuted.

19 See Dancy (2004). In this work, Dancy also discusses a third role, namely intensifiers/attenuators.However, to outline the general idea, I need only focus on the favourer/enabler distinction.


Favourers, we are told, are features (facts) that count in favour of acting in one way oranother. This cannot be said about enablers, though; they do not favour acting. Instead,what they do is to clear the ground for the favourer—they make it possible for a certainfeature to count in favour of acting. Thus, the idea is that the enabler allows the favourerto do its job. Here is one of Dancy’s own examples: “(1) I promised to do x, (2) mypromise was not given under duress, (3) I am able to do it, (4) There is no greater reasonnot to do it”. Of these, (1) is a favourer and (2–4) are different kinds of enabler.

The distinction between favourers and enablersmight well come in handy. It suggests away of meeting the PEF challenge: even if reasons imply ability, it does not follow that allreasons are agent-relative. We could draw a line between relative and neutral as follows:agent-neutral reasons consist of two kinds of fact: (i) the favourer, which is a fact that doesnot involve the agent (for whom the fact is a reason); and (ii) the enabler, which is the factthat the agent is able to perform the act in question. This is certainly a plausible response tothe positive argument. Agent-neutral reasons are favourers of a certain kind (they do notinvolve the agent). They have as enablers the ability of the agent, but the fact that the agentcan do something is not part of the favourer.Just how Dancy’s distinction should be understood in more detail is still very much

discussed. Raz, for instance, finds “the category of being an enabler . . . so diverse as tobe of little use, and likely to mislead” (2006, p. 106), More generally, Raz detects anumber of tensions in the distinction between favourers and enablers.20

Admittedly, Dancy’s distinction is fundamentally phenomenological, and so itshould come as no surprise that it gives rise to interpretative tensions. This notwith-standing, as far as I can see the idea that not all ‘reasons features’ play the same role pavesthe way for a defence of the neutral/relative dichotomy which is by far the mostpromising I am familiar with. However, I want to voice two worries right away. Thefirst has to do with Dancy’s view that all reasons are contributory reasons. I am inclinedto believe that there is at least another kind of reason, namely something we can referto as a complete reason; and I suspect that such (all facts inclusive) reasons are still open tothe challenge. Dancy comments on an idea closely related to the notion of a completereason. This is W. D. Ross’s view that “duty is toti-resultant”. He says: “Ross lumps allrelevant features together, as part of the right-making base (the ground, as one mightcall it), and this metaphysical picture of the situation is far too indiscriminate” (2004,p. 45; see also pp. 91–2).21 Here, Dancy’s choice of the term ‘indiscriminate’ is

20 Dancy’s account of favourers and enablers raises interpretative issues that I will not go into here. Razconsiders four interpretations. His final one makes room for a notion of a favourer that does not actuallyfavour anything (since there is no enabler present); see Raz (2006, p. 105). He might be right that there istextual evidence for this in Dancy’s book. However, as I read Dancy, the important part of his idea is thatsome features favour on condition, and that it is these features that are reasons, again on condition.

21 See also Dancy (2004, p. 39), where a related idea is discussed—namely, that the real favourer is acomplex consisting of features (1) and (2) (see his example). Dancy rejects this idea on intuitive grounds. Forcriticism of this defence, see Raz (2006) and Strandberg (2008, especially Section 7).

9.8 A POSIT IVE ARGUMENT 145

noteworthy. As indiscriminacy is not a feature of the world but more accurately of howit is described, Dancy is probably concerned with descriptions or explanations ofreasons. His objection to Ross’s idea does not, therefore, bear immediately on thenotion of reason I have been examining—the notion of reasons qua facts. We shouldcertainly agree that it is not very helpful to start enumerating all the relevant facts if wewant to explain, cite, or, more generally, describe a reason to someone. But this is quiteconsistent with the idea that there is one and only one thing I have a reason to do at t1,and that this is determined by how the world precisely is at t1.

Not that we ever aspire, or should aspire, to take into consideration all there is toknow about the world in the moment we think there is a reason to do something.Surely that would be preposterous. The world is epistemically muddy. Our knowledgeof, say, what our actions will result in is fallible. But if, at any given moment at whichwe are in the presence of a reason, there is a set (very probably limitless) of relevantfacts, and if it is true that what we have reason to do is, at least sometimes, determinednot by a single fact but a number of facts, then it is hard to see how we can rule out thenotion of a complete reason on purely conceptual grounds. That we cannot aspire toknow what facts constitute a given complete reason is something we have to live with;it does not render the notion of a complete reason unusable.

But, it might be objected, even if there are complete reasons in this sense, could westill not differentiate between complete reasons whose favourer(s) do not involve thereason holder and those that do? And so could we not just say, as before, that the firstkind is an agent-neutral reason and the second an agent-relative one?

This response is inadequately thought through. A complete reason is composed by allthe relevant facts, and on their own these do not constitute a favourer. At any rate, they donot do so inwhat I believe to beDancy’s sense, namely a favourer on condition. But,moreimportantly, even supposing that the complex fact constituted some sort of favourer, sucha favourer (what might be referred to as a complete or unconditional favourer) might wellbe constituted by facts which, taken on their own, constitute contributory reasons thatfavour things other than, or evendisfavour, the act that the complete favourer (reason) callsfor. The correspondence between the twokinds of favourer cannot, at least, be assumed tobe necessary without argument. It is therefore not obvious how the notion of a favoureron condition applies to complete reasons: Dancy’s favourer/enabler distinction is not veryhelpful, then, when it comes to one kind of reason.

We might tell apart those features that are favourers and those that are enablers,which together constitute a complete reason. However, it is not obvious that we canuse this to differentiate between agent-neutral and agent-relative complete reasons. Itremains unclear, at least, what the precise relationship is between these favourers/enablers that constitute different contributory reasons and the complete reason. Soeven if a given complete reason will, in a sense, encompass, say, a number of differentfavourers (in Dancy’s sense), some of which will involve the agent and some which willnot, it is far from clear what the relationship is between these favourers and the


complete reason. Hence, it remains obscure why the nature of a complete reasonshould eventually be determined by one favourer rather than another.This point needs to be amplified a little, however. We should first distinguish

between two kinds of complete reason, both of which can be characterized as reasonsthat are individuated (and thus constituted) by the facts at a certain time. Thus, it mightbe that the complete reason at t1 is (i) constituted by all the obtainings of states of affairsat t1. This would be one way of conceiving what is involved in the exemplification of acomplete reason. However, it is hard to think that there would ever be such a completereason.22 Among other things, it would mean that, at a given time, all facts wererelevant, which is hardly imaginable. A more realistic idea is that a complete reason is(ii) constituted only by all the relevant facts. However, since we are dealing with thenotion of a complete reason here, one of these facts would have to be somewhatpeculiar, namely the fact that all other facts are irrelevant, i.e. have nothing to do with thereason. The justification for this proviso is then the following observation: a reason at t1

would hardly exemplify the notion of a complete reason qua the facts if there werefurther facts that might alter, or influence, what we have a reason to do at t1.A complete reason of the ‘all the relevant facts’ type, or of the former, more

implausible, (i)-kind, should not be understood as the reason that results from weighingdifferent (contributory) reasons against each other. Rather, the idea is that the obtain-ing of the relevant states of affairs at a given time together equally constitutes the reason.Here is an analogy that might be helpful. A jigsaw puzzle J1 (depicting, say, some

prescriptive instruction) consists of equally shaped pieces that together constitute animage. Another jigsaw puzzle J2 (also depicting a prescriptive instruction) is not reallyfinished—there remain some pieces that we do not have access to. Still, in its presentstate J2 contains enough pieces to give us a good idea of what it would look like (or ofwhat is prescribed). However, J2 is made up of pieces all of which are unique in shape.Since the pieces are different, we might want to suggest that some pieces play roles thatother pieces do not. For example, we might suggest that some are better to begin withthan others: corners, for instance, can easily be spotted in the box at the outset. It wouldnot make sense to say this about J1, whose pieces are uniformly shaped.Now althoughwe never as a matter of fact seem to be in a position to say with certainty

what a complete reason is, the notion of a complete reason is not pointless. For instance, itmaywell bewhat best explains our motivation to look formore facts evenwhenwe thinkwe have a contributory reason for doing a certain thing. We are not—at least, notalways—just looking for yet another contributory reason. We are looking for (whatbest approximates to) the reason, i.e. the reason constituted by the complete picture.Now, my first worry is that such a complete reason squares badly with the idea that

reasons are favourers on condition. Each fact does its equal share, as it were, of thereason-constitutive work (witness puzzle J1). The explanation of why it does not seem

22 Notice that nothing hinges on the idea that reasons are thick entities; there is an analogous problemwith complete propositions.


to be a favourer on condition is that a complete reason does not in any obvious sensehave (or more accurately, admit of having) an enabler. The notion of a complete reasonoutlined here entails that any fact will be either (1) such that, together with all otherfacts, it equally constitutes the reason, or (2) such that it is either relevant or irrelevant.We are dealing with (1) if ‘complete reason’ is understood in the unrealistic senseimplying that all facts are relevant, and we are dealing with (2) if we read ‘completereason’ in the sense of (ii). Since it is not obvious how one and the same feature canbe constitutive of a reason and an enabler for this very same reason, there are grounds tobe concerned that this sort of reason-notion is still open to the challenge set by PEF.Dancy seems to recognize something to this effect:23 “Walter Sinnott-Armstrongsuggested to me that, in a case where the mere ability to act is a reason, it is also anenabler for itself. I see no reason to deny this amusing possibility.”

I am not quite sure what to make of Dancy’s admission (see footnote 24, though). Ifwe take facts to be ontological entities in their own right, it is hard to see how one andthe same ‘thing’ can be its own (condition or) enabler. Metaphysics apart, this footnoteof Dancy’s is interesting for another reason; if ability can be a favourer (on condition) aswell as an enabler, we cannot meet the challenge we are considering by claiming thatabilities appear only as enablers.

Now, Dancy discusses at length a number of reason notions that differ from his own‘contributory reasons’. For instance, Dancy objects to Roger Crisp’s notion of an‘ultimate reason’. Briefly, he objects to what he thinks is the idea underlying Crisp’snotion of a full explanation. The notion of a complete reason qua fact that I haveoutlined is not, in contrast with ‘explanation’, a success notion; so Dancy’s argumentagainst Crisp does not in any obvious way affect the notion I am deploying.

Dancy’s discussion of Raz’s notion of a complete reason is of particular interest:

[o]n Raz’s account of a complete reason, whose parts (which he thinks of as premises, and whichI would think of as reasons, some of them) are not reasons, it will be true that whatever is a reasonis always a reason, and always on the same side. (2004, p. 97)

Dancy takes Raz to endorse what he calls “cluster atomism”, and it is really this ideathat is bothering Dancy. A cluster atomist not only regards (at least) some reasons asclusters of features, but also maintains that “a cluster that plays a certain role in one case,must play that role wherever it appears” (p. 97). This means, then, that if something is areason for x in a certain situation, it will be a reason for x in any situation. This Dancytakes to be inconsistent with holism (his own view of reasons). Holism is the idea that afeature that is a reason to act in one case may be no reason, or even a reason againstacting, in another case.

Now, the atomism–holism issue is important. However, the belief that there arecomplete reasons in the sense I have outlined here commits one to neither atomism norholism. In fact it can quite plausibly be argued that the issue does not arise in the first

23 See Dancy (2004, p. 40, n. 1).


place. The argument proceeds on the basis that what is a complete reason at one pointin time will not be a complete reason at any other time (given the assumption, hard todeny, that the set of facts changes over time). In my view we should follow Dancy andreject atomism. That is, I share Dancy’s view that one should reject atomism when itcomes to contributory reasons. But not all reasons seem to be contributory. As far asI can determine, Dancy’s objections to complete reasons do not affect an idea such asthe one outlined above.The distinction between favourers and enablers is important. But even if we accept it,

which I think we should, we need not accept that all reasons are favourers on a condition.We must make room for the possibility of a non-contributory kind of reason, namely acomplete reason. Since such a reason cannot be depicted as a favourer on condition, wecannot avail ourselves of Dancy’s distinction to meet the PEF challenge.The categories ‘atomism’ and ‘holism’ apply more naturally to operative reasons, i.e.

to what we believe to be reasons. This brings me to my second worry. The anxietyI have here concerns an idea that several writers have had, namely that Dancy’sdistinction is fundamentally a pragmatic, rather than metaphysical, one. Favourersand enablers are notions we employ when we cite, or explain, what we believe tobe our reasons. Something that is a favourer in one context might therefore be anenabler in another, depending on what we are interested in emphasizing.24 Raz isready to read Dancy along these lines. Consider, for instance, the following:

[Dancy] protests against the claim made by Crisp (2000b, p. 44), that citing a favourer may begood enough an explanation of one’s reason for acting as one did (Dancy, 2004, pp. 47–8, 95–7).That seems to me true since explanations can have different objectives as well as different objects.Not infrequently citing the so-called favourer is, given one’s interlocutor’s interest, the bestexplanation of one’s reason. But this only illustrates the importance of distinguishing between theexplanation of a reason and the reason itself. The notion of a complete explanation probably doesnot make sense. There can always be additional puzzles calling for additional explanations. It doesnot follow that the notion of a complete reason does not make sense. One should not concludefrom the fact that human questions have no end that how things are in the world is equally openended and in the same way. (Raz, 2006, p. 110; my italics)25

24 Recall Dancy’s claim that one and the same feature might be a favourer on condition (reason) and itsown enabler. Perhaps this should be understood as follows: there are different ways of bringing out what issalient about the ability feature. Given the context, you might want to stress to someone that ability is whatfavours doing the act, but you might also point out that ability is a condition of doing the act. We are, in otherwords, talking about the feature in different ways. However, the fact that a person is able to do somethingremains the same fact whether or not we invoke it as a favourer or an enabler.

25 See also Caj Strandberg (2008), who gives such a pragmatic account of the distinction between “whatmakes objects have moral properties and enablers” (p. 150). Moreover, Strandberg thinks his distinction isgeneralizable to Dancy’s related distinction between favourers and enablers. He cites the following works asputting forward related suggestions: Raz (2000, p. 59), Broome (2004, pp. 32–5), and McKeever and Ridge,(2006, pp. 72–5). That a feature may be a favourer in one case and an enabler in another is something thatDancy himself points out (see his discussion of the ability to raise one’s arm, p. 40; see also n. 1). He also claimsthat “it is easy to find examples of cases where it is not clear which side of the favouring/enabling distinction agiven feature is to fall” (p. 51).


To the extent that there is something in these suggestions, we should take thedistinction between favourers and enablers to concern what reason considerations cando rather than what facts (features) can do.26 The distinction has more to do with howwe explain (our) reasons than it has with what (our) normative reasons are. What anexplanation will look like, in terms of favourers and enablers, might vary depending,say, on the person to whom the explanation is directed (e.g. a child or an adult; seeRaz, 2006, p.101). Since the challenge concerns what reasons are, rather than how weexplain them, it might be argued that Dancy’s distinction does not help us defend thedichotomy of agent-neutral and agent-relative reasons. So if these suggestions arecorrect, the favourer/enabler distinction lacks any metaphysical basis; hence, “enabil-ity” and “favourability” are probably no more than features we ascribe, or decline toascribe, to facts, depending on what appears to us as salient given our desires andinterests at the time. These ascribed features are, therefore, better characterized asbelonging to operative (normative) reasons, rather than to normative reasons qua factsthat are independent of our beliefs.

I confess I remain uncertain as to whether Dancy’s distinction is, or more impor-tantly must be, contextual in this sense. Perhaps it need not be. However, since it isdifficult to see what would settle this issue, it is surely best to adopt a cautious attitudeto the question whether Dancy’s distinction can be used to defend the dichotomy. Thisnotwithstanding, Dancy has supplied us with the tools needed to illuminate thatquestion.

The distinction between agent-relative and agent-neutral reasons attracts as muchphilosophical attention as it does in part, of course, because it involves several more orless complicated issues. Thus, there are:

1. Definitional Question (I): Is it possible to draw a distinction between agent-relative and agent-neutral reason-statements?

2. Definitional Question (II): Is it possible to draw a distinction between agent-relative and agent-neutral reasons?

3. Metaphysical Question: Are there metaphysical grounds for rejecting the dis-tinction?

4. Normative Question: Are there any agent-neutral or agent-relative reasons at all?27

The first two questions need to be held apart. Albeit that they are intimately related,they are not the same. From the fact that we have linguistic means to formulatestatements that are agent-relativized, it does not follow that we have to answer thesecond definitional question in the affirmative. And it is equally important to see that,even if we answer ‘yes’ to the second question, it might still be the case that themetaphysical underpinnings of this distinction are too obscure for us to identify ground

26 It should be mentioned that Dancy (2004) discusses the distinction between favouring/enabling/intensifying in terms of the different things that “relevant considerations can do” (p. 41).

27 I am grateful to Skorupski (personal communication) for reminding me of some of these questions.


on which we can draw the distinction in an adequately clear way. Finally, even if wehave no metaphysical objections to the distinction, we might insist that there are noagent-relative reasons—or, for that matter, agent-neutral reasons. We might well rejectone kind of reason on purely evaluative grounds.This chapter has drawn attention to Definitional Question (II), and in particular to a

feature of reasons that is an obstacle to the widely held view that normative agent-neutral reasons are facts.28 There is a general objection to my approach, however. Doesnot my insistence that all reasons are reasons for agents in effect make the questionwhether all reasons are agent-relative analytically true? And if that is indeed the case,does it not follow that I am not really in the business of arguing for, or clarifying,something? Am I not merely stipulating a definition?Surely it can be quite illuminating to draw attention to a feature of a distinction,

analytic or not. Hence I do not necessarily see myself as doing something as philo-sophically unexciting as solving an issue by definition. Rather, I see myself as under-lining what follows from a much neglected feature of all reasons.I have not established that there cannot be normative agent-neutral reasons, but it

ought to be clear that the notion of a normative agent-neutral reason is undermined bycertain views of reasons qua thick facts. Whether, in the end, this is a problem for theessentialist approach to the dichotomy, the definition of normative reasons, or forthe notion of agent-neutral normative reasons for action, is a further issue. It dependsin part on one’s attitude to the idea that reasons presuppose an ability to act. However,if you do not take Dancy’s distinction between favourers and enablers to solve theissue, it is bound to seem that reasons, qua facts, are all agent-relative—on theessentialist approach, at any rate.

28 See, for instance, Christine Korsgaard, ‘The Reasons We Can Share: An Attack on the Distinctionbetween Agent-relative and Agent-neutral Values’ (1993). She expresses doubts about the distinctionconnected principally with the Metaphysical Question and the Normative Question.


10

Value Bearers and Value Pluralism

The preceding chapter argued that we should be cautious about the dichotomy ofagent-relative and agent-neutral normative reasons. There are good grounds to believethat reasons are, by their nature, always agent-relative. Accordingly, there are goodgrounds to be cautious about understanding the distinction between personal andimpersonal values in terms of the dichotomy between agent-relative and agent-neutralnormative reasons. As we saw, if every reason is a reason for someone, PEFI (i.e. thepersonalizability feature implication) defies us to say what it is about a certain fact (or setof facts) that gives the agent a reason to act—if, that is, the fact does not involve theagent at all. If there is no such feature, the fact’s being a reason for the agent appears tobe inexplicable. Of course, we might not know, or we might be unsure, what therelevant feature is. We might also question whether the fact is in effect a reason in thefirst place. We might be speaking about operative reasons rather than normativereasons, or we may have theoretical rather than practical reasons in mind. Again, wemight not have distinguished between reason statements and true reason statements, orbetween reason statements and reasons, or we might have confused motivating reasonswith normative reasons. But if we are clear about these matters and have in mindnormative reasons, qua facts, then for someone taking an essentialist approach to thedichotomy the question is the following: Why is an allegedly agent-neutral reason areason for person a? To maintain that a fact that does not involve a might still be areason for a is to admit that there is no explanation of why the reason is a reason for a.That is, if the reason-constitutive fact in no way involves (or even relates to) a, itbecomes deeply puzzling that the fact is a reason for a. Jonathan Dancy’s distinctionbetween enablers and favourers provides some hope when it comes to contributoryreasons, but FA (i.e. fitting-attitude) analyses are not confined to such reasons. In fact, ifit makes sense to distinguish between complete and contributory reasons, the FAanalysis is naturally understood as being primarily about complete reasons.

The notion of a complete reason is not that straightforward, and so Dancy might beright in his criticism. Nor is the distinction between enablers and favourers clear-cut.However, even if we believe there is a metaphysical basis for this distinction, it remains

to be shown just how it provides us with a satisfying account of the relative/neutraldichotomy. Suppose it cannot do this, and that I am right that the notion of a reason is,in the way I have outlined, such as to necessarily introduce a relation to agents. In thatcase, on the FA pattern of value analysis we obviously need to differentiate betweenimpersonal and personal values in some way other than by employing the distinction inquestion.One important premise in this book is the idea that a formal analysis of value should

be accessible to as many plausible substantive value theories as possible. Two suchtheories are hedonism and preferentialism. Since these positions do not ascribe, or soI will argue in this chapter, non-derivative value to the same kind of value bearer, wehave reason to endorse a formal view of value that is consistent with value-bearerdualism. That is, given certain plausible assumptions about the nature of pleasure andpreference, the value accruing to pleasure (from a hedonist point of view) cannot beaccounted for by preferentialists. The two positions are most plausibly regarded asascribing non-derived final value to different kinds of value bearer (even if they ascribederived final value to the same sort of bearer).Second, the question of howmany kinds of bearers of value there are is a challenging

one, so in the remaining sections I discuss in more detail both monist and pluralistarguments, concluding that a pluralist position is preferable.1 Moreover, since mydiscussion in this chapter will be about final values in general, it should be recalledthat personal as well as impersonal values may be final values.

10.1 Extrinsic final valuesIn Chapter 1 I briefly considered some examples that are inconsistent with G. E.Moore’s views on intrinsic value. An object that is valuable because it is rare or uniqueis a good example. Rarity and uniqueness are externally relational properties and wouldnot be admitted in anMoorean supervenience base which only admits the value bearer’sinternal features. Many other objects seem to have a value in virtue of some externalrelational property. Objects related to persons of historical importance appear to carryimpersonal final value; the house where I have lived for more than twenty years is anexample of a personal extrinsic final value. It seems easy to come upwith an object, event,or person that is valued for its own sake because it has certain relational features.There is a somewhat different kind of extrinsic value, i.e. the kind of value that is

somehow conditional on something else, which I need to set aside here, although it doesseem to me to raise important issues. For instance, Dancy (2004, p. 172) has offered thepregnant example of a joke that is funny only when the butt of the joke is present; andDan Egonsson considers the beauty of a motorcycle he owns that he thinks is beautiful

1 In doing so I will unfortunately have to cover quite a bit of ground rapidly. Since there are a lot of tigersin the area, I realize I cannot reasonably expect not to be accused of excessive speed and inadequate care.

10.1 EXTRINS IC F INAL VALUES 153

“on condition of its having an instrumental value” (i.e. that it works).2 Shelly Kagan’sexample3 of a fast racing car seems also to belong to this group of examples.

One reason for resisting the notion of extrinsicfinal value is that it does not combinewellwith certain influential substantive views onwhat is valuable for its own sake. Consider, forexample, a theory like preferentialism. This position relates value to preferences, desires,and wants. Given certain plausible assumptions about these attitudes, preferentialism(at least, in a plausible form) can be depicted as containing the following core claims:

(1) Final value accrues only to the objects of (final) preferences (desires, wants).(2) A preference has as its object the realization, or obtaining, of a state of affairs.

From these two theses there follows a third claim about what the value bearers areaccording to preferentialism.

(3) Final value accrues only to obtainings of states of affairs.

If there are non-obtaining states of affairs, such as ‘that everyone is happy’ or ‘that I knowall there is to know about values’, preferentialism is most plausibly regarded as a view thatascribes value to the realization of such states—i.e. to what I have referred to as facts, orthe obtaining of states of affairs—rather than to the states of affairs themselves.

It is evident that there are different ways of understanding what a preference is, andhence that the claim that all preferentialists subscribe to (1)–(3) is debatable. Still, weneed a name for the kind of view that embraces (1)–(3), and preferentialism is, I think,quite suitable here. Assuming therefore that (1)–(3) are what preferentialists endorse,they cannot straightforwardly endorse the following claim:

H1: Final value accrues to concrete objects (not states of affairs or the obtainingsthereof ), e.g. stamps, wild nature, dresses, and experiences.

Hedonists actually support the stronger claim that final, non-derivative value accruesonly to experiences. However, the point I wish to establish requires only the weakerH1. It is not only hedonists, then, that might defend this thesis. An objective list theory,for instance, may well regard different experiences as the only value bearers there are,and another version might share with hedonism the idea that experiences are valuebearers, but nevertheless maintain that there are in addition other things (e.g. certainobtaining states of affairs) that have final value.

If preferentialists are faithful to (3), they will have to say that objects such as a uniquestamp or the Brazilian rainforest or (a personal value such as) one’s own pleasure experi-ence have a kind of value, if any, other than final value. It is not the stamp that is valuablefor its own sake; it is rather some realization of a state of affairs involving the stamp that isthe genuine bearer offinal value. For instance, itmight be that value accrues to the fact that

2 See his ‘Can Intrinsic and Final Preferences be Irrational?’ (2003).3 In ‘Rethinking Intrinsic Value’ (1998, p. 283). For a somewhat different and interesting approach, see

Rae Langton, ‘Objective and Unconditioned Value’ (2007).

154 VALUE BEARERS AND VALUE PLURALISM

a certain stamp is inmy possession or that having a certain dress ensures that I am related tosomeone famous. The stamp and the dress have at most derivative value. The value of theseconcrete objects is thus taken from some other entity that carries final value.4

I presume a reply like this one will work in some cases. However, it will not takecare of every case. In fact, a hedonist who regards certain experiences (notably pleasure)as having final value has an argument against the sort of preferentialism presented here.What the hedonist submits is that preferentialists, in saying that value accrues only tothe obtainings of states of affairs,5 miss what hedonists value for their own sake. So fromH1, the hedonist draws the conclusion:

C: Preferentialists are debarred from valuing what hedonists value for its own sake.

Notice by the way that the preferentialism which I consider here, what WlodekRabinowicz and Jan Osterberg (1996) have called object-preferentialism, should bedistinguished from what they refer to as the satisfaction version of preferentialism.The latter makes the following claim (rather than (1)): intrinsic value is assigned tothe circumstance that our intrinsic preferences are satisfied.6 But this version is even lesslikely to convince a hedonist about the falsity of H1 (i.e. the idea that final valueaccrues to experiences and not to the obtaining of states of affairs or circumstances).The simplest way of satisfying a preference is to make something the case—to see to itthat some states of affairs is realized.I will not defend the hedonist claim that final non-derived value accrues only to

experiences of a certain sort, namely those with hedonic properties. The hedonistperspective on value is restricted. However, whether or not one is a hedonist, and I amnot one, I think hedonism does capture something correct in insisting that value cannotsimply be understood as something accruing to the realization of a preferred, or desired,abstract state of affairs. But if this is true, there is reason to resist the attempt to view allfinal, non-derivative value as supervening on states of affairs. This issue is therefore ofgreat importance. A formal analysis of final value should, after all, be embraceable bythe proponent of any major substantive view on value, and hence by the hedonist andpreferentialist. In the next section I will briefly explain why I think the hedonist

4 Preferentialists might therefore accept the amendment H1*: final derivative value accrues to concreteobjects. This, however, would not settle the issue they have with hedonists who challenge their view on final,non-derivative value.

5 Very often hedonism is described as a theory about the value of abstract entities (e.g. states of affairs)rather than concrete objects (e.g. sensations). Consider, for instance, Richard Brandt’s comment “philoso-phers, from Epicurus to Bentham to Sidgwick to J. J. C. Smart and other contemporaries, have thought thatthere is one and only one state of affairs that is intrinsically good: pleasant (liked) experiences (and bad states:disliked experiences)” (1996, p. 36, my emphasis). In what follows I will not pay any attention to differentversions of hedonism that differ in their views on whether value is related to the duration of the pleasantexperience or its intensity or a combination of these. There are all sorts of issues involved here; these havebeen discussed at length by Michael Zimmerman (2001a). In particular, see his intriguing discussion ofevaluative inadequacy (pp. 142–8).

6 With regard to the point I wish to make here, nothing hinges on the fact that the two different versionsare formulated in terms of intrinsic, rather than final, preferences.

10.1 EXTRINSIC F INAL VALUES 155

conclusion C ought to force preferentialists to reconsider their position, and in effect,why a pluralist view on value bearers is to be preferred to a monist one.7 If I manage todo this, the idea that personal values accrue to many different sorts of objects, whichplays an important role in this book, will have been bolstered.

10.2 Hedonism and preferentialismPreferentialists, I imagine, can pursue various strategies to deal with the hedonist attack.For instance, a preferentialist could agree without difficulty that she is debarred fromvaluing what hedonists regard as pleasure as long as it is established that the hedonist has amistaken view of pleasure. Once the mistaken view is replaced by a more plausible one,C will lose its appeal; and so the hedonist argument does not in the end pose any realthreat to preferentialism.

This particular strategy tries to take the sting out of the argument by replacing thehedonist’s view of pleasure with one that would render her substantive claim—thatfinal value accrues only to pleasure, i.e. to experiences with hedonic qualities, or tones,being had at a certain time, in a certain place—less convincing.

We may question the scope of experiences with hedonic tones, but tocompletely deny that there are these kinds of experience is phenomenologicallycounter-intuitive.8 Of course, we need not claim that all pleasant experiences havehedonic tones. A mixed view between the ‘hedonic tone’ version and other versions—say, Richard Brandt’s well-known ‘desire-version’, or Fred Feldman’s recent attitudi-nal version of hedonism—may well be what, in the end, best captures our usage of‘pleasure’.9 But that is a merely a verbal matter.

Brandt’s position is, perhaps, particularly interesting in this context, since it explicitlyleaves room for hedonic tones. However, he thinks that an account of pleasure in termsof hedonic quality is elusive. Actually he makes it clear at an early stage that his way ofunderstanding ‘pleasure’ does not, and is not intended to, correspond to the commonusage of this term. His definition is rather intended to be “suited for a scientificpsychological explanatory conceptual framework” (1979, p. 25), and he seems tothink that, given this, a hedonic tone account is disqualified. I shall not here take astand for or against such a view. However, it is important to realize that the mere factthat hedonic tones may be difficult to fit into a scientific conceptual framework doesnot mean that we can leave them out of an axiological framework. If there areexperiences with hedonic tones, they should surely be strong candidates on anyone’s

7 I will not attempt to explain why, in combination with their own favoured view, hedonists shouldembrace the preferentialist idea that value accrues to the objects of at least some preferences.

8 Gilbert Ryle’s view of pleasure is another in which it is denied that pleasure is an experience with adistinctive positive tone. In his account, an activity that we find pleasant becomes an activity that we are proneto engage in (or to continue being engaged in). As T. L. S. Sprigge has put it, “This gives a strikingly joylesspicture of pleasure or happiness” (The Rational Foundations of Ethics, 1988, p. 132).

9 One possibility is that one analysis fits sensory pleasure and another fits non-sensory pleasure.


list of possible valuable objects. Thus, for the hedonist, the important thing to notice isthat value accrues to those experiences that have hedonic tones. How they areconnected to a person’s desires is another issue. Brandt may have put his finger on aconceptual link between desires and pleasant experiences. Even so, his view fails tobring out what is valuable about pleasure.Incidentally, this may not apply accurately to David Sobel’s more recent idea that a

pleasure is nothing more than an experience “which is intrinsically wanted for its ownsake and what is wanted is the way it feels when it is occurring”. Here the idea is that itis our desiring (or, more accurately, our liking) some present conscious state that makesthe state pleasant. Of course, a quite natural response to this is that we desire or likesomething not because doing so is pleasant, but rather because what we like, the objectof our liking, is something pleasant. Sobel rejects this phenomenological approach,though. Some of his reasons are familiar objections to such approaches. However, hismain idea seems to be that it is only if we understand pleasures as experiences that aredesired when they are occurring that we can account for their normativity. It is, in hisview, only the desire element in pleasure that makes pleasure into a reason-providingstate. This is no place to go into detail. I must simply state that I fail to see that Sobel hasreally established this conclusion.10

Brandt is right to say that the notion of a hedonic tone is elusive. For instance, itmight well be that if we try to direct our consciousness so as to single out this hedonicquality we actually lose sight of it. At least, this has been my own experience; trying tograsp and keep the pleasure, or pain, that I am focusing on is a bit like scooping up ahandful of sand and then trying to make a fist—the grains simply sift through thefingers as we tighten them. Some hedonic tones vanish as we focus on them. But totake this as evidence that there are no experiences that have hedonic properties is amistake too commonly made. Anyone who has tried focusing on a minor pain willprobably be able to confirm that they ended up being conscious of certain sensations—e.g. that something is burning or pulsating. The earlier experience—the one with, say,a dishedonic property—is replaced. Sometimes, this act of replacement is taken asevidence that there were no ‘hedonic tones’ involved in the first place. But I have notyet come across any argument that secures this point.11

The observation that we cannot necessarily hang on to the hedonic quality of anexperience by consciously focusing on it helps throw some light on another issuewhich has for some time been on the hedonist’s agenda: given the heterogeneousnature of our pleasant experiences, is it not wrong to talk about a hedonic tone? Itmight be claimed that even if we focused only on sensory pleasures, the pleasure weexperience when having, say, an orgasm has nothing (or very little) in common with

10 See Sobel, ‘Pain for Objectivists: The Case of Matters of Mere Taste’ (2005).11 That certain phenomenological states are not available to introspection has been confirmed by recent

scientific research. For discussion of the relevant literature see, for instance, Leonard D. Katz, ‘Pleasure’(2006).

10.2 HEDONISM AND PREFERENTIAL ISM 157

the pleasure of feeling a cool hand when you have fever. Pace Epicurus, Bentham, andSidgwick,12 this Mill-inspired13 observation might seem plausible. Obviously a lot ofthings happen in the first case that do not happen in the latter. They are experiences ofdifferent things. But it is not clear what we should conclude from this. If there issomething to the observation I made earlier—namely, that we tend to lose sight of thehedonic qualities when we try to focus on them—it is not so obvious that the twoexperiences need, after all, to be entirely different kinds of pleasures. The often-voicedclaim that there is no special quality of pleasure, but only various sensory experiences,needs to be backed up with something more than the observation that all we experi-ence are different bodily sensations. Different experiences need not have the samehedonic property to be cases of pleasure.

Let us not press this phenomenological point, though. Even if we grant (as seems notimplausible) that the two cases do not differ only in respect of how intense, or durable,the experiences are, the point has no really vital consequences vis-a-vis the presenthedonist argument. That different experiences may have different hedonic tones doesnot alter the fact that what has value, according to the hedonist, are certain experiences(whether or not there is a multitude of them) rather than different states of affairs.Another issue that it is not crucial to take a stand on is whether or not the pleasantexperiences should include moods. I see no reason to leave these out of the hedonistaccount and good reason to include them, but this need not be argued here.

10.3 PreferentialismThe disagreement between preferentialists and hedonists goes deeper than the questionwhether or not there are experiences with hedonic tones. Recall that what at least onecommon kind of preferentialist actually has in mind when she speaks of the value ofexperiences is the value accruing to the obtaining of some abstract proposition-likeentity involving, or being about, experiences. Hedonists, on their part (whether or notthey maintain that there are hedonic tones), would be ill-advised to give up the ideathat pleasure is a concrete entity.

The idea that pleasure is something proposition-like squares badly with the way weordinarily regard pleasant experiences, namely as particulars that exist in space and time.Still the distinction between what is abstract and what is concrete is, as we shall see inthe next sections, far from clear, and we should be careful not to put too much weighton it. Fortunately, the issue between preferentialism and hedonism about what are thefundamental bearers of non-derived final value can be illuminated further. For

12 Roughly speaking, the three writers mentioned here maintained that we should differentiate betweenso-called higher and lower pleasures by considering how long or intense the pleasure is. If it is very intense orlasts for a long time, the pleasure belongs to the higher category.

13 Mill would have pointed out that my examples were all examples of pleasures with an inferior quality.In Mill’s eyes pleasures of the intellect are superior quality pleasures. For an interesting discussion whichquestions Mill’s choice of superior pleasures, see Egonsson, Dimensions of Dignity (1998).


phenomenological reasons, I committed myself above to the existence of experienceswith hedonic tones. This gives me another reason to dispose of the idea that the valueof pleasure should be understood as value associated with the obtaining of a particularstate of affairs, namely one involving the pleasure. The argument sets out from theplausible assumption that hedonic tones are obviously not tones of proposition-likeentities such as states of affairs. A state of affairs may, of course, involve a pleasure (or ahedonic tone), but it would not make sense to say that a phenomenological featureaccrued to something like a propositional entity. (Indeed it might be argued, moregenerally, that states of affairs do not have features in the first place, hedonic orotherwise.) But if this is the case, we should ask the person who thinks obtainingsmight have such hedonic properties the following question: How can the realization ofsuch a proposition-like entity as a state of affairs give rise to an entity—an obtaining—that has a hedonic property if the state of affairs that it is a realization of does not havesuch a feature in the first place? Surely the only realistic difference there could bebetween an obtaining and the state of affairs it realizes is that the former, but not thelatter, is being realized. I see no reason why the obtaining of a state of affairs wouldsomehow acquire hedonic tones. The only entity I can reasonably imagine havingthese hedonic features is a concrete experience (or, more generally, the brain).As mentioned at the outset, I believe hedonists are wrong to think pleasure is the

only thing of value. However, you do not need to be a hedonist to accept that certainexperiences are valuable, namely those that have hedonic properties. In my view this ishard to deny, especially if we make it clear, as I have tried to do here, that this is not thesame as to say that they are experiences of hedonic tones. Moreover, I would add thatexperiences with hedonic tones are concrete objects—if, and to the extent that, anobject is something that can be described in different ways. Only our own limitationsseem to limit how the concrete can be depicted. This is not true of abstracta like the stateof affairs that Stockholm is the capital of Sweden in 2006. The same, I suggest, holds forobtainings of states of affairs. These obtainings may be slightly less abstract than states ofaffairs, but they are nonetheless too abstract to carry hedonic tones. Whether we shouldrefer to the concrete object as a physical thing, or perhaps confine ourselves to sayingthat it is a condition of the brain (leaving it open how to understand ‘condition’) is a lessacute issue for the value theorist—at least, as long as we keep in mind that it is thecondition that is valuable, not its obtaining.

10.4 Value bearersWe have seen, then, that the claim that bearers of value may be quite different innature, though it makes it hard to endorse any one particular view of value, remainsviable. However, other, more compelling arguments might show right away why weshould not recognize stamps, dresses, and other concrete entities as bearers of value. Sofar the discussion has concerned the obtaining of states of affairs, on the one hand, andconcrete objects, on the other. But these are not the only cases we need to examine.

10.4 VALUE BEARERS 159

Recent work in the philosophy of value has brought questions about the bearers of(final) value to the fore.14 Among the many accounts, here it is possible to detect twogeneral positions: value-bearer monists claim that (non-derivative) final value accruesonly to one kind of object; value pluralists deny this, claiming there is more than onekind of bearer of final value. Both preferentialism (at least, on the attitudinal readingof this theory) and hedonism are bone fide examples of the monist15 position. Thefollowing value bearers have recently been discussed: facts (the obtaining of states ofaffairs), properties (universalia), tropes and individual physical and mental objects(things, persons, experiences). As I have stated, by ‘fact’ I mean the obtaining of a stateof affairs. Regarding facts in this way allows me to distinguish between states of affairs asone kind of entity (such as that mermaids live in the sea) and obtainings of states ofaffairs, which are a second kind of entity. Since there are no mermaids, there is noobtaining of the state of affairs that they live in the sea.16

10.5 Value-bearer monismMonists have made a number of reductionist objections to pluralism. To makediscussion of these arguments more manageable, I shall begin by making some generalobservations about the issue without considering particular proposals (we shall turn tothose proposals in due course). Monists wish to reduce one thing, the reductandum, toanother, their preferred reductans. An example is:

Reductandum: a rare stamp is valuable for its own sakeReductans: that a rare stamp exists is valuable for its own sake.

This monist will argue that the statement ‘a rare stamp is valuable . . . ’ is somehowmisleading in that it is reducible to the statement ‘that a rare stamp exists is valuable . . . ’In Section 10.7 I will consider some proposed reductions in detail. Meanwhile, I needto differentiate between some different kinds of monistic approach, since there is one inparticular that I want to set aside.

A fully-fledged version of monism—what one might call ‘strong monism’—claimsthat only its favoured kind of reductans is strictly speaking true, and that any reductandumthat does not coincide with the reductans is false. Simplifying matters somewhat, onereason for adopting this sort of monism would be that there is no bearer of value such asthat referred to in the reductandum. If there are no such entities (pace the implicit

14 For some of the latest discussions of this topic, see Rønnow-Rasmussen and Zimmerman (eds.), RecentWork on Intrinsic Value (2005).

15 Hedonist monism should be understood as monism regarding ‘positive’ final value. An extreme andabsurd form of value monism is imaginable in which the view that final value accrues to the concrete object‘experience of pleasure’ is combined with rejection of a corresponding view about what has final negativevalue, i.e. experiences of pain.

16 Zimmerman (2001a, 49 ff.) goes a step further, suggesting that the obtaining of a state of affairs is bestunderstood as consisting in some individual(s) exemplifying a certain property at a certain time. For adiscussion of this suggestion see Rabinowicz and Rønnow-Rasmussen (2003a, especially the appendix).


suggestion of pluralists), then the reductandum cannot be taken at face value; it must beunderstood in some other way. It would seem that if this claim can be paired with aconvincing refutation of alternative pluralist suggestions as to the bearers of final value,monists will have a strong case for their own view.In its most extreme form, strong monism entails that there is, in fact, only one kind

of thing: the things that are the bearers referred to in the reductans. Supposed bearersthat do not coincide with what is depicted in the reductans are then consideredmetaphysically queer. A somewhat less breathtaking variant confines itself to sayingthat the valuable objects of the reductandum are not among the things that exist, i.e. donot obtain, or are not instantiated.I am not sure what the metaphysical rationale for the stronger version is, and I shall

not discuss it here. Suffice it to say that that version strikes me as much too radical aview. The milder view might, in certain cases, be true, i.e. there might be cases when acertain reductandum is capable of being reduced to an alleged reductans. But as a generalrule this does not seem to be the case.But not every form of value-bearer monism is based on metaphysics. Thus in

versions of ‘moderate monism’ one finds the idea that there is something fishy, butnot necessarily false, about the reductandum. To say that final value accrues to x might,according to the moderate monist, express some truth, but it is not the whole truth;only the reductans conveys the whole picture. If we were to admit that both thereductandum and the reductans expressed the whole truth of the matter, it would behard to see how this view could be described, strictly speaking, as monist. The pointmust rather be this: the truth there is to the reductandum (e.g. ‘a rare stamp is valuable forits own sake’) consists in the fact that there is some such thing as x and x is a valuebearer. However, the value that accrues to x is a derivative value that stems from the finalvalue accruing to the bearer suggested in the reductans, y (e.g. ‘that a rare stamp exists isvaluable for its own sake’). Saying that a value is derivative need not be, but often hasbeen, I suspect, a way of somehow downgrading ‘reductandum-value’. Just what thismeans is another question that will engage us later on.

10.6 SimplicityI suppose that one important driving force behind monism is its simplicity. A parsimo-nious ontology is preferred to one containing many different kinds of entity of the kindthat cannot be reduced to a fundamental single kind of entity. Still, it is difficult to seewhy simplicity in itself is a mark of truth. Pluralists will eventually counter by sayingthat the cost to monism is that their ontology is oversimplified.There is a version of the simplicity argument which, in standard textbooks on the

matter, is set aside as being flawed. I am thinking of the idea that there might be variousexternal reasons for wanting a simple theory. For one thing, such a theory might belooked upon as easier to deal with. Thus, giving priority to a theory that is simple mayon occasions be the rational thing to do—it will, for instance, be time-saving in that,

10.6 S IMPL IC ITY 161

most likely, it can be tested and eventually dismissed more rapidly than more complextheories. However, although simplicity may be an incentive to test one theory ratherthan another, it does seem like a wrong kind of reason for settling the truth-value of thetheory. That the life of the investigator, or the evaluator, will go more easily is hardly acompelling reason to become a monist.

David Alm has suggested (personal communication) that it might only be a badreason when it comes to empirical science, and that similar reasoning does notobviously apply to theories of value. Perhaps he is right when it comes to substantivetheories of value. However, I am quite sure that the fact that things are made easier forthe formal axiologist should not have an impact on his or her views on the monism/pluralism issue.

A more common argument nowadays has it that pluralism should be rejectedbecause it has a number of non-desirable consequences. Monists with regard to valuebearers have, for instance, claimed that pluralism makes value comparisons impossible.How can we compare, much less weigh one value against another, if these values arenot of the same kind? That would be just like weighing one metre against one litre.A version of this external (to truth) argument has also been voiced among value-bearermonists. The idea is that if there were only one kind of value bearer, we might be ableto provide an informative account of the computation of value. If pluralism is correct,and there are different kinds of value bearer, then such computation seems impossible.This objection is perhaps not as easily brushed away. Nonetheless, the pluralist’s basiccomment, I take it, will be that there is not much to computation if values are notcomputable.

Here is another explanation of why monism has been preferred. During the last fivedecades or so the development of a formal logic for operators such as “It is good that___” or “that___is better than that___” has continued steadily. The opportunity to leanon a well-established framework is probably one reason why some value theorists havebeen inclined to reduce all value to proposition-like objects. (Cf. Rabinowicz andRønnow-Rasmussen, 2000, p. 44). Again, this is a quite understandable motive thathas nothing to do with the monism/pluralism issue.

There is also an evaluative side to this issue. It might be argued that the pluralistposition rests somehow on the idea that there is a subject matter here to be discovered.But if what has value is in some sense up to us, it is surely better to have a value theorythat allows values to be compared and computed.

This is an intriguing argument. However, as I argued in Chapter 1, it is a mistake tothink that pluralists are somehow committed to objectivism, or realism, about values.Still, the gist of the argument is, I suppose, that a world that contains no moral conflictsis a better world than one with such conflicts, so if we want to systematize our thoughtson value, we might just as well opt for one that allows us to solve conflicts. But is thatreally that obvious? I, at least, am not sure what the substantive argument is for thenotion that a world in which values are computable is better, in some sense, than a


world in which value is not computable and where value conflicts may emerge. I mightbe wrong, but this is clearly a substantive issue.17

10.7 Some reductionsSuppose x refers to a pleasant experience. It may now be argued that the claim ‘x hasfinal value in virtue of its property F ’ (F referring to a hedonic quality) should actuallybe reduced to a claim about some existential fact like ‘what has final value is that x, whichhas F, exists’. An alternative to this might be ‘what has final value is the fact that there isa state of affairs of the form there exists some object that has F ’. Given an eliminativistreduction, the alleged value of the experience is now localized in the existential fact.This kind of reductionist manoeuvre is open to a very natural objection, namely that

it starts from the wrong end of the story—or, as Rabinowicz and I (2000) haveremarked: it puts the cart before the horse. The reason we think that the fact ‘theexperience x which has property F exists’ is valuable is surely that the experience isvaluable. The value of the experience grounds the value of the existential fact. Tosuggest otherwise by contending that it is the fact that grounds the experience is tomisplace what is of final value here. If anything, the fact derives its value from the valueof the experience. The reverse of this relation is unconvincing. The advocate of thisparticular reductandum—say, a hedonist—could, in other words, accept that valuemight also accrue to facts. In fact it would be strange if she denied them valuealtogether. However, these facts would have merely derivative value.In the case under consideration it seems natural to say that the existence of a pleasant

experience is valuable not because some existential state is non-derivatively valuable,but in virtue of the value that accrues to the object itself (i.e. the experience). Theremay be other examples in which it is conceivable that the value of the state need not bedrawn from the value of the concrete object. The latter might be of no value at all.A further possibility should be mentioned. Even if what we value does not accrue to

the fact that the experience that has property F exists, this alternative may still be regarded asbeing on the right track. Perhaps what is valuable for its own sake has something to dowith the experience having the property P.18 On the other hand, it does not seem veryplausible to say that it is properties, in the sense of universalia, that are value bearers.19

A different, but related, proposal does come with intuitive appeal, however. This is theproposal that it is rather the instantiation of properties that is valuable. So when person aexperiences pleasure, what is valuable is the instantiation of pleasure in a, and not some

17 Value-bearer monism is consistent with value pluralism. For instance, one might argue that happiness isone sort of value and freedom another, but that these values accrue to only one sort of bearer—say, to facts orto states of affairs. Amartya Sen, for instance, is often interpreted as being a value pluralist but a value-bearermonist (Development as Freedom, 1999).

18 Parayot Butchvarov (1989), for instance, argues that the bearers of final value are properties.19 Noah Lemos (1994) and Zimmerman (2001a) criticize this view.

10.7 SOME REDUCTIONS 163

universal which, perhaps, nobody exemplifies.20 Of course, that the instantiation ofF occurs in a rather than in another object (person) b is of no evaluative importance (atleast, as long as we have moral values in mind). It is not the fact that this object a hasF that is valuable. What is valued is rather the instantiation of F in a. Value accrues inthe same way to each instantiation of F, in whatever object it occurs.

On this proposal, then, we value various instantiations of P rather than the states that ahas F, that b has F, and so on. Some metaphysicians have suggested that the instantiationshould be regarded as an irreducible ontological entity sui generis—what, nowadays, isreferred to as a ‘trope’ (following DonaldWilliams, 1953). In contrast with properties (i.e.universalia) tropes are particulars. Whereas a single universale can be instantiated in manydifferent places, this is not the casewith a trope. Two ormore instantiations of a universaleare identical with each other. Tropes, on the other hand, are never identical with eachother. Thus a’s experiencing pleasure at t1 and b’s experiencing pleasure at t1 are two tropes(not the exemplification of one and the same pleasure property). There are two particularentities involved in the two cases. Moreover, unlike concrete individual things such aspeople, these particular entities are regarded as abstract particulars.21

This argument has a certain ring of plausibility. For one thing, if it is the instantiationof a property that is valuable, the cart-before-the-horse argument is inapplicable. Itwould be counter-intuitive to suggest that the value that this instantiation of property F att1 in a has derives from the value of a.

On balance, however, I think a hedonist should eschew this manoeuvre. The precisemetaphysical nature of these ‘instantiation-entities’ is still somewhat opaque. Are theyabstract or concrete entities? I am inclined to see them as abstract.

The notorious distinction between abstract and concrete entities is a vexed issue.However, recently Zimmerman has lent us a hand with it:

It’s been suggested, for instance, that abstract entities exist necessarily, whereas concrete entities existonly contingently. This is problematic, since God is often taken to be a necessarily existing concreteentity, while sets of contingently existing objects are often taken to be contingently existing abstractentities. It’s also been suggested that concrete entities have spatiotemporal location, whereas abstractentities do not. But this is again problematic, since individual souls are often taken to be concreteentities that do not occupy space, while times and places are often taken to be concrete and to haveno spatiotemporal location themselves; and again, God is often taken to be a concrete entity thatlacks spatiotemporal location. Another suggestion is that abstract entities can have instances, whereasconcrete entities cannot. But this too is problematic, since sets are often taken to be abstract and yet

20 Another possibility, suggested by Ingmar Persson (private communication) and Zimmerman (2001a;2001b), would be that it is rather that x has P, which is valuable. For criticism of this idea, see Rabinowicz andRønnow-Rasmussen (2000).

21 There is no consensus among trope theorists over what else there is besides tropes. Onwhatwemight referto as a nominalist position, no other entities exist apart from tropes. This is not classical nominalism; but in virtueof its denial of universalia and its insistence that there is only one fundamental ontological category (namely, ofparticular tropes) it is a kind of nominalism (contrast Keith Campbell, Abstract Particulars (1990, p. 27)).


are incapable of being instantiated, something which is also plausibly thought to be true of‘impossible’ properties (such as the property of being a square circle).22

Zimmerman next suggests that:

. . . [a]ll abstract entities (except sets, if they exist) exist necessarily, whereas no concrete entity(except God, if he exists, and states of God) does, that only abstract entities are instantiable, andthat only concrete entities can have spatiotemporal location.23

So where do tropes fit in? Take the claim that abstract entities exist necessarily. Tounderstand this claim we have to be aware of an idea that I think underlies it, namelythe idea that there is a conceptual distinction between the existence of an (abstract)entity and its instantiation or obtaining. Consider, for instance, properties—say, theproperty of being a mermaid. On this (platonic) account, this property qua universaleexists necessarily, but in this world it happens to be the case that no such property isinstantiated. Should we consider tropes as entities that exist necessarily?Modern trope theory developed as an attempt to throw light on our notion of a

property. It might therefore be thought that if the latter were assumed to havenecessary existence, so too should tropes. But it is in fact hard to detect anything thatspeaks in favour of this suggestion. The distinction between something’s existing andsomething’s being instantiated is, I believe, inapplicable to tropes. While properties quauniversalia are capable of instantiation, properties qua tropes (if there be such) are not.Given this, we cannot make use of this distinction in order to explain why tropes wouldexist necessarily. That tropes are not necessary is pretty obvious: my beard is greyish,but who would deny that I (in at least some sense) could have had a red beard.However, tropes can be abstract in some other sense than that given by the phrase‘entity existing but not obtaining’. I will return to such a possibility in a moment.Meanwhile, a word must be said about the possibility that tropes are concrete entities.Recall that, in Zimmerman’s sense of the term ‘concrete’, it was (with a few possible

exceptions) only concrete particulars that have spatio-temporal location.24 Are, then,all tropes spatio-localizable? This is of course speculation, but it seems to me that at leastsome tropes, if they exist, are more easily located in space and time than others. My haircolour would be an example, whereas my pain would not be just as easily positioned inspace. (Is it in my thumb? My brain?) And what should we say about dyadic tropes suchas my relation to the truth that ‘2 + 2 = 4’?25

22 Zimmerman (2001a, p. 34).23 Ibid.24 Zimmerman’s own view must be mentioned here. In his (2001a) he suggests that it is neither properties

nor abstract facts, but rather concrete states of individuals that are value bearers; see also Zimmerman (2001b).For a detailed examination of the idea that value can be reduced to such states of individuals, see Rabinowiczand Rønnow-Rasmussen (2003a).

25 In Rønnow-Rasmussen (1998, pp. 293, 294) I expressed doubts about using indexical terms such as‘here’ and ‘there’ about phenomenal qualities. Today my doubts are less insistent, at least when it comes tosome kinds of quality.

10.7 SOME REDUCTIONS 165

10.8 Separating the concrete from the abstractActually there is an alternative approach to the abstract/concrete division, and on thisapproach—which, I think, comes closer to the meaning of ‘abstract/concrete’—thereis no doubt that tropes are abstract particulars. Comparing different lollipops with eachother, Williams says:

To borrow now an old but pretty appropriate term, a gross part, like the stick, is ‘concrete’, as thewhole lollipop is, while a fine or diffuse part, like the color component or shape component, is‘abstract’. The color-cum-shape is less abstract or more concrete or more nearly concrete thanthe color alone but it is more abstract or less concrete than color-plus-shape-plus-flavor, and soon till we get to the total complex which is wholly concrete. (1953)26

In my view the main idea expressed here catches what we have in mind when weperform an act of abstraction: the colour of this lollipop, for instance, occurs togetherwith other qualities of the lollipop, and what I do, when I bring this quality before mymind, is set aside other qualities of the lollipop. An act of abstraction is, as KeithCampbell puts it, “an act of selective ignoring” (1990, p. 3). What trope theorists thendo is to claim that the object of such an act of abstraction has existence. Williams andCampbell also want to take a further step and say that tropes are located in the sameplace; for instance, the concrete lollipop is the totality of being where the colour,shape, (etc.) of the lollipop are. In other words, the doctrine that two different thingscannot be at the same place at the same time is either false or inapplicable to abstracta.But as far as I can see there is no need to accept this latter claim, even if we do regardabstract entities in the above way (i.e. as thin entities that constitute the buildingmaterial of thick entities). Nor, as far as I can see, need we endorse another claim that isimplicit in the above passage, namely that things can be more or less concrete/abstract.An object might be thicker or thinner, more or less diffuse, and in that sense we mayspeak of an abstract entity getting closer to, or further away from, being concrete. But,intuitively, to claim this is in no way to remove the gap between the concrete and theabstract.

What should we say, then, about the obtaining of a state of affairs? Are these abstractor concrete entities? I am not sure. FollowingWilliams, we might want to say that theyare more concrete than the features Williams mentions as abstract; but being moreconcrete is not necessarily the same as being a concrete object. I am inclined to denythat obtainings are concrete entities, but I cannot presently offer any good argumentfor this.

26 The passage comes from D. H. Mellor, and A. Oliver (eds.), Properties (1997, p. 113). Cf. Campbell(1990, pp. 2, 3). See also D. M. Armstrong, Nominalism & Realism (1978). Armstrong discusses the case of acoloured cube (pp. 120–1). This, as he calls it, ‘concrete’ cube he then compares with the tactual and visualcubes, which he refers to as abstract particulars. Notice that Armstrong does not endorse trope theory. Seealso Zimmerman (2001a, p. 66, n. 6), where Campbell’s use of ‘abstract’ and ‘concrete’ is criticized.


Let us return to the question of why we should accept that it is the instantiation ofpleasure that is valuable. Surely, in the end, the hedonist will say that what we value issomething concrete and not some thin, abstract entity such as a property-instantiation.Obviously it cannot merely be an instantiation that prompts the hedonist to ascribevalue to a sensation; after all, most property instantiations do not carry any valueaccording to hedonists. It is true, of course, that the sensation is valuable because it is acertain kind of instantiation, namely one involving a certain kind of experience. Butagain, the very fact that hedonists say that there is something about this kind ofsensation that makes it valuable explains why it still makes sense to ask “But what isso valuable about these experiences?”, and this suggests that “instantiating . . . ” onlygives us what has derivative value.The natural thing to say here is that the hedonic property P is a good-making

property that renders the complex experience valuable. To this it might be objectedthat if an object is valuable because it has certain properties, these properties (or, at least,the instantiations of them) must themselves be valuable. The idea underlying thisobjection would be that whatever is a condition of something of value must also bevaluable. However, as has been said before, this is a non sequitur: it is not logicallynecessary for the condition of the valuable to be valuable itself, either intrinsically orextrinsically. The good-making property need not itself be valuable to qualify as agood-making property.We now need to consider a further aspect of the hedonist position. So far I have

been speaking about its being a certain kind of experience that has value, according to(at least one kind of) hedonist. However, it might be pointed out, quite rightly, thatI have not said very much about what this experience is. Recently Ben Bradley hassuggested the following:

If there is a feeling of pleasure itself, akin to the feeling of warmth or sweetness, and if thepossession of such a feeling is intrinsically good, the bearers of intrinsic value need not be fine-grained. They might be coarse-grained events consisting of a person having that feeling ofpleasure at some time. Nevertheless, the bearers of value will be things that happen or obtain,rather than, say physical objects. And that is really the crucial distinction here: between, on theone hand, those who attribute intrinsic value to things and people, and on the other hand, thosewho attribute intrinsic value to events, tropes, states of affairs, property-instantiations, facts, orpropositions. (2006, pp. 115–16)

Bradley is right, I think, in suggesting that the value bearer in this sort of hedonism iscoarse-grained. And he is, of course, also right in thinking that such bearers are notphysical objects. It is in his further characterization of the bearer that I think he goeswrong. That is, what makes one experience valuable, rather than some other one,according to the hedonist, is that it is an experience with a hedonic property. That iswhat makes the experience into a feeling. An event, or happening, or obtaining,contains the hedonic property only indirectly, and these kinds of object will thereforeonly carry indirect or derivative value, according to the hedonist.

10.8 SEPARATING THE CONCRETE FROM THE ABSTRACT 167

Bradley gives us no reason to give up the idea that what has value, in classicalhedonism, are experiences of a certain kind, namely those with hedonic properties(which, incidentally, does not necessarily mean that they are experiences of hedonicproperties). But he is on the right track when he maintains that there is a crucialdistinction between two groups of value theorists, and that members of these groupsregard the nature of value bearers quite differently. Nevertheless, it seems to me thatthe real difference is not so much between those who think that physical things maycarry value and those who disagree (and who think that it is rather proposition-likeentities that are the value bearers). This simplifies the issue too much. The demarcationline runs rather between those who think that value bearers must be possible objects forattitudes that take propositional entities as their object, and those who think that otherkinds of attitude may well figure in the analysans of value. So I would not draw the linewhere Bradley does. In particular, I would reject the idea that pleasures are the directobject of attitudes such as desires and preferences. Pleasures, in my view, are concreteentities; they are concrete, not in the sense of being material objects, but rather in thesense that we can go on describing them in ways that seem, at least, inexhaustible. Thischaracteristic is not shared by abstract entities. It is shared by obtaining states of affairs,though. But they are not what a hedonist non-derivatively values. Although some ofthem certainly involve hedonic properties we would not say that a hedonic propertyaccrues to these entities in the way a hedonic property accrues to pleasant experiences.

Pleasures are nonetheless too concrete to be objects of attitudes that have abstractentities as their intentional objects. A hedonist may prefer that a certain person, or acertain consciousness, or brain, is in this sort of pleasure-state. But any value thataccrues to the fact that this consciousness is in this particular hedonic state, derives, onthe standard hedonist position, from the value of this sort of state.27

10.9 RecapitulatingThis chapter has addressed some quite complicated issues, and it might therefore be agood idea to go over the main points that have been argued for. It might even be worthrecapitulating the key ideas expounded in this book. This will help situate Chapter 10in its context.

In this work I have defended the idea that we have reason to expand our valuetaxonomy with a set of values that it is appropriate to refer to as personal values (ratherthan impersonal values like justice or equality). We refer to personal values withrelative terms such as ‘good-for’ and ‘value-for’. The notion of ‘good-for’ plays an

27 David Brax has recently developed an account of pleasure inHedonism as the Explanation of Value (2009).His central idea is that pleasures are experiences liked by the person having the experience. However, incontrast to, say, Brandt, he suggests that the liking is actually part of the experience liked. This, of course,makes pleasure into a self-reflecting attitude. A somewhat similar approach can be found in Murat Aydede,‘An Analysis of Pleasure vis-a-vis Pain’ (2000). See also Katz (2006) and Sobel (2005).


important role in many ethical discussions. However, there is little agreement amongphilosophers over its correct analysis, and so there is an obvious need for a new analysis.This work purports to meet this need by offering a novel way of understanding thenotion of personal value. For various reasons this kind of value has been assumed, inone very influential tradition in value theory, to be either incoherent or explicable interms of (and therefore less important than) what is good simpliciter. But the analysisdefended in this work shows that you can analyse personal value without appealing tosuch non-relative goodness. Moreover, the analysis supplies us with a notion of good-for which steers clear of standard objections. It does so by fine-tuning a certain patternof value analysis—one that has recently attracted much attention and which has roots inthe writings of the Austrian philosopher Franz Brentano.Personal final values come in two versions, depending on the nature of their value-

making properties. If all of these are internal to the object of value, we are dealing withfinal intrinsic value. If some of the value-making properties are external, i.e. externallyrelational, we are dealing with an extrinsic value for its own sake. As to the kind ofanalysis I have employed, one attraction is that it remains neutral when it comes to theissue of subjectivism versus objectivism. It is sometimes said that subjectivists cannotconceive of objects having intrinsic or final value. If this is true, there is not, it seems,much point in an analysis that remains neutral on the matter. However, sinceI believe subjectivists need not give up on final value, I have described in some detail(in Chapter 1) just how I conceive of this distinction. It will be recalled, I hope, thatwhat the subjectivist might find hard to accept is the following claim:

The Invariance thesis: The final value of an object is invariant over possible worlds.

But this thesis is not sacred, and indeed if it is the non-essential properties of an objectthat make it valuable, we should reject it. Not all value will then be invariant overpossible worlds. For this reason we cannot employ the thesis as a demarcation linebetween subjectivists and objectivists. Subjectivists can in fact choose between severalmore or less interesting ways of understanding value. To bring this out I distinguishedbetween value supervenience and value constitution.Understanding values in terms of reasons has a number of advantages. However, it

also faces some weighty objections. Chapter 3 was devoted to the most serious ofthese—the so-called ‘wrong kind of reason’ problem. Although, for obvious reasons,I would welcome a solution to this problem, I am not convinced by the solutionsI have seen proposed. In Chapter 4 I gave my reasons for not accepting the (naturalist)attempt to understand value-for in terms of the de facto attitudes of its beneficiary (i.e.the person whose value it is). My suggestion is instead the following: an object x’s valuefor a person a (i.e. x’s personal value) consists in the existence of normative reasons forfavouring/disfavouring x for a’s sake.The idea that something could be good for a person has often been criticized.

I considered Moore’s famous attack on the very notion of a personal good, arguing thatit is not very convincing.

10.9 RECAPITULATING 169

The for a person’s sake-attitude, which plays a leading role in my analysis, is difficult tocharacterize, certainly. In Chapter 5, in order to describe one of the more intriguingfeatures of this attitude, I introduced a distinction between two kinds of property.These differ from each other in the role they play in the intentional content of theattitude. I argued that this distinction between what I called justifiers and identifiershelps us to separate two kinds of for someone’s sake-attitude, both of which are pertinentto the analysans of personal value.

It is important to acknowledge that good-for has more than one sense. In particular,I admit that the idea that whatever is welfare-constitutive is good for the agent mightexpress a notion of good-for that differs from the one I try to illuminate. Anothercomplex question, which I addressed in Chapter 6, is whether universalizability is aproblem for an analysis like mine. It would seem to be so. For someone’s sake-attitudesthat take non-replaceable persons as their intentional objects are not obviously uni-versalizable. This may not be an insurmountable difficulty, though. It is important tokeep in mind that universalizability is a very strong requirement, and that value analysiscan manage with something less demanding: supervenience.

I have also distanced my views on good-for from some alternative suggestions, madeby other philosophers, and I considered the possibility that values in general, andpersonal values in particular, are so-called Janus values. If there are any such values Isuspect they are rare. The idea that personal values should be understood in terms ofagent-relative reasons, and that impersonal ones should be interpreted as agent-neutralreasons, seem plausible. However, in Chapter 9 I argued that we should be cautiousabout this blend of ideas.

Finally, in Chapter 10 I present an argument whose aim was to pave the way forvalue-bearer pluralism, i.e. the idea that entities belonging to quite different generalmetaphysical categories can be bearers of final value. The structure of the argument wasfairly simple. It set out from the idea that a common form of preferentialism and thestandard version of hedonism both make plausible claims about what is valuable.Briefly, preferentialists say that it is the satisfaction of our preferences and the obtainingsof certain states of affairs that have value. Hedonists, by contrast, claim that pleasure isvaluable. Of course, the argument will only gain traction with those who take thesetwo views to be credible, but since I feel confident that most people would be preparedto agree that pleasure and preference satisfaction (in the above sense) are valuable, I takethe argument to be of general interest.

The core idea of the argument is as follows. It is likely that pleasure will be amongthe many kinds of thing that preferentialists take to have value. However, the valuepreferentialists ascribe to pleasure will, the argument goes, always be regarded byhedonists as derivative in nature; and since hedonists think that pleasure has non-derivative value, they will conclude that preferentialists are debarred from valuingprecisely what they, as hedonists, value—at least, in the way hedonists value pleasure.As we saw, the reason for this is that although the obtaining of a state of affairs may beabout hedonic properties, its obtaining will not itself be an experiencing entity, and


hence it cannot have hedonic properties in the way that a pleasurable experience hashedonic properties. The value accruing to pleasure, from a hedonist point of view,cannot, therefore, be accounted for by the preferentialist. To think it can is—againfrom the perspective of the hedonist—to put the cart before the horse.28 So if we wantto retain substantial value dualism we have to give up the idea of value-bearer monism.The effect of this reasoning is to encourage abandonment of value-bearer monism.

Moreover, if we recognize that hedonists and preferentialists regard different kinds ofmetaphysical entity as the fundamental bearers of non-derivative value, and if we agreethat a formal analysis of value should be accessible to as many plausible substantivetheories of value as possible, the case for preferring pluralism to monism becomes evenstronger.

28 Notice the qualification here. If you are not a hedonist, the cart-before-the-horse argument might notapply to obtainings of states of affairs.

10.9 RECAPITULATING 171

Bibliography

Aiken, Henry D. (1950). ‘Evaluation and Obligation: Two Functions of Judgement in theLanguage of Conduct’. Journal of Philosophy, 47 (1): 5–22.

Alexander, Samuel (1920). Space, Time and Deity. London: Macmillan.Alfano, Mark (2009). ‘A Danger of Definition: Polar Predicates in Moral Theory’. Journal ofEthics and Social Practice, 3 (3): 1–13.

Allen, R. T. (1993). The Structure of Value. Aldershot: Avebury.Anderson, Elizabeth (1993). Value in Ethics and Economics. Cambridge, MA: Harvard UniversityPress.

Armstrong, D. M. (1978). Nominalism & Realism. Cambridge: Cambridge University Press.Aydede, Murat (2000). ‘An Analysis of Pleasure vis-a-vis Pain’. Philosophy and Phenomenological

Research, 61 (3): 537–70.Baron, Marcia W. (1997). ‘Kantian Ethics’, in Marcia W. Baron, Philip Pettit, and Michael Slote(eds.), pp. 3–91.

Baron, Marcia W., Pettit, Philip, and Slote, Michael (eds.) (1997). Three Methods of Ethics.Oxford: Blackwell.

Bradley, Ben (2006). ‘Two Concepts of Intrinsic Value’. Ethical Theory and Moral Practice,9: 111–30.

Brady, Michael (circulating ms.). ‘The Value of Valuing’.Brandt, Richard B. (1946). ‘Moral Valuation’. Ethics, 56: 106–21.——(1979). A Theory of the Good and the Right. Oxford: Clarendon Press.——(1996). Fact, Values, and Morality. Cambridge: Cambridge University Press.Brännmark, Johan (2001) ‘Good Lives: Parts and Wholes’. American Philosophical Quarterly,38: 221–31.

——(2002). Morality and the Pursuit of Happiness: A Study in Kantian Ethics. Lund: Media-tryck,Lund University.

——(2008). ‘Excellence andMeans: On the Limits of Buck-Passing’. The Journal of Value Inquiry,42: 310–15.

——(2009). ‘Goodness, Values, and Reasons’. Ethical Theory and Moral Practice, 12: 329–43.Brax, David (2009). Hedonism as the Explanation of Value. Lund: Media-tryck Sociologen, LundUniversity.

Brentano, Franz (1969 [1889]). The Origin of Our Knowledge of Right and Wrong. London andHenley: Routledge & Kegan Paul.

Broad, C. D. (1925). The Mind and Its Place in Nature. London: Routledge & Kegan Paul.——(1930). Five Types of Ethical Theory. London: Routledge & Kegan Paul.——(1942). ‘Certain Features in Moore’s Ethical Doctrines’, in P. A. Schilpp (ed.), ThePhilosophy of G. E. Moore (The Library of Living Philosophers). New York: Tudor PublishingCompany, pp. 43–67.

Broome, John (1995). ‘Skorupski on Agent-Neutrality’. Utilitas, 7: 315–17.——(2004). ‘Reasons’, in R. Jay Wallace, Philip Pettit, Samuel Scheffler, and Michael Smith,(eds), pp. 28–55.

Brülde, Bengt (1998). The Human Good. Goteborg: Acta Universitatis Gothoburgensis (PhDdissertation).

Burgess, J. A. (2009). ‘When is Circularity in Definitions Benign?’ The Philosophical Quarterly, 58(231): 214–33.

Butchvarov, Panayot (1989). Skepticism in Ethics. Indianapolis: Indiana University Press.Butler, Joseph (1900 [1729]). ‘Fifteen Sermons Preached at the Rolls Chapel’, in J. H. Bernard

(ed.), The Works of Bishop Butler, vol. 1 (London: Macmillan).Bykvist, Krister (2009). ‘No Good Fit: Why the Fitting Attitude Analysis of Value Fails’. Mind,118 (469): 1–30.

Campbell, Keith (1990). Abstract Particulars. Oxford: Basil Blackwell.Carlson, Erik and Sliwinski, Rysiek (eds.) (2001).Omnium-gatherum. Philosophical Essays Dedicated

to Jan Osterberg on the Occasion of his Sixtieth Birthday (Uppsala Philosophical Studies 50).Uppsala: Department of Philosophy, Uppsala University.

Chisholm, Roderick M. (1981). ‘Defining Intrinsic Value’. Analysis, 41: 99–100.Crisp, Roger (2000a). ‘Review of “Value . . . and What Follows”, by Joel Kupperman’. Philoso-

phy, 75: 458–62.——(2000b). ‘Particularising Particularism’, in Brad Hooker and M. Little (eds.), Moral Particu-larism. Oxford: Oxford University Press, pp. 23–47.

——(2005). ‘Value, Reasons and the Structure of Justification: How to Avoid Passing the Buck’.Analysis, 65 (285): 80–85.

——(2008). ‘Goodness and Reasons: Accentuating the Negative’. Mind, 117: 257–65.Dancy, Jonathan (1993). Moral Reasons. Oxford: Blackwell.——(2000a). ‘ShouldWe Pass the Buck?” in A.O’Hare (ed.), Philosophy, the Good, the True, and theBeautiful. Cambridge: CambridgeUniversity Press, pp. 159–75. Also inRønnow-Rasmussen andZimmerman (2005), pp. 33–45.

——(2000b). Practical Reality. Oxford: Clarendon Press.——(2004). Ethics without Principles. Oxford: Oxford University Press.Danielsson, Sven (2001). ‘The Supervenience of Intrinsic Value’, in Erik Carlson and Rysiek

Sliwinski (eds.), pp. 93–103.——(2007). ‘On Geach on Good’, in T. Rønnow-Rasmussen, B. Petersson, J. Josefsson, and D.Egonsson (eds.), pp. 1–11.

Danielsson, Sven and Olson, Jonas (2007). ‘Brentano and the Buck-Passers’.Mind, 116: 511–22.D’Arms, Justin and Jacobson, Daniel (2000a). ‘The Moralistic Fallacy: On the “Appropriateness”

of Emotions’. Philosophy and Phenomenological Research, 61: 65–90.——(2000b). ‘Sentiment and Value’. Ethics, 110: 722–48.——(2003). ‘The Significance of Recalcitrant Emotion (or, Anti-Quasi Judgmentalism)’, inA. Hatzimoysis (ed.), Philosophy and the Emotions. Cambridge: Cambridge University Press,pp. 127–47.

Darwall, Stephen (1983). Impartial Reason. Ithaca and London: Cornell University Press.——(2002). Welfare and Rational Care. Princeton: Princeton University Press.——(2003). ‘Moore, Normativity, and Intrinsic Value’. Ethics, 113: 468–89.Dickie, George (1979). Aesthetics: An Introduction. Indianapolis: Pegasus.Egonsson, Dan (1998). Dimensions of Dignity. Dordrecht: Kluwer.——(2003). ‘Can Intrinsic and Final Preferences be Irrational?’ in W. Rabinowicz and Toni

Rønnow-Rasmussen (eds.) (2003b), pp. 114–30.

BIBL IOGRAPHY 173

Egonsson, Dan, Josefsson, Jonas, Petterson, Bjorn, and Rønnow-Rasmussen, Toni (eds.) (2001).Exploring Practical Philosophy: From Action to Values. Aldershot: Ashgate Publishing.

Ewing, A. C. (1947). The Definition of Good. London: Macmillan.——(1959). Second Thoughts in Moral Philosophy. London: Routledge & Kegan Paul.——(1973). Value and Reality. London: George Allen and Unwin.Feinberg, Joel (ed.) (1969). Moral Concepts. Oxford: Oxford University Press.Feldman, Fred (2002). ‘The Good Life: A Defense of Attitudinal Hedonism’. Philosophy and

Phenomenological Research, 65: 604–28.——(2004). Pleasure and the Good Life: Concerning the Nature, Varieties, and Plausibility of Hedonism.Oxford: Oxford University Press.

Findlay, J. N. (1970). Axiological Ethics. London: Macmillan and St Martin’s Press.Frankfurt, Harry G. (1999a). ‘Autonomy, Necessity, and Love’, in Harry G. Frankfurt, Necessity,

Volition, and Love. Cambridge: Cambridge University Press, pp. 129–42.——(1999b). ‘On Caring’, in Harry G. Frankfurt, Necessity, Volition, and Love. Cambridge:

Cambridge University Press, pp. 155–66.Gauthier, David (1986). Morals by Agreement. Oxford: Oxford University Press.Geach, Peter (1956). ‘Good and Evil’. Analysis, 17: 33–42.Gert, Joshua (2004). ‘Value and Parity’. Ethics, 114: 492–520.Gibbard, Alan (1990). Wise Choices, Apt Feelings. Cambridge, MA: Harvard University Press.Grau, Christopher (2002). ‘The Irreplaceability of Persons’. Unpublished PhD dissertation,Johns Hopkins University.

Hare, R. M. (1952). The Language of Morals. Oxford: Oxford University Press.Hatzimoysis, Anthony (ed.) (2003). Philosophy and the Emotions. Cambridge: Cambridge Uni-versity Press.

Hartmann, Eduard von (1908).Grundriss der Axiologie oder Wertwägungslehre. Bad Sachsa: Haacke.Heyde, Erich (1926). Wert: Eine philosophische Grundlegung. Erfurt: Kurt Stegner.Heathwood, Chris (2008). ‘Fitting Attitudes and Welfare’.Oxford Studies in Metaethics, 3: 47–73.Hieronymi, Pamela (2005). ‘The Wrong Kind of Reason’. The Journal of Philosophy, 102 (9):437–57.

Humberstone, I. L. (1997). ‘Two Types of Circles’. Philosophy and Phenomenological Research,LVIII: 249–80.

Hurka, Thomas (1987). ‘“Good” and “Good for”’. Mind, 96: 71–73.——(2003). ‘Moore in the Middle’. Ethics, 113 (3): 599–628.Kagan, Shelly (1992). ‘The Limits of Well-Being”, in Ellen Frankel Paul, Fred D. Miller, Jr., andJeffrey Paul (eds.), The Good Life and the Human Good. Cambridge: Cambridge UniversityPress, pp. 169–89.

——(1998). ‘Rethinking Intrinsic Value’. Journal of Ethics, 2: 277–97.Katz, Leonard D. (2006). ‘Pleasure’, in Edward N. Zalta (ed.), Stanford Encyclopedia of Philosophy(Summer 2006 edition) <http://plato.stanford.edu/entries/pleasure>.

Kim, Jaegwon (1993). Supervenience and Mind. Cambridge: Cambridge University Press.Korsgaard, Christine M. (1983). ‘Two Distinctions in Goodness’. The Philosophical Review, 92:

169–95. Also published in Korsgaard (1996), pp. 249–74.——(1993). ‘The Reasons We Can Share: An Attack on the Distinction between Agent-Relative and Agent-Neutral Values”. Social Philosophy & Policy, 10: 24–51.

——(1996). Creating the Kingdom of Ends. Cambridge: Cambridge University Press.

174 BIBL IOGRAPHY

——(1998). ‘Motivation, Metaphysics, and the Value of the Self: A Reply to Ginsborg, Guyerand Schneewind’. Ethics, 109: 49–66.

Lang, Gerald (2008). ‘The Right Kind of Solution to the Wrong Kind of Reason Problem’.Utilitas, 20 (4): 472–89.

Langton, Rae (2007). ‘Objective and Unconditioned Value’. Philosophical Review, 116 (2):157–85.

Lapie, Paul (1902). Logique de la Volonte. Paris: Felix Alcan.Lemos, Noah (1994). Intrinsic Value: Concept and Warrant. Cambridge: Cambridge University

Press.——(1998). ‘Organic Unities’. Journal of Ethics, 2: 321–37.Mackie, J. L. (1977). Ethics: Inventing Right and Wrong. New York: Viking Press.McKeever, Sean and Ridge, Michael (2006). Principled Ethics. Oxford: Clarendon Press.McNaughton, D. and Rawling, P. (1991). ‘Agent-Relativity and the Doing/Happening Dis-tinction’. Philosophical Studies, 63: 167–85.

——(1993). ‘Deontology and Agency’. Monist, 76: 81–100.——(1995). ‘Value and Agent-Relative Reasons’. Utilitas, 7: 31–74.——(2002). ‘Conditional and Conditioned Reasons’. Utilitas, 14 (2): 240–48.Mellor, D. H. and Oliver, A. (eds.) (1997). Properties. Oxford: Oxford University Press.Morgan, C. Lloyd (1923). Emergent Evolution. London: Williams and Norgate.Moore, G. E. (1993 [1903a]). Principia Ethica. Cambridge: Cambridge University Press.——(1903b). ‘The Refutation of Idealism’. Mind, 12: 433–53.——(1922). ‘The Conception of Intrinsic Value’. Philosophical Studies, 253–74.Mulligan, Kevin (1998). ‘From Appropriate Emotions to Values’. The Monist, 84 (1): 161–89.——(2006). ‘Geist (and Gemüt) vs Life—Max Scheler and Robert Musil’, in R. Calcaterra (ed.),Le Ragioni del Conoscere e dell’Agire. Scritti in onore di Rosaria Egidi. Milan: Franco Angeli,pp. 366–78.

——(2008). ‘Scheler: Die Anatomie des Herzens oder was man alles fühlen kann’, in H.Landweer and U. Renz (eds.), Klassische Emotions-theorien von Platon bis Wittgenstein. Berlin:de Gruyter, pp. 589–612.

——(2009a). ‘Values’, in R. Poidevin, P. Simons, A. McGonigal, and R. Cameron (eds.), TheRoutledge Companion to Metaphysics. London: Routledge, pp. 401–11.

——(2009b). ‘Emotions and Values’, in P. Goldie (ed.), Oxford Companion to the Philosophy ofEmotions. Oxford: Oxford University Press, pp. 475–500.

Nagel, Thomas (1970). The Possibility of Altruism. Oxford: Oxford University Press.——(1986). The View from Nowhere. New York: Oxford University Press.Newton-Smith, W. (1973). ‘A Conceptual Investigation of Love’, in Alan Montefiori (ed.),Philosophy and Personal Relations. London: Routledge & Kegan Paul, pp. 113–36.

Nozick, Robert (1981). Philosophical Explanations. Oxford: Oxford University Press.Olson, Jonas (2003). ‘Revisiting the Tropic of Value: Reply to Rabinowicz and Rønnow-Rasmussen’. Philosophy and Phenomenological Research, 67: 412–22.

——(2004). ‘How to Pass the Buck’. (circulating manuscript).——(2005). Axiological Investigations. Uppsala: Acta Universitatis Upsaliensis.——(2009). ‘The “Wrong Kind of Solution to the Wrong Kind of Reason Problem”’. Utilitas,21 (02): 225–32.

Olson, Jonas and Svensson, Frans (2005). ‘Regimenting Reasons’. Theoria, LXXI, (3): 203–14.

BIBL IOGRAPHY 175

Olson, Jonas and Timmons, Mark (forthcoming). ‘A. C. Ewing’s First and Second Thoughts onMetaethics’, in Thomas Hurka (ed.), Underivative Duty: British Moral Philosophers from Sidgwickto Ewing. Oxford: Oxford University Press.

O’Neill, John (1992). ‘The Varieties of Intrinsic Value’. Monist, 75: 119–37.Parfit, Derek (1979). ‘Prudence, Morality and the Prisoner’s Dilemma’. Proceedings of the British

Academy, 65: 539–64.——(1984). Reasons and Persons. Oxford: Clarendon Press.——(2001). ‘Rationality and Reasons’, in Dan Egonsson, Jonas Josefsson, Bjorn Petterson, and

Toni Rønnow-Rasmussen (eds.), (2001), pp. 17–41.——(circulating ms). ‘Climbing the Mountain’.——(circulating ms). ‘On What Matters’.Pascal, Blaise (1995 [1867]). Pensees and Other Writings, translated by Honor Levi (Oxford

World’s Classics). Oxford: Oxford University Press.Paul, E. F., Miller, Jr., Fred D., and Paul, Jeffrey (eds.) (1992). The Good Life and the Human Good.

Cambridge: Cambridge University Press.Persson, Ingmar (2005). The Retreat of Reason: A Dilemma in the Philosophy of Life. Oxford: OxfordUniversity Press.

——(2007). ‘Primary and Secondary Reasons’, in T. Rønnow-Rasmussen, B. Petersson,J. Josefsson, and D. Egonsson (eds.), pp. 1–16.

Pettit, Philip (1987). ‘Universality Without Utilitarianism’. Mind, 72: 74–82.——(1988). ‘The Paradox of Loyalty’. American Philosophical Quarterly, 25: 163–77.Pettit, Philip and Smith, Michael (1990). ‘Backgrounding Desire’. The Philosophical Review,99: 565–92.

Portmore, Douglas W. (2001). ‘McNaughton and Rawling on the Agent-Relative/Agent-Neutral Distinction’. Utilitas, 13: 350–56.

Price, Richard (1897).AReview of the Principal Questions in Morals, ed. L. A. Selby-Bigge. Oxford:Clarendon Press.

Pucecetti, Roland (1993). ‘Does the Universe Exist Because it Ought To? A Critique of ExtremeAxiarchism’. Dialogue, 32: 651–57.

Rabinowicz, Wlodek (2008). ‘Value Relations’. Theoria, 74: 18–49.——(2009). ‘Values Compared’. Polish Journal of Philosophy, III (1): 73–96.Rabinowicz, Wlodek and Osterberg, Jan (1996). ‘Value Based on Preferences’. Economics andPhilosophy, 12: 1–27.

Rabinowicz, Wlodek and Rønnow-Rasmussen, Toni (2000). ‘A Distinction in Value: Intrinsicand for its Own Sake’. Proceedings of the Aristotelian Society, 100 (1): 33–51.

——(2003a). ‘Tropic of Value’. Philosophy and Phenomenological Research, 66: 389–403.——(eds.) (2003b). Patterns of Value: Essays on Formal Axiology and Value Analysis, vol. 1. Lund:Lund Philosophy Reports.

——(2004a). ‘The Strike of the Demon: On Fitting Pro-Attitudes and Value’. Ethics, 114:391–423.

——(eds.) (2004b). Patterns of Value: Essays on Formal Axiology and Value Analysis, vol. 2. Lund:Lund Philosophy Reports.

——(2006). ‘Buck-Passing and the Right kind of Reasons’. Philosophical Quarterly, 56: 114–20.Raz, Joseph (2000). ‘The Truth in Particularism’, in Brad Hooker and Margaret Little (eds.),Moral Particularism. Oxford: Clarendon Press, pp. 48–78.

176 BIBL IOGRAPHY

——(2006). ‘The Trouble with Particularism’. Mind, 115: 99–119.Regan, Donald H. (2004). ‘Why Am I My Brother’s Keeper?” in R. Jay Wallace, Philip Pettit,Samuel Scheffler, and Michael Smith (eds.) (2004), pp. 202–30.

Reisner, Andrew (2009). ‘Abandoning the Buck Passing Analysis of Final Value’. Ethical Theoryand Moral Practice, 12: 379–95.

Ridge, Michael (2005). ‘Agent-Neutral vs. Agent-Relative Reason’, in Edward N. Zalta (ed.),The Stanford Encyclopedia of Philosophy (Winter 2005 edition) <http://plato.stanford.edu/archives/win2005/entries/reasons-agent/>.

Roberts, Robert (2003). Emotions: An Essay in Aid of Moral Psychology. Cambridge: CambridgeUniversity Press.

Rønnow-Rasmussen, Toni (1993). Logic, Facts and Representation: An Examination of R. M. Hare’sMoral Philosophy. Lund: Lund University Press.

——(1998). ‘Moral Realists and Moral Experts’, in Carsten Bengt Pedersen and Niels Thamas-sen (eds.),Nature and Lifeworld: Theoretical and Practical Metaphysics. Odense: Odense UniversityPress, pp. 281–99.

——(2001). ‘L. W. Sumner’s account of Welfare’, in Juan Jose Acero, Francesc Camos Abril,and Neftalı Villanueva Fernandez (eds.), Actas del III Congreso de la Sociedad Espanola de FilosofıaAnalıtica. Granada: Universidad de Granada, pp. 281–85.

——(2002a). ‘Hedonism, Preferentialism, and Value Bearers’. Journal of Value Inquiry, 36:463–72.

——(2002b). ‘Instrumental Values–Strong and Weak’. Ethical Theory and Moral Practice,5: 23–43.

——(2003). ‘Subjectivism and Objectivism; An Outline’, in W. Rabinowicz and ToniRønnow-Rasmussen (eds.), pp. 246–63.

——(2004). ‘Recension av Stephen Darwalls Welfare and Rational Care’. Sats—Nordic Journal ofPhilosophy, 4 (2): 171–80.

——(2007a). ‘Analysing Personal Values’. Journal of Ethics, 11 (4): 405–35.——(2007b). ‘Dislodging Butterflies from the Supervenient’, in Stephen Voss (ed.), Philosophical

Anthropology, vol. IX, Proceedings of the 2003 Istanbul World Congress. Ankara: Philosophi-cal Society of Turkey, pp. 236–37.

——(2008a). ‘Buck-Passing Personal Values’, in David Chan (ed.), Values, Rational Choice, andthe Will: New Essays in Moral Psychology. Stevens Point: Springer, pp. 37–51.

——(2008b). ‘Love, Value and Supervenience’. International Journal of Philosophical Studies, 16 (4):495–508.

——(2009a). ‘Normative Reasons and the Agent-Neutral/Relative Dichotomy’. Philosophia,37: 227–43.

——(2009b). ‘On For Someone’s Sake Attitudes’. Ethical Theory and Moral Practice, 12: 397–411.Rønnow-Rasmussen, T., Petersson, B, Josefsson, J., and Egonsson, D. (eds.) (2007). Hommage aWlodek. Philosophical papers dedicated toWlodek Rabinowicz. <www.fil.lu.se/hommageawlodek>.

Rønnow-Rasmussen, Toni and Zimmerman, Michael J. (eds.) (2005). Recent Work on IntrinsicValue. Dordrecht: Springer.

Rosati, Connie (2008). ‘Objectivism and Relational Good’. Social Philosophy & Politics, 25 (1):314–49.

Ross, W. D. (1930). The Right and the Good. Oxford: Clarendon Press.——(1939). Foundations of Ethics. Oxford: Clarendon Press.

BIBL IOGRAPHY 177

Royce, Josiah (1891). ‘Discussions: The Outlook of Ethics’. International Journal of Ethics, I:106–11.

Scanlon, Thomas Michael (1998). What We Owe to Each Other. Cambridge, MA: HarvardUniversity Press.

——(2002). ‘Reasons, Responsibility, and Reliance: Replies to Wallace, Dworkin, and Deigh’.Ethics, 112: 507–28.

Scheffler, Samuel (ed.) (1986). Consequentialism and Its Critics. New York: Oxford UniversityPress.

Schilpp, P. A. (ed.) (1942). The Philosophy of G. E. Moore (The Library of Living Philosophers).New York: Tudor Publishing Company.

Schroeder, Mark (2007). ‘Reasons and Agent-Neutrality’. Philosophical Studies, 135: 279–306.——(2009). ‘Buck-Passers’ Negative Thesis’. Philosophical Explorations, 12: 341–47.Selby-Bigge, L. A. (ed.) (1897). British Moralists, vol. 2. Oxford: Clarendon Press.Sen, Amartya (1999). Development as Freedom. Oxford: Oxford University Press.Sidgwick, Henry (1907 [1874]). The Methods of Ethics, 7th edn. London: Macmillan.Skorupski, John (1995). ‘Agent-Neutrality, Consequentialism, Utilitarianism . . . : A Termino-

logical Note’. Utilitas, 7: 49–54.——(2007a). ‘Buck-Passing about Goodness’, in T. Rønnow-Rasmussen, B. Petersson, J.Josefsson, and D. Egonsson (eds.), pp. 1–15.

——(2007b). ‘What is Normativity?’ Diputatio, II (23): 247–69.——(2010). The Domain of Reasons. Oxford: Oxford University Press.Smith, Adam (1759/1790). The Theory of Moral Sentiments. London: A. Millar.Smith, Michael (2003). ‘Neutral and Relative Value after Moore’. Ethics, 113: 576–98.Sobel, David (2002). ‘Varieties of Hedonism’. Journal of Social Philosophy, 33: 240–56.——(2005). ‘Pain for Objectivists: The Case of Matters of Mere Taste’. Ethical Theory and Moral

Practice, 8: 437–57.Sprigge, Timothy L. S. (1988). The Rational Foundations of Ethics. London: Routledge.Strandberg, Caj (2004). Moral Reality: A Defense of Moral Realism. Lund: Media-tryck, LundUniversity.

——(2008). ‘Particularism and Supervenience’, in Russ Shafer-Landau (ed.), Oxford Studies inMetaethics, vol. 3. Oxford: Oxford University Press, pp. 249–60.

Stratton-Lake, Philip (2005). ‘How to Deal With Evil Demons: Comment on Rabinowicz andRønnow-Rasmussen’. Ethics, 115: 788–98.

Sumner, Leonard. W. (1996). Welfare, Happiness, and Ethics. Oxford: Clarendon Press.Swanton, Christine (1995). ‘Profiles of the Virtues’. Pacific Philosophical Quarterly, 27: 335–44.Urban, Wilbur M. (1909). Valuation, Its Nature and Laws, Being an Introduction to the General

Theory of Value. London: Sonnenschein/New York: Macmillan.Vlastos, Gregory (1969). ‘Human Worth, Merit and Equality’, in Joel Feinberg (ed.), pp.141–52.

Wallace, R. Jay (2002). ‘Scanlon’s Contractualism’. Ethics, 112: 429–70.Wallace, R. Jay, Pettit, Philip, Scheffler, Samuel, and Smith, Michael (eds.) (2004). Reason andValue: Themes from the Moral Philosophy of Joseph Raz. Oxford: Clarendon Press.

Wiggins, David (1987). ‘A Sensible Subjectivism?’ in D. Wiggins, Needs, Values, Truth: Essays inthe Philosophy of Value. Oxford: Blackwell, pp. 185–214.

Williams, Bernard (1972). Morality. New York: Harper and Row.

178 BIBL IOGRAPHY

——(1985). Ethics and the Limits of Philosophy. Cambridge: Cambridge University Press.Williams, Donald Carry (1953). ‘On the Elements of Being’. Review of Metaphysics, 7: 3–18.Zimmerman, Michael J. (2001a). The Nature of Intrinsic Value. Lanham: Rowman and Littlefield.——(2001b). ‘Intrinsic Value and Individual Worth’, in Dan Egonsson, Jonas Josefsson, BjornPetterson, and Toni Rønnow-Rasmussen (eds.) (2001), pp. 123–38.

——(2009). ‘Understanding What’s Good for Us’. Ethical Theory and Moral Practice, 12: 429–39.

BIBL IOGRAPHY 179


Index

abstract 4, 12, 20, 25, 26, 59, 158, 159, 166, 167entities 143, 155, 164, 165, 168

admirability-makers 69agent-neutral/relative reason distinction:essentialist sense 128–9number approach 129, 131–2

Alexander, Samuel 4 n. 5Alfano, Mark 29Allen, Richard T. 52 n. 12Alm, David xiii, 71 n. 16, 162amnesia 79analysis:fitting attitude analysis of value 20, 22, 24, 26,33, 90–1, 98–9, 121, 135, 140

naturalistic analysis 28, 105–6Anderson, Elisabeth 25 n. 21Anna v, 1, 130Armstrong, David M. 166, 184 n. 26atheism 112atomism 148–9attitude:and non-fungible persons 78–9, 80–1, 91constitutive ground of value 15dual role 37experienced as correct 21identity-involving attitudes 56in the background 15in the foreground 15non-discerning 37, 68, 73object-given reasons for 35, 38, 41, 43state-given reasons for 35, 41, 43

Aydede, Murat 168 n. 27

Baron, Marcia W. 25 n. 21Baudelaire, Charles 1Baumann, Peter xivbeing valued and being valuable 16 n. 29benefit(s) 2, 65, 84, 108Bentham, Jeremy 155 n. 5, 158better off 85, 88biconditional account 42–43Bjornsson, Gunnar xivBPA; see also buck-passing account 27Bradley, Ben 8, 167–8Brady, Michael xiv, 111 n. 1, 120 n. 7Brandt, Richard 21, 155n. 5, 156–7,

168 n. 27, 173–4Brännmark, Johan xiii, 4 n. 5, 6 n. 11, 7 n. 14,

52 n. 13, 53 n. 14, 61, 83 n. 7Brax, David xix, xiii, 168 n. 27

Braun, Eva 114, 117–19Brentano, Franz viii, x, 34 n. 1 and 2, 21–4,

39–40, 42British emergentists 4 n. 5Broad, Charlie D. 4 n. 5, 21, 49 n. 4Broome, John xiii, 44–5, 129 n. 8, 135 n. 14,

149 n. 25Brülde, Bengt 15 n. 29buck-passer 29, 43, 45, 97buck-passing account 27, 29 n. 28, 42–5Burges, J. A. 32 n. 44Butchvarov, Panayot 163 n. 18Butler, Joseph 124 n. 11

butterfly 11Bykvist, Krister xiii, 33 n. 1

Campbell, Keith 164 n. 21, 166Carlson, Erik xiiicart before the horse 163–4, 171‘Catch 22’ 34 n. 4Chan, David xiiichild molester 80Chisholm, Roderick M. 5 n. 9 and n. 10, 21 n. 6circularity 30 n. 36 and 37, 31–32, 48 n. 2cognitivism 23 n. 11Collin, Finn xivColonel Cathcart 34 n. 4Colonel Korn 34 n. 4concept of obligation 22concrete:

the distinction between abstract and 164–6entities 12, 25–6, 31, 59, 104, 124–5,135 n. 15, 154–5, 158

Conflation Problem 34 n. 2consciousness 157, 168

false 124constituted:

values 12, 16–17, 89, 115reasons 146–7

constitutive:attitude 14grounds of (final) value 9, 13, 19of reason 134, 147, 148, 152of welfare 84–5, 103, 107, 118, 170powers 54properties of the car 7

content-reason 40–1correct to prefer 22Cruft, Rowan xivCrisp, Roger 27–9, 34 n. 4, 42–4, 148–9

Dancy, Jonathan xiv, 10 n. 19, 20 n. 1, 24 n. 19,80 n. 2, 127, 133, 137, 144–6, 148 n. 152,153

Danielsson, Sven xiii, 7 n. 14, 11, 21, 24 n. 18,34 n. 3, 40–2

D’Arms, Justin 31, 34 n. 2, 72Darwall, Stephen 47, 88, 109, 114–18DEFA 53–4, 88De Facto Attitude analysis of Personal Value 53Deigh, John 25 n. 20delicacy and gracefulness 28Demon 34–42, 45deontic notion 24 n. 18derivative value 104, 109, 155, 161, 163, 167diachronic identity 57, 80Dickie, George 52dignity 88, 158distinctive feeling view 21 n. 5duplication 81 n. 3Dworkin, Gerald 25 n. 20

Egonsson, Dan xiii, 153, 158 n. 13Ehrenfels, Christian von 21Elly v, 70enabler 10 n. 19, 127, 144–6, 148–52end 2–3, 9, 15–16, 20–2, 26–7, 33, 55,

60–4, see also sake; objectiveEpicurus 155 n. 5, 158Ewing, A. C. xi, 21, 22–4, 25 n. 19, 30–1,

39, 42Ewing’s metaethical views 23 n. 11 and n. 13.Ewing’s notion of pro attitude 23

Explanatory reasons 79, 133–5, 138extrinsic final value 5–7, 15–16, 26, 53–4, 122,

153–4

FA 20, 32, 33, 102fact:as obtaining of state of affairs 143, 154–5,

159–60fact-independently 11FAP 109, 111–12, 114, 116–22, 124FAP* 78–9, 83–7, 92–3FAP-intrinsic 90Favour x for its own sake for b’s sake 58–62favourer 127, 144–51favourer on condition 146, 148–9

Feldman, Fred xiv, 21 n. 5, 156Fetishism 122–4FFSS attitude 58–61final value:extrinsic 5–7, 15–16, 26, 53–4, 122, 153–4the FA analysis of 27, 33is invariant over possible worlds 12, 169intrinsic 89, 91isolating extrinsic 5–7thick and thin varieties of 27

Findlay, John N. 21 n. 6fine-grained conceptual tools 19Fitting-attitude analysis 20, see also FAfounding father 21history of 22

Fittingness:as an ethical concept 23the ‘ought’ of 23as primitive notion 39

Foot, Philippa 7 n. 14Frankfurt. Harry J. 64, 72 n. 17Fritzson, Fritz-Anton xiv, 14 n. 27FSS attitude: 55dropping an 78–9

faces a reductio ad absurdum 78identity-involving attitudes 56Non-identity attitude 74–6two kinds of 57–8

Geach, Peter 6–7Gert, Joshua 24Gilbert, Margaret xiii, 121God 10, 66, 112, 133–4, 164–5GOL 99, 102Goldman, Alan xivGOLTP 99–102GOLTPC 102Good/goodness:absolutely 50intrinsic 3, 50, 52life 122–5like yellow 4plain 6–7relational 98–9, 103two distinctions in 9, 15

good-for:and welfare 46, 83–8, 103–4, 106–7, 114,

121–2Grau, Christopher xiv, 65 n. 9, 82 n. 4

H1 154–5H1* 155Haddock, Adrian xivHadji, Ish xivHare, Richard M. 4 n. 5, 92, 119Hartmann, Eduard von 3 n. 3Heath W. White xiv, 36 n. 11Heathwood, Chris xiv, 121–2hedonic tone 156–9, see also properties:

hedonicHedonismattitudinal version of 156

Heeger, RobertHeyde, Erich 17 n. 32Hieronymi, Pamela 44–5Hitler, Adolf 114–15, 117–20holism 148–9

182 INDEX

Humbertone, I. L. 32 n. 44Hurka, Thomas xiv, xxvi, 50–1, 93 n. 17,

96–8, 102–3, 105

Identifiers 56, 77, 82–3, 93 n. 16Identity attitudes 56, 75identity-involving attitudes 56I-justifies 71individual souls 164Instrumental valueIn the Strong evaluative sense 51In the Weak evaluative sense 52

interest 87intrinsic personal values 89, 90Invariance thesis 12–13, 169Isolation test:intentional version 5ontological version 5

Jacobson, Daniel 31, 34 n. 2, 72Janus values 110–11Juliet 65–6justice 1, 123, 168Justification:two sorts of 71identity-justification 71, 74, 93 n. 16, 94is called for 73property 71, 94

Kagan, Shelly 8, 16 n. 30, 53 n. 15, 154Kantians 8Kath 81 n. 3Katz, Leonard D. 157 n. 11, 168 n. 27Kay 81 n. 3Kim, Jaegwon 11Kirshin, Simon xivKorsgaard, Christine M. 9, 15, 16 n. 31, 61, 104,

151 n. 28Kupperman, Joel 34 n. 4

La Giaconda 13Lang, Gerald 41 n. 21Langton, Rae 26, 154 n. 3Lapie, Paul 3 n. 3Lemos, Noah M. xiv, 4–5, 12 n. 23, 25 n. 21,

163 n. 19Love:a volitional state 64as apprehension 64characterized as unconditional 80higher mode of taking pleasure 21occurrent 67Pascal's view of 65romantic view of 64, 80–1

Mackie, John L. 18 n. 34Maurin, Ia xiv

Massin, Olivier xivMcKeever, Sean 149 n. 25McNaughton, David 126 n. 2, 127, 129Meinong, Alexius 21, 30 n. 36Mellor, D. Hugh 166 n. 26metaethical 6–7, 23 n. 13, 52, 105, 134metaphysically queer 161Miller, Adrian xivMoore, G. E. 3–6, 8, 12, 17, 25 n. 19, 34 n. 3,

91 n. 14, 96, 98, 102, 103 n. 12, 47–50,153, 169

Moorean, 8–9, 22, 91, 93 n. 17, 96, 98–102,103 n. 12, 153

Moral:necessity 11obligation 23ought 24reason 23, 119thing to do 99 n. 7wrongness 32

Morgan, C. Lloyd 4 n. 5Mulligan, Kevin xiv, 27 n. 24, 30 n. 36,

31 n. 42, 55 n. 1, 64 n. 8, 83 n. 7Musschenga, Albert xivmy own good 49–50, 102

Nagel, Thomas 126–30Napoleon’s hat 53Newton-Smith, W. 67 n. 10nominalism 164 n. 21non-derivative value 62, 153–5, 160, 163,

168, 171non-fungible persons 67–8, 78–9, 91, 93–4non-natural property 6, 29normative concepts 24, 103normative reason: see also reason

Broome’s view of 44Hieronymi’s view of 44–5statements of 129ultimately desire- or attitude-based 116

Nozick, Robert 52 n. 12

objective list theory 154objective 26, 61–2, 64Objectivism 14–15, 17–18, 54, 89, 162, 169obtaining of states of affairs 12, 135, 143, 147,

154–5, 158–60, 166Oddie, Graham xivOlson, Jonas xiii, xiv, 7 n. 14, 34 n. 3, 40–2, 21,

23 n. 11, 24 n. 18, 25 n. 21, 28, 29 n. 27,120 n. 7, 133 n. 11

Oliver, A. 166 n. 26organic unity 61Osterberg, Jan xiii, 14, 155Ought:

of moral obligation 23of reasonableness 23–4

INDEX 183

parental love 80Parfit, Derek xiv, 35, 38–41, 121 n. 8, 126–7,

129 n. 7, 133, 135 n. 14Parfit’s approach 131, 132particularity 82Pascal, Blaise 65patchwork solutions 119PEF 132, 137, 139, 141–2, 145, 149PEFI 132, 137, 152personalizability feature 132, 141, 152

Personal value:Distinguished from impersonal value 55being intrinsic 90

Persson, Ingmar xiv, 38 n. 15, 133 n. 11,164 n. 20

Petersson, Bjorn xiiiPettit, Philip 128–30, 133 n. 11Piller, Christian xiv, 56 n. 2, 120pleasing in itself 21pleasure:distinctive feeling view 21 n. 5feeling of pleasure 21, 167higher mode of taking 21instantiation of 163, 167of the intellect 158 n. 13Ryle’s view of 156 n. 8sensory 156 n. 9, 157

pluralistic position 7, 22Portmore, Douglas W. xiv, 129, 141–2positive tone 156 n. 8, see also hedonic tonepractical reason 95 n. 1, 116, 144, 152Preferentialism, 153cannot endorse H1 154core claims 154object version 155satisfaction version 155

Price Richard 4 n. 7private value 47, 91, 127primitive notion:‘correctness’ as 39, 42‘fittingness’ as 39, 40, 42

Princess Diana’s dress 9Properties:attitude-justifying roles 70essential 35dependent 10hedonic 20, 155–9, 163, 167–8, 170–1instantiations of 67, 163–5, 167lower-order 27–8, 42–3non-essential 9, 30–1, 34, 169non-natural 6, 29reason-providing 25 n. 20, 29, 43–5relational 5, 6, 9, 36, 53, 83, 90, 93–4, 153subjacent 10, 24, 25 n. 20, 52, 69, 71,

94, 111supervenient 4, 10–11, 14–15, 94value-making 14, 25, 36, 69–70, 94, 169

Quante, Michael xivquasi-judgmentalism 31

R1 44–5R2 44–5Rabinowicz, Wlodek xviii, xix, 8–9, 14,

16 n. 30, 20 n. 1, 21 n. 6, 25 n. 21, 29,30 n. 32, 33, 34 n. 5, 35–7, 39, 41 n. 20,61, 140, 155, 160 n. 162–3, 164 n. 20,165 n. 24

Rawling, Piers 126 n. 2, 127, 129Raz, Joseph 136, 145, 148–50realism 73 n. 20, 162reason: see also moral reasoncontentdefeasible holding-reason 40–2holding 40–2qua facts 91, 135, 137, 139, 146, 150–2right kind of 35, 39, 41, 70, 90object-given 35, 38, 41, 43state given 35, 41, 43three varieties 134reason for, see personalizability;

recalcitrant emotions 72reductandum 160–1, 163reductans 160–1reductio ad absurdum 78Regan, Donald H. 48, 51 n. 10, 96–9, 101, 103,

104 n. 13, 105, 107Reisner, Andrew xiv, 27 n. 24, 39 n. 17, 42replica 81–3RIA 144Ridge, Michael 127, 129, 145 n. 25Roberts, Robert C. 70 n. 15Robinson, Paul xivRomeo 65R�nnow-Rasmussen, Toni 9, 12, 14, 16 n. 30,

20 n. 1, 21 n. 6, 25 n. 21, 32 n. 43, 34 n. 5,36 n. 10, 37, 39 n. 18, 52 n. 11, 53 n. 15, 56,61, 67, 98 n. 5, 99 n. 8, 160 n. 14 and 15,162, 164 n. 20, 165 n. 24 and 25

Rosati, Connie 48–9, 51 n. 19, 85, 96, 99–105,107. 109, 122 n. 9

Ross, W. D. 4, 26, 30–1, 145–6Ryle, Gilbert 156 n. 8

Sake 61–3saucer of mud 34–35, 36 n. 10, 37–8, 40, 45Scanlon, Thomas xiv, 21, 25 n. 20, 27 n. 23,

48 n. 2, 133Scheffler, Samuel 126 n. 2.Schroeder, Mark xiv, 45 n. 27, 132 n. 9self-destructive life 88 n. 11Sen, Amartya 163 n. 17sexuality 72Shah, Nishi xivShakespeare 1

184 INDEX

Sidgwick, Henry 21 n. 3, 47, 50 n. 9, 95 n. 1,102, 124 n. 11, 155 n. 5, 158

Simplicity 7, 161–2Sinnott-Armstrong, Walter 148Skorupski, John xiv, 21, 32, 38–41, 127–8, 129

n. 8, 135 n. 14, 150 n. 27Smart, J. J. C. 155 n. 5Smith, Adam 21 n. 3Smith, Michael 99, 103, 133 n. 11Smith’s being happy 4Sobel, David 157, 168 n. 27Sobel, Howard xiiiSocrates 3 n. 1, 11Sofia vsoldiers 87source 9, 15, 16, 104spatiotemporal location 100, 164–5Sprigge, T. L. S. 156 n. 8Stocker, Michael xiiiStrandberg, Caj xiv, 10 n. 19, 145 n. 21,

149 n. 25Stratton-Lake, Philip 36 n. 10subjacent properties, see properties,Subjectivism 12, 13 n. 25, 14–18subjectivists 12–13, 95Supervenience:base of 9, 13, 24 n. 16, 25, 36, 68–9, 90, 106consistency intuition 10–11go from intra-world to inter-world 11

Strong supervenience 11Svensson, Daniel xiiiSvensson, Frans 133 n. 11Swanton, Christine 25 n. 21Swedish priest 72 n. 18

Tersman, Folke xiv, 34 n. 5Thin-thin synonymity 84–4, 105–8Thin-thick synonymity 105–7Thomson, Judith Jarvis 7 n. 14Timaeus 3 n. 1Tropes 83, 160, 164–7dyadic 165

uniqueness 68, 153universale 164–5universalizability 79, 91–4Urban,Wilbur Marshall 3 n. 3

Value: see also derivative, extrinsic; final, intrinsic,instrumental; private

bearer monism 13, 160–1, 163 n. 17,171

bearer dualism 153bearer pluralism 170–1computation 162constitution 9, 14–15, 17, 169end 2–3, 9, 15–16, 25–6, 61–4realism 162source 9, 15–16, 104

Value judgements:being universalizable 79, 92

Value-for:Abel 48a teenager 111Cain 48a comatose person 100–1Hitler 114–15, 117–20more than one notion of 84

Vieth, Andreas xivVlastos, Gregory 3 n. 4Voorhoeve, Alex xiv

Wallace, Ray J. 25 n. 20Welfare 19, 84–8, 103–4, 106, 114–15,

117–18, 122, 124dependent on the attitudes of the welfaresubject 17

good-for 46, 83–5, 107intrinsic value of 121respect 88

Welfarists 20, 84–85, 99Well-being 19, 46, 86, 99, 106, 124White, Heath W. xiv, 36Wiggins, David 30 n. 37Wikforss, Åsa xivWilliams, Bernard 7 n. 14, 27 n. 22Williams, Donald. C. 164, 166WKR, see wrong kind of reason problemwrong kind of reason problem 33, 35, 37–8, 40,

41 n. 21, 44 n. 26

Yeats, W. B. 64Yossarian 34 n. 4

Zimmerman, Michael J. xi, xiii, xiv, 3 n. 1,4–5, 7, 12 n. 22, 21, 24 n. 16, 25 n. 21,81 n. 3, 84 n. 8, 97, 99 n. 7, 106 n. 15,107–8, 116 n. 5, 134 n. 12, 138, 155 n. 5,160 n. 14, 160 n. 16, 164 n. 20, 165,166 n. 26

INDEX 185

Documents

Personal Value