Value Theory 9781472530882, 9781472532923, 9781472594280, 9781472525307

Table of contents :
FC
Half title
Bloomsbury Ethics
Title
Copyright
Contents
Acknowledgements
Preface
1 Value and Normativity
1.1 Introduction
1.2 Which evaluations?
1.3 The idea of value theory
1.4 Value and normativity
1.5 Overview
1.6 Meta-ethical neutrality
1.7 Value theory: The questions
2 Meet the Values: Intrinsic, Final & Co.
2.1 Introduction
2.2 Final and unconditional value: Some philosophical examples
2.3 Intrinsic value and final value
2.4 The reduction to facts
2.5 Intrinsic and conditional value
2.6 Elimination of extrinsic value?
2.7 Summary
3 The Challenge against Absolute Value
3.1 Introduction
3.2 Geach and attributive goodness
3.3 Foot and the virtues
3.4 Thomson and goodness in a way
3.5 Zimmerman’s ethical goodness
3.6 A better reply: Absolute value and fitting attitudes
3.7 Summary
4 Personal Value
4.1 Introduction
4.2 Moore on good and good for
4.3 Good for and fitting attitudes
4.4 Moore strikes back?
4.5 Agent-relative value
4.6 Impersonal/personal and agent-neutral/agent-relative
4.7 Summary
5 The Chemistry of Value
5.1 Introduction
5.2 Supervenience and other relations
5.3 Organic Unities
5.4 Alternatives to organic unities: Virtual value
5.5 Alternatives to organic unities: Conditional value
5.6 Holism and particularism
5.7 Summary
6 Value Relations
6.1 Introduction
6.2 The trichotomy thesis and incomparability
6.3 A fitting attitude argument for incomparability
6.4 Against incomparability: Epistemic limitations
6.5 Against incomparability: Parity
6.6 Parity and choice
6.7 Parity and incomparability
6.8 Summary
7 How Do I Favour Thee?
7.1 Introduction
7.2 Three dimensions of favouring
7.3 Responses to value: Maximizing
7.4 Two concepts of intrinsic value?
7.5 Summary
8 Value and the Wrong Kind of Reasons
8.1 Introduction
8.2 The fitting attitude account and its rivals
8.3 The wrong kind of reasons problem
8.4 The structure of the problem and an initial response
8.5 Reasons for what?
8.6 Characteristic concerns and shared reasons
8.7 Circular path: No-priority
8.8 Summary
Bibliography
Index

Citation preview

Value Theory

BLOOMSBURY ETHICS Series Editors: Thom Brooks, Reader in Law, Durham Law School, UK and Simon Kirchin, Philosophy, University of Kent, UK Bloomsbury Ethics is a series of books on established and new areas in moral philosophy. Each book is designed both to introduce upperlevel undergraduates and postgraduates to a key field in ethics, and to develop a particular viewpoint within that field designed to appeal to researchers. All areas of moral philosophy are covered, from the theoretical to the practical. New proposals are always welcome. Please contact the series editors.

Titles available in the series: Intuitionism David Kaspar

Moral Motivation Leonard Kahn

Reasons Eric Wiland

Moral Realism Kevin DeLapp

Autonomy Andrew Sneddon

Virtue Ethics Nafsika Athanassoulis

Forthcoming: Ethics Without Intention Ezio Di Nucci

Luck Egalitarianism Kasper Lippert-Rasmussen

Trust, Ethics and Human Reason Olli Lagerspetz

Moral Skepticism Basil Smith

Moral Principles Maike Albertzart

BLOOMSBURY ETHICS

Value Theory FRANCESCO ORSI

Bloomsbury Academic An imprint of Bloomsbury Publishing Plc 50 Bedford Square London WC1B 3DP UK

1385 Broadway New York NY 10018 USA

www.bloomsbury.com Bloomsbury is a registered trade mark of Bloomsbury Publishing Plc First published 2015 © Francesco Orsi 2015 Francesco Orsi has asserted his right under the Copyright, Designs and Patents Act, 1988, to be identified as the Author of this work. All rights reserved. No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage or retrieval system, without prior permission in writing from the publishers. No responsibility for loss caused to any individual or organization acting on or refraining from action as a result of the material in this publication can be accepted by Bloomsbury or the author. British Library Cataloguing-in-Publication Data A catalogue record for this book is available from the British Library.

ISBN: 978-1-47252-530-7 Library of Congress Cataloging-in-Publication Data A catalog record for this book is available from the Library of Congress Typeset by Fakenham Prepress Solutions, Fakenham, Norfolk NR21 8NN

CONTENTS

Acknowledgements  viii Preface  ix

1 Value and Normativity  1 1.1 Introduction  1 1.2 Which evaluations?  2 1.3 The idea of value theory  6 1.4 Value and normativity  8 1.5 Overview  16 1.6 Meta-ethical neutrality  17 1.7 Value theory: The questions  22

2 Meet the Values: Intrinsic, Final & Co.  25 2.1 Introduction  25 2.2 Final and unconditional value: Some philosophical examples  27 2.3 Intrinsic value and final value  31 2.4 The reduction to facts  35 2.5 Intrinsic and conditional value  39 2.6 Elimination of extrinsic value?  41 2.7 Summary  43

3 The Challenge against Absolute Value  45 3.1 Introduction  45 3.2 Geach and attributive goodness  46

vi CONTENTS

3.3 Foot and the virtues  50 3.4 Thomson and goodness in a way  53 3.5 Zimmerman’s ethical goodness  56 3.6 A better reply: Absolute value and fitting attitudes  58 3.7 Summary  61

4 Personal Value  63 4.1 Introduction  63 4.2 Moore on good and good for  63 4.3 Good for and fitting attitudes  66 4.4 Moore strikes back?  69 4.5 Agent-relative value  73 4.6 Impersonal/personal and agent-neutral/agent-relative  77 4.7 Summary  80

5 The Chemistry of Value  81 5.1 Introduction  81 5.2 Supervenience and other relations  81 5.3 Organic Unities  85 5.4 Alternatives to organic unities: Virtual value  90 5.5 Alternatives to organic unities: Conditional value  93 5.6 Holism and particularism  98 5.7 Summary  100

6 Value Relations  101 6.1 Introduction  101 6.2 The trichotomy thesis and incomparability  101 6.3 A fitting attitude argument for incomparability  104 6.4 Against incomparability: Epistemic limitations  107 6.5 Against incomparability: Parity  109 6.6 Parity and choice  110 6.7 Parity and incomparability  113 6.8 Summary  115

CONTENTS

vii

7 How Do I Favour Thee?  117 7.1 Introduction  117 7.2 Three dimensions of favouring  117 7.3 Responses to value: Maximizing  126 7.4 Two concepts of intrinsic value?  130 7.5 Summary  134

8 Value and the Wrong Kind of Reasons  135 8.1 Introduction  135 8.2 The fitting attitude account and its rivals  136 8.3 The wrong kind of reasons problem  141 8.4 The structure of the problem and an initial response  143 8.5 Reasons for what?  145 8.6 Characteristic concerns and shared reasons  151 8.7 Circular path: No-priority  156 8.8 Summary  158 Bibliography  159 Index  169

ACKNOWLEDGEMENTS

I wish to thank those who have most inspired and encouraged me through the years: Eugenio Lecaldano, Tito Magri, Jonathan Dancy, Philip Stratton-Lake. A collective thanks also goes to fellow graduate students and colleagues in Rome, Reading, and Tartu, with a special mention for Valerio Salvi, with whom countless hours were spent discussing these issues. I am grateful to audiences and students for their feedback on parts of this book. I thank my Bloomsbury editors for their enthusiasm for this project. This book was written with the help of PUT grant 243 (Estonian Research Council). Finally, I thank my family and Julia for their patience and constant support.

PREFACE

We value things: relationships, particular people, particular events, or particular objects. This means that we feel and behave towards them as if they possessed a certain value, whether or not we consciously judge them to be valuable. Sometimes we also evaluate the same and other things as good, significant, worthy of our interest (or bad, worthless, and so on). In this case, we might well fail to feel and behave towards them as we think we should. But both in valuing and evaluating things not only do we express our attitudes or convictions; we do so by employing, more or less explicitly, a rich and sophisticated conceptual repertoire. This book is an attempt to introduce and examine some of this conceptual repertoire, as manifested both in ordinary evaluative practices and in philosophical theories of what is good: what is it for something to have value for its own sake? Can disparate sorts of things (say, people, hats, abstract objects) all have value? In what sense is, say, health good for me? And can something be good, but good for nobody at all? Can anything good come out of what isn’t good? Are there incomparable values? In order to navigate these different questions, and give unity to an otherwise diverse subject matter, I have followed a basic insight: in our evaluative practices not only do we express attitudes of preference or aversion, but also claim or at least feel that our preferences and aversions are justified by good reasons or, as some like to say using newly fashionable terms, are fitting, suitable, appropriate. Whether this insight provides a reliable and helpful compass is ultimately for the reader to decide.

CHAPTER ONE

Value and Normativity 1.1 Introduction We use value talk all the time. Most of the time, we don’t need to mention the word ‘value’ or similar terms like ‘worth’ or ‘merit’. More commonly we use evaluative expressions like ‘good’, ‘bad’, ‘better’, ‘the best’, ‘great’, ‘fine’, ‘excellent’, ‘poor’, ‘terrible’ and so on. To utter such evaluative words of course is not always to evaluate something positively or negatively. The exclamation ‘Good heavens!’ normally doesn’t say anything about the value of the heavens. Also, we can use ‘good!’ or ‘that’s good’ simply to express or report our satisfaction with something, regardless of whether we think it worthy of our satisfaction, and so ‘good’ in an evaluative sense. At other times still, evaluative words are used to register other people’s evaluations, without joining in ourselves: if I know nothing about row-crop tractors, except the ranking that an expert commission gave of models produced in 2013, I might call one 2013 model better than another simply to report the commission’s ranking. Of course, given its primary evaluative association, such a judgement might be misleading, since you might well infer that I do believe one model to be better than another (and thus, perhaps, that I am advising you to get that), whereas all I mean is really that the commission decided so. It is also important to keep in mind that evaluation need not be articulated via value terms, or indeed be verbally articulated at all. Often we can tell a person’s values and commitments from

2

VALUE THEORY

her behaviour and emotional reactions much more than from what she says (to others or even to herself). We can understand a given community’s unspoken and unwritten values or norms, or at least some of them, by interpreting their social practices, rituals, internal conflicts, and so on. However, in this book I will explore conceptual questions about value, rather than about evaluation as a social and psychological phenomenon. So I will treat the practice of evaluation or ‘people’s values’ as a window which reveals, or at least purports to reveal, putative truths about value. Therefore evaluations which are verbally expressed or at least expressible in value terms are the starting point. In this chapter I introduce the conceptual background of the book. First, I draw a distinction between different evaluations, in order to focus on the value concepts that will be discussed. Then I define and explain the project of value theory, as I pursue it here. In section 1.4 I lay out the central assumption or guiding thread: value is normative. I introduce two views intended to articulate this notion: the fitting attitude account and the buck-passing account of value. On this basis, I then give an overview of the contents of the book, showing how the diverse structural questions about value here investigated will be addressed through the lenses of the idea that value is normative. I close the chapter with a remark about how my approach will be neutral among different meta-ethical views, and with a graphical sketch of the contents of the book.

1.2 Which evaluations? Evaluation comes in so many linguistic forms and spans so many different categories. An exhaustive classification of evaluative judgements is a task for another occasion. But there is a conceptual distinction that can help to isolate the sort of evaluations that will keep us occupied in what follows. Some evaluations are pure, or purer than others, i.e. they carry minimal or no descriptive or informative content about the thing evaluated. Following Bernard Williams (1985), these can be called thin evaluations. When a certain concept is used in such an evaluation, it is a thin evaluative concept. As such, they contrast with thick evaluations (and thick concepts), which do carry more than minimal information about



Value and Normativity

3

the thing evaluated. It is not easy to find this theoretical distinction reliably mirrored in aspects of grammar or in characteristic words, but a few remarks may help to grasp the point. For instance, thin evaluation can be seen in unqualified uses of words such as ‘good’ and ‘bad’ and their comparatives. These can be applied to concrete objects or states of affairs: ‘this movie is bad’, ‘following her advice was good’. In philosophy, thin general evaluative statements are often made about abstract objects: ‘pleasure is good’, ‘knowledge is generally better than ignorance’. But one can also use abstract nouns like ‘value’ (or ‘good’ as a noun): ‘friendship is/has the greatest value’, and verbs like ‘to matter’: ‘friendship matters more than honour’. Now, these terms can be qualified by adverbs or adverbial phrases like ‘intrinsically’, ‘in itself’, ‘instrumentally’, or by the respective adjectives. Depending on the adverb, the resulting evaluation will be more or less thick, that is, will carry more or less descriptive or informative content about the thing evaluated. Sometimes it won’t carry any additional descriptive content, but simply stress the importance of a value: ‘helping others in need has incalculable value’. In other cases it will carry some minimal content: to say that knowledge is instrumentally good hints at the effects or consequences of knowledge, albeit not saying what these are, except that they are good. But in other cases, qualifications can carry quite a lot of informative content: ‘This photograph has personal value (for me)’, ‘Wedding rings have sentimental value (for many people)’, ‘This song has no artistic value’, ‘Your results are scientifically worthless’, etc. These qualifications assign the thing evaluated to some category or other of value, and so imply some substantial information about them, or about their relations to other things or people. Such information need not be very detailed, of course. To speak of personal or sentimental value implies something about the historical or affective connection of an object to people. To speak of artistic or scientific value turns one’s attention to the sort of features which justify the inclusion of an object in either category. In these cases, we are making thicker evaluations. Also, evaluative words can be used predicatively, as in the sentences used so far, which have the general form: x is good.

4

VALUE THEORY

But very commonly they are used attributively, i.e. they accompany a noun: ‘Sarah is a great dancer’, ‘Liquorice soup is a terrible entrée’, ‘WWII was the worst war ever’. In general: x is a good F. Often, these attributive evaluations are quite thick: for Sarah to be a great dancer, it must be true that she masters the skills of a dancer to a considerable degree. So attributive uses of value terms can convey a great deal of information together with the evaluation. But this need not be the case: it is not at all obvious what information is carried by the statement that WWII was the worst war ever, because it is not at all obvious what a war has to be like, as a war, to be worse than another. All we definitely know, of course, is that WWII was a war, and so that there was fighting and the like (for another grey-area example: ‘Valerie is a very good child’). Attributive evaluations then are not guaranteed, just in virtue of their form, to be particularly thick ones. There are two sorts of concepts which are often said to belong by their own nature to the thicker end of the thin-thick continuum, which I call: (1) value-maker concepts, (2) ‘-able’ concepts. Examples of (1) are concepts like courageous, honest, cowardly, corrupt, elegant, tacky, melodious, insightful and so on. Employing these concepts, at least normally, means both evaluating and describing the thing or person one way or another. For a woman to be courageous, she has to meet certain descriptive conditions, like being able to overcome fear in certain kinds of situations.1 While it is controversial how to best understand these terms, this much can be said: they bear an intimate relation to some relevant thin evaluative concept. A courageous woman has something good about her, i.e. merits a positive evaluation of some sort, at least pro tanto: as far as her courage goes, and ignoring countervailing factors. Likewise, a cruel action has something bad or evil about it, at least pro tanto. I call these ‘value-maker concepts’, because not only do they serve to evaluate, but, unlike thin concepts, and unlike other thick evaluations, they normally also present a reason Of course using further evaluative terms might be the best way to capture the relevant conditions, e.g. courage applies where there is something at stake that is worth the risk. So the conditions need not be purely descriptive. 1



Value and Normativity

5

for the evaluation: by saying the she is courageous, we are also saying that she is good (if she is) because she is courageous. Being courageous makes her good, at least pro tanto. However, as before, some of these terms need not be very thick: for instance, ‘virtuous’ (as in ‘possessing virtue’, rather than ‘chaste’) does convey the idea that the person is morally good because of certain personal qualities, but nothing more than that, partly because what makes a person virtuous is a much more contested issue than what makes her courageous or cruel. Examples of ‘-able’ concepts are: valuable, desirable, admirable, enviable, contemptible, etc. These concepts wear on their sleeves, so to speak, the idea that the thing so evaluated merits or is worth a certain attitude or response: valuing, desiring, admiring, etc. Obviously the ‘-able’ suffix need not appear: the concepts fearsome, trustworthy, amusing, shameful have exactly this same structure.2 The thickness of these concepts lies not in any particular descriptive condition the thing so evaluated has to meet, but rather in the fact that they describe a specific response as merited. By contrast, ‘good’, ‘bad’ and the like by themselves tell us nothing about merited responses – or at least, nothing as specific. The idea of merited, fitting, or appropriate response, as I explain in a moment, plays a central role in what follows. But note that, also in this case, there are varying levels of thickness: e.g. ‘valuable’ is a rather thin sort of evaluation, comprising itself a number of thicker possibilities (I can value x by admiring it, respecting it, etc.). Evaluations on the thinner end of the spectrum provide the background for most of the subject matter of this book. This is because most attention will be devoted to general properties of value as such: how is the value of something related to what we ought or have reason to do (in a broad sense)? How does value ‘depend’ on non-evaluative features? How do we work out the value of complex states of affairs? Must there be a common measure to compare different values? For answering these questions, thicker evaluations may well provide examples, but they will be interesting for what they can tell us about value in general rather than for their own sake. Having said that, there are two chapters in which Nor is the suffix sufficient for inclusion: e.g. ‘enjoyable’ normally doesn’t evaluate something as worthy of being enjoyed. Perhaps ‘response-worthy’ is a more accurate label for this category. 2

6

VALUE THEORY

thicker evaluation will be of more direct concern: Chapter 3, where I consider (and reject) arguments to the effect that thin predicative evaluation (‘x is good, period’) is based on a mistake, and Chapter 4, where I take some steps towards understanding what surely is a thicker type of evaluation, namely that involving what is ‘good for me’.

1.3 The idea of value theory The name ‘value theory’ doesn’t refer univocally. There are at least three different sorts of philosophical inquiries which go by that name.3 One is the general study of values, and includes the whole of moral philosophy, political philosophy, aesthetics, and possibly epistemology, regarded as a normative discipline. In this sense, it contrasts with all other ‘theoretical’ branches of philosophy, like metaphysics or philosophy of language. A second, much more restricted use, stands for all substantive views about what is fundamentally good and bad, and the debate among them. For instance, hedonism is a value theory, holding that pleasure is the one fundamental positive value. As such, it is opposed to other theories, e.g. perfectionism, holding that the excellent development of certain abilities is the fundamental positive value, or to some pluralist theory, holding that there is more than one fundamental value, for example pleasure, knowledge, and moral virtue. In this sense, often synonymous with axiology (from the Greek ‘axía’, ‘value’), value theory is usually contrasted with at least two different kinds of inquiries. One is normative theory, regarded as the elaboration of a view about what is fundamentally right or wrong. To the extent that value theory need not be specially concerned with moral values, their subject matters are distinct. But of course, normative theory and value theory often need to work in tandem: whether one is a utilitarian, a deontologist, or a contractualist, a great part of morality will have to do with morally appropriate responses towards what is good and bad. Moreover, the distinction tends to blur out in the case of elaborating a theory 3

I’m drawing here on Schroeder (2012).



Value and Normativity

7

of virtue: virtuous traits are both plausible candidates for fundamental values, and have an intimate connection to right and wrong action. Second, despite the names, normative theory and value theory are both normative inquiries, in the broad sense that to adopt a certain normative or value theory is to come to believe that one ought or at least has reason to behave (act, desire, feel, judge, etc.) in certain ways for certain fundamental reasons. (For example if hedonism is true, then in principle one should take pleasure, one’s own or others’, as the fundamental goal.) Together then they stand in contrast with meta-ethics, conceived as the philosophical study of linguistic, metaphysical, psychological, and epistemological questions about ethics, spanning morality and non-moral values alike. While certain meta-ethical views may ultimately have normative implications for conduct, the guiding aim of meta-ethics is not evaluative, prescriptive, or directive, but rather explanatory, reconstructive, or descriptive. However, this is a book in ‘value theory’ in a third and distinct sense. As already hinted above, I will ask certain general questions about value. I will not engage in axiology (value theory in the second sense), since I will not argue for or presuppose any particular view about what is fundamentally valuable. Only for the sake of argument will I assume in some places that a certain axiological view (e.g. hedonism) or a certain evaluative intuition (e.g. about the value of punishment) is correct. Nor will I argue for or against any particular normative theory. To this extent, this is a work in meta-ethics. But while some of the questions have a decidedly meta-ethical flavour (How should we define value? How is value related to non-value? Does the idea of absolute value make sense?), others may have a more direct normative import (How should we, in general, relate to value? Are certain values incommensurable?). And I won’t ask traditional meta-ethical questions about objectivity, realism, motivation, or the nature of value judgements. Value theory, in this third sense, purports to explore a host of structural questions about value, in the hope of understanding at least some salient properties that things have in virtue of being good, bad, and so on. Also, it is important to keep in mind a terminological issue. When we speak loosely of ‘value(s)’ as a countable term, we may refer to three different things: (i) the object or state of affairs that has value, (ii) the features that make an object or state of affairs

8

VALUE THEORY

valuable, and (iii) the fact that an object or state of affairs has value (is good, bad, etc.). An axiological theory (a ‘value theory’ in the second sense above) is typically a theory about the features that ultimately make anything valuable, that is, about (ii). As a result of the ‘value-making job’, we can then speak of values in senses (i) and (iii): there will be objects and states of affairs that are made valuable, or are bearers or carriers of value, and there will be facts or truths about such things being good or bad. In other words, the idea or fact of value-making is logically prior to the existence of valuable objects and of evaluative truths. But since here I am not doing value theory in the sense of providing an axiology, when I loosely speak about e.g. the value of friendship or of a wedding ring, I mean to refer to (iii): the fact or truth that friendship or the wedding ring is valuable, rather than friendship as a value-making feature, or friendship as a bearer of value (although it will of course be true that friendship is a bearer of value).4

1.4 Value and normativity Evaluative concepts in a very broad sense include not only good, bad, etc., but also ought, right, wrong, duty, obligation, practical or epistemic rationality, and good reasons for actions, attitudes, or beliefs. Certainly we can use such concepts to evaluate people and situations from diverse standpoints, like morality, prudence, rationality, and so on. However, in this book I am concerned with evaluation in a narrower sense: the sense at issue when we judge things to be good, bad, and the like, with or without qualifications, or when axiologists construct their ‘theory of value’. The other concepts mentioned above (ought, etc.) are sometimes grouped together as ‘normative’ concepts, with ‘ought’ or ‘reasons’ usually singled out as more basic than others. Other labels given are ‘deontic’ (but this seems to betray an excessive focus on duty, or ought in a moral sense), and ‘directive’ or ‘prescriptive’ (but to say A fourth sense of ‘values’ of course refers to the evaluative properties themselves: goodness, badness, etc. But since these properties are always attached to something (‘x is good’, ‘x is a good F’, etc.), then investigating the nature of such properties (as in the FA approach I explain below) is tantamount to investigating facts or truths about x being good, x being a good F, etc. 4



Value and Normativity

9

you have a reason to prefer a light comedy to a historical drama, if it makes for a better night out, while no doubt a normative statement, doesn’t have the force of a command or a prescription). ‘Normative’, however, should not be thought to refer to anything as norm- or rule-based as a quasi-legal normative system. In this broad sense, commonly used in philosophy, normative truths can bear the form of a rule as well as that of particular statements about what you should do here and now. Nor are actions the only normatively relevant objects: intentions, desires, emotions, beliefs are all just as relevant, insofar as it makes sense to hold that there is or there could be good reasons for and against them. T. M. Scanlon labels all these responses as ‘judgement-sensitive attitudes’, in the sense that ‘an ideally rational person would come to have [them] whenever that person judged there to be sufficient reasons for them and [they] would, in an ideally rational person, ‘extinguish’ when that person judged them not to be supported by reasons of the appropriate kind’ (1998: 20).5 Distinct as they may be, still the evaluative and the normative bear an intimate relation: value is normative. This can be said to be one of value’s fundamental properties. The exact articulation of such property is a matter of debate among philosophers, and we will be directly concerned with such debate in the final chapter. But this is a guiding assumption of this book: truths about value, at least, regularly entail normative truths of some sort about actions or attitudes. Here are some examples: that a certain hotel is overall better than another seems to entail that you should prefer it to the other; that a certain conduct was morally good seems to entail that there is some reason to admire it; that a movie is terrible seems to entail reasons to avoid going to see it, and so on. What is relevant is that the connection between value and reasons is entirely general: it applies to values or value claims and reasons as such. It is not because of what better hotels are like that they should be preferred, but simply because of what it is for something to be better than something else. Of course, the specific attitude or action that is appropriate to take will depend on the valuable object at stake: for instance, a great painting should be appreciated in a certain aesthetic way, and a poor one aesthetically censured, while morally

I’ll come back to this in Chapter 7.

5

10

VALUE THEORY

outstanding conduct is to be praised, and evil conduct blamed or condemned. But what remains constant across these diverse cases is that a positive value is to be responded with a positive attitude, and a negative value with a negative attitude of some sort. The most straightforward way to articulate the idea that value is normative is to say that what it is for something to be good is nothing else than for there to be reasons, in principle for anyone, to respond favourably to it, and likewise for what it is to be bad, better, etc. Goodness, or being good, simply consists in the existence of such reasons. Evaluative truths are really normative truths in a shortened form. In the philosophical literature, there are two existing accounts of this form. One has come to be known as the fitting attitude account of value (FA): FA: x is good = it is fitting to respond favourably to (‘favour’) x. FA has a long and distinguished history, starting with the German philosopher Franz Brentano.6 The crucial term is of course ‘fitting’. This is meant to be a fully normative term, expressing the idea of a normative or ideal match between the object x and the favourable response. But other terms have been used in this connection: appropriate, suitable, correct, worthy, deserved, merited, required, ought (x is good = x ought to be favoured). While these terms have different connotations, they are all supposed to be general normative terms. In particular, they should not be assumed to have a moral meaning. The other account has been called by Scanlon the buck-passing account of value (Scanlon 1998: 95–8): Buck-passing Account 1: x is good = x has the property of having other properties that provide reasons to favour x. This differs from FA in two respects. The first is that it uses the normative concept of a reason to favour. The second is that it A full list and history of FA advocates is beyond the present aims. But any list will include: Brentano (1969 (1889)), Sidgwick (1981 (1907)), Broad (1942), Ewing (1948, 1959), Chisholm (1986), Lemos (1994), Mulligan (1998), Zimmerman (2001). See Rabinowicz and Rønnow-Rasmussen (2004) for other names, and Dancy (2000) for a short but telling historical reconstruction. 6



Value and Normativity

11

makes clear a certain structural claim about value: x’s being good (and other value properties on the thinner end of the spectrum7) is not itself a reason to favour x, but rather is the fact that x has other features, i.e. other than goodness, which provide reasons to favour x. To talk of x’s goodness thus is not already to indicate one reason to favour x, but rather to announce that there are reasons to favour x. In this sense goodness ‘passes the normative buck’, i.e. the ability to provide reasons for attitudes, down to the features which make x good, rather than keeping the normative buck, i.e. providing a reason itself. We can represent the buck-passing structure as follows: x has certain properties (e.g. is pleasant) → (there are reasons to favour x = x is good). The arrow → signifies that being pleasant provides or grounds a reason to favour x, and correspondingly grounds x’s goodness. Scanlon’s account incorporates this aspect of value into the very definition, but also advocates of FA have been clear about this. So A. C. Ewing writes: [T]he reason why it is proper to admire anything must be constituted by the qualities which make the object of admiration good … The ground [for an attitude] lies not in some other ethical concept, goodness, but in the concrete, factual characteristics of what we pronounce good. Certain characteristics are such that the fitting response to what possesses them is a pro attitude, and that is all there is to it. We shall not be better off if we interpolate an indefinable characteristic of goodness besides, for it is no easier to see that it follows directly from the nature of things that they are good than it is to see that it follows directly from their nature that they are fitting objects of a pro-attitude. (1948: 158, 172) This is important. FA articulates the normativity of goodness and other thin concepts (and obviously of what I called ‘-able’ evaluative concepts). But is it applicable to what I called thick ‘value-maker concepts’? For example, courage makes a person good or admirable, but the property of being courageous may not be the property of having other properties which give reason to admire the courageous person. Courage doesn’t have to pass the buck to courage-making features: it is itself a perfectly good reason to admire somebody. The normativity of these thick evaluative concepts seems to lie in their immediate reason-providing (or good-making) ability, not in their being analysable in FA terms. 7

12

VALUE THEORY

Both FA and the buck-passing account then make two central claims: 1 Reduction Claim: value properties such as goodness are

reduced to normative relations of fittingness of, or of there being a reason for, attitudes; 2 Normative Redundancy Claim: value properties such as goodness do not themselves provide reasons for attitudes (make attitudes fitting) over and above the good-making features which already provide reasons for attitudes (make attitudes fitting). The two claims are related.8 If the Reduction Claim is true, then for x to be good is for x to have properties that provide reasons to favour x. Consequently, if x’s goodness were a reason to favour x, such reason would consist in the fact that x has properties that provide reasons to favour x. But what kind of reason would this be? First, suppose we ask ‘Why should I favour x?’ The answer ‘because x has properties that provide reasons to favour x’ is very unsatisfying. What we want to know is what reasons there are for favouring x, and all we are told is ‘the fact that there are reasons’. In a reason-giving exchange, at least, mentioning goodness is little or no use. Second, the complex fact about reasons in which goodness consists obviously depends on other properties already providing reasons to favour x: for instance, that x is pleasant. To hold that goodness provides a reason of its own over and above these other features, or on the same standing as them, is to ignore this fundamental dependence. Imagine again a reason-giving exchange: ‘Why should I favour x?’ ‘Because x is pleasant, and x is good.’ X’s goodness would not add any normative weight to the case for favouring x. The case for favouring x has already been made in order for x to be good. In this sense, if goodness is reduced to a fact about reasons (or about fitting attitudes), goodness would be redundant as a provider of reasons. On the FA/buck-passing picture, then, value is normative because it just is a normative relation, and not because it is a provider of good reasons or something that itself makes an attitude fitting. 8

See Stratton-Lake and Hooker (2006).



Value and Normativity

13

Is there any significant difference between the concepts of fittingness and reasons? Should one choose FA or the buck-passing account? Arguments can be made on either side. In favour of FA, one might say that to have a reason to do something, one must be able to do it (reason, like ought, implies can), whereas this doesn’t hold for fittingness, since it expresses a more ideal sort of normativity. And we can clearly imagine things being good or bad in a world where there are no agents capable of responding as they have reason to: e.g. the plight of dinosaurs’ extinction was arguably something bad, if non-human pain is bad at all, but nobody could do anything about it, and therefore nobody had reason to do anything about it. This doesn’t change the fact that the plight of dinosaurs merited, even back then, a negative response. However, it is still true that, in such situations, if there were agents capable of responding, they would have reason to do something about it. So one might simply qualify Scanlon’s account accordingly: Buck-passing 2: x is good = x has the property of having other properties that provide reasons to favour x, or would provide such reasons to suitably situated agents.9 Another possible reason in favour of fittingness comes from what is known as the wrong kind of reasons problem (WKR).10 There can be all sorts of reasons for positive attitudes towards things, people, and situations, which are not good. A commonly used example asks us to imagine that a demon will torture us or the whole humankind if we fail to admire him. It seems that, based on this threat, we have abundant reasons to admire him. If the buckpassing account is true, then it follows that the demon is good. But clearly the demon is evil rather than good, in part precisely because of the threat. His threat is a wrong kind of reason to admire him, because it doesn’t make him good or admirable. Such a reason to admire him does nothing to show the demon’s positive value. So, one reaction to this example might be that, whatever contingent reasons there might be to admire the demon, that doesn’t change See Dancy (2000), Suikkanen (2004), Bykvist (2009), Orsi (2013b). See Rabinowicz and Rønnow-Rasmussen (2004), who elaborate on previous work by Crisp (2000), D’Arms and Jacobson (2000a, 2000b). 9

10

14

VALUE THEORY

the fact that it is unfitting to admire him: a being like the demon simply doesn’t deserve admiration. Fittingness truths might seem to have the sort of fixedness that goes well with truths about value. But buck-passers can again modify their formulation, so as to pick out the right sort of reasons as those that stem from fittingness:11 Buck-passing 3: x is good = x has the property of having other properties that provide reasons of fittingness to favour x, or would provide such reasons to suitably situated agents. On the other hand, advocates of a buck-passing account might complain that the notion of fittingness really belongs to the evaluative rather than the normative realm. This might be for two reasons. First, the idea of an ideal match applies also outside of the domain of human responses: a key fits a keyhole, a chord fits a certain melody, a certain trait makes a species fitter for a certain environment, etc. These statements express evaluations of a functional, aesthetical, or biological kind, without directly addressing what anyone should do. Second, there is a similarity between reasons and fittingness that is worth bringing out: certain features of x provide reasons for an attitude towards x or an action, just like an attitude towards x or an action fits certain features of x. In other words, both notions relate objects and their properties with responses. But reasons seem to essentially relate to the agent or the subject of those responses in a way that fittingness need not. Reasons are always reasons for someone to do something, or they are not reasons at all, while a certain attitude might be called fitting prior to being fitting for someone to take. Of course, when appropriate, we can always add some ‘for A’ qualification, but so can we add like qualifications to claims about value: e.g. it is (would be) good of you to act generously. But intuitively we don’t think that goodness requires mention of an agent, even when it is the goodness of an action. So, since normativity has to do with claims addressed to agents, and reasons, but not fittingness, essentially apply to agents, reasons may seem better equipped to express the normativity of value. In Chapter 8 I will explore in detail this and other strategies to address the wrong kind of reasons problem. 11



Value and Normativity

15

However, these remarks are insufficient to make a case in favour of reasons over fittingness. First, the fact that the word ‘fitting’ and related words have other, non-normative uses shouldn’t give us pause, because the same applies to the word ‘reason’. When we talk about the reason why I’m ill or why my shoes got broken, we are not even in the business of evaluating anything, but rather explaining a certain event. Such polysemy is no obstacle to isolating a clear normative sense for these words. Moreover, the same concept can be expressed by other words like ‘deserved’ or ‘merited’, whose meanings are unmistakeably normative. Regarding the second remark, there can indeed be an intuitive difference between reason (essentially ‘agential’, belonging to agents) and fittingness. But, for one, it shouldn’t be overstated. The ‘agent’ argument in a reason-predicate often has to be left unfulfilled – whose are the counterfactual reasons to prevent the plight of dinosaurs’ extinction? Do we learn anything normatively significant by assigning those reasons to potential agents? Secondly, the fact, if it is such, that fittingness, like goodness, is not essentially agential might cut both ways. If reasons are relations to agents, and value is not, then a definition of value in terms of reasons looks less promising than one in terms of fittingness, because it would metaphysically require ‘more’ from value than we might have thought: the presence, albeit possibly only counterfactual, of agents. Perhaps we can accept that there is a level of normativity where the specification of an agent or a subject is an inherent possibility (after all, fitting responses will be someone’s responses), but no agent need feature as a separate term of the normative relationship. If this is true, and given added reasons of convenience (the modified buck-passing accounts above are complicated to keep in mind or state), in what follows I will mostly talk in terms of the fitting attitude account of value, or more precisely, I will speak as if FA provides the true account of the normativity of value. The question whether it is the right account will be directly addressed in the final chapter, where I will consider some objections against FA and possible alternatives to it. But this suspension of judgement will not undermine the explanations and arguments to follow. I will use FA as a clear articulation of the fundamental normativity of value, but it is really the latter idea that, depending on the context, will play the role of a constraint to be met, a desideratum, or a helpful explanatory tool.

16

VALUE THEORY

1.5 Overview So here is an overview of the contents in the light of this assumption. In Chapter 2, I put FA to work in order to highlight a number of different value concepts: intrinsic value, final value, unconditional value, and others. FA, as formulated above, should be taken as a definitional schema whose details need to be filled depending on what sort of value we are concerned with. For example, if final and instrumental value are different kinds of value, such difference will have to be captured at a normative level, e.g. as a difference in the attitudes that are appropriate to each kind of objects. FA will also have a role as the basis for an objection against the proposal to collapse the final/instrumental value distinction into the intrinsic/ extrinsic distinction. In Chapter 3, I use FA to reply to a sceptical challenge: the idea that only attributive value claims ultimately make sense, i.e. those expressible in statements like ‘x is a good F’. In Chapter 4, I show how the concept of personal value, or what is good for someone, makes good sense once it is understood along the lines of FA. In Chapter 5, I discuss various accounts of so-called ‘organic unities’: complex states of affairs whose value seems to be higher or lower than the sum of the values of the parts making up the state. FA here has a role as a constraint that, for instance, Moore’s account seems unable to meet (parts would have values but no corresponding attitude would be fitting), while other proposals (a ‘conditional’ account) have better resources to meet it. In Chapter 6, I explore value relations: are ‘being better/ worse than’ and ‘as good as’ the only ways in which we can compare things? Can we always compare things value-wise? Here FA will serve as a tool to illustrate the possibility of different value relations and the idea of value incomparability. In Chapter 7, I directly take up the question of ‘what we should do with value’, so I will delve deeper into the nature and diversity of fitting attitudes – what is behind the anodyne notion of ‘favouring’ something. In Chapter 8, as I said, I question FA as the true account of value, considering important objections to it, possible refinements, and sensible alternatives. But, whether ultimately correct or not as a definition of value, a fitting attitude approach is here shown to be a framework within which several questions about value can be asked and answered.



Value and Normativity

17

Apart from taking this approach as a necessary and fruitful assumption, this book advances no general thesis about value. It is rather meant to be a dispassionate investigation into the many forms that value can take and to address some hard questions raised by values. But readers will find an overall resulting picture which is by no means neutral, for two reasons. First, arguments in favour and against certain particular theses will be presented and assessed in each chapter, and naturally my treatment will incline towards one or the other thesis: while no conclusive argument will be offered, and no arguments will be conclusively refuted, the burden of proof will eventually be laid on some thesis or other. Second, it is typically arguments for certain reductive theses which will be criticized: the collapse of final and intrinsic value, the (connected) idea that only states of affairs are bearers of value, the elimination of absolute value or of personal value, the denial of a plurality of value relations, the reduction of fitting responses to one basic kind of response. So, a certain methodological pluralism will emerge as a result, usually justified by an application of the fitting attitude account to the specific question at hand. This is not to say that FA requires a pluralist picture of value: some of the reductive theses above have been advocated by philosophers who fully endorse FA. But if value is normative, then reducing possibilities for value means reducing normative possibilities, and – at times – it will appear that such a reduction needs more justification than it might otherwise have seemed.

1.6 Meta-ethical neutrality There is one respect in which this book, and its reliance on the idea that evaluative truths regularly entail normative truths, is meant to be completely neutral. The fitting attitude account (or some other articulation of the normativity of value), while no doubt a meta-ethical thesis, need not be regarded as a contender in certain meta-ethical debates about value, namely about the nature of value judgement, the existence of evaluative truths, or the question whether value is a natural or a non-natural property. This point might not be immediately obvious: doesn’t FA assume that (1) value judgement is cognitive rather than non-cognitive – a belief about an evaluative matter of fact, rather than a desire or emotion; (2) there

18

VALUE THEORY

are evaluative truths; (3) value properties are non-natural properties, since fittingness and the like are non-natural just as goodness and the like? Let’s consider these seeming assumptions in turn. 1. FA does not assume cognitivism, for two reasons. First, FA can be put forward as a general conceptual thesis about the content of value judgements: when we judge that something is good, we also judge (or are committed to judging) that it is fitting to favour it. Scanlon himself sometimes speaks in these terms, and it is notable that Allan Gibbard, a prominent non-cognitivist, accepts accounting for notions such as goodness or wrongness in terms of the rationality (or their being warranted) of certain feelings and attitudes, like approval, admiration, remorse, indignation, etc. (Gibbard 1990). It is then a further question whether these judgements of rationality (or fittingness) are to be understood in non-cognitive terms, e.g. as expressing a non-cognitive state of acceptance of a system of norms that permit certain actions and attitudes (as in Gibbard’s analysis), or as expressing beliefs about matters of fact concerning fitting attitudes (as on a cognitivist approach). The second reason why FA does not assume cognitivism is that, even if FA were an account of value properties, and thus in turn implicitly suggested a cognitive account of evaluative judgements as beliefs purporting to represent those properties, still non-cognitivism has over the years developed into what Simon Blackburn calls quasi-realism (Blackburn 1984). Quasi-realism is the project of reconstructing (or ‘earning the right to’) such notions as normative truth, properties, belief, and knowledge, working with the scant materials of a non-cognitivist approach and a generally empiricist worldview. Accepting a fitting attitude account of value properties doesn’t seem to make such project any harder than accepting a different account of value. If value is normative, a quasi-realist analysis will deploy whatever resources it already has to reconstruct such a basic feature of value. 2. John Mackie’s error theory holds that affirmative propositions about value, when understood as implying objective and categorical demands or reasons, are all false. To speak of objective value, or alternatively of attitudes that are fitting regardless of what anyone’s actual attitudes are like, is to commit oneself to a metaphysically and epistemologically queer view (Mackie 1977). However, FA doesn’t assume the truth of any evaluative



Value and Normativity

19

proposition. It only means to articulate what evaluative properties would have to be like, if there were any, and so, if any proposition about them were true. 3. Another remark worth making in connection with Mackie relates to the third question about the natural/non-natural distinction. Mackie takes common sense to be committed to the existence of value or fittingness conceived as entities constituting non-natural facts: facts which lie outside the world described by the natural sciences, including psychology, and which therefore require some non-empirical (and, for Mackie, queer) means of accessing and knowing about them. But in fact FA does not presuppose any of all this. FA says nothing about whether value, and therefore the fittingness of attitudes (or reasons for attitudes, as in the buck-passing account) can be understood in naturalistic terms, for instance as depending somehow on an agent’s actual or potential attitudes. The massive mistake imputed by Mackie to common-sense evaluative thinking lies not in FA, but in a certain meta-ethical (ontological and epistemological) understanding of the concepts and properties mentioned in the account. This point is easy to miss, since proponents of FA have themselves often hailed it as a welcome alternative to naturalist and non-cognitive analyses. C. D. Broad, defending FA’s reduction of goodness to fittingness, claims: I cannot see the need to have both a non-natural quality of goodness, grounded on these various natural characteristics, and a non-natural relation of fittingness grounded on this non-natural quality. (Broad 1940: 237)12 More recent advocates, such as John McDowell (1998a, 1998b) and David Wiggins (1987), propose FA (or something close enough) as a ‘sensibility theory’ of value with two alleged benefits: (i) unlike for instance Moore’s own view, which takes goodness to be an indefinable sui generis property, FA makes a connection to human attitudes central to understanding value, thus clearing somewhat the aura of extra-worldliness about it; (ii) unlike naturalistic analyses, and unlike otherwise analogous dispositional See also Ewing’s claim to provide ‘the minimum non-naturalist theory of ethics’ (1948: 169). 12

20

VALUE THEORY

accounts of colours and other mind-dependent entities, the human attitudes FA refers to are not actual or potential reactions caused by valuable objects, but responses merited by them – thus preserving the intuitive normativity of value.13 Further, Scanlon defends his buck-passing account by claiming it to back up Moore’s open question argument, which is aimed to show the failure of naturalistic analyses of value (Moore 1993: 66–9). For Moore, we can always sensibly ask, for any naturalistic property C (e.g. being pleasant, or being approved), whether an object that has C is also good or valuable, in a way that we cannot sensibly ask, for instance, whether a man who is an uncle is also someone’s brother. This conceptual openness is taken to be evidence for the irreducibility of goodness to any naturalistic property. For Scanlon, the buck-passing account draws a tight connection between value and reasons which explains the intuition of openness, and thus underwrites an anti-naturalist conception of value: Judgements about what is good ... generally express practical conclusions about what would, at least under the right conditions, be reasons for acting or responding in a certain way. Natural ... facts may provide the grounds for such practical conclusions ... Judging that these facts obtain need not involve explicitly drawing these conclusions however. Questions such as ‘This is C, but is it valuable?’ (where C is the term for some natural or ‘metaphysical’ property) therefore have an open feel, because they explicitly ask whether a certain practical conclusion is to be drawn. (Scanlon 1998: 96–7)14 However, while it is perfectly proper to advance FA as a model for a non-naturalist theory of value, such a strong meta-ethical thesis is not to be read straight off the mere formulation of FA. If value is normative along the lines of FA, this is no blow for naturalists. Both McDowell and Wiggins add to these a further point which I will consider in Chapter 8: some of these responses (e.g. admiration or approval) are not intelligible without a prior notion of value. Their sensibility theory is thus not meant as a non-circular definition of value. See also the way Jacobson (2011) introduces FA as an alternative to ‘robust realism’ and ‘dispositionalism’ about value. 14 See also Stratton-Lake and Hooker (2006). 13



Value and Normativity

21

Their job has always been to reconstruct normativity out of naturalistic (or naturalistically tractable) materials, and acceptance of FA will not make the naturalist programme any harder than it already was.15 For instance, the time-honoured ideal observer theory can analyse an attitude being fitting to x as the attitude that a fully informed, sympathetic, and imaginatively equipped observer would take towards x.16 Or, on the so-called internalist approach to normative reasons, there being a normative reason for someone to take a certain attitude towards x would be a matter of them being possibly motivated to take that attitude, coherently with their other existing attitudes (Williams 1981). Unlike the ideal observer approach, internalism does face a difficulty when presented in combination with a fitting attitude account of value: at least certain values seem to be such that there is reason for everyone, in principle, to respond to them in a certain way, but internalism makes such universality hostage to facts about the possible motivation of particular individuals. But it is noteworthy that internalists are aware of this problem. Some (like Michael Smith and Christine Korsgaard) accept it, and argue (or at least hope) for a thesis of convergence: qua informed, coherent, and rational agents, our motivational sets would overlap, at least on some issues. In turn, our reasons would converge. So values can be normative for

This is not to say that FA is completely neutral with respect to all naturalisms. FA, particularly in its buck-passing version, imposes a certain structure on value, whereby goodness is conceptually distinct from the good-making properties. This may be a problem for those naturalists who identify goodness with e.g. pleasure, or with a disjunctive list of whatever natural properties goodness supervenes on (see Jackson 1998). On these views being good is e.g. being pleasant, and being pleasant, in turn, seems to be what provides reasons, rather than being the property of having other reason-providing properties (i.e. other than pleasure). So being good is not the property of having other properties that provide reasons (the good-making properties), because one of these properties (pleasure) is goodness. This naturalist view sits better with a buck-stopping account of value, where goodness is itself a, or the, reason-providing property, or with a definition of reasons in terms of value (and so in turn in terms of pleasure) (see Chapter 8). Alternatively, the superveniencedriven naturalist can in primis accept FA, and then proceed to identify the property of being a reason with being pleasant or with a disjunction of naturalistic reasonproviders. What matters for this book is that these naturalists have to rely on some account of the normativity of value, and this is all that is needed here. 16 The idea goes back at least to Adam Smith (1759). See Firth (1952), Lewis (1989). 15

22

VALUE THEORY

everyone.17 Other philosophers (like Bernard Williams) seem to reject the idea that values can as such be normative for everyone, or at least that values imply the specific normativity of reasons. In sum, the stance of naturalist philosophers is to either do their best to accommodate FA, or (in Williams’s case) reject the assumptions that would make their account conflict with FA. In this sense, the fitting attitude account is importantly metaethically neutral. FA can be taken on board by naturalists, error theorists, and non-cognitivists, not as the ultimate meta-ethical truth about value, but as a general truism about value which one’s theory will aim to explain or otherwise come to terms with. To the extent that FA, or something close enough, is a guiding assumption of this book, there is no reason for all such philosophers to decline their interest in the value-theoretic questions explored here.18

1.7 Value theory: The questions The value-theoretic questions addressed in this book can be graphically represented as ‘branching out’ from the basic assumption. If x is valuable implies that it is fitting to favour x, then we can ask a number of questions:

See Smith (1994), Korsgaard (1996). I’m taking internalism to be a naturalist thesis about reasons. This can be disputed in Smith’s and Korsgaard’s cases, but the relevant point is that their views, while aiming to illuminate the notion of a reason in further terms (be they naturalistic or not), rather than taking it as a primitive, do not thereby conflict with FA. 18 Even the error theorist will be interested. If their stance is conservative, and ethics must be preserved as a useful fiction, then it will be important to understand the value-theoretic details of such fiction. If their stance is revisionary, then it will be clearer to them how to revise that part of ethics that deals with value and fitting attitudes. 17



Value and Normativity

FOR THE SAKE OF WHAT, OR WHOM? Final/Non-final Value (ch.2) Personal/Impersonal Value (chs 4, 7)

WHY? Intrinsic/Extrinsic Value (ch.2) Organic Unities (ch.5)

AS WHAT? Absolute/Relative Value (ch.3)

IT IS FITTING TO FAVOUR X

HOW AND HOW MUCH? Respect, Promote, Prefer … (ch.7)

Figure 1.1

BY WHOM? Agent-neutral/Agentrelative Value (ch.4)

MORE/LESS THAN Y? Value Relations, Comparability (ch.6)

23

CHAPTER TWO

Meet the Values: Intrinsic, Final & Co. 2.1 Introduction To the extent that we justify actions and attitudes in terms of some value they produce or express, we do not normally need to reach for fundamental, basic, or ultimate values. If a certain medicine promises to be effective against my headache, and accordingly I take it (or you advise me to take it), we need not question whether a headache-free condition would be good in itself, or because of some other good state related to it – say, having the capacity to think clearly. If a certain policy is thought to reduce unemployment or improve working conditions, we can support it without asking why having a decent job is a good thing: maybe work, in certain conditions, is good for its own sake, or maybe it is good because of other good things to which it connects – financial autonomy, or the possibility of achievement. But certainly there are plenty of contexts where our attention focuses on a deeper dimension of value. For instance, many of us would take a critical attitude towards a businessman intent on accumulating money exclusively for the sake of it, quite regardless of how money impacts on other aspects of his life. The feeling is that money, however important it may be, is the sort of thing to be valued not for its own sake, but for the sake of something else; at any rate, it cannot be the only thing to be valued for its own

26

VALUE THEORY

sake. Conversely, someone who appears to treat their own children exclusively as potential workforce for the family would be the target of blame: whether or not people can have some instrumental value, few of us believe that one’s own children can be regarded exclusively as potential workforce. In these and similar examples, our criticism is based on an intuitive distinction between what is valuable for its own sake (one’s own children) and what is instrumentally valuable (money). Similar distinctions are traditionally made by philosophers in their search for the basic elements of a good life, and for ultimate standards of action: the attempt is to pick out what, in some sense, matters fundamentally, as opposed to what matters for the sake of something else. The aim of this chapter is to make distinctions between value concepts which are often not distinguished, either within philosophy or in ordinary value talk: final value, exclusively final value, unconditional value, intrinsic value, necessary/essential value. The idea is to alert the reader to the importance of keeping these concepts separate. On the argumentative plane, I provide some defence for the idea that something can be valuable for its own sake (i.e. finally valuable, the sense in which one’s children are valuable and money arguably is not), and yet doesn’t have to be unconditionally valuable, nor must it owe its value to its intrinsic properties. This idea is worth exploring, since it has been assumed by most philosophers that what is valuable for its own sake can only be valuable in itself, that is, in isolation from other things, and not conditionally on other values. Final value, according to this tradition, must be the stopping place of our evaluations, or is not really ‘final’. I will argue that this tradition, which has a distinguished history and has recently been revived, considerably restricts the ways in which it is appropriate to value things, and depends on a questionable assumption on what it means to be ‘final’. In section 2.6 I also explain why we should distinguish merely instrumental extrinsic value from other forms of extrinsic value.



Meet the Values: Intrinsic, Final & Co.

27

2.2 Final and unconditional value: Some philosophical examples When philosophers build their substantive axiologies, i.e. their theories of what is good and bad, they generally seek to identify what is ultimately good and bad. In stating their views, they use various phrases. One is ‘good for its own sake’. For instance, after an examination of people’s opinions and values, Aristotle thus describes the sense in which eudaimonia (happiness as a state of flourishing) is the ultimate good: If, then, there is some end of the things we do, which we desire for its own sake (everything else being desired for the sake of this), and if we do not choose everything for the sake of something else (for at that rate the process would go on to infinity, so that our desire would be empty and vain), clearly this must be the good and the chief good ... Now such a thing happiness, above all else, is held to be; for this we choose always for self and never for the sake of something else, but honour, pleasure, reason, and every virtue we choose indeed for themselves (for if nothing resulted from them we should still choose each of them), but we choose them also for the sake of happiness, judging that by means of them we shall be happy. Happiness, on the other hand, no one chooses for the sake of these, nor, in general, for anything other than itself. (Nicomachean Ethics I 2, 7)1 Another commonly found phrase is ‘good in itself’, equally used by otherwise very different philosophers such as Kant and Bentham. Kant applies it to the good will, i.e. willing or intending to act according to the moral law: It is impossible to conceive anything at all in the world, or even out of it, which can be taken as good without qualification, except a good will ... A good will is not good because of what it effects or accomplishes – because of its fitness for attaining some

Aristotle is obviously not the first philosopher to draw such distinctions. For example, see Plato, Republic 357. 1

28

VALUE THEORY

proposed end; it is good through its willing alone – that is, good in itself. (Kant 1964: 61–2) Bentham proposes instead a hedonist view of value: Strictly speaking, nothing can be said to be good or bad, but either in itself, which is the case only with pain or pleasure; or on account of its effects, which is the case only with things that are the causes or preventives of pain and pleasure. (Bentham 1970: 11) Yet another phrase is ‘good/desirable as an end’, for instance to be found in Mill’s statement of hedonism (here misleadingly identified with utilitarianism, which includes but does not reduce to a theory of the good): The utilitarian doctrine is, that happiness is desirable, and the only thing desirable, as an end; all other things being only desirable as means to that end. (Mill 1861: IV 2) How are we sure that Aristotle, Kant, Bentham, and Mill are talking about the same thing? First, despite different terminologies, eudaimonia, pleasure, good will, or happiness are all contrasted with things that are ‘good for the sake of something else’, that is to say, things that are good because they stand in a certain (typically but not necessarily) instrumental relation to the former items. For example, from a hedonist point of view, health is good insofar as it produces, facilitates, or simply contains experiences of pleasure, or prevents painful experiences. Devoid of such relationships to pain and pleasure, health would be neither good nor bad. Second, they all mean to pick out what can be described as final values. Since in value theory this term is not always understood in the same way, here is the idea we are working with. Something is finally valuable when it is valuable for its own sake, that is, as suggested in Chapter 1, when we have reason or it is fitting to favour or value it for its own sake. The idea is fully normative: something that is finally valuable deserves, as such, a certain kind of treatment. However, beyond this shared concern for final values in this sense, other notions are mixed in the quotations above, and it is instructive to tease them apart. One is the concept of something



Meet the Values: Intrinsic, Final & Co.

29

that is not only finally valuable, but exclusively finally valuable: all the value it has or could ever have is final value. This is what after all for Aristotle sets eudaimonia apart from things like reason and virtue, which are finally valuable but not exclusively finally valuable, since they are valuable also for contributing to other things, such as happiness. Also Kant is explicit about this, as he claims that the usefulness of the good will could not add any value to it (1964: 62). And the same goes for hedonists: of course a particular experience of pleasure could give us further future pleasures, and to that extent be instrumentally as well as finally good, but there is no non-pleasure-related state which might make an experience of pleasure additionally valuable in this way. Pleasure could not be worth seeking for reasons other than pleasure: it could only be good for its own sake. So, alongside contrasting final to non-final values, these philosophers meant to pick out exclusively vs. non-exclusively final values. It is important to note, however, that the search for exclusively final value is not inevitable. Those who hold a pluralist axiology, whereby at least two things are finally valuable – e.g. pleasure and virtue – would have no problem admitting that virtue is both finally valuable, and also valuable when and because it produces pleasure, and vice versa for pleasure. One such example is W. D. Ross’s axiology, which includes innocent pleasure, knowledge, and virtue as final values (2002 [1930]). Exclusively final value thus is a notion that only monist axiologies require: if there is no other final value than V, then there is nothing else for the sake of which V could be valuable. Of course, if you are a pluralist, you might still believe that each of your chosen two (or more) final values is exclusively finally good: e.g. that the value of virtue cannot be enhanced by pleasure or vice versa. But the monist needs to believe in a value that is exclusively final. A second notion implicit in the quotations above is that those final values are also unconditional values, in the sense that their value does not presuppose the value of anything else. Again, if you are monist, e.g. you think that pleasure is the only final value around, then it follows that the value of pleasure cannot depend on the value of anything else. For an example of a conditional value, Kant famously mentions happiness: someone’s happiness is valuable only if they are worthy of it (due to their good will). In this sense, Kant denies that happiness as such is valuable for

30

VALUE THEORY

its own sake. It is a problematic question whether Kant thereby counts as a monist, since in some interpretations it appears that now two things are finally valuable: good will (unconditionally), and deserved happiness (conditionally on the value of good will). The possibility of such an interpretation already suggests that final value need not be unconditional. For a different example, consider a valuable relationship such as friendship. Friends are usually disposed to help each other out, and they take pleasure or derive happiness from their relationship. Plausibly, if their happiness and altruistic dispositions had no value, their friendship would have no value either. In this sense, the value of friendship is not unconditional. But this does not mean that we are to value friendship simply for the sake of these other things (happiness, altruistic action). Similarly, the value of altruistic action itself can plausibly be said to presuppose the goodness of happiness and the badness of suffering: if suffering were not bad, there would be no merit in alleviating others’ suffering. But this, again, does not mean that altruism is to be admired and encouraged simply for the sake of minimizing suffering. In sum, goods like friendship, altruism, or happiness (in Kant’s view) can be finally valuable and yet (unlike Aristotle’s eudaimonia or Kant’s good will) not unconditionally so. To prevent possible misunderstandings: of course, even values that are in this sense unconditional (i.e. not conditional on other values) may depend on the obtaining of certain enabling conditions; e.g. nobody can have a good will if they are not sufficiently mentally sophisticated creatures. Being a mentally sophisticated creature is a necessary condition for someone to have a good will, and therefore for the value of the good will to be instantiated. Moreover, plausibly nobody can be a mentally sophisticated creature in isolation from others (say, their parents): there being other people around is another necessary condition for the value of good will. But this is a dependence on factual, rather than evaluative, conditions: enabling conditions of this sort do not refer to the value of being mentally sophisticated or the value of there being people around. In this sense, an unconditional final value can and normally does depend on factual conditions. If we might be tempted to see the presence of evaluative enabling conditions as undermining the status of a given value as final (wrongly, as I have briefly suggested), no similar temptation should



Meet the Values: Intrinsic, Final & Co.

31

arise with regard to factual enabling conditions. Consider another example. Any experience we may have depends on us being alive: dead people have no experiences. Therefore also the value of our experiences (e.g. our pleasures) depends on us being alive. But it does not follow that experiences are valuable for the sake of being alive. Rather, being alive appears valuable, if at all, for the sake of the experiences it makes possible. The value of being alive (as an enabling condition) is therefore non-final (see Chapter 5 for further discussion of enabling conditions).2

2.3 Intrinsic value and final value The phrases ‘for its own sake’ and ‘in itself’ naturally suggest a further consideration: what makes these things finally valuable must somehow be found ‘in’ the valuable thing, rather than in something totally or partially ‘outside’ it. This is why such values have historically been called intrinsic, and therefore contrasted with extrinsic ones. G. E. Moore so defines intrinsic value: ‘To say that a kind of value is ‘intrinsic’ means merely that the question whether a thing possesses it, and in what degree it possesses it, depends solely on the intrinsic nature of the thing in question’ (1993 [1922]: 286). And in order to determine what has intrinsic value, he devised the so-called isolation test: ‘it is necessary to consider what things are such that, if they existed by themselves, in absolute isolation, we should yet judge their existence to be good’ (1993 [1903]: 236). Now, certain things just cannot exist in absolute isolation: e.g. a state of pleasure necessarily has a subject and (at least for some sorts of pleasures) an object (what it is pleasure at). So the test seems unserviceable as it stands: we cannot coherently imagine a world containing states of pleasures but no subject and no object of those pleasures. However, in the light of Moore’s definition of intrinsic value, we can propose an amended isolation test: in order to determine the intrinsic value of x, we have

Henry Sidgwick makes it clear that the factual or at least conceptual dependence of beauty and knowledge on minds is compatible with favouring beauty and knowledge as final ends – though he eventually does not take them to be so (1981: 114). 2

32

VALUE THEORY

to focus on x’s intrinsic nature alone, and on that basis see if we judge x to be good – presumably, this is something we can do with pleasures and other states.3 For instance, Kant’s good will is said to keep its value regardless of any relation to actions it may or may not originate. For a classical hedonist, it is the intrinsic qualities of pleasures (such as their sensory quality, intensity, or duration) which determine their value: if I feel good because I think I have won the lottery, then my state has the same intrinsic value whether in fact I have or have not won the lottery. And intrinsic value can belong, of course, also to goods, e.g. the beauty of a work of art, whose ‘intrinsic nature’ is a matter of certain relations among different parts (say, colours and shapes being in a certain arrangement): such relations make up the work of art for what it is. In this sense, they are ‘internal’ as opposed to external relations, such as the fact that the work of art provokes emotional experiences to viewers, or that it is a unique piece in its genre. Such external relations do not affect the intrinsic value of the object as a beautiful work of art. As for Moore himself, the isolation test delivered the result that the enjoyment of personal relationships and the appreciation of beauty are, not the only intrinsic values, but the highest ones. But is all final value intrinsic value? So did Moore assume. And until not long ago, such a question might have sounded otiose. If something is to be valued for its own sake, then it seemed obvious that what makes it worth valuing in such a way must be found in the intrinsic nature of the object. And if what makes something worth valuing is to be found partly ‘outside’ the object itself, then it seems that the value of the object cannot be final. However, precisely these statements should already make it clear that final/ non-final and intrinsic/extrinsic value are, at least, conceptually distinguishable pairs: final/non-final refers to the correct or appropriate way of valuing something, while intrinsic/extrinsic refers to the metaphysical location of the good-making properties of something. So a Moorean view which equates final and intrinsic value need not be obviously true. In particular, many believe that final values can be extrinsic:

3

See Lemos’s ‘intentional isolationism’ (1994: 10–11).



Meet the Values: Intrinsic, Final & Co.

33

objects which are valuable for their own sake partly thanks to their relations to other objects. This can happen in many ways.4 1 One is where final value depends on the value of other

things. As suggested above, if friends’ happiness were not valuable, and friendship did not contribute to friends’ happiness – both facts being conceived of as externally related to the friendship – probably friendship would not be valuable for its own sake. For a more concrete example: a particular fur coat might be regarded as valuable for its own sake, as an outstanding piece of handicraft, yet so only assuming an appropriate evaluative background. If fur coats were not in general instrumentally valuable for protection against the cold they provide, this particular coat could not have any value, let alone any final value.5 2 Another case is where final value might be enhanced by an object’s relational properties: a fine work of art, beautiful and thus already valuable on account of its intrinsic features, might have its value increased by its being a unique or rare piece (a kind of relational property, since it implies the absence of other things, or many other things, like it). And such uniqueness or rarity need not make the See Korsgaard (1983), Kagan (1998), Rabinowicz and Rønnow-Rasmussen (1999), Langton (2007). Korsgaard, who basically single-handedly started this line of thought, went on to argue for an ultimate dependence of final value on the extrinsic property of being valued by a rational agent. But this claim of hers is best understood as a general meta-ethical thesis about the nature of value, rather than as the denial that there is any real intrinsic value (see section 1.6). Being valued by a rational agent, after all, is not to be counted among the good-making properties on top of whatever intrinsic or extrinsic properties already make an object (e.g. a fur coat) finally valuable. In other words, intrinsic value need not be metaphysically mind-independent value (this is a confusion which arguably Moore himself caused in his 1922 essay). 5 Jonathan Dancy would claim that if the enabling conditions (factual or evaluative) for something to be a value are extrinsic properties, but the ground of value (the good-making feature) is an intrinsic property, then the value is still intrinsic (2004a: 172). In the example, friendship could still be an intrinsic value, even if happiness is an extrinsic enabling condition for its value. While I accept the distinction between enabling conditions and grounds, and hence the possibility that intrinsic (and final) value is variable (see Chapter 5), I prefer for illustrative purposes to use ‘intrinsic value’ in the way defined by Moore, whose notion of ‘dependence’ admittedly doesn’t discriminate between enabling conditions and grounds of value. 4

34

VALUE THEORY

object additionally valuable simply in instrumental terms, say because it increases its economic value, or because it provides particular pleasure to viewers or owners. 3 Yet another case is where the final value of an object does not

just counterfactually depend on the value of something else, but results from or is exhaustively grounded on some relation to other objects. Uniqueness or rarity can give an otherwise unremarkable item (say, a stamp) a new final value. A certain type of car might have the ability to race at unusually high speed: this relational, indeed causal, feature may conceivably make the car valuable (e.g. worth maintaining) for its own sake. Were it not for this feature, the car would simply have the instrumental, non-final value that most other cars have. Or consider Napoleon’s hat: a dull, worn-out accessory, but it is not absurd to believe it worth preserving for its own sake simply because it belonged to such an extraordinary person. In each of these cases, of course, there is also a long story to be told about counterfactual evaluative dependences. For instance, in the case of the car, we are dealing with something like the value of pushing physical boundaries; in the case of the hat, the value of keeping traces of historically crucial figures, and in turn the value of history. So in these cases, final value is extrinsic in two ways: first, it is grounded in relational properties of the object, and second, such relational properties make the object finally valuable only on the condition of something else than the object itself being valuable.

If these examples are persuasive, then final value can be extrinsic. It also follows that we cannot always apply Moore’s isolation test to discover what has final value. Remember that we had to imagine the object in isolation from any other thing. Then we would miss the final value of Napoleon’s hat, since we would ignore who the hat belonged to, and likewise the final value of a rare stamp, since we would ignore how many other stamps like it are around. We would miss the additional value of a unique artwork, since we wouldn’t know that there are no other exemplars of that. We would miss the final value of the fur coat, since we would have to ignore the point of producing and wearing fur coats. And we would possibly mistake the final value of any given friendship,



Meet the Values: Intrinsic, Final & Co.

35

since we wouldn’t know if it is actually good for the friends – we would ignore an external condition for its final value.

2.4 The reduction to facts Given the plethora of examples available, it seems that the burden of proof lies on those who take final value to be always intrinsic value, i.e. value which depends only on an object’s intrinsic features. How could one hope to show that? One strategy starts from the question: what are the real bearers of value, i.e. what sorts of entities really have value? Most of the examples assume that individual objects (coats, hats, stamps …) are the sort of entities that can have final value. The strategy invites us to reconsider this assumption. It seems that we could, in principle, always understand or translate the putative extrinsic final value of an individual object in terms of the final value of a fact, or a state of affairs, which includes the object together with the relevant relational properties. For example, the final value of Napoleon’s hat can be seen as really the final value of the fact that (F1) there is a hat which belonged to Napoleon. Alternatively, it could be seen as the final value of the fact that (F2) this hat belonged to Napoleon. Either F1 or F2 would be both worth valuing for its own sake, and in virtue of their intrinsic or internal properties: that the hat belonged to Napoleon is indeed what these facts are all about, be it F1 or F2. So this translation strategy, or move from objects to facts, would give the result that a Moorean needs: a picture where final value is always intrinsic. Similar moves can be made, in principle, with other putative cases of extrinsic final value.6

See Zimmerman (2001: chapter 3), who explicitly advocates this move. But the general idea goes back to Ross (2002 (1930): 112), and has been endorsed by many value theorists since (Chisholm (2005 [1972]), Lemos (1994: 20–31). See Rabinowicz and Rønnow-Rasmussen (1999, 2003) for discussion and criticism. 6

36

VALUE THEORY

It is interesting to note that the F2-style translation, which avoids the existential commitment, is perhaps preferable to F1 for two reasons. First, it is not obvious that to value something for its own sake always means to value its existence. This might work for artefacts which are worth preserving (and thus keeping into existence), but what about wishing a painless death to our sick dog as a way of valuing him for his own sake? Second, precisely this point shows that in general we value the existence of x only if we value x in the first place. F1-style translation instead sees valuing x’s existence as a prior condition on valuing x, so it seems to get things the wrong way around. But why adopt such a reduction or translation manoeuvre in the first place? Why can’t we be happy with final values that are extrinsic? One immediate reason might be simplicity. Extrinsic final values give us a less unified picture of how value comes into existence. And, as we have seen, we wouldn’t be able to use Moore’s isolation test, which might have given us a handy procedure to look for the things to care about for their own sake. A more theoretical reason, suggested by Michael Zimmerman (2005: 194), is that if there are long stories to be told about, e.g. why Napoleon’s hat is a valuable thing, i.e. stories that must refer to values other than the value of the hat itself, then what we have is at most a derivative value. But final values, in principle, should rather play the role of ‘endpoint values’: once we reach them, no helpful explanation of them can be given, except saying that they are good ‘as such’ – i.e. ‘for their own sake’. That’s why extrinsic values of the sort indicated above cannot be final values for Zimmerman: explanation (or rather justification) does not stop at them. Only values that are intrinsic can play such an endpoint role, because there’s no looking beyond them to explain why they are valuable. And it seems that only facts, and not individual objects, can be guaranteed to play this double role of carrying final, i.e. non-derivative, and intrinsic value. Moreover, the reduction to facts does correspond to a natural way of talking about value, so it’s not an ad hoc or merely technical move. We obviously do not only evaluate objects or persons, but also facts or situations: we say that it is good that this and that happened, or that it would be better if that did not happen, and so on. And it also matches with the idea that many of our fitting responses to value are what are generally called propositional attitudes, such as desire or preference, e.g. desiring that the war end. If, as claimed in Chapter 1,



Meet the Values: Intrinsic, Final & Co.

37

values must be the sort of things towards which fitting responses can be directed, then it is easy to see why values at least could consist in proposition-like entities, such as facts or states of affairs. However, the reduction manoeuvre is not easily accomplished. I will mention two challenges. First, since the strategy has to be carried out for the whole spectrum of final values, its proponents are committed to the claim that only propositional attitudes are fitting responses to final value. If final values are all facts such as F2, then it seems that attitudes, such as respect, preservation, or certain forms of love, which are thing- or person-oriented, rather than fact-oriented, cannot be appropriate responses to final values. We respect and love individual people (or animals), and this is a different attitude than loving that such people have this or that characteristic. We want to preserve Napoleon’s hat, rather than the fact that it belonged to Napoleon – indeed, preserving the latter seems to make almost no sense. According to the reduction manoeuvre, thing- or person-oriented attitudes might still be in some sense fitting to their objects, but they won’t match with final value: final value will rather belong to the sorts of states which we can value by a relevant propositional attitude. So the fittingness of preserving Napoleon’s hat will reduce, it seems, to the fittingness of valuing or cherishing F2, the fact that the hat belonged to Napoleon. While this latter example may give some plausibility to the strategy (after all preserving the hat is a way of valuing F2), it is not equally clear that, e.g. respect and love for individual persons can always be seen as fitting because of the fittingness of a kind of respect and love towards some fact about these persons. In a sentence, the reduction of finally valuable things to finally valuable facts commits one to a reduction of fitting responses (to final value) to fact-oriented attitudes, and the latter reduction might be hard to sustain, given the variety in the kind of responses that we normally identify as ‘valuing for its own sake’. Perhaps then bearers of final value are just as various as our responses: ‘We value many different kinds of things, including at least the following: objects and their properties (such as beauty), persons, skills and talents, states of character, actions, accomplishments, activities and pursuits, relationships, and ideals’ (Scanlon 1998: 95).7 See Anderson (1993) for a view at the other extreme: states of affairs are never fundamental bearers of final value. Only individuals (in particular, people) are. 7

38

VALUE THEORY

Second, the theoretical justification suggested by Zimmerman for regarding all final value as intrinsic is weak. There is no initial reason to think that final values, understood as things worth valuing for their own sake, also need to play the role of non-derivative or endpoint values, in the sense that no explanation of their value can be given, except pointing to their own intrinsic nature. Conceptually, we are again dealing with different notions: final value, which refers to the appropriate way of responding to something, and non-derivative value, a structural-metaphysical notion. Consider again friendship. The value of friendship, and thus the respect a particular friendship deserves from third parties, can be partly explained by reference to other values: the mutual desire for happiness that friends typically have is good; the happiness thus achieved is also good; and, at least on some views, friends must make each other happy by and large in morally permissible ways. The value of friendship then is not an ‘endpoint’ value, because appreciating the value of any given friendship is not independent of appreciating the value of other things (happiness, morality). But valuing, e.g. respecting, a particular friendship on the condition that it doesn’t involve immoral conduct, obviously does not mean valuing it for the sake of something else, namely moral permissibility. One can recognize a value as conditional on other values, and in this sense ‘derivative’, and still fittingly care about it for its own sake. What does seem to be true is that, if a final value is derivative or conditional on other values, then this reflects on the appropriateness of responses. If I care about Napoleon’s hat for its own sake, but show otherwise utter indifference towards Napoleon’s sword, or indeed to objects belonging to comparable historical figures (say, Sitting Bull’s war bonnet), then my valuing is not of the fitting kind, precisely because it is insensitive to the structure of the final value towards which it is directed (assuming, of course, that the hat is only valuable for its historical connection). But such interrelation among fitting attitudes does not subtract anything from the final value of the object – it is not as if its value gets ‘thinned out’ in the network of attitudes required towards similar objects. In the attempt to reduce final values to intrinsic, non-derivative values we can see again the concern with unconditional values which occupied Aristotle and Kant. Moreover, if such attempt is carried out by the reduction manoeuvre of construing the value



Meet the Values: Intrinsic, Final & Co.

39

of individual entities as the value of facts including those entities, what we get are essential values, i.e. values which depend on the essential properties of the valuable object. It is an essential property of F2 that it concerns Napoleon’s hat and its having belonged to Napoleon: a different hat or a different owner would make up a different fact. In general, a fact like ‘that x is P’ has the essential property of concerning x, or ascribing P to x. If a fact is valuable because of such intrinsic property, then it is also essentially valuable. In this sense, final values turn out to be incorruptible, that is, they remain constant for as long as their bearer remains what it is. It is a good question, to say the least, whether it is a defensible consequence of such a view that only unconditional and essential values are really worth valuing for their own sake, whereas values that are had conditionally or contingently do not deserve that same kind of response.8

2.5 Intrinsic and conditional value Having defended the possibility of final extrinsic values, we can now also clarify the relations between intrinsic/extrinsic and unconditional/conditional value. Let’s consider first unconditional value. Unconditionally good things do not depend on the value of anything else. Therefore whatever it is that makes them good must be looked for among their intrinsic properties: in their intrinsic nature. As said, even unconditional values might depend on factual enabling conditions: the unconditional value of pleasure depends on the subject being alive. Now, such enabling conditions can be intrinsic or extrinsic properties of the valuable object; e.g. there being other people around might be a factual extrinsic condition for the value of a good will. But this doesn’t mean that therefore the good will has an unconditional extrinsic value. This is because Moore’s definition of intrinsic value and the amended isolation test for it presuppose that whatever necessary conditions for x to exist

Bradley 2006 suggests that ‘Mooreans’ (for whom final value is intrinsic and necessary) and their opponents (whom he calls ‘Kantians’) might be talking past each other, and simply have in mind two different concepts of value. I discuss this in Chapter 7. 8

40

VALUE THEORY

and be intelligible as such, be they intrinsic or extrinsic properties of x, have already been counted in. So, for instance, it would make no sense to apply Moore’s test to the good will, and conclude that the good will has no intrinsic value because one instance of the good will in a world without other agents would strike us as valueless. Rather, when we apply the test, we should imagine a world where necessary conditions for a good will to exist and be intelligible as such are taken for granted (such as the existence of other people besides the agent). Factual dependence on external conditions doesn’t make a value extrinsic. What about conditional value? Here a certain object x is dependent for its value on the value ‘of something else’. Now, this ‘something else’ can be intrinsically or extrinsically related to x. Earlier I considered the value of a friendship, and suggested that it may depend not only on whether it produces happiness for both friends, but on whether such happiness is indeed good for them or more generally valuable. Happiness, and its value, are in this way extrinsically related to any given friendship: when we apply Moore’s test on friendship, its producing valuable happiness or not is one of those facts we should abstract from, because it is not a necessary condition for any friendship to exist – friendship can produce misery as well as happiness. So the value of friendship appears to be conditional and extrinsic (but not for these reasons non-final, as argued above). On the other hand, if the ‘something else’ is intrinsically related to x, then x has conditional but intrinsic value. A good example is offered by Thomas Hurka’s theory of virtues, in which virtues are attitudes of this form: ‘loving (desiring, etc.) the good for its own sake’ and ‘hating (avoiding, etc.) the bad for its own sake’ (Hurka 2001). Virtues are therefore second-order final values. Their structure makes their value at once conditional, because it presupposes the first-order value of the objects of love and hate, but also intrinsic, since it is an intrinsic feature of a virtue that it involves an attitude towards the good or the bad. An instructive contrast between intrinsic and extrinsic conditional value emerges from two different readings of Kant’s claims about deserved happiness. If we take the valuable state of affairs to be ‘S is happy and deserves being happy (i.e. has a good will)’, then its value is conditional on the value of the good will, and yet intrinsic, since S’s having a good will is a part of the state of affairs:



Meet the Values: Intrinsic, Final & Co.

41

it is intrinsically related to it. On the other hand, if the valuable state of affairs is simply ‘S is happy’, then its value is conditional on S’s having a good will, but extrinsic, because S’s having a good will is externally related to this state of affairs. The crucial point from this discussion is that intrinsic value can be conditional value. This is significant, since it is easy to confuse the two notions: it is natural to wonder, how can something’s value be intrinsic, belong to ‘the thing itself’, if it depends on the value of something else? But just like we shouldn’t assume final value to be necessarily intrinsic, so should we not assume that intrinsic value is always unconditional.9

2.6 Elimination of extrinsic value? There is a lingering concern with extrinsic values. If x owes its value to the value of something else y, then the worry is that x really has no value at all. Some philosophers are tempted to draw this conclusion when considering extrinsic value of the instrumental sort. Thus Ross on the value of acts: ‘Whatever value [an act] has independently of its motive is instrumental value, i.e. not goodness at all, but the property of producing something that is good’ (2002 [1930]: 133, my emphasis). There is a sense in which the act does not contribute any value to the world. In computing how much value the world contains, we are not going to add instrumental value on top of whatever value the act has caused. A world where the same valuable states of affairs occur through other means would contain the same amount of value (other things being equal). Hence the act is good-causing, but not literally good. To complete the picture: exclusively final values can be conditional values. For instance, Hurka’s virtues are attitudes whose final value is conditional on the value of their objects. But Hurka could argue that virtues are only valuable for their own sake. Can exclusively final values be extrinsic? Consider some of our examples. Friendship’s value depends on the value of happiness, but this by itself doesn’t mean that necessarily it is fitting to favour friendship (also) for the sake of happiness. Rather it means that one cannot value friendship without also valuing happiness. Of course, plausibly friendship can be favoured for its beneficial consequences, so probably it is not an exclusively final value. Presumably deserved happiness, for Kant, is a value that is both extrinsic (given the second reading in the main text) and exclusively final (it could only be worth favouring for its own sake). 9

42

VALUE THEORY

Similarly, we may call a medicine healthy, but the truth is that it is health-restoring (there is nothing healthy about a pill-shaped chemical aggregate). What is healthy is the state you are in upon taking the medicine. But could we draw eliminativist conclusions for all extrinsic value? For instance, some philosophers speak of ‘signatory value’: an X-ray is signatorily good if it indicates something else that is good (e.g. that the tumour has gone). (X-rays of course are in general also instrumentally good, as aids to medical knowledge.) Or there can be ‘contributory value’: a certain motif in a painting is good in this sense if it contributes to the (aesthetic) value of the painting as a whole. It is good ‘as a part’. It would be tedious to recount all forms of extrinsic value, because it would require drawing a list of all relevant relations: causing, being a sign of, being a part of a whole, being historically connected to (as in Napoleon’s hat), etc. But the question now is: if we grant that one sort of relation (instrumental) to value means that an object (an act, say) really is valueless, why not generalize and conclude that the concept of extrinsic value is, at best, a handy way of talking about value, but does not capture a genuine evaluative reality? And if extrinsic value doesn’t capture a genuine evaluative reality, then a fortiori all final value must be intrinsic. While not everyone agrees with Ross, it might be worth ending this chapter by suggesting why, even if he is right about instrumental value, not all extrinsic value can be eliminated like that. The argument for elimination assumes that what has instrumental value does not contribute any value over and above the value of its causal consequences. However, this does not seem to apply to the cases of extrinsic final values considered above. First, it is true that Napoleon’s hat has no value apart from its historical connection. But Napoleon’s hat contributes value precisely in being an exemplar of a supposedly valuable category of things – objects (maybe of a certain kind) that belonged to important historical figures. Destroy the hat, and you have directly reduced the amount of value contributed by this category. On the other hand, once you imagine away acts and in general things that only have instrumental value – while keeping constant the amount of valuable states of affairs they would otherwise produce – you will not have diminished in the slightest the amount of value in that world. And this is after all what makes sense of why it is



Meet the Values: Intrinsic, Final & Co.

43

appropriate to value the former for its own sake, and the latter for the sake of something else. Second, regarding values such as friendship which are conditional on external factors, the elimination argument would work if it were true that the fact of friendship would contribute no value, over and above the value contributed by other facts which are conditions for friendship’s value (e.g. mutual happiness produced within a friendship). But it is hard to see how to show this much. While it might be held that a given friendship loses in positive value if it makes one or both friends worse off, what loses in value is the complex of expectations, mutual feelings, shared history, etc., which define friendship for what it is, and which – prior to philosophical arguments to the contrary – constitute its final value and its specific valuable contribution ‘to the amount of value in the world’. Of course some philosophers (e.g. hedonists) would be ready to make the substantive claim that friendship is only instrumentally valuable, e.g. insofar as it promotes the general happiness. But this move would not be acceptable, since here we were looking at possible reasons why extrinsic values are eliminable (and a fortiori non-final) qua extrinsic. It seems that no general argument for elimination can be found.

2.7 Summary In this chapter I have distinguished several pairs of value concepts: MM

final vs. non-final value: what is fitting to favour for its own sake vs. what is fitting to favour for the sake of something else;

MM

exclusively final vs. non-exclusively final value: what is fitting to favour only for its own sake vs. what is fitting to favour for its own sake and for the sake of something else;

MM

unconditional vs. conditional value: what is fitting to favour (for its own sake) independently of whether it is fitting to favour something else vs. what is fitting to favour (for its own sake or not) not independently of whether it is fitting to favour something else;

44

VALUE THEORY

MM

intrinsic vs. extrinsic value: what is fitting to favour because of its intrinsic properties vs. what is fitting to favour (for its own sake or not) partly because of its extrinsic properties;

MM

essential/necessary vs. contingent value: what is fitting to favour in all possible worlds where it occurs vs. what is fitting to favour in some but not all possible worlds where it occurs.

I have argued that we should keep an open mind as to how these concepts relate to one another. In particular, I have suggested that final value need not be unconditional value, and criticized a strategy to equate intrinsic and final value based on the idea that only facts or states of affairs are bearers of final value. Since the resulting view is that there can be final extrinsic values, then I had to defuse a worry that the very category of extrinsic value could be dispensed with.

CHAPTER THREE

The Challenge against Absolute Value 3.1 Introduction Things can be good or bad relatively, i.e. in relation to other things: good for people or animals, good as objects of a certain kind, good to buy/enjoy/know ... But the philosophical search for general final values has traditionally been the search for values which are not in this sense relative, but absolute: what is (finally) good simpliciter, or plain good. With the possible exception of Aristotle, for whom the concept of what is plain good overlaps with the concept of what is good for human beings, the philosophers mentioned in the previous chapter (Kant, Mill, Moore, Ross) believe that their chosen final values (good will, deserved happiness, pleasure, beauty enjoyed, etc.) are, in this sense, things that are absolutely good. Moreover, while our everyday evaluative talk no doubt makes extensive use of ‘good’ and ‘bad’ in their relative senses (more or less explicitly), it is not uncommon to find oneself assessing general or particular situations and outcomes with no specific relativization in mind: ‘military intervention in Syria is a really bad idea’, ‘human extinction would be the worst possible outcome’, ‘critical thinking is better than passive acquiescence’.

46

VALUE THEORY

Notice: we are not necessarily thinking that these things would be good or bad for or relatively to someone or something, nor that military intervention is really bad as an idea, or that human extinction is bad as an outcome, or that critical thinking is better as … (what?).1 Some philosophers have argued that all such talk of absolute value is meaningless. In this chapter I explain the challenges brought by P. Geach, P. Foot, and J. J. Thomson. Then I present one reply by M. J. Zimmerman, notice one problem with it, and briefly suggest how to remedy it.

3.2 Geach and attributive goodness Peter Geach starts from looking at how we ordinarily use terms such as ‘good’ and ‘bad’. Like most other adjectives, in English we use ‘good’ both in predicative ways: ‘the meal was good’, ‘it’s good that you came early’ (= ‘that you came early is good’), etc., and in attributive ways: ‘that was a good meal’, ‘this is a good knife’, etc. This is a fact about grammar. But what Geach focuses on is the logical behaviour of our ordinary terms: what kind of inferences are legitimate, given our ordinary use of these terms? Only this kind of analysis will show something philosophically interesting about the concept of goodness. Geach defines as logically predicative an adjective A if ‘x is an A y’ (where ‘y’ is a noun) splits up logically into ‘x is A’ and ‘x is a y’; otherwise it is logically attributive (1956: 33). The intuitive idea is clear enough: a logically predicative adjective allows certain intuitive patterns of inference that an attributive one does not. For example: 1 This is a square photograph. 2 This is a postcard. 3 Therefore, this is a square postcard.

I should add that claims of absolute value are also often expressed by terms such as ‘matters (more than)’ or ‘is (more) important (than)’, where the speaker rejects any suggestion that the thing in question matters or is important only to them or to anyone else in particular. 1



The Challenge against Absolute Value

47

This is an acceptable inference. Therefore ‘square’ is predicative. But consider an analogous inference with the adjective ‘fast’: 1 Sue is a fast turtle. 2 Sue is an animal. 3 Therefore, Sue is a fast animal.

The conclusion is false. But the point is not just that the conclusion is false, but rather that it is not warranted, given the premises. This can be seen by considering the following inference, which shows how ‘good’, like ‘fast’, is attributive: 1 Sarah is a good lawyer. 2 Sarah is a chess player. 3 Therefore, Sarah is a good chess player.

Now, Sarah is a good chess player (let’s suppose). But clearly the premises are insufficient to establish that. The inference is invalid, even if premises and conclusion happen to be true.2 According to Geach, what explains the intuitive difference between the inferences with ‘x is a square y’ and those with ‘x is a fast (good) y’ is that the premises have a different logical structure: the former is decomposable (it can split), the latter is not (it can’t split). ‘X is a square y’ has the following structure: x is a square y = x is a y and x is square. This structure makes explicit why the inference above is acceptable. It is acceptable because it is valid: 1 This is a photograph and this is square. 2 This is a postcard. 3 Therefore, this is a postcard and this is square.

Geach uses this sort of case to contrast ‘red’ and ‘big (small)’, but a rather different one with ‘bad’: ‘we cannot infer e.g. that because food supports life bad food supports life’ (1956: 33–4). See Rind and Tillinghast (2008) for a recent analysis of Geach’s actual examples. See Szabo (2001) for a discussion of Geach’s argument (and its consequences) in the context of semantic theory. 2

48

VALUE THEORY

Now, if ‘x is a good y’ had a similar logical structure, it would produce equally acceptable inferences. But it doesn’t. So ‘x is a good y’ must have a different structure – it cannot split into ‘x is a y and x is good’. Therefore, in this sense, Geach claims that: 1. The adjective ‘good’ is logically attributive. But presumably an analysis of the logical behaviour of ‘good’ will not just show us something about the word ‘good’, but about the concept of goodness itself: 3 2. If ‘good’ is logically attributive, then the concept of goodness is attributive. 3. Therefore, the concept of goodness is attributive. If the concept of goodness is attributive in the sense that ‘x is a good y’ is not decomposable into ‘x is good’ and ‘x is a y’, this has obvious epistemological consequences (ibid.: 34). Knowing that a Volvo is a good car doesn’t involve joining two separate pieces of information: that a Volvo is a car and that it is good. It’s not even clear what it means to know the latter as such. Likewise, knowing that Sarah is a good lawyer doesn’t involve knowing that Sarah is good. If this is understood as ‘Sarah is a good person’, this information may be quite irrelevant to knowing whether Sarah is a good lawyer. Compare with ‘this is a red car’. In this case there are two relevant pieces of information to be learnt separately: that it is a car (suppose we are colour-blind), and that it is red (if we don’t know what a car looks like). And if the concept of goodness is attributive, this constrains the ways in which we can meaningfully think about the property of goodness: 4. If the concept of goodness is attributive, then ‘there is no such thing as being just good or bad, there is only being a good or bad so-and-so’. (ibid.: 34) Note that Geach assumes that the logical behaviour of an adjective A in its grammatically attributive uses (‘x is an A y’) decides the question whether the adjective is as such logically attributive or predicative. This assumption requires defence. 3



The Challenge against Absolute Value

49

That is, if ‘good’ is attributive, i.e. it is a mistake to think of ‘x is good’ as a free-standing proposition detachable from ‘x is a good y’, then it is a mistake to think there can be a free-standing property of goodness that can equally attach to different kinds of objects. If Sarah being a good lawyer and the Volvo being a good car were a matter of Sarah being good and the Volvo being good, then we could think of Sarah and the Volvo as possessing a common property. But the idea that goodness is attributive rules this out.4 Positively, goodness is always relative to some specification of the thing or person we call good or bad: good as a lawyer, good as a car, good as food. In general, the only acceptable form is ‘x is a good K’, where ‘K’ stands for some appropriate specification of the sort of thing that x is, or that x is good as (a folded sweater can be good as a pillow without being a pillow). Goodness is always goodness of its, or of a, kind. But what to do with predicative uses of ‘good’? 5. If the concept of goodness is attributive, then any occurrence of predicative uses of ‘good’, e.g. in ‘Julien is good’ or in ‘helping others in need is good’, can be explained away as an attributive use in disguise, where an appropriate accompanying substantive can be found, or else it is meaningless (i.e. doesn’t correspond to a legitimate concept). Therefore all talk of values, final or intrinsic, either needs to be understood as elliptical for talk of attributive value, or else is meaningless. Consider claims traditionally made in axiology: ‘pleasure is good’ is either meaningless, or else might be understood as ‘pleasure is a good state of mind’; likewise, ‘justice is good’ perhaps can be interpreted as ‘justice is a good arrangement’. However, Geach warns that the substantive to which ‘good’ can logically attach must be contentful enough as to ‘convey either a criterion of identity or a standard of goodness’ (ibid.: 41), and unlike functional or role terms such as ‘knife’ or ‘hitman’, it is not clear that the notions of a state of mind or an arrangement Note that the fact that ‘good’ is attributive doesn’t mean that good cars and good lawyers have nothing in common from an evaluative point of view. They certainly do have this in common: both good cars and good lawyers serve well the purpose or function associated with being what they are (see e.g. von Wright 1958). 4

50

VALUE THEORY

can convey by themselves a standard of goodness for things of that kind. Likewise for terms like ‘state of affairs’ or ‘(possible) world’.5 Geach’s argument thus presents a challenge to substantive value theory as classically understood, because it condemns as meaningless the application of value concepts to objects or states of affairs in the absence of appropriate standards of evaluation for those kinds of things.6 But such is the use of value concepts in philosophical axiologies which aim to tell us what is just finally valuable, or provide a list of such just finally valuable states of affairs, as we have seen in the previous chapter by considering claims by Kant, Mill, Moore, and Ross.

3.3 Foot and the virtues Philippa Foot targets the notion of an unqualifiedly good (and better, best, etc.) state of affairs as used, in particular, by act-consequentialists in trying to construct their moral view that an action is right if, and only if, there is no alternative action which brings about a better state of affairs. Consequentialism owes its appeal to the compelling thought that ‘it can never be right to prefer a worse state of affairs to a better’ (Foot 1985: 198). If talk of preferences is turned directly into talk of actions, we arrive at act-consequentialism’s moral view.7 Foot thinks we can dispense with consequentialism in all of its forms only by dispensing with its compelling thought. In turn, we can do that only by throwing into doubt the very legitimacy of talk of simply better or worse states of affairs. Her argument is based partly on (1) some kind of semantic continuity, and partly on (2) eliminating alternatives. 1 Foot offers the sketch of a meta-ethical view of what it is

to claim that things are good, better, etc., when no further

As Thomson argues, these are not ‘goodness-fixing’ kinds (2008: 25). Meaningless rather than just false, since the argument targets the very intelligibility of claims of absolute goodness. Compare: ‘x is big’, in the absence of standards of size for xs, is similarly unintelligible, since we don’t know what would make it true or false. 7 It should be mentioned that indirect forms of consequentialism, e.g. rule-consequentialism, refrain from directly translating talk of preferences into talk of actions, but they too owe part of their appeal to the ‘compelling thought’. 5 6



The Challenge against Absolute Value

51

specification is given, e.g. ‘that’s a good thing’. When we use such unqualified evaluations about teams in a contest, or about people we care about, it seems clear to her that our claims have only speaker-relative truth conditions. They signal, or rather report, that the speaker is interested in whatever they claim to be good, whether for their own sake or for the sake of others. So, Foot argues, why suppose that we use evaluative terms in a completely different sense (i.e. non-speaker-relative) when we comment on some event that doesn’t involve us personally, e.g. ‘that was very bad’ when talking about the disasters of some distant earthquake (ibid.: 201)? At least, any proposal to interpret such statements in a non-speaker-relative way – e.g. the proposal that we mean to ascribe to the event some property such as being simply bad – needs special justification. 2 Foot considers and eliminates two alternative interpretations. According to the first, what we mean to say in contexts where we are not personally involved is that things are good, bad, better, etc., from an impersonal point of view. Foot quickly dismisses this, by arguing that the concept of an impersonal point of view can only be understood by reference to the concept of maximum welfare. But the claim made by welfaristic consequentialism that ‘the (impersonally) best state of affairs is the one which contains maximum welfare’ is not supposed to be true by definition.   The second interpretation is: some state of affairs is good, bad, better, etc., from the moral point of view. Foot claims that simply adding this qualification will not make the concept intelligible. The qualification must be given some content. And the most plausible content refers to the concept of a moral person: ‘there are some things that a moral person must want and aim at in so far as he is a moral person and ... he will count it “a good thing” when these things happen and “a good state of affairs” either when they are happening or when things are disposed in their favour’ (ibid.: 204). What this interpretation shows, however, is that the concept of a good state of affairs presupposes moral concepts, and so cannot be used as part

52

VALUE THEORY

of a morally neutral premise to arrive at moral conclusions. For instance, the idea of happiness as a good state of affairs from a moral point of view presupposes the idea of a benevolent agent who must aim at such a state. Likewise, other states of affairs we might be inclined to call ‘good’ will refer back to some virtue or other. In this sense, the best interpretation we can give of ‘a good state of affairs’ is parasitic upon moral concepts.8 If Foot is right, then the claims about unqualified value made in axiology are really speaker-relative, or are simply a by-product of our moral thinking, or else are meaningless. In this sense, Foot’s challenge is somewhat less radical than Geach’s: she accepts the legitimacy of talk of goodness that is not reducible to talk of attributive goodness, but points out its dependence on moral concepts. While less radical, Foot’s view has two consequences which significantly reduce the importance of traditional axiology. First, if the only acceptable claims about good states of affairs are a by-product of moral thinking, then such evaluative claims cannot serve as an independent basis for all or even some moral claims, contrary to what both consequentialists and some non-consequentialists (e.g. Ross) believe. Axiology would lose its role as a theoretical guide for practical decisions. Second, Foot must regard as meaningless any evaluative claim that doesn’t obviously correspond to what the moral person would want. For example, if the advancement of knowledge or the enjoyment of beauty for its own sake are not intelligible aims of a moral person as such (what kind of moral virtue would have that as its aim?), then accordingly the claims that knowledge or the enjoyment of beauty are valuable for their own sake are meaningless, or at best only have speakerrelative meaning. It is easy to see how this undermines the sort of claims that are ordinary currency in axiology. Like in Geach’s view,

One important consequence is that the sorts of outcomes one can meaningfully call ‘better’ will be limited by an assessment of what the moral person would aim at. Foot suggests that killing one in order to prevent the killing of two cannot be called ‘better’ than letting the two be killed, since the benevolent person would not aim at the former: the scope of benevolence is limited by other virtues, e.g. justice (ibid.: 206). And it seems that, in the absence of an independent concept of ‘good state of affairs’, one could not argue in favour of specifically consequentialist virtues. 8



The Challenge against Absolute Value

53

it is not that such claims are false, but rather that the very debate on whether they are true or false is based on a misconception.

3.4 Thomson and goodness in a way Following Geach, Thomson accepts that ‘good’ is an attributive adjective. But she rejects the implication that ‘good’ must always be accompanied by an appropriate substantive, or else it cannot be applied. She thinks that Geach’s conclusion is both not sufficiently radical and too narrow. It is not sufficiently radical, since it implies that goodness still is a respectable property, only a relative one like bigness. If goodness is relative in this way, one could still in principle construct piecemeal axiologies for the kinds of things which admit of standards of evaluation (and notably, Geach thinks one can do this for ‘a good person’ and ‘a good action’). But Geach’s conclusion is also too narrow, because ‘x is a good K’ is only one of the many ways in which one can meaningfully use ‘good’ and its cognates: things can be good to eat, drink, read; good for use in doing something; good for someone or something; good as a thing of that kind, and so on. Indeed, Thomson argues that understanding ‘x is a good K’ depends on knowing, from the context of utterance, in which of these ways of being good one means that x is a good K. One, but only one among the possibilities, is that x is good as a K, which is what Geach seems to mean by the idea that the accompanying substantive must be able to convey standards of goodness. In general, ‘x is good’ is always elliptical for ‘x is good …’, where ‘…’ is replaced by an ‘adjunct’, i.e. an answer to ‘in what way(s) is x good?’. Otherwise put, ‘x is good’ is a structurally incomplete expression, which needs to be ‘resolved’ in order to appear in a meaningful sentence. In Thomson’s view, ‘all goodness is goodness in a way’ (1997: 276), or alternatively, something’s being good is nothing over and above one or more of the different ways of being good. It follows that goodness is neither an absolute property, nor a property simply relative to a certain kind of thing. Metaphysically speaking, Thomson’s view can be interpreted in two rather different ways: 1 Being good is a relational property which relates an object

54

VALUE THEORY

(say, a book) or a state of affairs, a person, etc., and some appropriate ‘adjunct’ (an activity: ‘to read’, a person: ‘for you’, a comparison class: ‘compared to other philosophy books’, etc.): x is good [insert appropriate adjunct] = x G adj.a (x G adj.b, etc.)

In a sense, this would not be much different from Geach’s thesis, except for the larger variety of ‘adjuncts’ which Thomson admits. For this reason, this interpretation doesn’t seem to be what Thomson has in mind. 2 Being good is not a relational property which, so to speak, remains stable across different kinds of relata, but rather there are as many ‘goodnesses’ as there are ways of being good. In this sense, e.g., being good to read, being good for your soul, and being good as a philosophy book are three different relational properties: (subscribed letters: a = activity, f = for, k = relative to the kind). x is good to read = x Ga read x is good for your soul = x Gf your soul x is good as a philosophy book = x Gk philosophy books ... This seems closer to Thomson’s intentions. If Thomson is right, then there is no room for unqualified value claims such as ‘pleasure is good’, ‘pain is bad’, ‘deserved pain is better than undeserved pain’ and so on, nor could one make similarly unqualified claims such as ‘acting so and so will make things better’, or ‘this and that will make the world a better place’. These formulations seem to intentionally evade the question, good, bad, better. etc., in what way?9 Consequently, we cannot hope to build normative claims, i.e. claims about what we have reason (morally or otherwise) to do, at least in part on the basis of unqualified value claims (this is what some deontologists, such as Ross, think they can do). Much So they are meaningless, if taken literally. Thomson’s challenge is clearly also more radical than Foot’s, for whom such unqualified claims can still have meaning as referring to what a moral person would want for its own sake. 9



The Challenge against Absolute Value

55

the less can one hope to build a full normative theory on such basis (this is what classic consequentialists, like Mill and Moore, think they can do). Thomson does provide her own basis for (moral) normative claims: the virtues, which in her view constitute second-order ways of being good. They are ‘second-order’ since they depend on one particular way of being good: being good for. A second-order way of being good/bad, such as being just/unjust, being kind/cruel, etc., is such only insofar as being just, kind, etc., is, in more or less direct ways, good for people, i.e. insofar as ‘whatever else may be true of the people among whom we live, it is better for us that they have the trait than that they not have it’ (ibid.: 284). But here we are concerned not with Thomson’s normative views, but with the idea itself of goodness in a way, which Thomson presents as an alternative to ‘Moorean’ or unqualified goodness. Despite the differences between ways of being good, Thomson accepts that they share a core feature which makes them all ways of being good. Thomson claims: ‘Intuitively, for a thing X to be good in one of the first-order ways is for X to benefit someone or some thing Y (which might or might not be X itself) in the appropriate way, or to be capable of doing so’ (ibid.: 289). Examples: something is good to look at, if looking at it pleases people; someone is good at playing chess, if she is capable of winning (often enough), thus benefiting herself in this sense; a frying pan is good for use in making omelettes, if it answers to what people making omelettes want from it. Finally, some X is good for Y, if X conduces to Y’s doing what it is designed to do (if Y is an artefact), or if X conduces to Y’s reproductive success (in case of animate objects), or, in case of human beings, if X conduces to Y’s reaching one or more of the goals Y would have under ideal conditions (that is, were Y fully informed about relevant facts, and their goals were not the result of some inappropriate influence) (ibid.: 296).10

On this aspect, though, Thomson (2001: 52–6) seems to reject her former 1997 views. 10

56

VALUE THEORY

3.5 Zimmerman’s ethical goodness Michael Zimmerman is among the few who have directly responded to the Geach/Foot/Thomson challenge.11 He makes three main points, taking Thomson as his principal opponent. The first point is that Thomson’s view seems incoherent. On the one hand, Thomson claims that if goodness is always goodness in a way, then there is no such property of being just good. On the other hand, she clearly believes that also being good for someone or something is always being good for someone or something in a way, and we have just seen such ways of being good for. But she doesn’t therefore conclude that there is no such property of being just good for. So, at the very least, she owes an explanation of why being good is different from being good for (Zimmerman 2001: 20). Zimmerman’s second point is that even if goodness is always goodness in a way, it simply doesn’t follow that goodness is not a property, but only that goodness is a determinable property, i.e. a property that needs to be ‘determined’ or specified in one way or another. Such is the case for being shaped or being coloured: nothing can be shaped without being shaped some way (square, circular, etc.), but this seems no reason to deny being shaped or being coloured the status of properties (ibid.: 19–20). Zimmerman’s central reply, however, starts from noticing that the property (or concept) traditionally assumed by Moore and other axiologists is not some form of pure, unadorned goodness, but intrinsic goodness. And being intrinsically good seems certainly one way in which things can be good. Moreover, ‘intrinsically good’ seems to be logically predicative rather than attributive: from ‘Judith being pleased is an intrinsically good state of mind’, and ‘a state of mind is a state of affairs’, we can infer ‘Judith being pleased is an intrinsically good state of affairs’ (ibid.: 21). But, as Zimmerman notes, Thomson could still ask: what kind of way of being good is that? And two natural answers are not satisfactory: 1 x is intrinsically good when x’s goodness depends only on

x’s intrinsic properties (see Chapter 2). This answer, true

11

See also Smith (2010), Arneson (2010), Sturgeon (2010).



The Challenge against Absolute Value

57

or not in its own right, seems to appeal to some concept of pure, unadorned goodness, which Thomson would reject. 2 x is intrinsically good when x is non-derivatively good. But

x’s being derivatively good means that x’s goodness is owed to some extent to the goodness of something else. And this (despite its adverbial form) is not really a way of being good, but rather a way of being related to something else which is good (in some way). (Recall the doubts in Chapter 2 about whether instrumental value is a form of value at all.) A fortiori, being non-derivatively good is not a way of being good either.

Zimmerman’s answer to Thomson is that being intrinsically good, in the sense that interests value theorists, coincides with being ethically good: common candidates for intrinsic goodness, such as pleasure, knowledge, or beauty are such that ‘there is a moral requirement to favor them (welcome them, admire them, take satisfaction in them, and so on) for their own sakes’ (ibid.: 24, author’s emphasis).12 Being ethically valuable in this sense seems indeed to be a distinctive way of being good: what is good for people, or good as an object of its kind, might or might not be also ethically valuable. Moreover, the notion of ethical value doesn’t make any appeal to pure, unadorned goodness. If so, it seems we can restore the idea that when we talk of what is ‘just’ good, or of good states of affairs, and so on, we are ascribing a specific evaluative property: being ethically good. However, it can be easily seen that Zimmerman’s proposal, as it stands, and as he explicitly notes (ibid.: 26, 28), is not distant from Foot’s idea that any meaningful concept of ‘good state of affairs’ must implicitly refer to what a moral person would aim at. On the reasonable assumption that a moral person would favour (or aim at) x for its own sake if and only if there is a moral requirement to favour x for its own sake, the two proposals really overlap. 12 Zimmerman doesn’t conflate the concepts of intrinsic value and ethical/final value. As seen in section 2.3, he recognizes a distinction between the two notions, while at the same time arguing on metaphysical grounds that all ethical/final value is intrinsic value. His answer to Thomson therefore shifts the attention from intrinsic value to ethical value, but given his views being ethically valuable is a necessary feature of what is intrinsically good.

58

VALUE THEORY

But as we have previously remarked, if Foot’s thesis is correct, then axiology would lose its role as a guide for moral claims, nor would it be possible to debate the merits of things like beauty or knowledge in the absence of a concept of a moral virtue which has it as its point to aim at or favour such things for their own sake. At the very least, we need to convince ourselves that the moral person as such aims at or favours such things as knowledge or beauty for their own sake, or these things won’t have intrinsic (i.e. ethical) value. Perhaps she does so aim, but the point is that axiology is thus made hostage to such a question. And this seems to invert the order of explanation in traditional axiology, in which first certain intrinsic goods are specified, and then certain attitudes towards them are declared to be virtues and vices.13 Therefore, pending further explanations, it is not clear that Zimmerman’s answer to Thomson can reinstate the ambitions of substantive value theory.

3.6 A better reply: Absolute value and fitting attitudes However, Zimmerman’s insight can be salvaged by removing the reference to a specifically moral requirement.14 In effect, this amounts to answering Thomson’s challenge by offering a fitting attitude account (FA) of the way in which things like pleasure, knowledge, etc. are claimed to be simply good: x is good if and only if it is fitting to favour x. In axiology, we will be interested in final values, i.e. in what it is fitting to favour for its own sake. It is indeed peculiar that Thomson never seriously considers this long-standing idea about value. Perhaps this can be explained by her focus on Moore: Moore famously rejects any such definition of intrinsic value.15 But he sometimes remarks that ‘x has intrinsic For instance, Zimmerman’s own account of virtue presupposes that there are intrinsically (ethically) good and bad states (ibid.: chapter 6.5), whereas in Foot’s view it is the theory of virtue that has priority over intrinsic value (though virtues for Foot, like for Thomson, are based on another kind of value, i.e. goodness for). 14 See Ewing (1948: 150–1). See also Zimmerman (2007: 329–31, 1999). 15 See Thomson (2006: 253). As Michael Smith nicely brings out in Smith (2010), Thomson’s most recent views seem to acknowledge the concept of correct (or fitting) attitudes as a meaningful evaluative concept (e.g. Thomson 2008: 119). Her only 13



The Challenge against Absolute Value

59

value’ and some such phrase about worthy or fitting attitudes are logically equivalent, and this is all that matters for present purposes: whether definitional or not, the connection of final value to fitting attitudes tells us something relevant about the way of being good that being finally good is.16 Recourse to FA as specifying a way of being good would reinstate the role of axiology also against Foot’s challenge. Evaluative claims will be conceptually independent of moral ones: that it is fitting to favour x doesn’t entail that it is morally fitting to favour x, or that a morally sensitive person would favour x. Of course, for some xs it is morally fitting to favour them for their own sakes: people’s deserved happiness, morally good traits, etc. But this will be a fact pertaining to the specific nature (i.e. the supervenience base) of the value in question, and not a truth about final value as such. Conceptual independence allows one to construct some or all of one’s moral claims on the basis of axiology. For instance, the compelling moral thought which Foot ascribes to consequentialism: it cannot be right to prefer a worse state of affairs to a better seems to simply derive from a non-moral thought that is equally compelling (indeed self-evident, given a fitting attitude account of ‘better’): it cannot be fitting to prefer a worse state of affairs to a better, plus the plausible assumption that if F-ing is not fitting, then F-ing is not morally right either.17 It is not clear how a defender of Thomson or Foot might object remaining opposition would then be against understanding goodness simpliciter in such terms. (She would also oppose her naturalistic story to an intuitionist account of fitting attitudes, but that’s a matter for substantive meta-ethics which is orthogonal to FA’s truth. See section 1.6.) 16 For example, Moore (1993: 150, 1959: 94, 1942: 600). See also Ross (1939: 275–6), who rejects an analysis of one sense of ‘intrinsically good’, while still holding it equivalent to ‘worthy of admiration’. More on this in Chapter 8. 17 Given the FA framework, the task for non-consequentialists is to neither deny the compelling moral thought, nor take Foot’s route, but rather find a way to avoid saying that e.g. letting two killings happen is worse than killing one person; plus,

60

VALUE THEORY

to this amendment of Zimmerman’s rejoinder. Perhaps it will be said that a not-further-specified concept of fittingness presents the same difficulties as goodness simpliciter. One idea might be that something must be always fitting to favour one way or another. But this demand is satisfied by FA: final values are fitting to be favoured for their own sake. However we understand this phrase and the associated attitude, it seems to provide a characterization of one way of being fitting to favour.18 Another idea might be that we should specify not only the mode or ‘direction’ of being fitting to favour (final or non-final), but also its normative dimension, like Zimmerman does by talking of moral fittingness. And if we do this, then again we won’t have an independent notion of final value which we can use in constructing normative claims. In reply, we can say that, even if it were possible to always assign a normative dimension to fittingness claims, e.g.: x (beauty/knowledge/humour …) has final value = it is aesthetically/epistemically/comically fitting to favour x for its own sake when final value is concerned, we assume that favouring these things is also and simply normatively fitting. In terms of reasons, the question is whether the various aesthetic, epistemic, comic, and moral reasons stand as genuine reasons to favour x for its own sake. This is what matters, if final value is supposed to be a guide for attitudes and action. Mere assignment to a putatively normative dimension doesn’t add anything significant in this respect. Yet another objection might be that fittingness, like other normative notions, doesn’t hang in the air, but must attach to agents. Again, FA can accept this: final values are fitting to be favoured for their own sake, in principle, by any agent who is capable of favouring them. If, however, the demand is that FA provide an answer to ‘why this agent?’, then it is a demand too of course, they can restrict the sorts of things they are willing to admit as intrinsic goods and evils (that is, as goods or evils prior to moral evaluation). 18 Thomson briefly suggests that ‘‘prefer” and “favour” are arguably themselves attributive’ (2010: 756). But by analogy with goodness the main point should be that these attitudes are always ‘in a way’ or need specification (favour x how?), and being attributive is only one such mode of specification (favour x as a K).



The Challenge against Absolute Value

61

many. For all that FA says, it might be that there is some kind of further story to tell about the relation between fittingness and particular agents (perhaps something along the lines of Williams’ internalism about reasons), but providing such a story involves taking a step into meta-ethical analysis. FA, on the other hand, is supposed to hold true regardless of whether evaluative and normative judgements are to be explained in realist (naturalist or non-naturalist) or non-realist terms (see section 1.6).19

3.7 Summary In philosophy and ordinary life, it is not uncommon to evaluate things absolutely: from no particular point of view, or relative to no particular standard. Many philosophers have argued that all such talk is ultimately meaningless (Geach, Thomson) or at best derivative upon moral concepts (Foot). By discussing and refining Zimmerman’s reply to this critical challenge, I have tried to show how a fitting attitude account, or at least elucidation, of what is meant by absolute value can make such a concept both intelligible and useful in the construction of moral theories.20 Richard Kraut’s recent attack against what he calls ‘absolute goodness’ explicitly acknowledges the cogency of a fitting attitude sort of reply to Thomson’s challenge (Kraut 2011: 7–8). His target is rather the thesis (attributed to Moore and Ross) that absolute goodness itself plays the role of a (or the) reason-giving consideration (on which see Chapter 8). His arguments are both structural and based on moral intuition. His positive view assigns the only relevant practical role to what is good/ bad for someone (see Kraut 2007). So, as I see it, Kraut doesn’t deny the intelligibility of a concept of final or intrinsic value, but rather locates such value in the property of being good for someone: there are things which it is fitting to value for their own sake, and these things are facts such as that x is good for someone. His ‘ethical objection’ against absolute goodness (2011: 81ff.) is directed at those value theorists who take certain things (e.g. knowledge) as finally good even if not good for anybody. But the latter are substantive views about value which, correct or not, are not required by the concept of goodness simpliciter as such. Otherwise said: it is perfectly coherent to believe both that all that matters is whether things are good for people – or sentient beings – and that this matters for its own sake, i.e. as a final value. 20 Questions for further study: What is the best way to understand goodness of its kind? What is its normative status? (see Schroeder 2010, Skorupski 2007, for a FA-style approach, Brännmark 2008 for criticism, and von Wright 1958, Mackie 1977, and Thomson 1997 for what I take to be a different approach). 19

CHAPTER FOUR

Personal Value 4.1 Introduction In ordinary conversation, it is extremely common to evaluate how good or bad things are for determinate people, how they affect their interests, welfare, or well-being, whether they benefit or harm them. We can assess experiences, events, relationships, or whole lives as being good in this sense. We can also regard particular objects as having personal value for some people: a photograph, a ring, etc. And philosophers have engaged in theorizing about what makes a life good in this personal or prudential sense: hedonism, eudaimonism, perfectionism, ‘objective list theories’ all try to give answers to this question, quite independently of whether pleasure, eudaimonia, etc., are also good in some absolute sense. In this chapter, I stand aside from such substantive questions, and investigate the very concept or property of being good for someone. Starting with G. E. Moore’s critical view, I will then explore how a fitting attitude account of value can help here. I will then discuss the distinct concept of agent-relative value, and understand its connection with personal and impersonal value.

4.2 Moore on good and good for The history of philosophical approaches to the concept of being good for someone, like for the concept of intrinsic value, can

64

VALUE THEORY

be made to start with Moore. In conjunction to his enthusiasm for intrinsic, and indeed absolute, goodness, we find him deeply sceptical about the viability of goodness for. Moore’s discussion takes place in the attempt to refute ethical egoism, i.e. the thesis that everyone ought to pursue their own good, or what is good for them, as the only end. Moore’s aim is to show that good for me or my own good are unserviceable concepts, at least if one wants to build an ethical theory out of them. Moore’s view then is nicely symmetrical to the attempt, by Geach and others, to discredit absolute good, at least if one wants to build an ethical theory or certain ethical claims out of it. Therefore it is worth keeping in mind his critical argument. Moore’s central claim is that ‘when I talk of a thing as “my own good” all that I can mean is that something which will be exclusively mine, as my own pleasure is 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 can exist privately or for one person only’ (1993: 150). In the terms introduced by Geach (see previous chapter), Moore seems to treat ‘good for A’ as a logically predicative adjectival phrase, because it can be logically split up: if x is good for A, then x is good (therefore, good absolutely) and x is somehow related to A.1 Given this premise about the meaning of ‘good for me’, it can be seen how ethical egoism collapses. If my own good is something absolutely good, then anyone in principle ought to pursue it, and the same goes for anyone’s own good: any agent in principle ought to pursue any other agent’s own good. The distinction between my own good, or what is good for me, and the good of others, or what is good for others, no longer holds any fundamental normative significance. There are a number of replies one could give to Moore’s argument. One is this: the ethical egoist need not make any claim about goodness. She could simply claim: everyone ought to pursue their own happiness (or pleasure, or success, etc.) as the only end. Whatever else might be said against such a view, it is left untouched by Moore’s argument. A second reply might be that all they mean Moore also acknowledges that ‘x is good for A’ sometimes means ‘A thinks that x is good’ (compare ‘p is true for A’). But this presupposes again an absolute notion of goodness. 1



Personal Value

65

by ‘x is good for me’ is ‘x benefits me’, and then interpret the latter as a non-ethical claim like ‘x increases my chances of getting what I (really) want’, or some hypothetical claim about what I would desire if I were in ideal conditions of information and psychologically well-functioning. Again, this sort of view would make egoism immune from Moore’s criticism. However, such replies explicitly turn the concept of good for into a non-normative concept: what is good or bad for me would not be immediately related to what I or others have reason to do. Moreover, consider the second reply: x is good for A = A would desire x if A were in ideal conditions of information and psychologically well-functioning. Couldn’t A desire what is absolutely good or good for others as well as what is good for her? Being desired by such an ideal subject might be a necessary but not a sufficient condition for something to be good for her.2 So a more interesting reply to Moore would try to rehabilitate the intrinsic ethical significance of the concept of good for as distinct from absolute goodness. Henry Sidgwick, writing a few years before Moore, thought the concept made perfect sense if we take ‘‘ultimate good on the whole for me’ to mean what I should practically desire if my desires were in harmony with reason, assuming my own existence alone to be considered’ (1981: 112). That is, Sidgwick appears to explain the concept of good for someone in terms analogous to the fitting attitude account: good for me is what it is fitting for me to desire (or I have reason to desire), assuming my own existence alone to be considered.3 Now, this latter qualification does not quite sound right: if we abstract from the existence of other human beings, probably a lot of what we usually think is good for us would be left out (e.g. personal relationships). Moreover, the qualification makes Sidgwick’s definition rather close to Moore’s approach: what is good for me is what I ought to desire that occur in my life considered in isolation from others. As in Moore, the ‘for me’ in ‘good for me’ then seems to refer to a particular location See Zimmerman (2009: 433–4). In general, writers who favour a ‘full-information’ or ‘ideal observer/advisor’ account of value make it clear that they are not addressing goodness for in the sense of personal welfare (Railton 1986: 30, fn.10, Smith 2003). But not all do (e.g. Thomson 1997: 296). 3 See Darwall (2002: 31–8) for a penetrating discussion of Sidgwick. 2

66

VALUE THEORY

where good things (things one ought to desire) take place, rather than being a way for things to be good.4 Therefore, there is no guarantee that all I ought to desire that occur in my life are things that are good for me. Some such things might be the possession of virtuous character traits which, no matter how good absolutely, might conceivably fail to be good for me in the relevant sense of benefiting me. The difference with Moore is that Sidgwick takes the fact that, so to speak, I am that location to be normatively significant: from the point of view of the universe (a favourite Sidgwickian phrase) my good (e.g. my happiness) is as important as others’, all else being equal, but from the point of view of my existence as such my happiness acquires an additional significance. Therefore ethical egoism makes rational sense. What is more, one cannot be rationally compelled to take the point of view of the universe as the dominant one.

4.3 Good for and fitting attitudes Despite issues of formulation, Sidgwick’s idea paves the way to a more general reply to Moore.5 Suppose that value is what it is fitting to favour. If x is good for A, then this suggests that A is implicated one way or another in the fact that it is fitting to favour x. Here are some proposals: a) It is fitting specially for A to favour x. b) It is fitting to favour x for A’s own sake. c) It is fitting for A, and/or for those who care for A, to

favour x for A’s own sake. d) It is fitting, for those for whom it is fitting to care for A, to favour x for A’s own sake. See Hurka (2003: 611). Moore does consider a fitting attitude account of ‘x is good for him’: ‘it is peculiarly appropriate that a thing which will belong exclusively to him should also by him be approved or aimed at’, but dismisses it since ‘by saying that a certain relation between two things is fitting or appropriate, we can only mean that the existence of that relation is absolutely good in itself’ (1993: 152). Appropriateness, like deontic predicates (rightness), is analysed by Moore in terms of intrinsic goodness. See Chapter 8. 4 5



Personal Value

67

In each proposal A has a different role. In (a), she is the subject to whom a certain normative predicate applies. In terms of reasons, we can say A has special or personal reasons to favour x. While these reasons might be shared with others, it is crucial that they need not be so shared. However, just having special reasons to favour some x doesn’t entail that x benefits the subject of those reasons. I might have a special reason to favour repaying my debt to the bank (I incurred it!), but we can easily imagine how repaying my debt might fail to be good for me.6 In (b), A has the role of a patient rather than that of an agent: someone for whose sake it is fitting to favour x.7 Here we get closer to the idea that x somehow benefits A, or makes A better off (and conversely, if x is bad for me, then x harms me in some way). Laura’s recovery from illness is fitting to favour for Laura’s sake: it is good for her. However, consider a thief getting away with theft. We can easily imagine how getting away with the crime might be good for her. But is it fitting to favour such a state of things for her own sake? Some things that are good for people might be unfitting to favour, or fitting to disfavour, and a fortiori not fitting to favour for those people’s sake. Or at least, it is not fitting for just anyone to so favour them. Proposal (c) builds on this idea: only for or by some people is it fitting to favour certain things for a person’s sake. In (c) ‘A’ is a term of the relation both as a patient and as the, or an, Zimmerman suggests a view similar to (a): ‘x is good for P =df. P has a prudential reason to favor x’ (2009: 436). While this might get around the problem of special reasons unconnected to the agent’s welfare, the notion of a ‘prudential reason’ seems too close to that of ‘good for’ in order for the definition to be informative. Also, this definition ignores reasons that others might have to favour x insofar as x is good for P. Maybe the idea is that, whatever such reasons others may have, they are asymmetrically dependent upon P’s (actual or possible) prudential reasons. For example, parents have reason to favour what’s good for their infant child, only because the child would have reason to favour what’s good for her, if she could (and not vice versa). Compare with proposal (e) below, which doesn’t give such priority to A’s reasons (or to what it is fitting for A to do). Compare also with Schroeder (2010: 46–8), whereby the right kind of reasons need not be A’s reasons, but rather those which would be shared by anyone engaged in the activity of ‘watching out for A’. 7 Compare Rønnow-Rasmussen: ‘An object x’s value for a person a (i.e. x’s personal value), consists in the existence of normative reasons for favouring/disfavouring x for a’s sake’ (2011: 47). 6

68

VALUE THEORY

agent from whom a certain response is called for. By whom is it fitting to favour getting away with crime for the criminal’s sake? Well, probably only by the criminal, although it is an interesting question whether the sympathy felt by those who care for her is to some extent a fitting reaction. For instance, one might say that a complete stranger has no business caring for criminals, and to this extent their sympathizing is simply not fitting or even positively unfitting (think about a stranger showing their sympathy by praying for the criminal’s success).8 Therefore a better idea is that only those for whom it is fitting to care for A will also fittingly favour things for A’s sake. This is proposal (d). The proposal has two virtues: first, it restricts the pool of ‘favouring agents’: what is good for A need not concern just anyone who could want or do something for A’s sake; second, it maintains the idea that what is good for A has to do with what benefits A (something one should want for A’s sake). One problem is that (d) makes goodness for hostage to there actually being somebody whose caring for A is fitting. And we want to allow that things can be good and bad for A, even if A herself cannot favour or disfavour them (because she is an infant, or in a coma, or dead, etc.), and even if there happens to be nobody else around to care for A. But a simple modification might help: e) x is good for A = it is or would be fitting, for those for

whom it is fitting to care for A, to favour x for A’s own sake, if they could.

If we take (e) as our favourite understanding of goodness for, then ethical egoism will be the thesis that everyone ought to pursue what it is fitting to favour for their own sake (by those for whom it is fitting to care for them). This is a recognizable form of egoism, if a little bit complicated in its formulation. And even if one doesn’t follow the egoist in thinking that this is the only ultimate ethical principle, it seems that (e) can make sense of good for as a valid normative category quite independent of absolute goodness.9 Proposal (c) is close to Darwall’s thesis: ‘[W]hat it is for something to be good for someone just is for it to be something one should desire for him for his sake, that is, insofar as one cares for him’ (2002: 8). 9 A worry about (e) might be that the notion of caring can only be understood in 8



Personal Value

69

4.4 Moore strikes back? Moore’s attack on personal goodness was based, as seen, on the notion that what is good for me must be good (in some other, non-personal, sense), and must be related to me, in an ownership relation of some sort. The fitting attitude account that has emerged denies both conditions. First, x can be good for A without being good in some other non-personal sense. The thief getting away with theft is one such case. Second, x need not be ‘owned’ by A, or even be a state of affairs in which A is essentially involved. For instance, that my favourite football team wins the league is such a state: I don’t need to exist in order for them to win the league. Let’s assume it is good for me. This is because it is fitting that those, who fittingly care for me, favour such a state for my own sake. Of course, the reason why so favouring this state is fitting will be something like: ‘he’s going to be happy for his team’. Wishing x for A’s sake makes sense only if x is somehow thought to benefit A.10 But the bearer of personal value need not be any particular state belonging to the person.11 But it is hard to shake the feeling that nothing can be good for me or contribute to my well-being, if it’s not already good.12 For example Scanlon (who in many respects lies at the opposite of Moore) writes: It would make sense to say that I work hard at philosophy because I believe it is worthwhile, or because I enjoy it, or even because I long for the thrill of success. But these things in turn are not desirable because they make my life better. Enjoyments, terms of good for, so that (e) would be circular (Fletcher 2012a: 88–9). 10 As a friend of yours, it could be fitting for me to respect your desire to engage in something that is bad for you but morally or otherwise good. And it seems that I would do this for your sake, though your desire is something admittedly bad for you. But this is no objection to the account, if we think that while your desire is bad for you, my respecting it is to some extent good for you. In this case, respect has the role both of a fitting ‘for someone’s sake’ attitude (me respecting you) and of a state of affairs that is good for you (you being respected by me). 11 I note agreement with Rønnow-Rasmussen (2011: 20), although he thinks these (and the case of respect in the previous footnote) are cases where goodness for doesn’t necessarily mean welfare in a ‘mental state-sense’ (see 2011: 86–8). 12 See Regan (2004).

70

VALUE THEORY

success in one’s main aims, and substantive goods such as friendship all contribute to well-being, but this idea of wellbeing plays little role in explaining why they are good. This might be put by saying that well-being is what is sometimes called an ‘inclusive good’ – one that is made up of other things that are good in their own right, not made good by their contributions to it. (1998: 127) Scanlon seems to express two ideas here. The first is that the way we think about what is good for us is not independent of what we think is simply good or worthwhile. We desire what is good for ourselves under the guise of the simply good, or at least not in contrast with it. An account that makes personal value independent cannot make sense of this fact. However, this point just is not very convincing. The intellectually honest thief (and those who care about her) may well recognize that her getting away with crime is, simply, bad, and yet still believe it is good for her. She might be wrong as a matter of fact (maybe being caught would be better for her), but she is not conceptually confused. The other idea is that contributing to well-being doesn’t explain why things like friendship or enjoyment are good – if there is any order of explanation, it goes from goodness to well-being. This remark, however, is compatible with thesis (e) above. Defending a non-Moorean account of goodness for doesn’t commit one to taking good for or well-being as somehow prior to goodness. Indeed, we should not expect that the fact that x contributes to someone’s well-being explains why x is good non-personally, since the point stressed is precisely that the two concepts are independent.13 All such an account entails is that (e.g.) friendship is (generally, by no means always) both simply good, and good for those who participate in it. Indeed, this very case suggests another example of conceptual independence: particular friendships might be good for the participants, but simply bad – say, because they revolve around sadistic activities against strangers. The Moorean might insist at this point that if getting away with crime and immoral friendships are really good for these people, then they must at least be pro tanto simply good, even if In fairness to Scanlon, he is here arguing against taking well-being as the ‘master value’, like utilitarians typically do. 13



Personal Value

71

they are simply bad overall: that is, there is at least some respect under which they are simply good.14 However, in the light of the general connection between value and fitting attitudes, this means or entails that there is some feature which makes it to some extent fitting for anyone to favour immoral friendships and getting away with crime, even if overall it is fitting to disfavour them. It is not clear, to say the least, that it is fitting even pro tanto for the victims of such states of affairs to favour them. Moreover, there is a general worry here: it is difficult to believe that whether something is good for me should conceptually depend on what it is fitting for anyone to do, no matter whether or how related to me they are. It is a conceptual truth that if x is good for me, and x, y, you, and I are similar in all relevant respects, then y is good for you. In this sense what is good for me conceptually depends on what it is fitting for others to do for your sake (and vice versa). But whether what is good for me is also simply good seems a further question, not to be decided on a priori grounds alone.15 Of course this doesn’t mean that there are no connections between good and good for. But these will be rather indirect. First, it might be plausibly said that what is good for A must at least be a default good: i.e. it must be the sort of thing that is in general good (e.g. freedom, friendship), but could in given circumstances explicably fail to be even pro tanto good, as in the cases cited of freedom from a deserved punishment, and immoral friendships.16 Second, there are structural affinities. The distinction between final and non-final values holds also for personal value. We draw distinctions between things that are good for us for their own sake and those that are good for us for the sake of something else that is good for us.17 Being rich, arguably, is good for me to the extent See Fletcher (2012b: 20–1). Rosati argues that good for a person depends on the (non-personal, intrinsic) value of the person: ‘[with] X is good for P, we indicate that there is a reason to promote X with P as the beneficiary of the action in light of or out of regard for the value of P’ (2008: 344). I have no space to discuss this interesting proposal. 16 Or at least it must be a default good under some relevant specification. A team winning the league is certainly not a default good, but as argued it might be good for someone. See section 5.6 for a brief discussion of default value. 17 Can final personal value be intrinsic, i.e. based on x’s intrinsic properties alone? That depends: if x is good for A, and x is not already a state of A’s, then such value will be extrinsic (i.e. result from some relation between x and A). 14 15

72

VALUE THEORY

that it allows me to have the sort of things that directly contribute to my well-being (e.g. enjoyment, achievement, etc.). Note: just to say that being rich is good for me because it contributes to my wellbeing by itself doesn’t mean that being rich is instrumentally good for me, since the same could be said about any x that is good for me. Indeed, the ‘because’ here doesn’t introduce an explanation, but rather a reformulation. The question is whether being rich contributes directly or indirectly to my well-being. In turn, the question is whether being rich is a condition that I, and those who fittingly care for me, should favour as such for my own sake. In general, then: x is finally good for A = it is or would be fitting, for those for whom it is fitting to care for A, to favour x for A’s own sake for its own sake, if they could. One should not be misled by this welter of ‘for the sake of’ locutions. The idea is not that x is to be favoured both for A’s own sake and for its own sake. For that would make non-personal final value a necessary condition for personal final value. Rather, ‘for A’s own sake’ is the logically primary specification of ‘it is fitting to favour x’, and ‘for its own sake’ acts as a modifier of ‘for A’s own sake’:18 it is fitting, for those for whom it is fitting to care for A, to favour x (for A’s own sake [for x’s own sake]). This formulation should make it clear that, while final values can be personal or impersonal, final impersonal value, or what we identified in the previous chapter as absolute value or value simpliciter, is distinct in two ways from personal value: (1) it takes ‘for its own sake’ as the only necessary minimal specification of fittingness, i.e. final impersonal values might be worth favouring for someone’s sake, but need not, because the question ‘for whose sake?’ need not always apply; (2) if fittingness is a three-place relation between an object (state of affairs), an attitude, and an agent, then in the case of final impersonal value it is left unsaid who Heathwood (2008: 55) seems to disagree. See also Rønnow-Rasmussen (2011: Chapter 5). 18



Personal Value

73

or what fills in the agent term of the relation, whereas in the case of personal value, the agent term is filled by ‘those for whom it is fitting to care for A’, which might or might not include A herself. I turn now to this second aspect.

4.5 Agent-relative value Personal value, as we have discussed it so far, must be distinguished from another concept of which much has been made recently: agent-relative value. The distinction has in fact already emerged in discussing proposal (a) above: x is good for A = it is specially fitting for A to favour x. The objection against (a) was that it might be specially fitting, for moral reasons, for A to favour his repaying a debt, while repaying a debt might not be good for him – it will make him poorer, for instance. However, the idea that a normative demand attaches especially to an agent (and not to another), while unable to capture the concept of good for A, has been thought to configure a different category of value: x as being good relative to A. Agent-relative value, unlike personal value, appears to be a technical concept, created for specific reasons having to do with the credentials of consequentialism. In its classical form, consequentialism is roughly the view that we morally ought to maximize final value, where the latter is non-relative, absolute final value. One implication of this is that, if faced with the choice between letting two murders happen and murdering one person ourselves, we morally ought to murder the one person. This is because we have to choose between two evils of the same sort (two murders, one murder), and for consequentialism this is all that morally matters. It would be wrong to choose the greater evil (two murders). Committing murder is the only way to maximize value here.19 Non-consequentialists have complained that it cannot be right to murder to prevent two murders. From a classical consequentialist Here I am talking about classical, or act-consequentialism. A rule-consequentialist will instead say that though committing the murder is the only way to maximize value, what matters is the value maximized by the internalization of and conformity to certain rules, rather than directly by a single act. Since one such rule might well be ‘don’t kill an innocent human being’, rule-consequentialism might not require committing the murder. See Hooker (2000). 19

74

VALUE THEORY

point of view, such complaint appears groundless: how is it not right to prefer the least bad state of affairs? However, some consequentialists have taken a different line, one that allows at least in principle for the non-consequentialist claim. They do not abandon the central idea, i.e. that we morally ought to maximize value, but relativize the notion of value to particular agents: each of us morally ought to maximize (final) value relative to them.20 Going back to our example: if we grant the substantive claim that one murder by A is worse relative to A than the murders by B and C together, then A ought not to murder in order to prevent murders by B and C. When seen as absolute evils, all murders have the same disvalue (ceteris paribus): therefore the more, the worse, period. But when their disvalue is relativized to agents, it seems both possible and plausible to assert that the murder I commit is worse relative to me than the murders committed by someone else. Therefore I ought not to murder, since if I murdered someone I would maximize disvalue relative to me. This agent-relativized structure can make sense of a range of situations where we think it morally significant that the agent is involved. I need to repay my debt to my creditor, but I learn that I could help B repaying her higher debts to her creditors. But if I do this, I will be unable to repay my debt to my creditor. Ceteris paribus, it seems I have an overriding duty to repay my debt. According to agent-relative consequentialism, this is because repaying my debt is better relative to me than repaying others’ debts. Analogous examples can be multiplied at will. But introducing agent-relativity is not just a move to make consequentialism more plausible. In fact, it affects the way we conceive of value. If all value is always relative to some agent, then it is a mistake to think of value that is not relative to anyone. The most one can say is that certain things have agent-neutral value: they are good- (or bad-) relative to anyone, and their value does not vary from agent to agent. The classical consequentialist then believes that there are only agent-neutral values. Ceteris paribus, my murder and someone else’s murder have equal value relative to me (and relative to anyone else, of course). The agent-relative Again, this does not mean: each of us morally ought to maximize what is good for them. This is egoism, not agent-relative consequentialism. See Schroeder (2007: 272–3). 20



Personal Value

75

consequentialist will on the other hand insist that at least certain kinds of goods and evils cannot be assumed to have agent-neutral value. The agent-relativist and the agent-neutralist could and perhaps should agree on a given list of universal final values – A’s murder is bad relative to A just as B’s murder is bad relative to B, and so on. Relativization does not mean private or individual values, for the reason why A’s murder is bad relative to A is the same reason why B’s murder is bad relative to B. But the crucial question is, how good or bad, relatively to each, are these things? If murder is an agent-relative evil, then B’s murder might be bad relative to A, but not as bad relative to A as A’s own murder, and vice versa as regards B. In other words, given two instances x and y of the very same value V, there is no single ranking on which we can say that x is better than, worse than, or as good as y. Rather, there are as many rankings of x vis-à-vis y as there are agents.21 Authors have noted that the idea that value is essentially relative to agents had better be something more than just a technical device, on pain of introducing a purely arbitrary and ad hoc solution to the problems of consequentialism. It has to express a concept of value which somehow resonates with our understanding of value. But some options in this respect are blocked off: (1) Agent-relative value is not personal value, as seen above. (2) Agent-relative value is not ‘value relative to my evaluative point of view’, i.e. relative to what I happen to value. This would make agent-relative consequentialism the subjectivist view that we ought to maximize what we in fact value. But this is not what we are looking for.22 Also, agentrelative and -neutral value cannot be seen as simply the evaluative counterparts of agent-relative and -neutral reasons. Agent-relative reasons, on an intuitive understanding, are reasons that apply to some agents only, while agent-neutral reasons apply to everyone. In other words, agent-neutral value means agent-invariant ranking, and agentrelative value means agent-variable ranking. A radical agent-relativist can of course hold that all value is agent-relative in this sense. But while in principle variable, rankings belonging to ‘equally uninvolved’ agents can converge: relative to C and relative to D, where C and D are total strangers to A and B, A’s and B’s murders should rank equally. Likewise for ‘equally involved’ agents: ceteris paribus, the wellbeing of a son vis-à-vis the well-being of a daughter should rank equally in their parents’ respective rankings. 22 Also see Schroeder (2007, 2006) for why agent-relative value cannot stem from a speaker-relative semantic view of ‘good’. 21

76

VALUE THEORY

Now, there is a reason for everyone not to murder. This reason is agent-neutral. But the act or state of affairs of murdering is held to be an example of an agent-relative value. So, the two ideas are different.23 However, there is a natural home for the notion of agent-relative value. The thesis that value and fitting attitudes are in broad correspondence seems to allow or perhaps even require a notion of value relative to. Fitting attitudes will be some agent’s attitudes. So, if x is valuable, it always makes sense to ask for whom it is fitting to favour x. And if it is fitting for me (but not necessarily for others) to favour x more than y, then we can say that x is better relative to me than y. This seems to be what agent-relative consequentialists mean when they talk about ‘evaluator-relative’ value (Sen 1983): not what each of us happens to value, but what each of us should value or, better, should favour. While FA makes room for agentrelative value, it must be stressed that it does not follow from FA alone that there actually are agent-relative values: that is, as far FA goes, it could be that whatever is valuable is simply valuable relative to anyone, and exactly to the same degree. Some may feel that appealing to what we ought to favour presupposes what consequentialism must explain. But notice: for agent-relative consequentialism each morally ought to maximize what they, and possibly only they, ought to favour the most, i.e. what ranks highest in their rankings. These concepts are all different: (1) ‘ought (or ‘fitting’) to favour’ doesn’t mean ‘morally ought to favour’ (see Chapter 3); (2) ‘ought to favour the most’ doesn’t mean ‘ought to maximize’ (see Chapter 7). Thus understood, agent-relative consequentialism states an interesting and distinctive connection between moral concepts and evaluative ones. A non-consequentialist still has plenty of room to deny (1) that all we morally ought to do has to do with value (be it relative or neutral), and (2) that all we morally ought to do with value is to maximize it.

Schroeder shows how it is Thomas Nagel’s own technical understanding of reasons for action as reasons to promote a certain state of affairs that leads him to assume a tight fit between the two distinctions (2007: 276–7). For the debate and literature on agent-relative and agent-neutral reasons, see Ridge (2011). 23



Personal Value

77

4.6 Impersonal/personal and agent-neutral/agent-relative Impersonal value poses no restriction on the class of favouring agents. If FA is broadly correct, impersonal value is relational in its structure just like personal value, but as noted above, it is left unsaid for whom it is fitting to favour stuff for its own sake. Importantly, it is misleading to think that the class of favouring agents is ‘everyone’. While it may be true that impersonal value makes claims on everyone, it is still not the same as value relative to everyone. Impersonal value is meant to generate agent-invariant rankings, rather than being the result of uniformly converging agent-variable rankings. The class of favouring agents is perhaps better expressed by ‘anyone’, that is, no one in particular. Thus impersonal value is properly agent-neutral value. Also, if ‘agent-neutrally better’ is understood as ‘better relative to everyone’, we will have to rule out that things can have conflicting agent-neutral and agent-relative values. For it cannot be the case both that x is better relative to me than y (because x is better relative to everyone than y, and that includes me) and that y is better relative to me than x (because y has a particular agentrelative value relative to me). That is simply inconsistent. On the other hand the idea of ‘anyone’ behaves differently, for it refers, so to speak, to an ‘anonymous’ agent: as such, I should prefer x to y; while as this particular involved agent, I should prefer y to x. There is conflict between agent-neutral and agent-relative value but no contradiction. As an agent among others, I should prefer my one murder to two murders; as the agent who would commit the murder, I should prefer the two murders to my one murder.24 The agent-relative consequentialist solves such conflicts by giving priority to the agent-relative value.25 Suikkanen (2009) similarly speaks of a conflict between the agent as an impartial spectator and the agent as a situated agent. Smith (2003) arguably suffers from the contradiction problem, once we formulate his view in terms of betterness rather than goodness. See Schroeder (2007: 292–4) on this issue. 25 The radical agent-relativist will simply deny that there is any agent-neutral value, i.e. any agent-invariant ranking, and so will solve the conflict by rejecting its very possibility. This doesn’t mean that according to such a view when the agent is not involved as an agent there is nothing she ought to do: it’s just that similar things will 24

78

VALUE THEORY

What about the relation between personal and agent-relative values? Well, some agent-relative values neatly overlap with personal values. What is good for A, in the definition given above, addresses a restricted class of agents whose favouring for A’s sake is fitting: those who fittingly care for A. So what is good for A will rank higher for these agents than what is good for B (that is, provided that it is not fitting at all for them to care for B, or that it is fitting for them to care for A more than for B). And vice versa for those who fittingly care for B. Personal values thus seem to generate agent-variable rankings. But not all putative agent-relative values are personal values. Indeed, typical examples of agent-relative value are not well understood as personal values. Paying my debts is better relative to me than paying others’ debts even if the latter is better for me (say, because it will get me a lot of friends). Murder by my own hands is supposed to be evil relative to me even though it might in given circumstances be good for me (say, it’s my only way to safely escape imprisonment). Agent-relative values need not be states of affairs that are good for the agent or for other people involved. As such, then, they are not reducible to either impersonal or personal values, though they have something in common with both. Like impersonal value, agent-relative value need not presuppose fitting attitudes of care or ‘for-someone’s-sake’ attitudes towards anybody. Like personal value, agent-relative value keeps track of who is the agent and how she is related to the relevant state of affairs. One might complain that if agent-relative value is neither personal nor impersonal, then it is not a species of value. But such a complaint can only come from someone who rejects the whole fitting attitude approach to value. If we think of value in FA’s terms, then there will be as many putative species of value as the results that we can generate by (sensibly) ‘toggling the controls’ within the FA schema. The agent-neutral/agent-relative value distinction is the result of toggling the ‘agent’ control one way or another – indeed, we could compare agent-neutral value to the muting of such control. Having distinguished these four forms of value, I do not mean to deny possible overlaps. Precisely how these values overlap is a happen to rank similarly within her ranking. If I can choose between 100 murders and 1 murder by others’ hands, I ought to choose the 1 murder, because relative to me 100 murders by others’ hands are simply worse than 1.



Personal Value

79

matter of substantive dispute, but the following propositions should help to give an idea of their possible relations. I have suggested that (1) if x is agent-neutrally good, then it is impersonally good, and vice versa, and (2) if x is personally good, then it is agent-relatively good (where the agent might or might not be the beneficiary of x), but the converse doesn’t hold. Beyond this, it can be argued that (3) some xs are both impersonally and personally good (and thus agent-neutrally and agent-relatively good); (4) some xs are only agent-neutrally/impersonally good; (5) some xs are only agent-relatively good. An example of (3) might be my own health: it matters as both something absolutely good and good for me, and thus something that has both a position in an agent-invariant ranking of states of health, and one in a ranking relative to me (and certain others). Examples of (4) might be certain aesthetic or logical properties (harmony, coherence), where there is simply nobody to care about for their own sake. Examples of (5) might be either values that can only be personal (e.g. a killer getting away with crime, as discussed above), or agent-relative values that are not personal, e.g. the honour of my family, which might conceivably fail to correspond to my or anyone’s well-being. The following diagram should help to illustrate these possibilities, while keeping in mind that one might on substantive grounds reject some or all of them:

Impersonal Agent-neutral

Personal

Agent-relative

Figure 4.1

80

VALUE THEORY

4.7 Summary This chapter explored the common-sense notion of something being good for someone. Pace Moore’s worries, this concept is not easily reducible to impersonal goodness, and a fitting attitude approach can make sense of its distinctive normative import. I have then examined the relations between personal value and the distinct idea of agent-relative value.

CHAPTER FIVE

The Chemistry of Value 5.1 Introduction In this chapter, I examine what we might call the ‘chemistry’ of value: how the value of a valuable object gets formed from its other features, and from other values possibly present in the object. First I clarify the notion of supervenience, and introduce a distinction between the value-making features of an object and other conditions that enable such features to be value-making. Then I discuss the so-called principle of organic unities: the idea that the value of a complex object need not be just the sum of the values of its parts. Noting some problems with it, I contrast it with two alternative views, virtualism and conditionalism. Finally, I briefly present the ideas of invariant and default value, with reference to the recent debate on particularism in moral theory.

5.2 Supervenience and other relations Value essentially depends on something else. If a certain object is valuable, then something other than its being valuable must make it so. This is reflected in our evaluative practices: while it might be obvious in many cases why a certain decision or a certain state of affairs is a good or a bad one, one is always in principle entitled to an answer as to why it is good or bad. Nothing can be just good, just bad, or just better than something else. The concept

82

VALUE THEORY

of supervenience is an attempt, in part, to articulate rigorously this intuitive idea. To say that value properties supervene on non-evaluative properties is to say that value properties depend on non-evaluative ones in a certain way: if two objects (states of affairs, experiences, actions, or what have you) have exactly the same non-evaluative properties, then they must have the same value properties. For example, if my toothache is just as painful as your toothache, then (other things being equal) they must be equally bad, i.e. bad to the same degree. But this positive way of formulating supervenience, while perfectly accurate, might lead astray into thinking that if two objects have different non-evaluative properties, then they cannot or do not have the same value properties. But the latter claim is no part of supervenience (‘if p, then q’ does not entail ‘if not p, then not q’!). Moreover, it is obviously false. My toothache might be caused by poor toothbrushing, while yours might be caused by a wisdom tooth, but this non-evaluative difference might not make any evaluative difference: both experiences can still be bad to the same degree. In order to avoid such misunderstandings, it is common (though not more accurate) to describe supervenience in negative terms: there cannot be a difference in evaluative properties without a difference in non-evaluative ones. If my toothache turns out to be worse than yours, then this must be because of some non-evaluative difference between our aches: maybe mine lasts longer, or is more intense than yours. It is instructive to point out a few features of supervenience: 1 It is a principle about what necessarily is the case: notice

the ‘cannot’. This means that any difference in value properties requires a difference in other properties. It’s just not possible for two things to be evaluatively different (say, one good and the other bad) and be exactly similar in other respects. 2 Supervenience does not tell us which things or states are good and bad. It is a logical constraint or requirement on things that are or could be already good or bad (note that its logical form does not even entail that there are good or bad things). It is rather about the relation of properties of a certain kind – whatever they are, actually instantiated or not – to properties of a different kind. Nor



The Chemistry of Value

83

does supervenience tells us how to go about deciding what is good and bad, since it doesn’t tell us for any specific value property (say, the badness of my toothache) which are the non-evaluative respects on which it supervenes. In part, this is because it is a metaphysical relation, not an epistemological one: it is about how things could (not) be. But even the epistemological derivates of supervenience (for instance: ‘you cannot assert that there is an evaluative difference between x and y if x and y are exactly similar in other respects’) do not tell us what these other respects are – except that they are not evaluative ones. In this regard, supervenience of value is less informative than the supervenience claims commonly made about other sorts of properties; e.g. it is generally accepted that mental properties (at least for humans) supervene not just on any non-mental ones, but specifically on physical properties of brains (and not, say, on geological or astronomical properties!). 3 Supervenience does not tell us much about the sort of

relation between value and non-value properties, except of course that they stand in a modal relation: any change in the former necessarily entails a change in the latter. But we are not told whether value properties reduce to, consist in, are grounded on, are caused by ... etc. non-value properties.

4 The supervenience claim leaves it open whether some

evaluative properties also supervene on other evaluative or anyway normative properties. For instance, if we are to believe the fitting attitude account of value presented in Chapter 1, then evaluative properties like being good and being bad supervene on normative properties such as an attitude being fitting. Likewise, if intra-evaluative analyses are correct, then e.g. the property of being good will supervene on (roughly) the property of being better than nothing. Also, very often the only manageable way of stating parts of the supervenience base of particular evaluative facts will be in terms of other evaluative, or anyway not simply descriptive or naturalistic, facts: the action was admirable because it displayed the proper

84

VALUE THEORY

amount of courage; a policy is better than another because it allows people to get what they deserve; a pleasure has positive value if it is innocent, and so on. However, the supervenience claim does imply that all these further evaluative facts in turn supervene on non-evaluative ones: nothing can be just proper, or just deserved, or just innocent, even if there may be no finite way to spell out in non-evaluative terms the full supervenience base of these properties. The idea of supervenience is intuitively understood in spatial terms: evaluative properties stand on top of (super-vene on) non-evaluative ones (which therefore sub-vene under them). This is why it is common to talk of the supervenience base of evaluative features. The principle of supervenience does not tell us what is included in any supervenience base; all one can say is that any given supervenience base includes all the non-evaluative properties which are able to make an evaluative difference. In other terms, it includes the properties without which the object or state in question would be evaluatively different: say, not as good as it is, or not good at all. Therefore the base of any given value property may be a very diverse and possibly not entirely discernible set of properties or facts. By way of illustration, just consider all the negative facts which are able to make an evaluative difference: e.g. a certain pleasant experience is good, as long as it does not lead to painful experiences, it is not a response to your children being tortured, and so on. For this reason, some philosophers have thought it useful to distinguish, within any supervenience base, at least three different sorts of features or properties: (1) those from which a value results, or which ground value, or make something good (also known as the ‘resultance base’); (2) those which enable the former properties to make something good, and whose absence might disable the good-making role of the former properties: the enabling conditions; (3) those which intensify or diminish the value of something (Dancy 2004a: 170–2, 178–81). As an example, consider an experience of pleasure, say, obtained from a warm bath. Let’s assume that it is good to a certain degree. On what does its goodness supervene? (1) On what its goodness results from, what makes it good: arguably, the very sensation of pleasure. (2) On a number of enabling conditions, such as the



The Chemistry of Value

85

negative facts recalled above: that the pleasant sensation does not lead to subsequent painful sensations, it does not distract me from more important tasks, it is an innocent pleasure, and so on. The idea here is that if any of these things were not the case, the pleasantness of the sensation would not count in favour of the experience, that is, would not make the experience good. But of course, enabling conditions can also be positive facts: recall the discussion of evaluative and factual enabling conditions in section 2.2. (3) Possibly, on a set of intensifying factors: e.g. that the pleasures of the bath come after a long day of work makes the experience better, or have a higher degree of goodness than it would have, say, after a lazy Sunday, though that would still be quite good. Besides purposes of conceptual clarification, these distinctions are particularly suited to express certain views about the computation of value – i.e. about how, given a certain value assigned to a complex state of affairs, we should understand the contribution made by the different elements composing that state. Originally devised by Jonathan Dancy to support his particularism (see section 5.6), we will see how the distinctions can also be usefully employed by less radical views (section 5.5).

5.3 Organic Unities In Chapter 2 we discussed Moore’s idea that final value depends only on intrinsic properties of an object or state, that is, that final value is intrinsic value. We had examined some difficulties for this thesis, which now we can put in the following terms: the supervenience base of some final values seems to include extrinsic properties, for instance as enabling conditions, or as part of the resultance base, or again as intensifiers, so that final value cannot always only depend on intrinsic properties. But Moore’s view seems to face another difficulty. If final value depends on intrinsic properties only, then extrinsic facts or properties won’t be able to affect final value. Freedom of movement, arguably, has final value, and the loss of it has final disvalue: we should regret our own or others’ loss of freedom of movement, for its own sake. But is there anything to regret about a jailed serial killer not being free to move around?

86

VALUE THEORY

Well, if the thesis that final value is intrinsic value is correct, then the killer’s lack of freedom is still finally bad, and just as bad as an innocent person lacking freedom. The fact that he lacks freedom as a result of his crimes, being an extrinsic fact, does not matter for its final disvalue. So, if we consider this state of affairs: ‘killing people and being jailed for that’, we will have two intrinsically bad things: people being killed, and someone being jailed. We could then evaluate it by just summing these intrinsic disvalues: say –10 (people being killed) and –2 (someone being jailed) = –12. But then this state will turn out to be worse than the state in which people are being killed and the killer gets away with the crime: -10, for simplicity. For many, including Moore, this is unacceptable: that a serial killer is jailed is not worse, but better than if he were on the loose, and finally so – not just because he is then unable to harm others, or the victims’ relatives feel satisfied. Using Moore’s isolation test (section 2.3), it seems we would prefer a world where he is jailed to a world where he runs free, even ignoring all other consequences. But the thesis that final value is intrinsic value cannot underwrite this intuition. Is there a way around it? The answer, given by Moore, is that the final value of a state of affairs composed of parts (a whole) need not be equal to the sum of the final values of the parts – i.e. the values they have intrinsically, or on their own (Moore 1993: 79). Some wholes are organic unities in this sense. Another example might be the following. If pleasure as such has final intrinsic value, then being pleased as such has some positive value. Now, cats being tortured is rather bad. If cats being tortured = –10, and being pleased by something = +2, then we might expect that the value of ‘being pleased by cats being tortured’ equals –8, and as such is at least a bit better than cats being tortured. But that’s clearly unacceptable (at least for many, Moore included): the whole ‘being pleased by cats being tortured’ is definitely not better, and likely worse, than cats being tortured. So its value cannot result from a sum of the intrinsic values of the parts. Or take artistic examples. The final value of the Mona Lisa must result from its intrinsic properties. However, arguably the painting would have little to no aesthetic value if we simply summed the final values of each brushstroke. Conversely, if we join together two pieces of music, each beautiful on its own, we might get something quite cacophonic. As we sometimes in fact say, it is the whole that counts.



The Chemistry of Value

87

But how can positive (negative) value result from a whole composed only of bad (good) parts, or even parts which are neither good nor bad? Stating the principle of organic unities is one thing, but accounting for how value gets formed in an organic unity is quite another. One idea is that, in such cases, the relevant state of affairs S is composed by two sorts of entities: the parts of the whole, and the whole as a distinct and additional entity, e.g.: S: (a) people being killed, (b) killer being jailed, (c) (a&b). It is important here to understand that (c) is not quite the same thing as the state of affairs we are evaluating: although obviously overlapping with it, for the purposes of computing value (c) is to be understood as a constituent of the state of affairs under consideration, so that its value contributes to, but in principle does not exhaust, the value of the state of affairs.1 This approach is suggested in the distinction that Moore draws between the value of a whole as a whole and its value on the whole: the value which a thing possesses on the whole may be said to be equivalent to the sum of the value which it possesses as a whole, together with the intrinsic values which may belong to any of its parts. (Moore 1993: 263) The overall value of a state of affairs like ‘a serial killer being jailed’, that is, its value on the whole, results from the sum of three factors: the values of the two parts (a) and (b), plus the value of the overlapping whole as a whole ((c) above), which is determined organically – i.e. doesn’t need to correspond to the sum of the values of (a) and (b). Moore thus makes it clear that he does not deny the additivity of value, i.e. the idea that value must ultimately be computed by performing the operation of summing.2 Only, The difference between the state of affairs evaluated and the whole can also be appreciated by thinking of the state of affairs as a complete world where only the whole and its parts exist, or at least where only the whole and its parts have value (see Bradley 2002: 35). 2 See Carlson (2001) and Bradley (2002: 35–6). For this reason I have avoided, in expounding organic unities, talk of contrast between intrinsic and contributive/ ory value, value as a part. For example, Dancy (2004a: 176ff., 2007) points out that, in an organic unity, parts seem to contribute a value which, intrinsically, they 1

88

VALUE THEORY

when we are dealing with complex states of affairs, we should take into account as a further addend the value of the organic unity (should there be one), on top of the values of the parts. So the value of the state of affairs will be something like this: (i) S’s value on the whole = (a) (–10) + (b) (–2) + (c) (4) = –8. By giving (c) an ad hoc value of +4, we get the result that a serial killer being jailed is at least better than people just being killed (–10), though it is still a bad state of affairs. If one instead wants to claim that retributive punishment has positive value, then the value of (c) will have to be increased accordingly, say +14: (ii) S’s value on the whole = (a) (–10) + (b) (–2) + (c) (14) = 2.3 The doctrine of organic unities, so understood, seems rather problematic for two reasons. The first issue is what I call attitude mismatch. The introduction of the value of the whole (c), as an additional element to be computed in calculating the total value of S, has strange implications. Remember that (c) is in effect overlapping with S. But how can the value of (c) (‘value as a whole’) diverge so much from the value of S (‘value on the whole’)? The problem becomes more evident when we understand final values as the objects of fitting attitudes. Consider example (i). Here it is fitting to have a negative attitude towards S, but also fitting to have a positive attitude towards (c). Or consider example (ii). Here lack. For example, in deserved pain, pain seems to contribute a value (makes the whole better) which intrinsically it lacks, if we assume that pain is intrinsically bad. However, if we understand organic unities as Moore himself did, we see that, like for all complex objects, in computing the value of an organic unity we are not done until we compute its value on the whole: and what makes things better on the whole is not any contributory value of pain, but rather the intrinsic value of (c) as a whole. Of course, the value of (c) still supervenes on pain and its being deserved, but this doesn’t mean that pain itself contributes a positive value to (c). The positive value of (c) simply results from the non-additive combination of (a) and (b): what value each part contributes is, in a sense, a wrong question to ask about (c). For the same reason, it cannot be said that in (c) pain doesn’t contribute its intrinsic disvalue, i.e. that, paradoxically, such disvalue is at the same time still in place but not contributing to the value of (c) (Zimmerman 2001: 145). 3 This captures the implausible view that punished crime (+2) is better than absence of crime (0). The example is merely illustrative.



The Chemistry of Value

89

it is fitting to have a positive attitude towards S, but a much more intense positive attitude towards (c). Such mismatches of fitting attitudes are, at best, very surprising, since S and (c) come down to the same metaphysical thing: a serial killer having killed people and being jailed. S and (c) overlap, since (c) contains whatever is included in S. There seems to be no non-evaluative difference between them which would warrant having attitudes differing in kind or even in degree towards the two states.4 When we consider the relation of values to fitting attitudes, a second problem emerges, which I call normatively idle values. Recall that Moore is drawn to the doctrine of organic unities due to the thesis that final value is intrinsic value. This means that, e.g., if pleasures of a certain intensity have some positive final value, then the pleasure felt by the sadist in seeing cats being tortured carries the same positive final value as an equally strong but innocent pleasure. Occurring in different contexts or wholes does not affect their intrinsic value. As Moore explicitly says: ‘The part of a valuable whole retains exactly the same [final] value when it is, as when it is not, a part of that whole’ (1993: 81). However, it is not clear whether, while admitting that it is overall fitting to condemn someone being pleased by torture, anyone would be prepared to say that it is also fitting, at least to some extent, to respond favourably to the fact that they are pleased, e.g. by viewing the pleasure as such as improving things somewhat. In this instance, it seems, favouring pleasure is just not a fitting response at all. However, if so much is in fact granted by the organicist, then we have a normatively idle value: a putative positive value unrelated to positive fitting attitudes. This is an implication that even someone, like Moore, who rejects a fitting attitude analysis of value should view with suspicion. As we saw in Chapter 1, nobody can really deny the existence of a regular connection between value and fitting attitudes. Of course, the organicist might bite the bullet, and concede that even the sadist’s pleasure has some positive value and merits Note that we might agree that the appropriate attitude towards S outweighs the attitude towards (c). But the question is how they can be different. Also, things don’t change if we think of S as a complete world only composed of (a), (b), and (c). While S and (c) are in this sense notionally different, still whatever differences there might be they do not warrant being adverse to world S (since S has negative value) but being happy for (c), since (c) contains exactly the same things included in world S. 4

90

VALUE THEORY

some kind of positive response, albeit one that is outweighed by the overall negative response to the overall state of affairs ‘A takes pleasure in torture’. But this is not an attractive view. Or the organicist might reject this particular example, i.e. reject the suggestion that being pleased by something has some final positive value regardless of what pleases one. This can mean that being pleased, as such, has neither positive nor negative value, but is neutral (equals 0). This answer will not do. If something is evaluatively neutral, then intuitively it is fitting to be indifferent to it – neither favour nor disfavour it. But is there any room for indifference in a case like this? Indifference seems rather unfitting here. And if the organicist agrees on this, then we have again a kind of value (admittedly, a neutral one) which doesn’t tie up with fitting attitudes.5

5.4 Alternatives to organic unities: Virtual value These might not be knock-down objections to the doctrine of organic unities, yet they do motivate the search for alternative approaches.6 One such approach starts exactly from examples such as sadistic pleasure, and claims that states like being pleased, as such, are evaluatively inadequate: not good, bad, or even neutral. It follows that being pleased never features as a part whose final value has to be computed to arrive at a whole’s value, since it simply has no value. It also follows that no question of fitting attitudes arises. To be more precise: faced with an evaluatively inadequate state, the fitting response to it is to withhold favour, disfavour, or indifference, just like a movie one hasn’t seen yet. What we have, then, is not a bad whole with some good or neutral and some bad parts, See Stratton-Lake (2002), Olson (2004a), Dancy (2007) for versions of this objection. 6 In response to the attitude mismatch problem, the organicist might try to say that S is different from the whole (c) to the extent that the value of (c) results, non-additively, from a relation of appropriateness between (a) and (b), but the value of S doesn’t. So (c) merits a positive response even if S doesn’t. But not all organic unities need involve appropriateness between parts (see the value of Progress below). 5



The Chemistry of Value

91

but an undivided and uniformly bad state of affairs: being pleased by seeing torture. There is no room for favouring the state of being pleased, since this state as such has no actual final value. If it has no (actual) value, then a fortiori it is not a normatively idle value. Nor is there any necessity to postulate the value of a state of affairs as a whole as opposed to its value on the whole. So, no problem of unwarranted attitude mismatches can arise. But how can we explain the resultance of value from a state which essentially contains evaluatively inadequate elements? People being tortured is pretty bad, but that plus other people taking pleasure in that is even worse. How does this approach explain the difference clearly made by the sadists’ pleasure? Zimmerman (1999, 2001: Chapter 5) claims that being pleased has virtual value: if it is fitting to e.g. disfavour a state of affairs whose existence entails being pleased (e.g. the state ‘being pleased by torture’), then being pleased is virtually bad. By the same token, if the state ‘being pleased by others’ happiness’ is actually good, then being pleased (there) is virtually good. Virtual value is a derivative and therefore extrinsic form of value, since it depends on the actual value of the state of affairs the object takes part in.7 One problem, however, is that if pleasure only has virtual value, we cannot say that pleasure is what makes the evaluative difference between torture and sadistic torture. Any evaluative weight that pleasure carries is logically derived from, or owed to, the value of the larger state of affairs. In the order of explanation, actual value is prior to virtual value. So this approach has to take the difference between torture and sadistic torture somehow as a given. Sure enough, the virtualist can say that sadistic torture is worse because of the presence of pleasure (supervenience: evaluative differences require non-evaluative ones). However, the virtualist cannot say that pleasure acts as an additional bad-making feature. Pleasure somehow makes things worse without adding any negative weight of its own. In general, since virtual value presupposes the actual value of a state of affairs, virtual value cannot contribute to explaining actual value. 7 Zimmerman speaks of pleasure as having intrinsic virtual value: of course it is intrinsic to the extent that such value supervenes on the nature of pleasure. But since virtual value also depends on the actual value of the state of affairs in which it figures, on Moore’s definition of intrinsic value it should count as extrinsic.

92

VALUE THEORY

The virtualist approach is designed to avoid recourse to organic unities, while endorsing the main thrust of Moore’s philosophy: actual final value is intrinsic value, and as such doesn’t vary in kind or degree from context to context. But can the approach be generalized across all putative cases of organic unities? Here a second problem emerges. Where the value of intentional states is in question, the idea may seem plausible enough: sadistic pleasure is an intentional attitude towards something, and like for other attitudes (belief, desire, fear …), we cannot really pass an evaluative or epistemic judgement on them in abstraction from the object of the attitude. But now consider other alleged cases. For example, the claim that evaluative progress (for a life, a career, a relationship …) is better than evaluative decline, even if the relevant good and bad ‘stages’ have precisely the same degree of value: Progress: Going from B (bad) to G (good) is better than Decline: Going from G to B. The doctrine of organic unities, as we have seen, will compute the values of G and B for each state, and then place an extra value in the very wholes, so that Progress comes out better than Decline. While that move has its problems, the virtualist, on the other hand, seems forced to ascribe no actual value to the parts of either Progress or Decline.8 And we should withhold responses towards the parts, since these only have virtual value. But this amounts to rejecting the terms of the example. In Progress and Decline there must be things (states of affairs) with actual positive and negative values – otherwise how could Progress be actually better than Decline? Progress and Decline stand in an actual second-order evaluative relation (‘it is better that good follows bad than that bad follows good’), but as such they must be able to relate to actual first-order values. Moreover, suppose we define virtual value as follows: Necessarily, S is virtually intrinsically good [bad] to a certain degree if and only if, for some state S’ whose existence entails Zimmerman (2001: 148) indeed makes it an explicit clause that what has virtual value doesn’t have actual value. 8



The Chemistry of Value

93

that of S, the contemplation of S’ and S’ alone requires that one favour [disfavour] S’ to a corresponding degree. Even conceding for the sake of argument that G and B have merely virtual values, it turns out that, in Progress, B (by assumption something bad) is virtually intrinsically good, since presumably we are required to favour Progress. On the other hand, in Decline, G (by assumption something good) is virtually intrinsically bad, since presumably we are required to disfavour Decline. This is surely a surprising implication: whether actual or virtual, the values of G and B cannot switch polarity from one case to the other without changing the very terms of the example. In sum, when we move to computing the value of complex non-intentional states, the virtualist approach loses some of its appeal.

5.5 Alternatives to organic unities: Conditional value We have seen that organic unities are problematic for a number of reasons. We have also seen that organic unities are a natural consequence of the following theses: A) Actual final value is always and only dependent on

(supervenes on) the intrinsic nature of an object. B) (following from A) An object retains its actual final value across contexts and wholes in which it may appear. If virtualism does not work as an alternative, then it seems that the easy solution is to give up A, and B in turn. Indeed, in Chapter 2 we have seen some reason to doubt A. The alternative – let’s call it conditionalism – claims that: C) Actual final value can depend on (supervene on) an object’s

intrinsic and extrinsic properties alike.

D) (therefore) An object need not retain its actual final value

across contexts and wholes.

How does conditionalism explain the value of the two states of affairs we have been using as examples?

94

VALUE THEORY

S: A serial killer being jailed for his crimes. T: Taking pleasure in others’ pain. In both cases, one need not suppose that the parts or components which would or might be good (or bad) on their own, namely freedom and pleasure, carry over their value to these contexts (this is what D tells us). Therefore there is no need for an additional entity, overlapping with the states of affairs, whose value as a whole is able to make the overall values of S and T (‘on the whole’) turn out different from the sum of the values of their parts. Regarding S, remember that we want to be able to say that punished crime is, as such, better than unpunished crime. The conditionalist can arrive at this claim by using the distinctions sketched in section 2. The fact of desert acts here as a condition: it changes the value-making contributions of punishment and freedom. For instance, we can say that in the serial killer’s case desert not only disables the intrinsic evil of being punished and the intrinsic positive value of freedom, but also enables punishment to switch from bad to good, or freedom from good to bad. So punished crime is better than unpunished crime because the former contains one bad feature (crime) and one feature that is neutral or even good (being punished), while the latter contains the same bad feature (crime) and one feature that is bad or neutral (freedom): Crime Punished>Crime Unpunished: Bad+Neutral>Bad+Bad, or Bad+Good>Bad+Neutral. In these explanations, we can see that the kind (polarity) or the degree of the values of punishment and freedom are conditional on an extrinsic factor, namely desert. Regarding T, we can say that pleasure, when enjoyed by sadists, is simply and finally bad. We can explain this because sadism disables pleasure to be good-making, and enables it to be a bad-making feature in this context.9 Note that this is not the same This is better than: ‘sadism is a condition that disables pleasure from providing positive value’. In our example, we want sadist pleasure to be evil and not just not good. 9



The Chemistry of Value

95

as saying that pleasure as such is an evaluatively inadequate state, but rather that the degree and kind of value it does actually take may be conditional on its object, and thus cannot be assumed to be good or bad or neutral across all contexts. Because of this different analysis, the problems of attitude mismatch and normatively idle values do not arise. We can simply be happy, or at least satisfied, for S on the whole without any competing (in kind or degree) fitting attitudes towards the whole as a whole. And we can agree that pleasure, in the case of the sadist, has no idle positive value, but indeed has a non-idle negative value, and accordingly we can have an unmixed fitting attitude of condemnation towards it. However, the conditionalist analysis invites two objections. The first we can call the arbitrariness objection. Consider retributive punishment again. On what grounds do we decide that desert makes punishment a good thing (or neutral, or less bad), rather than making a bit less bad the evil caused by the punished criminal, and worse the evil caused by the unpunished criminal? In this case, we probably rely on an intuitive sense that the degree of evil caused by the crime just cannot be improved or made worse by anything external to it, so we are led to focus on the other component (the punishment). But in other cases, the choice seems more arbitrary. Take sadistic pleasure. Why does sadism affect the value of pleasure, rather than making worse the fact of people being tortured? Put in another way, why does sadism disable the good of pleasure, rather than intensifying the evil of people being tortured? Or take a case of disproportionate attitudes. An extremely selfish person is disproportionately concerned for herself more than for others. But which of the two concerns does the lack of proportion affect, turning it bad from good or at least from neutral? Is great concern for oneself a bad thing, when accompanied by little concern for others, or is it the other way round? The vice of selfishness, it seems, lies precisely in the combination, or the whole, rather than in the degree of intensity of either concern (Hurka 1998).10 While the conditionalist might find more or less The difference between Progress and Decline, likewise, is subject to the same arbitrariness problem. Is a good somehow better than an otherwise equal good, because it follows a past evil? Is an evil somehow worse than an otherwise equal evil, because it follows a good? 10

96

VALUE THEORY

satisfying answers to each case, adopting the doctrine of organic unities has an apparent advantage in that it simply does not require an answer to such ‘location’ issues. Any value to be explained can be placed in the whole rather than in one of its parts. In this sense, organic unities treat their parts symmetrically, whereas conditionalism needs to justify its asymmetric treatment of parts. A second problem with conditionalism can be called the adjusted attitude objection. Consider yet again retributive punishment. If desert makes the deprivations suffered by the criminal a (finally) good thing, then the appropriate response to them should be pure happiness (for its own sake). But, as Hurka points out, presumably even for fans of retributivism ‘retribution involves or should involve a distinctively sombre or subdued emotional tone, one suffused with regret’ (1998: 310). Organic unities go some way towards justifying such reaction, since the evil of punishment has not been cancelled or reversed by desert. While the overall fitting response is one of satisfaction, there is room for the sombre tone to which Hurka alludes, given that the evil of punishment is still one of the addends. But on the conditionalist analysis, there seems to be no justification for adjusting one’s feelings based on what the criminal is suffering. Analogously, consider the good of compassion, understood as being pained at someone else’s (undeserved) pain. The organicist makes sense of why, while admiring compassion, we may think twice before getting people to be compassionate: after all, while the whole has positive value, pain felt by the compassionate preserves its intrinsic negative value as pain here as elsewhere. But the conditionalist will locate only positive value in such pain (Olson 2004a: 40) – the final value of compassion must after all result from something good in it. Let’s consider briefly two possible replies to this second objection. One is that the conditionalist need not take retribution to be good, in order to explain why it is better as such than absence of retribution. This easily explains the sombre tone to be adopted: even on a conditionalist analysis, the state of affairs is a bad one. And if one were to insist that retribution is good (and not just better as such than its absence), then it is less clear why we should adjust our feelings based on what the criminal suffers – of course, since the state of affairs implies the evils of crime, one should not be too cheerful about it.



The Chemistry of Value

97

This reply, however, may not work for the case of compassion, where one wants to say that being compassionate is a good character trait. Why think twice before encouraging something good? Naturally there may be all sorts of contingent reasons: e.g. compassion may make some people very weak or vulnerable. But the objection is that there is something intrinsic to compassion which justifies such hesitation, namely the pain in which compassion consists. Here the conditionalist can allow that the pain felt makes compassion somewhat bad, only, not finally bad (bad for its own sake) but personally bad: bad, to some extent, for the subject who feels it. The overall appropriate attitude to a compassionate person, particularly for those who are supposed to care about what is good and bad for her, will take this into account, by moderating admiration (based on the positive final value of compassion) with a cautionary attitude (based on its partly negative personal value) (see Chapter 4). A similar explanation might also be applied to retribution. Even when retribution is held to be finally good, the criminal’s loss of freedom can be personally bad, i.e. bad for him. This implies that the criminal, and those who are specially related to him, have some reason to disfavour (e.g. be saddened by) the punishment. A sombre emotional tone by a bystander would in turn be justified not by the final disvalue of the criminal’s loss of freedom, but as a way of respecting the sadness appropriately felt by friends and relatives of the criminal.11 Such explanations have a theoretical burden, in that they require introducing a notion of personal value.12 And we have not found a general solution to the arbitrariness problem.13 The virtualist analysis also has its limitations. It remains to be seen whether organicists can do any better, saddled as they are with the issues of attitude mismatch and normatively idle value, plus general doubts

Even if the criminal has no-one left to care for him, adopting a sombre tone can be an appropriate reaction to an imagined group of such people. 12 But see Olson (2004a) for a different conditionalist analysis of such examples. 13 Dancy (2007: 341) suggests that it is both true that punishment is less bad, in the context of crime, and that crime is less bad, in the context of punishment. Both aspects have their disvalues decreased, so that punished crime is better than unpunished crime (and of course better than punishment for no crime). 11

98

VALUE THEORY

about the Moorean view of final value, which is what motivated the doctrine of organic unities in the first place.

5.6 Holism and particularism Remember the claim made by the conditionalist: D) An object need not retain its actual final value across

contexts and wholes.

This claim allows for contextual determination of both the degree and the kind or polarity of a certain value. As such, it stands against the claim (B) that an object retains its kind and degree of final value across all contexts or wholes, which has been variously labelled as the thesis of universality (Lemos 1994) or invariability (Zimmerman 2001). However, (D) as such doesn’t make value a radically ‘local’ (i.e. contextual) or variable affair. A conditionalist might say that while certain things change their final value from one context to another, other things never do, and these things might well be wholes. For instance, sadistic torture might be always finally evil, regardless of context. Or the virtues, even if ‘built from’ objects of conditional or variable value (pain, fear, or other attitudes), might always be finally valuable. The conditionalist may go on to produce a coherent list of such invariable values, and thus offer a pluralist axiology. Or they may even claim that only one sort of thing has final value, although it doesn’t always have it, because its value can be disabled in certain conditions, or doesn’t always have it to the same degree, because of intensifying or diminishing conditions. This is what a conditionalist hedonist might say: only pleasures have, i.e. ground, final value (and pains disvalue), but not all pleasures are valuable or valuable to the same degree. For example, malicious pleasures are not good, and some social or intellectual pleasures are better than some bodily pleasures (a claim that could turn out crucial for J. S. Mill’s ‘qualitative hedonism’, see Olson 2004a: 48). Conditionalism appears tantamount to what Dancy defines as holism in the theory of value: a feature or part may have one value in one context and a different or opposite value in another (Dancy 2005: 333). And Moore’s thesis (B) is equivalent to what Dancy calls atomism: what is a (final) value in one context is the



The Chemistry of Value

99

same (final) value in all contexts.14 However, Dancy does think of value as essentially a variable affair, and in analogy with his theory of reasons, argues from holism to particularism in the theory of value, which would sound like this: the possibility of evaluative judgement does not depend on a suitable provision of evaluative principles (Dancy 2004a: 7). Value particularism stands opposed to value generalism, the view that evaluative judgement cannot do without a suitable provision of principles, i.e. invariant evaluative truths, as exemplified for instance by pluralism and conditional hedonism above. More precisely, particularism doesn’t deny that there might be invariant values, only their existence is not required for evaluative thought, and so we shouldn’t expect to find any. It is indeed a chief question in the debate over particularism whether, once Dancy’s holism is accepted, there is any room or motivation left for a principled picture of a pluralist or monist sort.15 Dancy’s particularism grants the existence of what he calls ‘default values’ alongside purely contextual values: some features [e.g. ‘causing needless pain’] come switched on already, as it were, though they can be switched off by other features; others do not come switched on, but they can be switched on by a suitable context. (2004a: 185) As a matter of fact, some default values might be invariant across contexts. But, as values, they always susceptible to variation, though any such variation in value would require explanation. Default values thus cannot form a fixed list of final values, as in the pluralist view sketched above. What about conditional hedonism? It might seem that pleasure behaves precisely as a default value there. However, important differences remain between a conditionalist axiology and particularism. First, if the idea is to provide suitable guidance for evaluative judgement, we would expect conditional hedonism to keep variability under control, namely by specifying in broad terms the sorts of conditions that can ‘switch off’ (or increase and decrease) the value of pleasure. For particularists, contextual Of course this is a stipulated meaning. Elsewhere ‘holism’ is associated with the idea that the nature of an entity or phenomenon cannot be reduced to its individual parts – in this sense Moore’s organic unities are an holistic approach in the theory of value. 15 See Dancy (2004a), McKeever and Ridge (2006). 14

100

VALUE THEORY

variability cannot, and anyway doesn’t have to, be controlled in such ways. Second, even if conditional hedonism adopted an open-ended conception of such conditions, it could not accept the idea of purely contextual values: simply, context cannot give rise to other final values besides states of pleasure. A monistic (or for that matter pluralistic) structure, while compatible with Dancy’s holism, must stop short of embracing the full range of possibilities opened up by holism.

5.7 Summary Understanding the chemistry of value requires understanding the many ways in which value is explained by and related to its supervenience base. Moreover, when it comes to complex objects of value, there are competing accounts of how their value depends on the values of the components or parts. I have presented three approaches (organicism, virtualism, conditionalism), and noted how they all suffer from difficulties, notably (but not exclusively) having to do with how it is fitting to respond to such complex states of affairs. Still, conditionalism might seem a better approach, to the extent that it stems from a less rigid picture of intrinsic and final value than the other approaches assume.

CHAPTER SIX

Value Relations 6.1 Introduction In the previous chapter we explored different views about how the value of a complex state of affairs results from the values of its parts. Among other things we saw some reasons to prefer an account which doesn’t assume the value of objects to be a fixed, permanent, or invariant matter, even when it comes to their final or intrinsic value. In this chapter, I examine another phenomenon which appears to be at odds with a tidy picture of value: the idea that some values are incomparable or incommensurable with others. I will present arguments from Raz and Anderson meant to establish that value comparisons are not always possible, and then discuss arguments from Regan and Chang meant to defend value comparability. This will hopefully clarify at least one of the ways in which values are possibly related to each other.1

6.2 The trichotomy thesis and incomparability It is natural to assume that, given two objects or states of affairs X and Y, if we can compare them at all, then we can always (in In a longer work, I would also address salient properties of the betterness relation, such as transitivity (if x is better than y, and y is better than z, then x is better than z), and the concept of lexical superiority (x is lexically superior to y when any amount of x is better than any amount of y). 1

102

VALUE THEORY

principle, if not in practice) rank them against each other in one of these three ways: X is better than, worse than, or as good as Y. For example, other things being equal, an hour of innocent pleasure is better than an hour of undeserved pain; unpunished crime is worse than punished crime; one printed copy of Van Gogh’s Sunflowers is as good as another exactly identical printed copy. This has been called the trichotomy thesis (Chang 1997b: 4). But sometimes we seem to be less confident making such comparisons. Joseph Raz gives as an example the choice between a career as a lawyer and a career as a clarinet player (1986: 342). Neither career seems better than the other, all considered. Are they therefore equally good? If they were, then a slight improvement in either career would make choosing that career better than the other. So, suppose you will earn $1,000 more as a lawyer. Raz suggests that this slight improvement is not sufficient to make the lawyer’s career better than the clarinet player’s. But this is how a tie between equally good alternatives can normally be broken: e.g. the fact that a dress at shop A costs even a little bit less than the same style of dress at shop B makes it better to choose shop A, other things being equal. So, he concludes, the two careers are not even equally good.2 If the trichotomy thesis is true, then they are incomparable in value: no positive ranking or comparison between the two is possible. Note the structure of the problem: the two items are incomparable under a certain aspect, i.e. their value, or more precisely their value as careers. The two careers can of course be compared under other aspects: maybe the clarinettist’s career is longer than the lawyer’s career, possibly the lawyer’s career allows more leisure time than the clarinettist’s, etc. So they need not be incomparable under every aspect (are any two things as a matter of fact totally incomparable?). But, as we know by now, value supervenes on non-value. So a full survey of the features which make either career valuable should also bring out the reasons why the two are incomparable in value. There are two possibilities here. One is that, in fact, the two careers are comparable under every non-evaluative aspect, but there is no way of settling which of these comparisons matter more,

2

See also Chang (2002).



Value Relations

103

nor whether they equal each other out. The idea is that the same values, or rather the same value-making features are present in both cases, albeit to different degrees in each. As a clarinet player, you will earn enough money to live on, but not as much as you would as a lawyer. As a lawyer, you will cultivate your intellectual skills, but not as well as you would as a musician. And so on. In this case, each value-making feature is shared, but there is no overarching common measure or method to balance their contributions against each other. We can call this overall incommensurability. The other possibility, perhaps more realistic for this sort of case, is that each career contains some value-making feature that the other simply lacks. Perhaps the kind of aesthetic sensibility you develop as a clarinettist finds no match in a lawyer’s career, and the same can be said mutatis mutandis for the dialectical skills required for a lawyer. Further, it is not clear that the one value-making feature contributes more value than the other. Nor is it plausible to say they contribute value in equal measure: aesthetic sensibility and dialectical skills are just very different abilities. So, in this case, the items are incomparable because they are partly made valuable by distinct features whose values are, at least as far as they go, incomparable.3 This we can call pro tanto incommensurability.4 So incomparability between two objects (choices, states of affairs, etc.) can happen either because there is no way to overall measure the importance of their non-evaluative features, even if shared to some extent, or because their non-evaluative features have not enough in common for their evaluative contributions to be compared with each other. What if there were only one fundamental value, e.g. pleasure? The lawyer’s career and the clarinet player’s career would have This, in turn, might be a feature of the specific case, or the result of a constitutive incomparability between these values. The latter case is sometimes illustrated by saying that e.g. friendship cannot be compared with any amount of money, or it ceases to have the value it has. If any trade-off between them takes place, it won’t be justified by their relative values (see Raz 1986, and Chang 2001 for criticism). Aesthetic sensibility and dialectical skills are different values, but presumably not constitutively incomparable: ceteris paribus, to possess great dialectical skills is better than to possess poor aesthetic sensibility, and vice versa. 4 Of course the items must have enough in common to be at least candidates for comparability. For example, a legal career and the number five are not incomparable, but rather non-comparable. 3

104

VALUE THEORY

a common value-making feature: the pleasant experiences they contain (ignoring for argument’s sake the pleasure of other people involved). This would rule out what I call pro tanto incommensurability, since both careers would contain the one value-making feature. But it is interesting to note that, by itself, such a monistic view wouldn’t rule out overall incommensurability. For suppose the clarinet player’s career contains less intense pleasures, but longer lasting ones than the lawyer’s. How do we balance intensity and duration of pleasure against each other? There may be no way to do this for each case, nor is there any pressure to conclude that such careers must therefore be equally good. The availability of a common measure, in other words, doesn’t guarantee comparability between any two items. And on the other hand, a pluralist view, whereby there is more than one fundamental value (e.g. pleasure and virtue), makes pro tanto and overall incommensurability conceptually possible, but not inevitable. Spending money on a morally good cause might be better than spending it for one’s entertainment, even if these choices concern completely different values, with nothing in common. In other words, absence of a common measure doesn’t make value comparisons impossible.5

6.3 A fitting attitude argument for incomparability Apart from an appeal to intuition in particular cases, can a general argument from the nature of value be made for the possibility of incomparability? So argues Elizabeth Anderson. She starts out from a premise that fitting attitude minded philosophers should agree on: ‘there is nothing else for a value to be but something that guides our deliberations and attitudes in practice’ (1997: 91). To speak of options whose values are comparable or incomparable is thus to speak of the attitudes one should take towards them. If two Pluralism might be even consistent with what Richardson calls ‘weak commensurability’ (1994: 105): the thesis that in any given conflict of values, there is a true ranking of the realization of one value against the realization of the other value in terms of some value, which may be one of the values in question.

5



Value Relations

105

options are comparable in value, then it makes sense (or it is fitting) to express attitudes of a comparative sort towards them: desire (admire, etc.) the one more than the other, or favour them equally. And conversely, if it makes no sense to express such attitudes towards any two given options, then the goods are incomparable. Anderson calls this the ‘no good reasons principle’ for incomparability (ibid.: 99). She argues that the principle generates many plausible cases of incomparability, ranging from aesthetic cases involving crosscategory items (e.g. novels and sculptures) to tragic dilemmas involving choices between two loved people. In all these examples, it seems, it is positively unfitting to hold comparative attitudes. The project of universal aesthetic cross-category comparison can only be for ‘philistines, snobs, and prigs’ (ibid.: 99), that is, for people whose sense of aesthetic appreciation is distorted by various factors. The proper development of an aesthetic sensibility is incoherent with the notion of a universal art ranking, and the attitudes that would go with it. In the case of a value comparison between loved ones, it would need to be true both (i) that there is one single common attitude to be expressed e.g. both towards one’s mother and towards one’s friend, and (ii) that there is a possible answer to the question ‘how intense such an attitude does each deserve?’. But these assumptions are not always justified: (i) different people deserve different sorts of attitudes (e.g. filial love and friendship), (ii) even when two people deserve the same sort of attitude, e.g. respect, the notion of ‘deserving respect of intensity n’ may not (always) be applicable (ibid.: 103). A decision between incomparable options might still be possible, but it will have to be grounded on something else than their comparative values. Moreover (although Anderson doesn’t mention this possibility), when decision is not at stake, it might be fitting (for me as well as for others) to refuse to compare one’s actual career as a musician to a merely potential, alternative career as a lawyer. Anderson’s argument has a certain force, when reconstructed this way.6 Note, however, that it cannot simply depend on the ‘no I’m interpreting her claims charitably, since part of her argument is also that it would be pointless (rather than unfitting) to e.g. make cross-category aesthetic comparisons. But the fact that it would be pointless to compare them is arguably a wrong kind of reason to show that the options are incomparable (see Chapter 8). 6

106

VALUE THEORY

good reasons’ principle. That principle should be acceptable also to a philosopher who denies incomparability, because it expresses a general truism about the relation between value concepts (including comparability) and fitting or appropriate attitudes. What her argument depends on is a certain substantive interpretation of this truism. In particular, she denies that there is always some single kind of attitude fittingly directed to different valuable objects or people. This is the crucial point: there is good reason to compare two items only if there is good reason to hold the same sort of attitude towards each, with a different or identical intensity or polarity as the case may be. If x and y are comparable, then it must be fitting to have e.g. a desire-like attitude towards each – and then, towards either, a stronger, weaker, positive, negative, or equal desire-like attitude, as the case may be. Now, defenders of FA might well agree that different kinds of responses are in general fitting towards different objects of value (see Chapter 7), but in this connection Anderson needs a stronger claim, namely this: similar kinds of responses are unfitting towards different objects of value; at least in certain cases, that is, similar responses are ruled out. For instance, Anderson implies that it is unfitting to desire one’s mother’s staying alive more than to desire to save a friendship, and vice versa, or to desire these things equally. Even if it might turn out that we ought to save our mother, this would not be because saving our mother is what we ought to desire the most. A different FA theorist might instead believe that, even if different objects do merit different responses (filial love, friendly loyalty, etc.), comparison might at least in principle always have a point, to the extent that even very different objects also merit similar kinds of responses, such as desire or preference.7 This is not meant per se as an objection to Anderson’s view, but it points out the sort of argumentative burden she has to face.

Interestingly, if FA adopted desire as a master fitting attitude (as presumably Sidgwick did, and Oddie (2005) has recently done), then FA would not just allow but in fact imply value comparability, at least in principle: any two valuable states of affairs would be comparable, to the extent that each deserves to be desired with a desire of a given intensity. See Kelly (2008), and Klocksiem (2011) for criticism. 7



Value Relations

107

6.4 Against incomparability: Epistemic limitations Donald Regan defends both the trichotomy thesis, and what he calls the ‘complete comparability’ view. In one single formulation: given two items sufficiently specified as far as their value is concerned, it will be true that either is better than or as good as the other (1997: 129). On this view, then, putative instances of incomparable values are really cases where one of the three relations holds, but we do not know which one. Regan, like Anderson, starts out from a general view about the relation of value and attitudes. Regan’s central claim, at least for one of his arguments, is that ‘[v]alues, properly apprehended, motivate’ (ibid.: 141), and motivate to the appropriate degree. So, consider the choice between the clarinettist’s and the lawyer’s career: since both are valuable to some degree, an ideal agent, who properly apprehends values, would be motivated by each to some degree. But then the ideal agent’s choice according to the stronger motivation will thereby reveal which one is better, i.e. how they compare (if there is a motivational tie, i.e. the ideal agent has equally strong motivations, then they are equally good). In other words, when we consider any putative case of incomparability, we should not assume that we know that neither option is better than the other: as fallible and less than ideal agents, we might have failed to properly apprehend the relevant values. There are two weak points in Regan’s argument. The first is that the limitation attributed to actual agents cannot be simply epistemic or propositional, i.e. such that we do not know whether a career is better than the other. Our predicament, as should be clear from the explanation in terms of pro tanto and overall incommensurability, is rather such that we don’t know how to balance or weigh the pros and cons of the two careers against one another. This is a practical limitation. But if this is so, then we cannot rule out, as a candidate explanation of our predicament, that there is no way to balance them: there might be no way to ‘operate the weighing mechanism’ in this case, because no such mechanism is available. To use another metaphor: if I don’t know how to ride a certain bike, it might simply be that the bike is not fit for riding.

108

VALUE THEORY

Regan might at this point retort that we don’t know how to compare, not because no comparison is possible, but because we are not ideal agents. This brings us to the second worry. Regan assumes that the attitudes and dispositions of an ideal agent are always comparative in kind. Faced with any valuable state of affairs, an ideal agent is essentially willing (and of course able) to assign it a position in a ranking of options. But this is a substantive assumption, which doesn’t simply follow from the idea that ‘values, properly apprehended, motivate’. To be sure, Regan is in no worse dialectical position than Anderson on this score. However, his assumption does follow from a further claim he makes early in the article: ‘the proper ultimate aim of practical reason is to produce the best state of affairs possible’ (ibid.: 131). If the ideal agent is the (ideal) personification of ‘practical reason’, then her ultimate aim is to produce the best state of affairs, and this aim requires taking an essentially comparative attitude towards options. However, now that the source of Regan’s assumption has been exposed, it is clear that it needs further defence on two counts: (1) the ultimate aim might be producing not the best, but the sufficiently good;8 (2) even if ‘the best’ is a legitimate ultimate aim, it might not be a mandatory aim – Regan seems to think that values always impose what have been called peremptory reasons, such that we ought to act on them or respond in a certain way to them.9 But many have suggested alternative roles or sorts of reasons: some reasons are enticing rather than peremptory (Dancy 2004b), they make options eligible rather than required (Raz 1999), or they justify rather than require action (Gert 2004a). It remains an open question whether an ideal sensitivity to reasons means that an agent ought to aim at producing the best state of affairs, and in turn ought to hold the essentially comparative attitude required for that aim. Regan might be right, but more argument is needed to establish complete comparability.

See e.g. the debate on ‘satisficing’ as an alternative to maximizing: Slote (1989), Byron (2004). 9 For example, talking about reasons for wanting, he says: ‘wants that are to be rendered intelligible by their dependence on reasons are not optional. An agent ought to be moved by the perception of a reason’ (1997: 147). It is perhaps noteworthy that he stresses this point while discussing a choice between pears and bananas. 8



Value Relations

109

6.5 Against incomparability: Parity Even if Regan is wrong, still one should not therefore jump to the conclusion that the legal career and the musical career are incomparable. Ruth Chang has forcefully argued for the existence of a fourth comparative relation besides being better than, worse than, and equally good: being on a par with. Chang agrees with the Razian argument that, if a small improvement on the legal career doesn’t make it better than the musical career, then the two careers are not equally good (nor is the one better than the other). But she argues that a further possibility needs to be considered before declaring them incomparable. So she defends comparability precisely by denying the trichotomy thesis. Her ‘chaining argument’ starts from the idea that some comparisons can be made even between very diverse items: in her example, Talentlessi, a very bad sculptor, is clearly worse (with respect to a relevant value such as creativity) than Mozart, a great composer. Then she appeals to the principle that small unidimensional differences cannot change two comparable items into incomparable ones. So a slightly better sculptor than Talentlessi is still comparable to Mozart. Apply this in turn to ever more creative sculptors, and you will be forced to hold that Michelangelo is comparable to Mozart, despite their diversity. If neither is better than the other, nor are they equally good, then we need a fourth comparative term to describe their relation (Chang 2002: 674–5). There are two worries about Chang’s position. First, the structure of Chang’s chaining argument looks suspiciously like the famous sorites paradox. Consider John, who has a head full of hair, and Jack, who has only one hair. John is hairier than Jack: this is the relation they are in. Now, add one more hair to each successive Jack. If the principle of small unidimensional differences is true, then the relation between the two items would be preserved through any small unidimensional change in one of the two. It follows that John must be hairier than even a very hairy Jack – who has the same number of hairs as John. But this is absurd. The sorites paradox is supposed to show that certain predicates (‘being hairy’, ‘being bald’) have no precise or determinate boundaries: they are vague predicates. By the same token, one could say that the predicate ‘being comparable to’ is vague as well. Talentlessi

110

VALUE THEORY

is worse than Mozart, but it might still be indeterminate whether and how Michelangelo is comparable to Mozart. There might be no fact of the matter about that: it is neither true nor false that one is better than the other or as good as the other. The chaining argument thus seems to motivate the view that comparative terms are vague, rather than showing the need for a fourth comparative relation.10 Chang replies to this worry by offering a general reason why comparative terms such as ‘better than’ cannot (always) be treated as vague predicates (ibid: 682–6). The comparison between Mozart and Michelangelo is a hard one to make. If comparative terms were vague terms, this comparison would be analogous to the decision over a borderline case of, e.g., being bald. Chang points out that we proceed in very different ways in these two cases. In cases of borderline application of vague terms, we can ‘resolve’ a dispute as to whether Jack is bald essentially by flipping a coin. This is in effect tantamount to stipulating a new non-vague predicate (‘being bald*’) and applying it to Jack. But, Chang observes, we cannot do this in the case of value comparisons between Mozart and Michelangelo: there seems to be a substantive disagreement which cannot be resolved by flipping a coin, nor (equivalently) by stipulating a new term of comparison (‘being better than*’) and applying it to either. This would be to change the topic of our disagreement.

6.6 Parity and choice So, despite having a sorites-argument form, Chang’s chaining argument does not require that ‘comparable’ be a vague predicate, and can support the idea that we need a fourth comparative relation. A second worry about her argument, however, is that she has not done enough to support the existence of a new fourth comparative relation as basic as being better, worse, and equally good. Joshua Gert (2004b) has articulated this worry. He starts from an assumption which readers of this book will find congenial (and which Chang has no reason to reject either): talk of values See Broome (1997) for such a view. Space prevents me from discussing it properly.

10



Value Relations

111

is or is closely tied to talk about justified (fitting, appropriate, etc.) attitudes. In particular, for Gert, talk of value comparisons is closely tied to talk of permissible choice: A is worse than B when choice of A over B is rationally impermissible. So, given what Chang has told us, parity can in principle be defined as follows: A and B are on a par if and only if ‘it is not a mistake, or irrational, to choose A over B, or B over A, and ... this may continue to be true even if one of the items is slightly improved’ (2004b: 506). Given the value/rational choice link, this means that A and B are on a par when neither is better than the other, nor are they equally good (remember that if they were equally good, any slight improvement on either would make it better than the other). However, Gert’s point is not just that parity is definable in terms of other value relations. Rather, the category identified by As and Bs that are neither better than one another nor equally good, and yet comparable, is not well described by the relation of parity. The cases at issue can be represented as involving overlapping ranges of strength in rationally permissible preferences (or choices), e.g.: 1 A, B: 8–12, 8–12 (= we can have a preference for A to any

degree between 8 and 12, no less and no more, and the same for B). 2 C, D: 9–10, 9–10. 3 E, F: 10–12, 10–20. 4 G, H: 5–10, 9–15.

In (1), the ranges are perfectly overlapping, and this for Gert we can call a case of parity.11 In (2) the ranges are narrower and A and B are not equally good for Gert, presumably because he thinks a slight improvement in A will increase A’s range, say up to 9–13, without thereby making A and B disjoint, i.e. without thereby making A better than B (e.g. that would happen only when A’s lower range point goes up to 13). But Chang rejects Gert’s interval model (2005: 340–1). First, she thinks (1) and (2) are cases of equality, not parity. Second, in order for A to be even slightly improved (to become A+), we should increase the range uniformly so that A+’s range is disjoint from A’s range (since that’s what it is to be better, according to Gert himself, and A+ is supposed to be better than A). But then any slight but real improvement on A will make A+ better than B. So Gert’s model (and indeed, for Chang, any interval model) cannot really capture the idea of two items that are not equally good. See Rabinowicz (2008) for a different modelling based on permissible preference orderings. 11

112

VALUE THEORY

approximate one unique value (a one-point range). Since a unique value would mean equality, Gert calls this case ‘rough equality’.12 In (3) and (4) the permissible ranges are quite markedly different: F has a much higher top bound than E; G and H are almost disjointed. So, even if still partially overlapping, the relative items cannot be said to be on a par. The moral is that Chang’s parity covers too many different cases in which it is true that ‘it is not a mistake to choose A over B, or B over A, and this may continue to be true even if one of the items is slightly improved’. Parity is, at most, just one possibility in a class of cases that can be described using the usual comparative relations. Chang (2005: 344–7) has replied to Gert’s objection by defending parity as involving distinctive constraints on the sense in which it is permissible to prefer A to B and vice versa. Also equally good items are such that it is OK to prefer either (at least in the choice-related sense of preference: otherwise, equally good items are such that one should be equanimous towards them, i.e. not prefer one to the other). What makes parity different, however, is that over a series a choices among items on a par we might end up with less value than we could have had. So it’s not always permissible to choose either of two items on a par. This is due to the ‘immunity’ to slight improvement which is characteristic of parity. Suppose A and B are on a par: A is a decent cup of espresso, B is a good cup of tea. A+ is by definition an improvement on A (a slightly better cup of espresso), but still is on a par with B. The first choice is between A+ and B: they are on a par, so we can choose either. Suppose we choose B. The second choice is between B and A: even if the items are again on a par, now it would be a mistake to choose A over B, because we would end up with less value than we could have had, had we chosen A+ at first. We would end up with a worse cup of espresso than we could have had. In this sense, the first choice constrains the second, even if both choices are between items on a par. No such constraints apply to equality, since serial choices between equally good items are guaranteed to result in identical amounts of value.

For this notion, see Griffin (1986), Parfit (1984: 431).

12



Value Relations

113

6.7 Parity and incomparability Chang’s defence of parity stands against the trichotomy thesis: if items are comparable, then there will be four and not only three possible ways of comparing them. But it is also meant to remove ground for the thesis that incomparability is possible and indeed inevitable. The idea is that, before declaring two items incomparable, we now need to make sure that they are not on a par. And this might prove rather difficult, since Raz’s argument from slight improvements was shown to be compatible with the possibility of parity. Moreover, the choice among incomparable items seems to be subject to the same constraints as the choice among items on a par that we have seen just seen. If incomparable items are such that it is always OK to prefer either, then over a series of choices we would get the same undesirable results as with items on a par. A pattern of unconstrained successive choices between incomparable items, distributed across a lifetime, might land us with a life barely worth living, that is, a life from which value has been ‘pumped away’ by each subsequent choice, and justifiably so, because at each choice situation we were permitted to choose either item (Chang 1997b: 11). However, by way of defending the possibility of incomparability against the ‘comparativist’ alternative offered by parity, two things should be kept in mind. First, if parity holds, then it is not clear what it means to compare the value of items on a par. On top of ascertaining the values of each item, comparing involves an attitude (or an act) of relating their values to each other in such a way as would yield some novel fact about them. When we compare the weights of two objects, we get to know which one weighs more. When we compare the ages of two people, we get to know who is older. This is not only true of numerical comparisons: when we compare the taste of two chocolate bars, we get to know which one is sweeter. And so on. However, when we compare items that are on a par, we don’t get to know any such novel fact, except what we already knew, e.g. that Mozart is great and Michelangelo is great. Of course, we do get to know something negative about them: that neither is better than or as good as the other. But it is not clear why we should interpret this negative finding as a successful comparison rather than as a failure of comparison.

114

VALUE THEORY

Second, even if we can make some sense of the notion of comparing items on a par, there is still plenty of room left for incomparability. The remark by Chang above, about series of permissible choices resulting in value being ‘pumped away’, is indeed evidence that incomparable items, like those on a par, are not such that it is always OK to choose either. But in itself it doesn’t show that the only way around ‘value pumps’ is to impose constraints on permissible choices among incomparables, as it is for items on a par. Another reaction might be to simply deny that the idea of incomparability is tied at all to that of permissible choice. Anderson’s cases above pointed to a rather different understanding: two items are incomparable when it is unfitting to compare them – when there is sufficient reason against comparing them. In a similar vein, Rabinowicz suggests that incomparable items are such that the absence of a disposition to choose either is required (2008: 25–7).13 Parity, on the other hand, even when subject to the necessary constraints to avoid ‘value pumps’, is essentially related to the idea that it is OK to prefer either of two items on a par, and thus that it is not unfitting to hold a comparative attitude towards them. These definitions leave it open, of course, that other ‘non-comparative’ attitudes or dispositions might still be fitting towards either item, and thus that we might be able to make a justified decision among incomparables, particularly if and where deciding is indeed better than not deciding at all. This is clearest in the career choice. Even if neither career is better as a career, we can still legitimately embrace either on the basis of considerations that do not pertain to their comparative values, that is, that do not show the two careers to be positively related: for example, that one career is not worse than the other could be such a justifying consideration (see Hsieh 2007). In some range of cases, we can even flip a coin or let non-rational motives decide, provided that our decision does not reflect an evaluative comparison or an intention to choose one option over another. Or as Rabinowicz says: ‘We do make a choice if we have to, but this choice is made without the conflict of reasons being resolved’ (2008: 26).

According to Chang herself, when it comes to incomparable items, the ‘choice function’ is simply undefined (Chang 2005: 346–7). 13



Value Relations

115

6.8 Summary In this chapter, I have focused on one value relation, besides the usual ones of being better/worse than, and as good as, namely the negative one of two values or value instances being incomparable. I have considered arguments by Raz and Anderson in favour of the possibility of value incomparability, and two different rejoinders from the ‘comparativist’ camp, one defending the traditional trichotomy (Regan), the other pointing to a fourth possibility of comparison (Chang). In all these cases, the relation of value to fitting attitudes helped to shed light on the issues, or even provide considerations in favour of either side.

CHAPTER SEVEN

How Do I Favour Thee? 7.1 Introduction The fitting attitude approach to value appeals to the idea of favouring valuable things for their own sake or for someone’s sake. But what exactly does this mean? Here I try to articulate somewhat this concept along different dimensions (‘kind’, ‘telos’, and ‘degree’ of favouring), while bearing in mind that a full account of fitting attitudes can be given only simultaneously with a substantive value theory or axiology, which is not our present aim. But I will critically discuss a substantive monistic proposal, namely the idea that promotion or maximization must be the fundamental fitting response to value. In the end I consider a thesis which, if true, would undermine the value-theoretic debate about the nature of fitting attitudes, as well as in other areas, as a case of philosophers talking past each other.

7.2 Three dimensions of favouring If value and fitting attitudes are in broad correspondence with each other, as we have been assuming so far, then the full story about particular values or valuable states of affairs will require a specification of the relevant fitting attitudes. So far we have been working with the notion of favouring (and disfavouring) an object or a state of affairs. But this is a generic notion, meaning nothing

118

VALUE THEORY

more than an attitude of being in favour of (or against) something, a pro-attitude (or a con-/anti-attitude) towards something.1 Thin as it may sound, the basic idea is not completely empty: favouring must be, at least, a positive intentional state, a favourable response towards a certain object of thought, or towards a fact. This rules out two sorts of mental states from figuring in the relevant class: (i) non-intentional mental states such as: feeling hungry (itchy, sleepy, etc.), being in a certain mood (bored, euphoric, etc., but not about anything), sensing pleasure (as opposed to taking pleasure in something), and so on; (ii) intentional but not valenced (positive, negative, neutral) mental states: mere believing, contemplating, imagining, supposing, remembering, and so on (believing that something is good should count as a possible pro-attitude here). It seems to me that favouring can be determined along several dimensions: 1 What kind of attitude it is: a desire, a practical stance, an

emotion? 2 Its ‘direction’ or ‘telos’: for the sake of what or whom? 3 Its intensity: how strong, and how stronger than attitudes towards other things? These three dimensions, while perhaps not exhaustive, are typically present, though often implicitly, when we make ordinary value judgements. Here is a rough illustration. 1. The first dimension depends on a number of features of the valuable object. For example, we can consider its existential, modal, or probabilistic status. Something is intrinsically good, Ewing suggests, when, on its own account, ‘we ought to welcome, rejoice in it if it exists, seek to produce it if it does not exist. We ought to approve its attainment, count its loss a deprivation, hope for and not dread its coming if this is likely, avoid what hinders its production, etc.’ (1948: 149, my emphasis). Or again, it can matter which ontological category a value belongs to. For Ewing, an intrinsically good experience (e.g. an innocent pleasure) is fitting to be pursued, but an intrinsically good action is as such admirable or worthy of approval, though not necessarily fitting to The term ‘pro-attitude’ was introduced by Ross in (1939: 237), used in this connection by Ewing (1948: 148–50), and later made famous by Davidson (1963). 1



How Do I Favour Thee?

119

be pursued or chosen for its own sake (ibid.: 153–5). Furthermore, as hinted in section 2.4, there seem to be attitudes that are fittingly directed towards individual objects and people, and only indirectly towards states of affairs or proposition-like entities containing those individuals: preserving, protecting, cherishing, respecting, trusting, loving and hating people. And conversely there are attitudes primarily directed at states of affairs rather than individuals, like desiring, taking pleasure in, preferring (at least in one sense), promoting. According to some philosophers, there is one kind of practical attitude (promotion or maximization) that enjoys some kind of normative priority among others. I will come back to this in section 3.2 2. I used the ‘for the sake of’ locution to distinguish final from non-final value (Chapter 2), and personal from impersonal value (Chapter 4). Final value is to be favoured (in some of the ways above) for its own sake. Personal value is to be favoured for someone’s sake. Let’s examine final value first. In Chapter 2, I discussed and tentatively argued for the claim that final value can be intrinsic or extrinsic, i.e. can supervene on internal or relational properties of the valuable thing, where these also include relations to other valuable things. In the case of extrinsic final value, relational properties can play different roles within the supervenience base (see section 5.2). Consider our familiar Napoleon’s hat. Its final value results from its having belonged to a historically important man. And the significance of this fact is conditional on, or enabled by, history having some sort of final value (at least for the sake of argument). If history didn’t matter, Napoleon’s hat would be just another worthless accessory. For more abstract sorts of final value, e.g. the value of a certain friendship, likewise some enabling evaluative conditions need to hold, such as its not being immoral, or its being good for the participants. While it seems clear that intrinsic and unconditional values are to be favoured for their own sake, it might not be immediately obvious how we can favour extrinsic and conditional values for their own sake. For example, Zimmerman claims that we should Another question under this heading regards the intentional content of the fitting attitude: does favouring x require thinking that x is good or fitting to be favoured? I discuss this in section 8.7. 2

120

VALUE THEORY

favour final value ‘for its being what it is ... and not for the sake of its relation to some other valuable state’ (Zimmerman 2001: 91). Is Zimmerman’s conception of favouring for its own sake also applicable to extrinsic and conditional values? I think so. Two things should be clear from the discussion in Chapters 2 and 5: (i) extrinsic value is not necessarily instrumental value; (ii) an enabling condition for value is not a ground of value. Let’s start with (i). An object like Napoleon’s hat is to be favoured (e.g. preserved) for the sake of its relation to Napoleon, but this is not an instrumental relation to Napoleon or to any valuable state. The object is fitting to preserve because of the historical role of Napoleon, and ultimately, because of the value of history. Now, in a sense, this puts the hat in a relation to ‘some other valuable state’: Napoleon’s deeds, and ultimately world history. But note that it makes no sense to say that the hat itself produces or is a means to history, as if history were a separable outcome caused by the hat. Rather, the connection to Napoleon forges a unique link between the hat and history, analogous to the link between a wedding ring and a relationship, such that the loss of the hat (or of the ring) constitutes a loss in value even in the presence of other hats (or rings) similarly historically connected. Therefore, to say that the hat has to be preserved for the sake of history, which is surely correct, implies (rather than contradicts) that it has to be preserved for what it is, as required by Zimmerman’s definition.3 Now to (ii). If the value of friendship requires the value of other things connected to friendship, this doesn’t mean that it is for the sake of its relation to these other things that it is fitting to favour (engage in, respect, etc.) a given instance of friendship. It only means that our favouring friendship should not occur in isolation from our favouring those other things. This is because the conditions for the value of friendship affect but do not exhaust its value. Friendship is valuable as the kind of relationship it is, whose Of course if Zimmerman understands ‘for what it is’ as ‘for its intrinsic properties’, then Zimmerman’s criterion cannot be applied to extrinsic value. But this is precisely what should be left open, or there would not be common ground between those who think that final value is always intrinsic and those who deny it. Also, in the case of desire, some philosophers use the label ‘intrinsic desire’ for the attitude fitting to final value (Lewis 1989). But they don’t necessarily mean ‘desiring something for its intrinsic properties’. The idea is just to contrast desiring as an end vs. desiring something as a means (to the satisfaction of another desire). 3



How Do I Favour Thee?

121

value doesn’t simply result from being morally OK or good for the participants. So we can, and should, value friendship for what it is, when those conditions obtain. The cases above can be contrasted with extrinsic relations which do exhaust a thing’s value, and therefore make it fitting to favour it for the sake of a relation to some other valuable state. The obvious example is instrumental value. The radiator in my room, or the fact that it works, should certainly be favoured (during winter), but our evaluative attention or focus is or should be confined to its benefits to me and others. This is the value for the sake of which we should favour the radiator’s working. I haven’t delved as deep into ‘for someone’s sake’ attitudes. We shouldn’t assume that these have something more substantial in common with ‘for its own sake’ attitudes than their being positive responses towards something. Their being essentially person-oriented might make them rather different in kind. First, there is a notable structural difference. In impersonal final value, the valuable object is that for the sake of which an attitude is fitting. In personal value, it isn’t: we favour an object for someone’s sake, because we are interested in its being good for a person, and ultimately in the person herself. Second, it seems clear that desiring for your own sake that you live a happy life is not the same as desiring, for its own sake, that you live a happy life. The level and kind of concern displayed for you is different in the second case, even if you ‘are within’ a state I desire finally. The difference shows up, at least in part, in the emotional reaction to the frustration of my desire: sadness in the first case, but only disappointment in the second.4 Third, final and personal value can also be combined, as explained in section 4.4: when we correctly favour something for someone’s sake, we can still make a mistake as to how we favour it. For example, it might be a mistake for me to favour, for your sake, your getting as rich as possible for its own sake. So it seems that we are dealing with related but different attitudes. An analysis of for someone’s sake attitudes is beyond the scope of this book. The basic idea is that ‘we favour x with an eye to someone (else)’ (Rønnow-Rasmussen 2011: 56), and in particular with an eye to how certain properties of x relate to certain

4

See Darwall (2002: 1–2, 47, 68–9).

122

VALUE THEORY

properties of a person. If we think that pleasure is finally good for someone, then this value will supervene on some relation between pleasure and the person: intuitively, pleasure directly contributing to her well-being. The problem with this is that, in order to explain which properties of a person’s are relevant, we have referred back to the concept of well-being or goodness for. Perhaps a non-circular account of good for is unavailable. Or perhaps we could appeal to some psychological description of the attitude of caring for someone. In any case, this point serves again to underscore the difference between ‘for its own sake’ and ‘for someone’s sake’ attitudes: in the former case, we are not necessarily concerned about anybody’s well-being. Perhaps we should conclude that the word ‘sake’ is semantically ambiguous (ibid.: 60). 3. Value comes in degrees, and so should fitting attitudes. Some finally valuable things might be much better or much worse than others, and our fitting responses should be sensitive to these differences. Likewise, a minor evil is fittingly responded to with mild disfavour, while in response to a major evil the same mild disfavour is not a fitting response, or anyway less of a fitting response than strong disfavour. One could suggest that talk of milder or stronger favour and disfavour is really reducible to talk of preferences. That the object has a given degree of value entails that the object has a certain position in a fitting ranking of things, but not that it should be favoured or disfavoured with a corresponding strength. And preference, in turn, may not be reducible to ‘greater favour’: e.g. to prefer the lesser evil to the greater is not to favour the former more than the latter, because that would be to take a positive attitude towards evil.5 One problem with this reductive approach is that talking about degrees of value is simply not tantamount to making value comparisons, and thus to preferring something: we are not always interested in betterness when e.g. we judge a certain state of affairs to be very good (e.g. we need not imply that it is better than most states of affairs of that kind). A second problem is that, even if we accept preference as a distinctive attitude, there is still an issue about by how much one should prefer one thing to another. For See Zimmerman (2001: 104), and his references to Chisholm, Lemos, and Brentano. 5



How Do I Favour Thee?

123

a schematic example: ceteris paribus, preferring one minute of pain to two hours with the same intensity as one prefers one hour of pain to two hours seems intuitively wrong. Both one minute and one hour of pain are better than two hours, but one minute is much better, and our preferences should reflect that. For a less schematic example: we should prefer a good pianist to a bad one more than we should prefer a mediocre to the same bad one. Betterness comes in degrees, and so should the relative intensity of preferences. It is worth mentioning another challenge to the idea that fitting attitudes come in degrees. Degrees of value might be said to correspond to degrees of fittingness, rather than degrees of attitudes. A good (10 units of pleasure), it could be said, is better than another (5 units of pleasure) when it is more fitting to favour the former than it is to favour the latter, rather than when it is fitting to favour the former more (i.e. more intensely) than the latter. What is gradable here is the property of being fitting, not the intensity of attitudes. Even admitting this conceptual possibility,6 the proposal has two significant difficulties. First, let’s think about the notion of ‘more fitting than’. In one case, we have a good x of degree 10, a generic attitude of favour towards it, and (let’s suppose) a property or relation of fittingness between the two to a degree of 10. In another case, we have a good y of degree 5, the same generic attitude of favour towards it, and a property or relation of fittingness between the two to a degree of 5: (1) xG10 & xFavour [→ F10] (2) yG5 & yFavour [→ F5] If this is what is meant by (1) being more fitting than (2), i.e. something like ‘five times more fitting’, then it borders nonsense. This way of putting things might suggest that (1) contains more fittingness than (2). Or alternatively, that (1) is a better specimen of fittingness than (2). But such claims (whatever exactly they mean) are clearly unjustified. In the two cases we have a good of a certain degree, and the same kind of response to each good. Provided that Fittingness may not always admit of degrees. Compare: a key either fits or doesn’t fit a keyhole. If a key fits a hole ‘better’ than another hole, then it fits the one and not the other. 6

124

VALUE THEORY

the response is of a fitting kind (and it is), (1) and (2) can only be equally fitting. Maybe ‘more fitting’ is more charitably interpreted as: the reason to favour x is stronger or weightier than the reason to favour y. Degrees of fittingness should be understood as degrees of normative strength or weight, a notion familiar from talk about reasons and deontic language. But this proposal is also untenable, as it stands. If reasons here are simply another way to talk about fittingness, the idea then must be that ‘reasons of fittingness’ to favour x are stronger than ‘reasons of fittingness’ to favour y. But as we have just seen, as far as fittingness goes, (1) and (2) are on an equal footing. So favour as such is no more fitting or appropriate to x than to y. So there cannot be stronger ‘reasons of fittingness’ to favour x. Now, the case for x is stronger than the case for y, but this normative truth should be expressed differently. While it is not more fitting to favour x than to favour y, it is fitting to choose x over y, should it come to that. And this is because, given how good x and y are compared to each other, it is fitting to favour x more than y, and in turn it is fitting to choose what we should favour the most. The second difficulty with ‘more fitting than’ is that it cannot hope to replace the relevant distinctions. Quite simply, we do evaluate attitudes, feelings, and emotions on the basis of their intensity, and other relevant features such as duration, resistance to certain challenges, and so on. With regard to different objects or states of affairs, some degrees and levels of concern, desire, satisfaction, fear, frustration, etc., are more or less proportionate than others. This often has to do with our particular position as situated agents (e.g. normally we should care more for the near and dear than for strangers) or as epistemic subjects (e.g. normally we should fear the likely more than the unlikely). But just as often it has to do with differences in the personal and impersonal values we are responding to. It would be a mystery why we should respond to value by having certain attitudes, while a central feature of attitudes, namely their strength, in a broad sense, is not to be involved in this process.7 Likewise, we should respond to evidence not only by acquiring or discarding relevant beliefs, but by adjusting our degrees of credence appropriately. 7



How Do I Favour Thee?

125

Once we admit degrees of attitudes, there arise intricate questions regarding how symmetrical relations between degrees of value and degree of fitting attitudes should be. For example, suppose I am enthusiastic about a minor achievement (say, ironing a shirt). My favouring response is of a fitting kind, but not ‘maximally’ fitting: I should be satisfied rather than enthusiastic. Is my response thereby unfitting or only less than maximally fitting? On the other hand, notice that when a response is of the unfitting kind, e.g. if I am happy about an earthquake in Japan, there seems to be no limit to how unfitting my response can be – there is no maximal unfittingness. Again, another question is whether favouring the neutral (what is neither good nor bad) is positively unfitting, or simply not fitting, and whether disfavouring the neutral is similarly unfitting or simply not fitting.8 There seems to be no reason to expect a perfect symmetry, based on conceptual considerations alone. These are substantive issues, which should not be decided in advance by a fitting attitude account of value. If we want to leave these questions open to substantive disagreement, though, we had better formulate FA accordingly. For instance, we cannot define finally good to degree n simply as fitting to be favoured for its own sake to some degree, because this would allow any degree of favouring towards the finally good to be fitting, regardless of the degree of goodness. But we have seen that not only different degrees of favouring can be more or less fitting, but excessive favouring of some good could be even unfitting (and the same might be true for defective favouring, and mutatis mutandis for excessive/defective disfavouring of the bad). On the other hand, we don’t want to define good to degree n as fitting to be favoured precisely to degree n, because we want to allow at least some degrees of favouring n to be fitting rather than ‘not fitting’, albeit less fitting than favouring to degree n. It seems then that all we can say is that if x is finally good to degree n, then x is fitting to be favoured for its own sake to a degree sufficiently close or proportional to n, where room is left for determining this ‘sufficiency’ from a very inclusive range (favouring x to degree n is ‘perfectly’ fitting, but any favouring will

See Zimmerman (2001: 103–13).

8

126

VALUE THEORY

still be fitting) to a very exclusive one (only favouring x to degree n is fitting). Moreover it is important, in addressing these questions, to separate the (un)fittingness of a response from its value. Being enthusiastic about a minor achievement might be unfitting, but still good, or at any rate better than being indifferent towards it. Likewise, favouring what is neither good nor bad might be unfitting, but again still good, or at least not bad. Questions about the value of fitting and unfitting attitudes will be decided by one’s substantive theory of value. There does not seem to be a conceptual requirement that good responses, i.e. responses it is fitting to favour, need to be themselves fitting responses.9

7.3 Responses to value: Maximizing Let’s go back to the first dimension: what kind of (dis)favouring is appropriate to what? On the one hand, we can be as liberal and pluralistic as we like. There will be appropriate responses to some things and not others, in virtue of their different properties: appreciating an artwork, praising a person’s moral qualities, preserving a forest, and so on. And some responses might be common to many things: e.g. admiration, care, respect seem to apply to people as well as to valuable objects and states of affairs.10 However, for some philosophers all such responses are appropriate only to the extent that they express some single ‘master’ fitting attitude to final value. In the broadly consequentialist tradition, final value is, first and foremost, something to be 9 In Hurka’s view (2001) virtues, qua intrinsically good responses, are fitting responses to the good and the bad. But this is a substantive claim. 10 Also, even on a pluralist approach there can be a broad structure to appropriate responses. For instance Raz (2001: 161–9) describes three stages of correctly relating to value: recognition, preservation (both of which fall under ‘respect’ for Raz), and personal engagement. The first two stages are not valuespecific, and have the role of protecting the possibility of the third stage, in which values are ‘realized’ by people engaging with them in the right ways (preserving a Picasso painting protects the possibility of people enjoying it, admiring it, being moved by it, etc.). Pluralist philosophers in this sense also include, among others: Anderson (1993), Swanton (1995), Baron (1997), Scanlon (1998). See Nozick (1981: 429–30) for a list of 57 (!) appropriate responses.



How Do I Favour Thee?

127

maximized: we should make it the case that the world contains as much value and as little disvalue as possible.11 How we do this is largely an empirical question: admiring, respecting, preserving value are usually all fine ways to maximize it, nor do we need to be constantly and actively engaged in producing value, like workers in some sort of value factory. But sometimes maximizing makes it necessary to sacrifice one instance of a value for a bigger amount of the same value, or for another weightier value. We have seen this in Chapter 4 while discussing agent-relative value: it might be necessary to kill one in order to save two from being killed. It might be necessary to wage war in order to protect peace in the long run, or to prevent some major injustice. It might be necessary to destroy Napoleon’s hat in order to raise historical awareness. Notice: the philosophical issue here is not (just) whether all things considered this is what one should do. Rather, it is about the best way to construe such value conflicts. If maximization is the only appropriate response to value, then killing, waging war, or destroying Napoleon’s hat are prima facie wrong or unfitting only because they are actions which prima facie fail to maximize the relevant value (human life, peace, history), and not because they express some other unfitting attitude towards those values (say, disregard or disrespect). The choice then is not between respecting human life (peace, history) and maximizing human life (peace, history), but rather between two potentially maximizing strategies: maximizing a certain value by respecting it or maximizing it by not respecting it. And once the conflict is seen from this angle, it becomes obvious that one should simply choose the option which does (or is expected to) maximize the relevant value, and in this sort of case, respecting the value simply is not a way to maximize it. But is this the right way to view such conflicts? The answer depends on whether there is any reason to assume maximization as the fundamental way of favouring value. Philip Pettit suggests that universalizability provides such an argument (1997: 142–3). Suppose I claim that respecting, though not necessarily maximizing, peace is the fitting attitude for me (or more realistically for a country) in a given circumstance: it is Normally this is meant as a moral ‘should’, since consequentialism is a moral theory. I treat it here instead as a generic normative ‘should’ of fittingness. 11

128

VALUE THEORY

fitting to abstain from waging or engaging in war. Now, fittingness claims, like other normative claims, are universalizable: in making this claim, I commit myself to the claim that it is fitting for anyone else in the same relevant circumstances as mine to respect peace. If this is so, Pettit argues, I am committed to subscribing to one single kind of consideration that would call for all agents (similarly situated) to respect peace. So, universalizing an attitude of respect towards peace is tantamount to being committed to or desiring a world where everyone respects peace. But a world where everyone respects peace is a world where peace is maximized. This is because Pettit defines respecting or rather ‘honouring’ or ‘instantiating’ a value as follows: ‘to instantiate a value is to behave in the way that would promote [maximize] the value in a world, roughly, where others were equally compliant’ (ibid.: 127). So respect really commits one to maximization, rather than being a distinct and possibly incompatible attitude. Those who think they can respect value but eschew a commitment to maximization are simply confused. There seem to be two problems with Pettit’s argument. One is this. Even if it is true that a world where peace (or some other value) is universally respected is a world where peace is maximized, it doesn’t follow that I, as a ‘respecter’ of peace, am thereby committed to maximizing it, period. All I am committed to, it seems, is desiring a world where peace is maximized by being universally respected. In turn, this commitment seems compatible with a refusal to accept maximization by other means, e.g. engaging in war to protect long-term peace, which may involve attitudes and behaviour contrary to respect for peace. A second problem with Pettit’s argument is that it simply assumes that the values to which respect is an appropriate response are always, also, values which it is appropriate to maximize, whether or not maximization dominates respect. But arguably not all such values are ‘maximizables’, such that it always makes sense to desire, and in turn bring about, more of them rather than less.12 Scanlon hints at this point regarding friendship: ‘it seems The problem is not that maximizing might not be the appropriate response to e.g. certain forms of artistic excellence, which depend for their value on their relative rarity, so that producing them as much as possible is nonsense. Pettit might say that 12



How Do I Favour Thee?

129

overblown to say that what is important about friendship is that it increases the value of the state of the universe in which it occurs’ (Scanlon 1998: 88). Anderson sums this up in general terms: ‘Since there are no generally valid practical maximizing principles for intrinsic value, we say that intrinsic value cannot be increased by increasing the number of its bearers’ (Anderson 1997: 97).13 Honouring friendship for its own sake has nothing to do with preferring a world where there are 10 friendships to one where there are 9. Respecting humanity has nothing to do with wishing that more human beings are created. Note: the point is independent from the question, addressed in Chapter 2, of whether individual objects or proposition-like entities are the primary bearers of final value. Even if we agree that only facts carry final value (e.g. that A and B are friends; that she is a human being, etc.), it does not follow that preferring or producing more rather than less of them is always appropriate. The advocate of maximization might be tempted to reply that if increases in the number of value bearers do not increase value in general (as Anderson says), or the value of the state of the universe (as Scanlon says), then maximizing a certain value simply does not require increasing the number of value bearers of that kind, but rather it requires holding the whole panoply of attitudes Scanlon and Anderson recommend. This answer, however, seems to trivialize the notion of maximization as a distinctive kind of response to value. If maximizing a given value V amounts to making sure that things go as well as possible with respect to V, or that the world is as good as can be as far as V is concerned, then we are back to a generic sort of concern or favouring towards V. To be sure, the notion won’t be completely empty of significance: after all, it is a concern that things go as well as possible, as opposed to ‘well enough’.14 But what has disappeared is the distinctive, and therefore more controversial claim

if rarity is a condition for artistic excellence, then maximizing the value contributed by excellence will simply involve attitudes and actions which do not undermine the rarity of the relevant skills. 13 Anderson defends the more radical view that intrinsic value cannot ever be maximized. Here I’m discussing the claim that it cannot always be maximized. 14 See Slote (1989) and Byron (2004) on the debate about ‘satisficing’ as an alternative to maximizing.

130

VALUE THEORY

that the more final value (and the less final disvalue), the better, and with it the corresponding idea that our attitudes should be properly sensitive to this principle.15

7.4 Two concepts of intrinsic value? Throughout the book, we have explored a number of contrasting views: 1 final value is always intrinsic (and unconditional) value vs.

final value can be extrinsic (and conditional);

2 final value bearers are always states of affairs (or

proposition-like entities) vs. final value bearers can be individual objects or persons; 3 value is always impersonal vs. value can be personal; 4 if something has final value, it has it (and to the same degree) no matter where it occurs vs. something can alter its final value in different contexts; 5 final value is fundamentally comparable vs. there can be incomparability; and in this chapter: 6 there is one fundamental kind of fitting response to final

value (maximization) vs. there can be different kinds of fitting responses.

Ben Bradley (2006) has argued that these disputes are to a large extent merely apparent, since there is no common concept of final or intrinsic value shared by the disputing parties prior to taking a stand on these issues. Rather, there are two distinct concepts of value, which Bradley calls Moorean and Kantian: Moorean value is by and large defined by the left-hand claims in 1–6, while

Pettit also argues for a constraint on appropriate responses which only maximization would satisfy: non-iteration. This is an interesting idea, but it hasn’t been sufficiently elaborated. See Pettit (1997: 261–2), and McAleer (2008) for a critique.

15



How Do I Favour Thee?

131

Kantian value is defined by the right-hand ones.16 If the theses and antitheses in 1–6 define different concepts of value, then the disputing parties are really talking past each other. By way of an analogy, the dialectical situation here might be similar to a disagreement on God’s properties. Theologian A argues that God is all-benevolent, while theologian B denies that. They seem to be disagreeing over a common subject matter. But it turns out that theologian B also denies that God has a personal nature – for B, God is rather a force of nature, energy, or something of that sort. Now it seems that A and B are really talking past each other: if they don’t agree on whether God has personal nature, they must be having two different concepts of God to begin with, and the dispute over God being all-benevolent dissolves. Likewise, if the Moorean and the Kantian don’t agree on something so basic as, say, the nature of final value bearers, then any further disagreement about value must be treated as merely apparent. Now, among the claims defining Moorean value Bradley includes a normative claim about practical reasons (or fitting attitudes and actions): PRO: When something is intrinsically good, someone has a good reason to try to promote it, or preserve it, or make it true, or bring it into existence. (Bradley 2006: 120) If Bradley were right, then the foregoing disagreement between the maximizer (or ‘promoter’) and the non-maximizer would be based on a mistaken assumption. There simply isn’t one common value concept or property on which we can disagree regarding its normative implications for conduct. Rather, there are two concepts of value, each defined, inter alia, by what responses are appropriate Among Mooreans: Moore, Ross, Chisholm, Lemos, Zimmerman. Among Kantians: maybe Kant, Scanlon, Anderson, Raz, and the philosophers who argue in favour of extrinsic final value, at least to the extent they do so argue: Korsgaard, Kagan, Rønnow-Rasmussen and Rabinowicz. Note that for Bradley the two concepts, being different, can be combined within the same normative view: Kant himself arguably applies Kantian value to the good will and to persons, but Moorean value to the state of being deservedly happy. Kantian value can be an essential element in a Mooreanly valuable state of affairs. The label ‘Kantian’ is not ideal, though understandably inspired by Kant’s claims about the value and worth of people and the good will. Nonetheless I cannot avoid using it in discussing Bradley. 16

132

VALUE THEORY

to the value bearer: there is Moorean, i.e. maximizable value, to which PRO applies, and there is Kantian, i.e. non-maximizable (or at least, not necessarily maximizable) value. In this book I have proceeded contrary to Bradley’s claim. In Chapter 1 I presented the general framework under which disputes such as those 1–6 should be embedded: to be valuable is to be the object of fitting favourable attitudes. And in Chapter 2 I defended the assumption that different philosophers such as Aristotle, Kant, Mill, and Moore are discussing about a common subject matter, by pointing to the fact that they all accept, in one form or another, a notion of what is valuable for its own sake, that is, final value, as opposed to what is valuable for the sake of something else that is valuable. On the basis of this assumption, disputes involving 1–6 have been treated as genuine disagreements regarding the articulation and specification of further features of final value. For instance, dispute no. 6 clearly flows from the general fitting attitude characterization: which favourable attitudes, if any, are fundamentally fitting towards final value? Bradley doubts that FA provides sufficient common ground among the disputing parties. He writes: What would be problematic for my thesis would be if there were a unifying value concept V such that the disputes between Mooreans and Kantians discussed in this paper could plausibly be seen as disputes over the nature of V. But FA does not provide, and is not intended to provide, any such unifying concept. (ibid.: 127) I have three replies to give here. First, it is not clear that FA does not provide a unifying value concept. Certainly FA is intended to articulate a central, non-negotiable fact about value to which both Mooreans and Kantians are responsive. At each juncture of the debate over 1–6, FA can reasonably be expected to stand as a common constraint which determines the benefits and the costs of a given thesis. In particular, we have seen how FA determines some costs for the Moorean theses 1–4 (Chapters 2, 4, 5).17 Moore did not endorse FA as an analytical thesis about intrinsic value. But he did accept it as a synthetic necessary truth. See Chapter 8, and Olson (2006). And the other Mooreans mentioned in the previous footnote do seem to accept some version of FA as a definition of value. 17



How Do I Favour Thee?

133

Second, it is true that FA is an abstract schema, and as such leaves open many fundamental questions about the nature of final value. And Mooreans and Kantians answer these questions – fill in the schema – in their own distinctive ways, as claims 1–6 show. In doing so, each develops a view about the logic and structure of final value. It is also true and interesting that claims 1–6 form rather unified clusters: if you accept the view that final value is always intrinsic, then you will look for the sort of value bearers which deliver that result, and in turn will accept only fitting responses which could logically be directed at these value bearers; or instead you could start with a view about fitting responses to value, and work your way up from there through the metaphysics of value. However, it’s essential that each of the claims in 1–6 needs to or at least can be argued for, and this can only be done (and is typically done) by, inter alia, arguing against the seemingly opposing view. Bradley’s point would imply that such argumentative attempts are misguided. But this would take the ground away from both Mooreans and Kantians. How else can Mooreans build their picture of value, if not by, at each step, showing the difficulties of the contradictory thesis? Third, note a certain asymmetry between Mooreans and Kantians. The latter typically (though not always) hold views which recognize part of the truth as lying in the corresponding Moorean thesis. For example, Kantians do not need to deny that some value bearers are states of affairs, or that the value of some things remains constant across contexts, or that maximization is an appropriate response to certain goods. What they deny is that the Moorean claims hold necessarily. By Bradley’s lights then, these Kantians are guilty of misunderstanding their own views. They would be mistaken in thinking that what they are doing is proposing a more generous metaphysical and normative picture of the same thing (i.e. final value), about which the Moorean gives a more rigid or unified account. While it can be argued that a Kantian view may suffer from such a pluralistic stance (how can the same thing, i.e. value, come in so many different shapes and sizes?), it is less plausible to think that Kantians are wrong to understand themselves as pluralists in value theory.

134

VALUE THEORY

7.5 Summary In this chapter I have tried to clarify in general terms the nature of fitting attitudes. I have defended to some extent a pluralistic view about the ways in which value can be favoured. Finally, I have claimed that there is a genuine debate here, as elsewhere in value theory, while agreeing with Bradley that answers to certain questions (e.g. about value bearers) may constrain answers to other questions (e.g. about fitting attitudes). The FA approach, even when not taken as a definition of value, provides sufficient common ground among discussants.

CHAPTER EIGHT

Value and the Wrong Kind of Reasons 8.1 Introduction In Chapter 1, I introduced the idea the value is normative as a general and essential property of value. I used the fitting attitude account as a historically and theoretically important example of how the normativity of value can be articulated: FA: x is good = it is fitting to respond favourably to (‘favour’) x. I have relied on FA as the working assumption throughout the book. In this chapter, I reconsider the merits of FA as a definition of value, as an answer to the question ‘what is value?’. The main problem for FA is that not any fitting favouring attitude towards x, or not any reason to favour x, implies that x is good. I will explore in some detail different strategies FA theorists have devised to solve this problem. The conclusion will be that further work is needed, and possibly we should consider the merits of alternative accounts.

136

VALUE THEORY

8.2 The fitting attitude account and its rivals FA is not the only account that aims to make sense of how value is connected to normative concepts such as reasons, ought, or fittingness. Remember FA’s two central tenets: 1 Reduction Claim: value properties such as goodness are

reduced to normative relations of fittingness of, or of there being a reason for, attitudes. 2 Normative Redundancy Claim: value properties such as goodness do not themselves provide reasons for attitudes (make attitudes fitting) over and above the good-making features which already provide reasons for attitudes (make attitudes fitting). Any genuine alternative to FA must deny either or both central claims. So here is a list of alternatives: Early Moorean Account: denies (1), because it turns it around. Normative relations are reduced to evaluative properties. In Principia Ethica, Moore holds there that while intrinsic goodness is indefinable, deontic concepts such as rightness or duty can be defined in terms of value: To ask what kind of actions we ought to perform, or what kind of conduct is right, is to ask what kind of effects such action and conduct will produce … Every judgement in practical Ethics may be reduced to the form: This is a cause of a good thing … ‘[R]ight’ does and can mean nothing but ‘cause of a good result’. (1993: 196) It is rarely noticed, but very relevant in this connection, that Moore also analysed non-moral concepts like fittingness in terms of intrinsic value: [B]y saying that a certain relation between two things is fitting or appropriate, we can only mean that the existence of that relation is absolutely good in itself. (ibid.: 152)1 See also Moore (1993: 239).

1



Value and the Wrong Kind of Reasons

137

How to precisely and satisfactorily spell out an Early Moorean Account may require some effort,2 but the general idea should be clear enough: evaluative concepts are basic, and normative concepts are to be explained in evaluative terms. On this approach, value is normative because normativity is already evaluative: you cannot understand what a reason or an ought is, without an understanding of what is good, bad, and the like. It should be noticed that (2), the Normative Redundancy Claim, follows from the Early Moorean Account just as it does from FA: if being good serves to define the notion of a reason, then it cannot also be a provider of reasons. The existence of a reason to favour x would mean, inter alia, that x is good, intrinsically or otherwise. For x being good to provide a reason to favour x would be for x being good to make it the case that x is good. But that is an empty claim. Therefore, also on this account it is not goodness, but good-making properties which provide reasons. Necessary Equivalence Account (or Later Moorean Account): Moore came to reject his own definitions of deontic terms in evaluative terms, settling for the view that there is a necessary equivalence between evaluative and normative facts, but not any relation of identity or reduction.3 Whenever a certain normative claim is true, so is a relevant evaluative claim, and vice versa. In the same way, we could say that a triangle having three sides and a triangle having three angles are necessarily equivalent but not identical properties: they necessarily occur together (no possible triangle has one but not the other property), but they still are different properties of triangles. So (1) is rejected because normative and evaluative truths are tightly related, but not in any reductive way. What about (2)? Here there are two possibilities:4 I have proposed a formulation in (Orsi 2013a). In Ethics he makes it clear that the relation between what is our duty and what produces maximum intrinsic value is not analytical (1966 [1912]: 89). In his Reply to My Critics he explicitly acknowledges his previous view as a mistake (1942: 558–9). Moore was influenced by Bertrand Russell’s The Elements of Ethics (1910), where Russell both takes good to be indefinable, even in normative terms (1910: 5–6), and rejects ‘x’s consequences will probably be the best possible’ as an analysis of ‘x is right’, though it is ‘a mark which in fact attaches to all objectively right actions and to no others’ (ibid.: 26). 4 To my knowledge Moore himself doesn’t explicitly endorse or reject either. See Olson (2006). 2 3

138

VALUE THEORY

Necessary Equivalence with Redundancy: this view accepts (2). Fittingness and value are necessarily related, but not because value makes attitudes fitting, or provides reasons for attitudes. Of course, such a view has its challenges to face. First, it needs to establish (2) on grounds other than the identity or reduction of one sort of property to the other. Second, it needs to provide an alternative explanation of why the necessary equivalence holds: it cannot be a mere coincidence that value is always normative. Proponents of this view may say that the equivalence is explained by goodness and fittingness sharing their resultance bases (see Chapter 5): whatever makes an attitude fitting also makes a state of affairs good, e.g. a certain experience being pleasant both makes it fitting to favour it and makes it good. And if this is true, then goodness cannot be a reason-provider or fitting-maker: if it were, then it would also be a good-maker, but that is again an empty claim.5 Necessary Equivalence without Redundancy: according to this view it is fitting to favour x whenever x is good, and vice versa, because x’s goodness makes it fitting to favour x, or provides reason to favour it. The necessary equivalence is explained by goodness and other like evaluative properties being themselves reason-providers. Value is normative in the sense that it is normatively significant:6 it provides reasons. So this view denies (2). In Scanlon’s terms, goodness or being valuable doesn’t pass the buck: ‘When something has the right natural properties it has the further property of being valuable, and that property gives us reason to behave or react in certain ways with regard to it’ (Scanlon 1998: 97). To some extent, common sense supports this view. We often mean Roger Crisp (2005, 2008) holds a view close to Necessary Equivalence with Redundancy: (at least some) reasons are provided by evaluative properties, though not by goodness. Still, the property of goodness serves to separate the evaluative from the non-evaluative, and thus good things will be necessarily reason-providing. See also Dancy (2000) for a view close to this one. What both Crisp and Dancy leave open is whether all reasons to favour pair up with goodness, since not all reasons to favour need be grounded on evaluative properties. So they hold a weaker thesis: necessarily, if x is good, then there are reasons to favour x. (Indeed, Dancy would even drop the necessity claim and opt for a regularity thesis.) See Stratton-Lake (2009) for criticism of Crisp’s view. 6 See Parfit (1997) for the distinction between normative facts and normatively significant facts. 5



Value and the Wrong Kind of Reasons

139

to justify our choices, preferences, or attitudes by simply appealing to the goodness or betterness of an action, or the badness of the alternatives, and so on. And some philosophers also have resisted the pressure to ‘pass the buck’. For instance Ross argues against FA on the following basis. If good were defined as fitting to be admired, then good would be defined as fitting to be thought of as good. This would already involve moving in a circle. But, further, the circle would be a vicious one, because ‘if we ask on what ground a thing is worthy of being thought to be good, only one answer is possible, namely that it is good’ (Ross 1939: 278–9). So goodness has a role to play as an ineliminable fitting-maker.7 Also, when it comes to other normative claims, e.g. moral ones, Ross seems to argue that the prima facie duty to promote intrinsic value is grounded on the fact that something has intrinsic value (Ross 2002: 24).8 The structure of this view then is: x being pleasant (e.g.) → x being good → it is fitting to favour x. According to this view, there is a certain division of explanatory or grounding labour: certain properties make things good, and goodness in turn grounds normative truths. Once the structure of the view is clarified, it cannot be argued against it that goodness does not add any normative weight to an already given list of reasons to favour x: until we mention goodness, simply there are no items on that list.9 Further, in a reason-giving See also Blanshard: ‘If saintliness and generosity are such as to merit favouring, it must be because there is something in them that goes beyond their ‘factual characteristics’ and equally goes beyond a mere blank cheque on our favour. What is this? I think we must answer, a goodness that they have already’ (1961: 286). 8 Ross’s overall view is more complicated, since he also acknowledges a distinct sense of ‘goodness’ that is reducible in FA terms. And his distinctive deontological position crucially seems to depend on negating the necessary equivalence claim: some basic normative claims, i.e. most of the prima facie duties, involve relations of fittingness, but are not about responses to the good. 9 For instance Scanlon: ‘[N]atural properties provide a complete explanation of the [practical] reasons we have for reacting in [certain] ways to things that are good and valuable. It is not clear what further work could be done by special reasonproviding properties of goodness and value, and even less clear how these properties could provide reasons for action’ (1998: 97, my emphasis). For doubts about the strength and significance of Scanlon’s redundancy thesis, see Crisp (2005), Väyrynen (2006), Schroeder (2011). 7

140

VALUE THEORY

exchange ‘because it is good’ seems to be an acceptable answer to the question ‘why should I favour x?’ Of course, it won’t be a very informative answer, but this is because it needs to be supplemented by a specification of the good-making features, rather than of further reasons to favour x. So it seems that any good objection to this view must be directed not so much at its structure, however baroque it may seem, but rather at its implications: e.g. that there is only one kind of ultimate reason in favour of actions and attitudes, namely goodness and its cognates. A pluralistic account of reasons might perhaps be preferable. It is indeed this kind of objection that Scanlon mentions as one of his arguments against such a view: [M]any different things can be said to be good or to be valuable, and the grounds for these judgements vary widely. There does not seem to be a single, reason-giving property that is common to all these cases. (1998: 97–8) Note that proponents of Necessary Equivalence without Redundancy can accept the first part of the quotation: many different things can make things ultimately good. The objection lies rather in the second part: it is difficult to believe that, given pluralism about the good (about ultimate good-makers), there is only one and always the same reason to favour these different things. What is odd, according to one way of reading Scanlon, is the idea that once we have learnt that innocent pleasures are good, and on this basis it is fitting to favour them, we have thereby also learnt why it is fitting to favour other disparate good things like virtuous characters or friendship – as if understanding virtue or friendship were not necessary to appreciating the reasons to favour them, and how we should respond to them, because understanding goodness suffices for that. In sum, we can have different models of the relation between fittingness (or reasons) and value. Supposing that (a) x is good if and only if (b) it is fitting to favour x,10 This is not an innocent supposition (and of course it needs to be qualified, but here I am proceeding schematically). As I mentioned in previous footnotes, one can deny necessary equivalence altogether (e.g. Dancy 2000), even with the proper 10



Value and the Wrong Kind of Reasons

141

there are competing explanations of why it is so. Schematically: 1 Fitting Attitude Account: (b) defines (a); 2 Early Moorean Account: (a) defines (b); 3 Necessary Equivalence with Redundancy: (a) and (b) both

result from x having certain other properties;

4 Necessary Equivalence without Redundancy: (b) results

from (a).

8.3 The wrong kind of reasons problem These alternatives to FA are well worth mentioning, because there is one major stumbling block to FA’s success, namely the wrong kind of reasons problem (WKR), which I briefly introduced in Chapter 1. There I gave the example of the evil demon who will torture us (or the whole humankind) if we fail to admire him. It seems that, based on this threat, we have abundant reasons to admire him. If the buck-passing account is true, then it follows that the demon is good. But clearly the demon is evil rather than good, in part precisely because of his threat. His threat is a wrong kind of reason to admire him, because it doesn’t make him good or admirable. Such a reason to admire him does nothing to show the demon’s positive value. In Chapter 1, I suggested that buck-passers might approach a solution to the problem by adverting to fittingness: whatever reasons there might be to admire him, it seems that admiration just doesn’t fit a being like the demon. So the buck-passing account could be formulated as follows: Buck-passing 3: x is good = x has the property of having other properties that provide reasons of fittingness to favour x, or would provide such reasons to suitably situated agents. Or alternatively, and more simply, one could adopt FA. But, qualifications. Or one can deny that truths like (b) always require truths like (a), e.g. when truths like (b) are about moral normativity, even if (a)s always require (b)s (e.g. Ross).

142

VALUE THEORY

arguably, this is more like the promise of a solution than a genuine solution to WKR. Until we obtain an account of fittingness, or of reasons of fittingness, we are not saying much more than this: x is good when there are reasons of the right kind to favour x. The crucial and thus far unanswered question is: what makes reasons of fittingness reasons of the right kind? On the other hand, alternatives to FA do not seem to face WKR with the same urgency. Of course, WKR presents itself as a counter-example to the claim of necessary equivalence: for some xs, there are reasons to favour x, but it is not the case that x is good. However, on the Early Moorean Account, value is conceptually primary: we start out with value judgements, such as that the demon and his threats are evil, and in turn there are reasons to despise the demon and against admiring him. If there are also reasons to admire him, then these must stem from some other value; plausibly here, the value of minimizing suffering. On the thesis of Necessary Equivalence without Redundancy, likewise, value judgements are prior. If there are reasons to admire the demon, it doesn’t yet follow that the demon is valuable, because those reasons might be grounded in the goodness of something else (again, minimizing suffering). Also the thesis of Necessary Equivalence with Redundancy seems to avoid WKR: true, a ground for reasons to favour x is, in principle, also a ground for x’s value (a value-maker of x), but we can consult our intuitions about value to determine that the demon is not valuable, and thus whatever reasons to favour him there might be are not also considerations that make him valuable. In other words, on these accounts we are free to appeal to the whole range of evaluative truths in order to explain away apparent counter-examples to necessary equivalence, because we can have an independent purchase on the notion of value. Not so with FA: an appeal to values here would be circular, since value is what FA is trying to understand.

8.4 The structure of the problem and an initial response Given the logical possibility of scenarios like the evil demon, WKR is generated by a set of assumptions about such a scenario:



Value and the Wrong Kind of Reasons

143

MM

Value: some thing x has a certain value (good, bad, better than y, etc.).

MM

Attitude: there are reasons of some kind to hold an attitude that is unfitting to x’s value.

MM

Non-circularity: these reasons cannot be defined as being unrelated to x’s value.

Several philosophers have pointed out one obvious feature of the evil demon case: the reason to admire the demon is not provided by a feature that belongs to the demon. It is provided, rather, by some state of affairs which will come about if we admire the demon: the minimization of suffering. The fact that this state of affairs is in this case, so to speak, under the demon’s control, does not mean that it provides reasons only in connection with the demon’s threat – we already have reasons, independent of anyone’s threat, to seek minimization of suffering. In one word, the reason to admire the demon is not object-given: not provided by the demon or any of his properties. Generally object-given reasons are contrasted with state- or attitude-given reasons: reasons provided by features of the very attitude for which there is a reason. In this case, one can say that the reason to admire the demon is provided by the instrumental properties of admiration: the fact that admiration will spare us from suffering. But, as I just implied, a better grounding for this reason seems to be given by the absence of suffering, which in this unusual case we would achieve by admiring the demon.11 In presenting this reply I am in fact combining two separate lines of response to WKR. First comes (1) the idea that reasons to admire the demon are derivative on other reasons (to minimize suffering), and so they only have a contingent relation to the object of admiration (i.e. the demon). If so, then (2) such reasons cannot be said to be object-given. Putting things in this sequence avoids the complications mentioned by Rabinowicz and Rønnow-Rasmussen (2004: 407) about neatly distinguishing object-given and state-given reasons. See Stratton-Lake (2005) for (1), and Parfit (2001), Olson (2004b), Piller (2006), Lang (2008) for the object-given/ state-given distinction and solutions inspired by that. See Rabinowicz and RønnowRasmussen (2004, 2006), Olson (2009b) for criticism. For a different solution based on the dual role of reasons as grounds for attitudes and subject’s motives for those attitudes, see Rabinowicz and Rønnow-Rasmussen (2004) and Cook (2008) (the idea being that the fact that admiring the demon will save us from torture cannot be a subject’s motive for admiring the demon for his own sake, and so is a wrong kind of reason to show the demon’s value). 11

144

VALUE THEORY

The next claim then is that only object-given reasons are reasons of fittingness. This seems a plausible enough claim: fittingness between attitudes and objects requires a match between the attitude and some relevant properties of its object, rather than properties of something else. If the reason to admire the demon is not to be searched in the demon, then admiration is not a fitting response to the demon. Therefore it is not the case that the demon is good or admirable. This initial reply, however, does not work for the whole range of scenarios which present WKR problems. I have three sorts of cases in mind. First, consider a very cruel joke being made about somebody.12 On this basis, let’s suppose that there is overall reason not to be amused by the joke. On the basis of FA, it would seem to follow that the joke is a bad one. But, barring a strong form of moralism about the comic, this need not be so: the joke can still be very funny, albeit cruel, and so a good joke, though one that we ought not to be amused by. The fact that it is cruel is a wrong kind of reason to make it bad as a joke. But note that the fact that it is cruel is an object-given rather than attitude-given reason against being amused by it. It is not a matter of what would happen if amusement at the joke were shown in the face of the targeted person. Cruelty is a property possessed by the joke, i.e. by the object of amusement. In other words, some object-given reasons for or against amusement need not correspond to the value of the joke. Second, there are possible (indeed, quite realistic) scenarios where object-given reasons to prefer an object to another do not necessarily coincide with its being better than another. A father might have reason to be happier about his own daughter’s recovery from illness, than about someone else’s recovery from an even worse illness.13 The father’s reason for this stronger attitude is object-given rather than attitude-given, because it is provided by

See D’Arms and Jacobson (2000a, 2000b). Another example they use is this: there can be moral or prudential reasons against envying somebody’s fortunes, but this doesn’t suffice to show that her fortunes are not enviable. But in this case it seems to me that such moral or prudential reasons would be at least partly attitude-given rather than object-given, whereas what’s wrong with being amused at the joke is that the joke itself is cruel (or morally problematic in other ways). 13 See Olson (2009a), Bykvist (2009), Zimmerman (2011), Orsi (2013b). 12



Value and the Wrong Kind of Reasons

145

his daughter’s recovery and by her being in a closer relationship to him, rather than by some property that his having a stronger attitude would possess. But such a reason is of the wrong kind to make his daughter’s recovery better than the stranger’s. In this and the previous example, it seems that morality or personal relationships can offer lots of object-given reasons for or against attitudes which do not necessarily bear on the value of the object. There is an element of fittingness in such responses, but not the sort of fittingness which goes along with the value of their objects. It seems that we need finer distinctions than the object-given/attitude-given one to capture the relevant reasons of fittingness. The third case is one where the distinction simply doesn’t apply. Incentives like the demon’s threat need not give reasons for attitudes at all, but simply reasons for actions, like cheating in a game of chess: if you cheat, the demon will spare us from torture.14 Intuitively this is a reason of the wrong kind to make cheating an admissible move. Cheating, like admiring the demon, remains unfitting, despite there being an overall reason in favour of it. But what rules out such incentives as reasons of fittingness cannot be anything about a mental state as opposed to its object, since we are not being asked to have any mental state in the first place. Again, we need a different criterion if we want to explain the general difference between reasons of fittingness and other reasons.

8.5 Reasons for what? Some philosophers address WKR by denying or qualifying one of the assumptions of the evil demon scenario, namely that in these cases there really are reasons to admire the demon, or that there are reasons to primarily admire him. These solutions focus on the nature of the relevant response rather than on the reason-giving features. Let’s consider them in turn. Practical Reasons. It may be argued that in the evil demon case we do not really have a reason to admire the demon, but rather a reason to want to admire him, or to try to admire, or to bring it about that one admires him. That is, the threat of being tortured 14

See Schroeder (2010).

146

VALUE THEORY

if we don’t admire the demon provides a reason to have a secondorder attitude such as wanting to admire him, but not a reason to admire him. After all, it may be said, this is perfectly coherent with FA: having the attitude of admiration is an (instrumentally) good thing, for it saves us from torture, so we have reason to have a pro-attitude towards having this attitude. On the other hand, there being no reason to admire the demon for his own sake (to have a first-order pro-attitude towards him), the conclusion is avoided that the demon is intrinsically good. The claim that in the evil demon case we have reason to want to admire him is correct. The proponent of this solution however has the burden of explaining why we don’t also have a reason to admire the demon. Both John Skorupski (2007) and Ingmar Persson (2007) introduce a parallel with reasons to believe. When it is useful to believe that p (e.g. that God exists) – as it is useful to admire the demon – does this give both a reason to want to and make yourself believe that p and a reason to believe that p? Both authors answer in the negative, although for different reasons. Skorupski offers two considerations. First, he illustrates a general distinction between what he calls epistemic reasons (reasons to believe), evaluative reasons (reasons to feel or have certain emotions), and practical reasons (reasons to do something). Then he points out that evil demon cases offer in the first place reasons to do something, namely bring it about that one admires the demon. Likewise, if the evil demon threatens to torture us if we don’t believe that p, we have reason to bring it about that we believe that p. But these are all practical reasons, not reasons to admire or to believe. Second, if the usefulness of admiring x or believing that p were also a reason to admire x or believe that p, then we would need to defend a pragmatist theory of reasons to admire (or to believe), which Skorupski rules out as ‘unusual’ (2007: 12). Skorupski’s two points are connected. There are different kinds of reasons, and the considerations that bear on one kind of reason (e.g. the usefulness of an attitude, which bears on practical reasons) cannot bear on another kind of reason (epistemic or evaluative). The usefulness of admiring x cannot be a reason to admire x but only a reason to bring it about that one admires x. However, apart from ruling out a pragmatist theory of reasons to admire, Skorupski does not tell us exactly why we should not admire the demon although we should bring it about that we



Value and the Wrong Kind of Reasons

147

admire him. This is quite a paradoxical situation: why should we bring it about that we do x if x is not what we should do? After all, we will be safe from the demon’s threat if we admire him, not just if we make ourselves admire him. Note that there is an important difference here between actions such as bringing one’s admiration about or making oneself admire and, for instance, trying or intending to admire. It is true that, in principle, one could have a certain prize attached to merely the trying or the intention, without this implying that one should do what one should try or intend to do, as in the famous Toxin Puzzle, where one gets rewarded for intending to drink a toxin but has no reason to drink it.15 If we could be saved from the demon just by trying to admire him, then it would not follow that we have any reason to admire him. But, as Skorupski recognizes, we are safe only if we bring it about that we admire him. And ‘bringing about’ is, arguably, a success-term: we bring x about only if x occurs. In other words, the case is set up so that we have reason to succeed; and since succeeding here means admiring the demon for his own sake, it seems to follow that, if we have reason to bring it about that we so admire him, then we also have reason to so admire him. Secondary Reasons. Persson doesn’t deny that we also have reasons to admire the demon, but these are secondary reasons. Consider Pascal’s wager. The probability of being amply rewarded after death if we believe that God exists, plus the negligible costs of believing in God, overall constitute a reason to want to believe that God exists. Is this also a reason to believe that God exists? It is a pragmatic reason to so believe. But what is characteristic of these reasons is that we cannot directly respond to them by forming a belief, because we cannot see these reasons as justifying the relevant belief by supporting its truth. The fact that, if we believe, we might be rewarded after death does not make it true or more probable that God exists. So pragmatic reasons are only secondary reasons to believe (Persson 2007: 4–5). They are, primarily, reasons to want to believe, as the direct response to the fact that believing that God exists will reward us is to want to have that belief. Analogously, the prospect of torture gives us only secondary reasons to admire the demon. This is because we cannot directly respond to the belief

See Kavka (1983).

15

148

VALUE THEORY

that the demon will torture us if we don’t admire him by admiring him, for we cannot see that belief as justifying admiration towards the demon (ibid.: 7). Our direct response will rather be a desire to admire him (a second-order attitude). Consequently, reasons of fittingness should be identified with primary reasons only, that is, those considerations the direct response to which is a certain pro-attitude, and that primarily justify such a pro-attitude.16 Persson’s central claim is that we cannot see admiration for the demon as justified by the belief that he will punish us if we fail to admire him. Persson thinks that this is because ‘in all probability, there is no practical reasoning attributable to you which has this [admiration] as its conclusion and the belief as its premise’ (ibid.: 7–8). Such a reasoning would have the form: (i) I believe that the demon will punish me if I don’t admire

him; (ii) I desire not to be punished; (iii) Therefore I admire the demon.

Persson thinks that only (iv) is the proper and direct conclusion: (iv) Therefore I desire to admire the demon.

This may be true. However, it remains to be explained why our practical reasoning should be as Persson says it is. That is to say, why is it that admiration is at best only an indirect and secondary response to (i)? In the epistemic case, believing that God exists is only an indirect response to the fact that we will be rewarded after death if we so believe, because we cannot see the fact of reward as supporting the truth of the belief that God exists. This is in line with the function of belief, which is to fit the facts. In the practical and emotional case, our pro-attitude, say, admiration for the demon, is an indirect response to the fact that if we don’t admire him he will punish us, but why? What we need here is something analogous to the specification in the epistemic case: p epistemically justifies our belief that q if p supports q’s truth or makes it more probable. With the justification of pro-attitudes, the criterion may be: p practically justifies a pro-attitude for q if p makes q valuable. 16

See also Raz (2009).



Value and the Wrong Kind of Reasons

149

Since the demon’s determination to torture us if we don’t admire him clearly does not make the demon valuable, neither can it justify admiration for the demon. But this criterion is not available to somebody upholding FA. Remember that here we are trying to explicate the notion of a primary reason for a pro-attitude, which in turn is needed to analyse the notion of x being valuable. If we adopt the following as a criterion: ‘p practically justifies (thus is a primary reason for) a pro-attitude for q if p makes q valuable’, then the analysis would be patently circular. Those who propose to solve the wrong kind of reasons problem by distinguishing between primary and secondary reasons for pro-attitudes, as Rabinowicz and Rønnow-Rasmussen remark: need to clarify, without taking the notion of value for granted, what makes something a reason for wanting (or trying) to have a certain attitude towards an object rather than a direct [i.e. primary] reason for having the attitude in question. (2004: 413–4) Persson tries to give such a clarification in terms of what attitudes are justified by what, but the analogy with epistemic reasons leaves him without a similarly deep explanation of such an idea, apart from a circular one. Epistemic justification can appeal to truth, but practical justification cannot appeal to value.17 There is also a more general problem for these approaches: they don’t seem to work for the other WKR scenarios mentioned above, namely the moral reasons against being amused by a cruel joke, the partial reasons to prefer one’s daughter’s recovery, and the rewarded cheating scenario. In all these cases of wrong kinds of reasons, it is simply implausible that the relevant considerations do A proposal in this similar spirit is also Hieronymi (2005). She argues that right kinds of reasons for a given attitude A are those which bear on a question, the settling of which amounts to having or forming A. But can the relevant questions for certain attitudes be specified in non-evaluative terms? Coming to admire the demon on the basis of his threat is nothing like having settled the question of whether he is admirable, but this we knew already. Hieronymi’s other notion is that the right kinds of reasons for A must be reasons that support the commitments constitutive of A, but she doesn’t specify what these commitments are in the case of favouring and the like. 17

150

VALUE THEORY

not provide reasons (or primary reasons) for the relevant attitudes (rejecting the cruel joke, preferring one’s daughter’s recovery) and actions (cheating under the demon’s threat). A Brentanian Solution? Danielsson and Olson (2007) argue for a different distinction. There are reasons to hold an attitude, which they call holding-reasons, and reasons for the correctness of that attitude, which they call content-reasons. The crucial notion here is correctness. There is a difference between a belief being correct, i.e. true, and the fact that we have reason to have the belief. Pascal’s wager provides such an example: the cost-benefit analysis of belief in God gives us a reason to believe that God exists (a holdingreason), but doesn’t make it correct to believe that God exists (it’s not a content-reason). The same point applies to pro-attitudes: pace Skorupski and Persson, the demon’s threat does give us a perfectly good reason to admire the demon, but still doesn’t make it correct to admire the demon. In turn, FA should understand x’s value in terms of there being content-reasons to favour x, i.e. considerations that make it correct to favour x. Despite the attitude-oriented labels, this approach can also be extended to actions: the demon’s threat gives us a holding-reason to cheat in the game, but not a content-reason – it doesn’t make the cheating moves correct. However, this much we knew already from Chapter 1: we knew that to solve WKR we had to focus on reasons of fittingness, or of correctness, as in Brentano’s terminology. Danielsson and Olson’s contribution points to a solution without providing one.18 Moreover, it doesn’t seem to apply to the other WKR scenarios. There are perfectly good content-reasons for rejecting the cruel joke (or for not being amused by it): it is, from a moral point of view, a correct or fitting response to it. There are content-reasons to prefer one’s daughter’s recovery to a stranger’s: it is a correct response, from a parent’s point of view. But these content-reasons don’t show that the joke is a bad or unfunny one, or that one’s daughter’s recovery is better than a stranger’s.19

See for instance Reisner (2009). Olson (2009a) argues that FA can avoid such WKR cases, since they are based on moral reasons, and these are different from reasons of fittingness. However, it seems hard to deny that some reasons of fittingness are moral reasons (isn’t keeping a promise fitting to having made one?). What Olson needs, again, is a specification 18 19



Value and the Wrong Kind of Reasons

151

8.6 Characteristic concerns and shared reasons The next two solutions I consider aim to say something more about reasons of fittingness, without relying on an object-given/stategiven distinction, and without ruling out wrong kinds of reasons as merely practical reasons or secondary reasons for attitudes. The first is by Justin D’Arms and Daniel Jacobson. They propose to isolate reasons of the right kind for an attitude (more precisely, an emotion) as reasons given by the features which are the emotion’s characteristic focus: our proposal is that to think an emotion a fitting response to some object is to think there is (pro tanto) reason, of a distinctive sort, for feeling the emotion toward it. Such judgements differ from other sorts of appraisals of an emotion, in virtue of the kind of reasons they concern. Reasons of fit are those reasons that speak directly to what one takes the emotion to be concerned with, as opposed to reasons that speak to the advisability or propriety of having that emotion. So reasons of fit for fear are roughly those that speak to whether or not something is a threat. (2006: 107–8) Analogously, reasons of fit for admiration are those that speak to what admiration is concerned with, as opposed to reasons that speak to the advisability or propriety of admiring. For example, the prudential or moral reasons we have for admiration in the evil demon case are of the wrong kind, because the threat to torture people is not a characteristic concern of admiration. Likewise, the cruelty and more generally the immorality of a joke may give decisive reasons against being amused by it, but still are not a characteristic concern of amusement: the joke is not thereby shown to be unfunny. In order to understand the relevant notion of fittingness of a response, we need some story as to what is the distinctive concern of that response. Fear is concerned with threats, so it is fitting when of the relevant reasons of fittingness, not just a bare appeal to fittingness (or correctness).

152

VALUE THEORY

it is a response to genuine threats; shame, as D’Arms and Jacobson also suggest, is concerned with what we recognize as a social disability, so it is fitting to be ashamed of one’s social disabilities, and so on. The notion of a response’s characteristic concern seems to advance our understanding of reasons of fittingness. In order for it to work as an analytical proposal, we should be able to grasp a response’s characteristic concern in a sufficiently independent manner from the value to be analysed: we should be able to understand something to be a threat or a social disability quite independently from its being fearsome or shameful. Even if this desideratum is satisfied for certain responses, however, there are two outstanding issues that require attention. First, for other responses it is not clear whether we can identify their distinctive concern without employing the very value we are supposed to analyse, or a generic value term. Take admiration. Its distinctive concern (the kind of things it is typically directed to) may be said to be, roughly, human strengths and achievements. But not just any achievement will prompt and justify admiration: rather, those achievements that manifest a sufficiently high degree of value, because of their being excellent, uncommon, or challenging for the person to realize. The problem is not so much that the notion of an achievement is evaluative – FA need not be committed to reducing thick as well as thin values – but rather that there might no way to pick out significant from minor achievements other than by saying they are better or higher in value.20 There is a threat of vicious circularity. Second, some responses are so generic that they do not even seem to have a distinctive concern. At least prima facie, attitudes such as desire or preference, unlike specific emotional states, may be prompted by far too many diverse circumstances for us to be able to pick out a set of features as their distinctive concern – unless by employing generic value concepts such as the good, the better, and the best. D’Arms and Jacobson are aware of this difficulty, but they argue that an FA analysis must primarily be concerned with specific value concepts and properties (the fearsome, the shameful, the funny, etc.) and thus with specific responses (2000b: 747). However, this reply assumes that WKR cases framed in terms of

Rabinowicz and Rønnow-Rasmussen (2004: 421).

20



Value and the Wrong Kind of Reasons

153

generic value concepts and generic responses must be intelligible in more specific terms, or else are spurious. But then consider again the father and daughter case. D’Arms and Jacobson might say that the ‘preference’ for a daughter’s recovery is intelligible as a distinctive emotion, such as parental love. However, the facts that justify the father’s preferential response are within the distinctive concern of that emotion, since they include his daughter’s wellbeing. If so, the father’s response is of the fitting kind according to D’Arms and Jacobson’s criterion. But still, his daughter’s recovery is not better than a stranger’s recovery from an equal or worse illness. Moreover, as seen above, a solution to WKR should throw light on a notion of fittingness that applies to actions as well as attitudes, whereas D’Arms and Jacobson’s view seems limited to emotions. Mark Schroeder has recently proposed a different criterion, based on the idea that activities (and not just attitudes) are often governed by standards of correctness: games have rules specifying which moves are correct, etiquette specifies correct ways of setting the table, and so on. This notion of correctness, in turn, is understood in terms of right kind of reasons, which for Schroeder are shared reasons in the following sense: The right kind of reasons with respect to any activity, A, are all and only those reasons which are shared by necessarily every able person engaging in A, because they are engaged in A, together with all reasons which are derivative from such reasons. (2010: 39) If the evil demon threatens torturing unless you put out fish knives for a meat-based meal, this is a wrong kind of reason with respect to the activity of setting the table, in the sense that it wouldn’t be shared by every table-setter because they are engaged in tablesetting. Even if every possible table-setter were threatened by the evil demon, and so there was a necessarily shared reason to put out fish knives, such a reason wouldn’t stem from the fact that they are engaged in table-setting. Therefore putting out fish knives is not a correct way to set the table for a meat-based meal, no matter how strong (wrong kind of) reasons there might be. How does Schroeder’s view apply to the other WKR scenarios? (i) The reason to admire the demon for his own sake might be

154

VALUE THEORY

likewise shared by all ‘admirers’, but still not because they are engaged in admiration. In this sense, admiring the demon is not a correct way of admiring demons or anyone. So the demon is not admirable. (ii) The reason not to be amused at the cruel joke might well be actually shared by everyone, but again not because we are engaged in the activity of appreciating jokes, or something like that. Engagement in this activity does not require lack of amusement towards cruel jokes. From this point of view, it is not a correct response, and therefore doesn’t show the joke to be unfunny.21 (iii) The father and daughter case is however trickier, because it is not clear what is the relevant activity. Suppose it is ‘showing love for Laura’ (this particular child). Since not everybody who is engaged in this activity would necessarily prefer Laura’s recovery to the stranger’s (suppose you are a friend of Laura’s, but the stranger’s son), we could say that it is not correct to prefer it, and hence whatever reasons her father has, they cannot show that Laura’s recovery is better than the stranger’s. This would solve WKR, but it seems based on a very narrow reading of the relevant activity. Mention of individuals in the specification of an activity might be appropriate in the analysis of what is good for someone, but we cannot assume that there is only personal value at stake here (indeed, her father’s preference doesn’t have to be good for Laura, who might not know about it).22 Suppose then the relevant activity is ‘showing love for one’s child’ or ‘being a decent parent’. Now, this activity does seem to generate shared reasons, for those engaged in it, to prefer one’s child’s recovery over a stranger’s, in any situation like the father and daughter case. And these are not entirely general reasons (like Engagement with other activities, such as moral censorship, might require lack of amusement as a correct response from everybody, but this would show that the joke has a different, moral, kind of disvalue. Still, Schroeder’s solution to (i) and (ii) is not without problems. (i) He explicitly appeals to evaluations in defining standards of correctness for admiration (e.g. admiration for x is correct if it would not be ‘a bad idea’ to emulate x (ibid.: 42)). (ii) Here the solution seems smoother. But it’s crucial to its success that Schroeder can describe the relevant activity, engagement in which would make a moralistic reaction at best optional and at worst incorrect, in terms other than ‘appreciating the comic value of jokes’, since this would presuppose the evaluative notion at issue. 22 See Chapter 4 for the analysis of personal value. 21



Value and the Wrong Kind of Reasons

155

moral reasons not to laugh at a cruel joke), because they only apply to those who are parents, and insofar as they are parents. But still, despite the preference being in this way correct, the daughter’s recovery is not better. Schroeder might pick ‘preferring’ as such as the relevant activity. Clearly, not anyone engaged in the bare activity of ‘preferring’ has a reason to prefer Laura’s recovery – presumably the stranger and his associates don’t. The father’s reason is not one necessarily shared, hence it is not of the right kind. The problem with this reply is that it is not clear that preferring constitutes an activity with its own standards of correctness. True, transitivity may provide one such standard: if we are engaged in preferring A to B, then we have reason to prefer A to C, if we also prefer B to C. But is there anything more substantial one can say about preference? Clearly there is a notion of a correct preference which includes more than honouring transitivity. But what this ‘more’ consists in cannot be gleaned just by reflecting on the nature of preference – except of course the truth, unhelpful here, that in general we have reason to prefer the better to the worse. Preference is too thin an ‘activity’ to deliver (significant) standards of correctness, and in turn shared reasons. Of course, Schroeder could reply that if there are no (significant) standards of correctness for preference, and a fortiori no reasons of the right or the wrong kind to prefer, then this case simply doesn’t pose a WKR problem, if formulated in terms of preference. But this move, like D’Arms and Jacobson’s rejection of generic responses, would greatly reduce the appeal of FA as a general account of the evaluative in terms of the normative. All we would be left with are piecemeal analyses of those values which pair up with specific emotions or with activities governed by standards of correctness. Moreover, not only would this approach be limited in scope, but it would also be far less interesting: it only takes a little bit of semantic competence, and not a philosophical discovery, to know that e.g. the admirable must be what it is correct to admire. The philosophical work then would be merely one of trying out different criteria to isolate the relevant notion of correctness: reconstructing the path that led us where we already are. This is why Schroeder claims we should be confident that a solution to WKR can be found (ibid.: 26–8). But such confidence can be had only based on either of

156

VALUE THEORY

two assumptions: (1) a reduction of value in general is not our aim, (2) value in general won’t present additional difficulties once the piecemeal analysis is carried out successfully. The first assumption is, arguably, too modest; the second, as the discussion above is meant to show, is unjustified – whatever solution may work for ‘admirable’, it may not suit concepts like ‘preferable’ and in turn ‘good’ or ‘better’.

8.7 Circular path: No-priority If the WKR problem should prove resistant to general solutions, one should indeed reconsider the merits of those alternatives to FA, outlined above, which do not suffer as greatly from WKR, because they don’t reduce value to reasons or fittingness. But there is one methodological move still available: reject or modify the assumption of non-circularity. The idea is suggested – though not as a reaction to WKR – by both McDowell and Wiggins. For instance, Wiggins writes: [T]here will often be no saying exactly what reaction a thing with the associated property will provoke without direct or indirect allusion to the property itself. Amusement for instance is a reaction we have to characterize by reference to its proper object, via something perceived as funny (or incongruous or comical or whatever). There is no object-independent and property-independent, ‘purely phenomenological’ or ‘purely introspective’ account of amusement. (Wiggins 1987: 195) McDowell puts it hypothetically: ‘If there is no comprehending the right sentiments independently of the concepts of the relevant [evaluative] features, a no-priority view is surely indicated’ (1998b: 160). One way to elaborate on this is to say that the relevant responses, and in turn the notion of them being fitting, are not prior to and independent of the concept of value, because they include as part of their content a relevant value judgement. And vice versa: the notion of value is not prior to the notion of fitting attitude, since it can only be understood in these terms. Moreover, it is not the case that truths about fittingness (or reasons) need to be in general definable as truths about value, as in the Early Moorean Account.



Value and the Wrong Kind of Reasons

157

Nor is this just a Necessary Equivalence Account, since the claim made is one of identity rather than necessary co-occurrence: No-Priority FA: x is good = x has properties that make it fitting, inter alia, to judge that x is good. If we think of value in this way, then what we need is simply an account of the right kinds of reasons to judge that p, which will deliver the result that the demon’s threat is not a right kind of reason to judge that the demon is good: for instance, it doesn’t make the proposition ‘the demon is good’ probable or provide evidence for it. Above, I noted that a solution such as Persson’s relied on a problematic analogy between beliefs (or in general truth-aimed states) and pro-attitudes. But on a no-priority view one would only need to adopt an account of reasons to believe (or to judge), without worrying whether a similar story can be told about admiration and the like.23 However, even leaving aside the thorny question of whether the kind of circularity of this view is acceptable or not,24 a no-priority view seems to misconceive the normativity of value as theoretical or epistemic normativity: value essentially requires judgements of value, rather than pro-attitudes or actions. Of course, for any given valuable object, there will be a fitting emotion or attitude, of which the value judgement is a component. But the way in which a no-priority view helps in a solution to WKR makes the judgement the crucial element common to all relevant fitting attitudes. And conversely, any sort of attitude which doesn’t include such a judgement (and a fortiori, actions) will not be a fitting response to value. This is a substantial constraint on the range of fitting attitudes. So a no-priority view of this kind cannot provide a satisfactory solution to WKR.25 Though see Gregory (2013) for the further doubt whether the notion of a reason (to believe or whatever) is itself evaluative. 24 One reason why it is acceptable might be that, in fact, in order for a no-priority view to help in a solution to WKR, the value judgement might as well be understood itself as a judgement that a given attitude is fitting to x, thus eliminating a direct reference to ‘good’. Values would be understood in terms of the fittingness of attitudes including fittingness judgements. 25 In fairness to McDowell and Wiggins, they do not claim that all fitting responses involve value judgements, nor are they concerned with WKR in the first place. 23

158

VALUE THEORY

8.8 Summary This book has, at many junctures, proceeded on the assumption that a fitting attitude account of value, or some other account of the normativity of value, serves as a constraint to bear in mind while pursuing value theory. In this chapter I have questioned whether FA is actually a superior account, by discussing the wrong kind of reasons problem and alleged solutions to it: object-given vs. attitude-given reasons; evaluative/primary vs. practical/secondary reasons; fittingness/correctness accounts. While we are on the search for a completely satisfactory solution, it might be worth considering the merits of alternative accounts of value.

BIBLIOGRAPHY

Anderson, E. (1993). Value in Ethics and Economics. Cambridge, MA: Harvard University Press. —(1997). ‘Practical Reason and Incommensurable Goods’, in Chang 1997a. Aristotle (350 bc). Nicomachean Ethics (trans. W. D. Ross). http://classics.mit.edu/Aristotle/nicomachaen.1.i.html/ Arneson, R. (2010). ‘Good, period’. Analysis, 70(4): 731–44. Arrhenius, G. (2005). ‘Superiority in value’. Philosophical Studies, 123: 97–114. Arrhenius, G. and Rabinowicz, W. (2005). ‘Millian superiorities’. Utilitas, 17: 127–46. Baron, M. (1997). ‘Kantian Ethics’, in Baron et al. 1997. Baron, M. et al. (1997). Three Methods of Ethics. Oxford: Blackwell. Beardsley, M. C. (1965). ‘Intrinsic value’ Philosophy and Phenomenological Research, 26: 1–17. Bentham, J. (1970 [1789]). An Introduction to the Principles of Morals and Legislation. London: Athlone Press. Blackburn, S. (1984). Spreading the Word. Oxford: Clarendon Press. Blanshard, B. (1961). Reason and Goodness. London: George Allen & Unwin. Bradley, B. (2002). ‘Is intrinsic value conditional?’. Philosophical Studies, 107: 23–44. —(2006). ‘Two concepts of intrinsic value’. Ethical Theory and Moral Practice, 9: 111–30. Brännmark, J. (2008). ‘Excellence and means: on the limits of buck-passing’. The Journal of Value Inquiry, 42: 301–15. Brentano, F. (1969 (1889)). The Origin of Our Knowledge of Right and Wrong (ed. R. M. Chisholm). London: Routledge & Kegan Paul. Broad, C. D. (1930). Five Types of Ethical Theory. London: Routledge & Kegan Paul. —(1940). ‘Review of Foundations of Ethics (by W. D. Ross)’. Mind, 49: 228–39. —(1942). ‘Certain Features in Moore’s Ethical Doctrines’, in Schilpp 1942, 41–68.

160 Bibliography

Broome, J. (1997). ‘Is Incommensurability Vagueness?, in Chang 1997a. Bykvist, K. (2009). ‘No good fit: why the fitting attitude analysis of value fails’. Mind, 118: 1–30. Byron, M. (ed.) (2004). Satisficing and Maximizing: Moral Theorists on Practical Reason, Cambridge: Cambridge University Press. Carlson, E. (2001). ‘Organic unities, non-trade-off, and the additivity of intrinsic value’. The Journal of Ethics, 5: 336–60. Chang, R. (ed.) (1997a). Incommensurability, Incomparability, and Practical Reason. Cambridge, MA: Harvard University Press. —(1997b). ‘Introduction’, in Chang 1997a. —(2001). ‘Against constitutive incommensurability or buying and selling friends’. Philosophical Issues, 11: 33–60. —(2002). ‘The possibility of parity’. Ethics, 112: 659–88. —(2005). ‘Parity, interval value, and choice’. Ethics, 115: 331–50. Chisholm, R. M. (2005 [1972]). ‘Objectives and Intrinsic Value’, in T. Rønnow-Rasmussen and M. J. Zimmerman (eds) (pp. 171–9), Recent Work on Intrinsic Value. Dordrecht: Springer. —(1981). ‘Defining intrinsic value’. Analysis, 41: 99–100. —(1986). Brentano and Intrinsic Value. Cambridge: Cambridge University Press. Conee, E. (1982). ‘Instrumental value without intrinsic value?’. Philosophia, 11: 345–59. Cook, P. (2008). ‘An augmented buck-passing account of reasons and value: Scanlon and Crisp on what stops the buck’. Utilitas, 20: 490–507. Crisp, R. (2000). ‘Review of value … and what follows (by J. Kupperman)’. Philosophy, 75: 458–62. —(2005). ‘Value, reasons and the structure of justification: how to avoid passing the buck’. Analysis 65: 80–5. —(2008). ‘Goodness and reasons: accentuating the negative’. Mind, 117: 257–65. Dancy, J. (2000). ‘Should we Pass the Buck?’, in A. O. Hear (ed.), The Good, the True and the Beautiful (pp. 159–73). Cambridge: Cambridge University Press. —(2004a). Ethics without Principles. Oxford: Clarendon Press. —(2004b). ‘Enticing Reasons’, in R. Jay Wallace et al. (eds), Reason and Value: Themes from the Moral Philosophy of Joseph Raz (pp. 91–118). Oxford: Oxford University Press. —(2005). ‘The Particularist’s Progress’, in T. Rønnow-Rasmussen and M. J. Zimmerman (eds), Recent Work on Intrinsic Value (pp. 325–47). Dordrecht: Springer. —(2007). ‘Was Moore Right about Punishment?’, in S. Nuccetelli

Bibliography

161

and G. Seay (eds), Themes from G. E. Moore: New Essays in Epistemology and Ethics (pp. 325–42). Oxford: Oxford University Press. Danielsson, S. and Olson, J. (2007). ‘Brentano and the buck-passers’. Mind, 116: 511–22. D’Arms, J. and Jacobson, D. (2000a). ‘The moralistic fallacy: on the “appropriateness” of emotions’. Philosophy and Phenomenological Research, 61: 65–90. —(2000b). ‘Sentiment and value’. Ethics, 110: 722–48. —(2006). ‘Anthropocentric Constraints on Human Value’, in R. Shafer-Landau (ed.), Oxford Studies in Metaethics, Vol. 1. Oxford: Clarendon Press. Darwall, S. (2002). Welfare and Rational Care. Princeton: Princeton University Press. Davidson, D. (1963). ‘Actions, reasons, and causes’. Journal of Philosophy, 60(23): 685–700. Ewing, A. C. (1948). The Definition of Good. New York: Macmillan. —(1959). Second Thoughts in Moral Philosophy. London: Routledge & Kegan Paul. Feldman, F. (1998). ‘Hyperventilating about intrinsic value’. Journal of Ethics, 2: 339–54. Firth, R. (1952). ‘Ethical absolutism and the ideal observer’. Philosophy and Phenomenological Research, 12: 317–45. Fletcher, G. (2012a). ‘Resisting buck-passing accounts of prudential value’. Philosophical Studies, 157(1): 77–91. —(2012b). ‘The locative analysis of good for formulated and defended’. Journal of Ethics and Social Philosophy, 6(1): 1–27. Foot, P. (1985). ‘Utilitarianism and the virtues’. Mind, 94: 196–209. Geach, P. (1956). ‘Good and evil’. Analysis, 17: 33–42. Gert, J. (2004a). Brute Rationality: Normativity and Human Action. Cambridge: Cambridge University Press. —(2004b). ‘Value and parity’. Ethics, 114: 492–510. Gibbard, A. (1990). Wise Choices, Apt Feelings. Cambridge, MA: Harvard University Press. Gregory, A. (2014). ‘A very good reason to reject the buck-passing account’. Australasian Journal of Philosophy, 92: 287–303. Griffin, J. (1986). Well-Being: Its Meaning, Measurement and Importance. Oxford: Clarendon Press. Heathwood, C. (2008). ‘Fitting Attitudes and Welfare’, in R. ShaferLandau (ed.), Oxford Studies in Metaethics, vol. 3 (pp. 47–73). Oxford: Oxford University Press. Hieronymi, P. (2005). ‘The wrong kind of reason’. The Journal of Philosophy, 102: 437–57.

162 Bibliography

Hooker, B. (2000). Ideal Code, Real World: A Rule-Consequentialist Theory of Morality. Oxford: Clarendon Press. Hsieh, N. (2007). ‘Is incomparability a problem for anyone?’. Economics and Philosophy, 23: 65–80. —(2008). ‘Incommensurable Values’, in E. N. Zalta (ed.), The Stanford Encyclopedia of Philosophy (Fall 2008 edn). http://plato.stanford.edu/ archives/fall2008/entries/value–incommensurable/ Hurka, T. (1987). ‘“Good” and “good for”’. Mind, 96: 71–3. —(1998). ‘Two kinds of organic unity’. Journal of Ethics, 2(4): 299–320. —(2001). Virtue, Vice and Value. New York: Oxford University Press. —(2003). ‘Moore in the middle’. Ethics, 113: 599–628. Jackson, F. (1998). From Metaphysics to Ethics. Oxford: Oxford University Press. Jacobson, D. (2011). ‘Fitting Attitude Theories of Value’, in E. N. Zalta (ed.), The Stanford Encyclopedia of Philosophy (Spring 2011 edn). http://plato.stanford.edu/archives/spr2011/entries/ fitting–attitude–theories/ Kagan, S. (1998). ‘Rethinking intrinsic value’. Journal of Ethics, 2(4): 277–97. Kant, I. (1964 [1785]). Groundwork of the Metaphysic of Morals. New York: Harper and Row. Kavka, G. (1983). ‘The toxin puzzle’. Analysis, 43: 33–6. Kelly, C. (2008). ‘The impossibility of incommensurable values’. Philosophical Studies, 137: 369–382. Klocksiem, J. (2011). ‘Moorean pluralism as a solution to the incommensurability problem’. Philosophical Studies, 153: 335–49. Korsgaard, C. M. (1983). ‘Two distinctions in goodness’. Philosophical Review, 92: 169–95. —(1996). The Sources of Normativity. Cambridge: Cambridge University Press. Kraut, R. (2007). What is Good and Why: The Ethics of Well-Being. Cambridge, MA: Harvard University Press. —(2011). Against Absolute Goodness. Oxford: Oxford University Press. Lang, G. (2008). ‘The right kind of solution to the wrong kind of reason problem’. Utilitas, 20 (4), 472–89. Langton, R. (2007). ‘Objective and unconditioned value’. Philosophical Review, 116: 157–85. Lemos, N. M. (1994). Intrinsic Value. Cambridge: Cambridge University Press. —(1998). ‘Organic unities’. Journal of Ethics, 2: 321–37. Lewis, D. (1989). ‘Dispositional theories of value’. Proceedings of the Aristotelian Society (suppl.), 63: 113–37.

Bibliography

163

Louise, J. (2004). ‘Relativity of value and the consequentialist umbrella’. Philosophical Quarterly, 54: 518–36. McAleer, S. (2008). ‘Pettit’s non-iteration constraint’. Utilitas, 20: 59–64. McDowell, J. (1998a). ‘Values and Secondary Qualities’, in Mind, Value, and Reality (pp. 131–50). Cambridge, MA: Harvard University Press. —(1998b). ‘Projection and Truth in Ethics’, in Mind, Value, and Reality (pp. 151–66). Cambridge, MA: Harvard University Press. McKeever, S. and Ridge, M. (2006). Principled Ethics: Generalism as a Regulative Ideal. Oxford: Oxford University Press. Mackie, J. (1977). Ethics: Inventing Right and Wrong. London: Penguin. Mason, E. (2011), ‘Value Pluralism’, in E. N. Zalta (ed.), The Stanford Encyclopedia of Philosophy (Fall 2011 edn). http://plato.stanford.edu/ archives/fall2011/entries/value–pluralism/ Mill, J. S. (1861). Utilitarianism. http://www.utilitarianism.com/mill1.htm/ Moore, G. E. (1993 [1903]). Principia Ethica (ed. T. Baldwin). Cambridge: Cambridge University Press. —(1966 [1912]). Ethics (2nd edn). London: Oxford University Press. —(1993 [1922]). ‘The Conception of Intrinsic Value’, in Moore 1993 (1903). —(1942). ‘A Reply to My Critics’, in Schilpp 1942: 535–677. —(1959). ‘Is Goodness a Quality?’, in Philosophical Papers (pp. 89–101). London: George Allen & Unwin. Mulligan, K. (1998). ‘From appropriate emotions to values’. The Monist, 81: 161–88. Nozick, R. (1981). Philosophical Explanations. Cambridge, MA: Harvard University Press. Olson, J. (2004a). ‘Intrinsicalism and conditionalism about final value’. Ethical Theory and Moral Practice, 7: 31–52. —(2004b)‘Buck-passing and the wrong kind of reasons’. Philosophical Quarterly, 54 (215): 295–300. —(2006). ‘G. E. Moore on goodness and reasons’. Australasian Journal of Philosophy, 84: 525–34. —(2009a). ‘Fitting attitude analyses of value and the partiality challenge’. Ethical Theory and Moral Practice, 12: 365–78. —(2009b). ‘The wrong kind of solution to the wrong kind of reason problem’. Utilitas, 21: 225–32. Orsi, F. (2013a). ‘What’s wrong with Moorean buck–passing?’ Philosophical Studies, 164: 727–46. —(2013b). ‘Fitting attitudes and solitary goods’. Mind, 122: 687–98. Parfit, D. (1984). Reasons and Persons. Oxford: Oxford University Press. —(1997). ‘Reasons and motivation’. Proceedings of the Aristotelian Society (suppl.), 71: 99–130.

164 Bibliography

—(2001). ‘Rationality and Reasons’, in D. Egonsson et al. (eds), Exploring Practical Philosophy: From Action to Values. Aldershot: Ashgate. Persson, I. (2007). ‘Primary and Secondary Reasons’, in D. Egonsson, J. Josefsson, B. Petersson and T. Rønnow-Rasmussen (eds), Hommage à Wlodek. Philosophical Papers Dedicated to Wlodek Rabinowicz. www.fil.lu.se/hommageawlodek Pettit, P. (1997). ‘The Consequentialist Perspective’, in Baron et al. 1997. Piller, C. (2006). ‘Content-related and attitude-related reasons for preferences’. Philosophy, 81: 155–82. Portmore, D. (2005). ‘Combining teleological ethics with evaluator relativism: a promising result’. Pacific Philosophical Quarterly, 86: 95–113. Rabinowicz, W. (2008). ‘Value relations’. Theoria, 74: 18–49. Rabinowicz, W. and Rønnow-Rasmussen, T. (1999). ‘A distinction in value: intrinsic and for its own sake’. Proceedings of the Aristotelian Society, 100: 33–51. —(2003). ‘Tropic of value’. Philosophy and Phenomenological Research, 66(2): 389–403. —(2004). ‘The strike of the demon: on fitting pro-attitudes and value’. Ethics, 114: 391–423. —(2006). ‘Buck-passing and the right kind of reasons’. Philosophical Quarterly, 56: 114–120. Railton, P. (1986). ‘Facts and values’. Philosophical Topics, 14(2): 5–31. Raz, J. (1986). The Morality of Freedom. Oxford: Clarendon Press. —(1999). Engaging Reason: On the Theory of Value and Action. Oxford: Oxford University Press. —(2001). Value, Respect, and Attachment. Cambridge: Cambridge University Press. —(2009). ‘Reasons: Practical and Adaptive’, in D. Sobel and S. Wall (eds), Reasons for Actions. Cambridge: Cambridge University Press. Regan, D. (1997). ‘Value, Comparability, and Choice’, in Chang 1997a. —(2004). ‘Why Am I My Brother’s Keeper?’, in R. Jay Wallace et al. (eds), Reason and Value: Themes from the Philosophy of Joseph Raz. Oxford: Oxford University Press. Reisner, A. (2009). ‘Abandoning the buck passing analysis of final value’. Ethical Theory and Moral Practice, 12: 379–95. Richardson, H. (1994). Practical Reasoning about Final Ends. Cambridge: Cambridge University Press. Ridge, M. (2011). ‘Reasons for Action: Agent-Neutral vs. AgentRelative’, in E. N. Zalta (ed.), The Stanford Encyclopedia of Philosophy (Winter 2011 edn). http://plato.stanford.edu/archives/ win2011/entries/reasons-agent/

Bibliography

165

Rind, M. and Tillinghast, L. (2008). ‘What is an attributive adjective?’ Philosophy, 83(1): 77–88. Rønnow-Rasmussen, T. (2011). Personal Value. Oxford: Oxford University Press. Rosati, C. (2008). ‘Objectivism and relational good’. Social Philosophy and Policy, 25(1): 314–49. Ross, W. D. (2002 [1930]). The Right and the Good (ed. P. Stratton– Lake). Oxford: Clarendon Press. —(1939). Foundations of Ethics. Oxford: Clarendon Press. Russell, B. (1910). ‘The Elements of Ethics’, in Philosophical Essays (pp. 1–58). London: Longmans. Scanlon, T. M. (1998). What We Owe to Each Other. Cambridge, MA: Harvard University Press. Schilpp, P. A. (ed.) (1942). The Philosophy of G. E. Moore. Evanston and Chicago: Northwestern University. Schroeder, M. (2006). ‘Not so promising after all: evaluator-relative teleology and common-sense morality’. Pacific Philosophical Quarterly 87(3): 348–56. —(2007). ‘Teleology, agent-relative value, and the “good”’. Ethics 117: 265–95. —(2010). ‘Value and the Right Kind of Reasons’, in R. Shafer-Landau (ed.), Oxford Studies in Metaethics, Vol. 5 (pp. 25–55). Oxford: Oxford University Press. —(2011). ‘Buck-passers’ negative thesis’. Philosophical Explorations, 12: 341–47. —(2012). ‘Value Theory’, in E. N. Zalta (ed.), The Stanford Encyclopedia of Philosophy (Summer 2012 edn). http://plato.stanford. edu/archives/sum2012/entries/value-theory/ Sen, A. (1983). ‘Evaluator relativity and consequential evaluation’. Philosophy and Public Affairs, 12: 113–32. Sidgwick, H. (1981 [1907]). The Methods of Ethics (7th edn). Indianapolis: Hackett. Skorupski, J. (2007). ‘Buck-passing about goodness’, in D. Egonsson, J. Josefsson, B. Petersson and T. Rønnow-Rasmussen (eds), Hommage à Wlodek. Philosophical Papers Dedicated to Wlodek Rabinowicz. www.fil.lu.se/hommageawlodek/ Slote, M. (1989). Beyond Optimizing. Cambridge, MA: Harvard University Press. Smith, A. (1759). The Theory of Moral Sentiments. Indianapolis: Liberty Fund. Smith, M. (1994). The Moral Problem. Oxford: Blackwell. —(2003). ‘Neutral and relative value after Moore’. Ethics, 113: 576–98.

166 Bibliography

—(2010). ‘On normativity’. Analysis, 70(4): 715–731. Stratton-Lake, P. (2002). ‘Pleasure and Reflection in Ross’s Intuitionism’, in P. Stratton–Lake (ed.), Ethical Intuitionism: Re-evaluations (pp. 113–36). Oxford: Oxford University Press. —(2005). ‘How to deal with evil demons: comment on Rabinowicz and Rønnow-Rasmussen’. Ethics, 115: 788–98. —(2009). ‘Roger Crisp on goodness and reasons’. Mind, 118: 1081–94. Stratton-Lake, P. and Hooker, B. (2006). ‘Scanlon versus Moore on Goodness’, in T. Horgan and M. Timmons (eds), Metaethics after Moore (pp. 149–68). Oxford: Oxford University Press. Sturgeon, N. (2010). ‘Normativity’. Analysis, 70(4): 744–53. Suikkanen, J. (2004). ‘Reasons and value—in defence of the buck-passing account’. Ethical Theory and Moral Practice, 7: 513–35. —(2009). ‘Buck-passing accounts of value’. Philosophy Compass, 4: 768–79. —(2009). ‘Consequentialism, constraints and the good–relative–to: a reply to Mark Schroeder’. Journal of Ethics & Social Philosophy, March 2009: 1–9. Swanton, C. (1995). ‘Profiles of the virtues’. Pacific Philosophical Quarterly, 76: 47–72. Szabo, Z. (2001). ‘Adjectives in Context’, in R. Harnish and I. Kenesei (eds), Perspectives on Semantics, Pragmatics, and Discourse (pp. 119–46). Amsterdam: John Benjamins Publishing Co. Tannenbaum J. (2010). ‘Categorizing Goods’, in R. Shafer-Landau (ed.), Oxford Studies in Metaethics, Vol. 5. New York: Oxford University Press. Thomson, J. J. (1997). ‘The right and the good’. Journal of Philosophy, 94: 273–98. —(2001). Goodness and Advice. With commentary by Philip Fisher et al. Princeton: Princeton University Press. —(2006). ‘The Legacy of Principia’, in T. Horgan and M. Timmons (eds), Metaethics after Moore (pp. 233–55). Oxford: Oxford University Press. —(2008). Normativity. Chicago: Open Court. —(2010). ‘Reply to critics’. Analysis, 70(4): 753–764. Väyrynen, P. (2006). ‘Resisting the Buck-Passing Account of Value’, in R. Shafer-Landau (ed.), Oxford Studies in Metaethics, Vol. 1 (pp. 295–324). Oxford: Oxford University Press. Wiggins, D. (1987). ‘A Sensible Subjectivism?’, in Needs, Values, Truth: Essays in the Philosophy of Value. Oxford: Blackwell. Williams, B. (1981). ‘Internal and External Reasons’, in Moral Luck. Cambridge: Cambridge University Press. —(1985). Ethics and the Limits of Philosophy. Cambridge, MA: Harvard University Press.

Bibliography

167

Wright, G. H. von (1958). The Varieties of Goodness. http://www. giffordlectures.org/ Zimmerman, M. J. (1999). ‘The moral aspect of nonmoral goods and evils’. Utilitas, 11(1): 1–15. —(2001). The Nature of Intrinsic Value. Lanham: Rowman and Littlefield. —(2005). ‘Intrinsic Value and Individual Worth’, in T. RønnowRasmussen and M. J. Zimmerman (eds), Recent Work on Intrinsic Value (pp. 191–205). Dordrecht: Springer. —(2007). ‘The good and the right’. Utilitas, 19(3): 326–353. —(2009). ‘Understanding what’s good for us’. Ethical Theory and Moral Practice, 12: 429–439. —(2010). ‘Intrinsic vs. Extrinsic Value’, in E. N. Zalta (ed.), The Stanford Encyclopedia of Philosophy (Winter 2010 edn). http://plato. stanford.edu/archives/win2010/entries/value–intrinsic–extrinsic/ —(2011). ‘Partiality and intrinsic value’. Mind, 120: 447–83.

INDEX

additivity 87 agent–relative/agent–neutral value 73–9, 127 doubts about the distinction 75–6, 78 Anderson, E. 37n. 7, 104–6, 114, 129 appropriateness see fittingness and unfittingness Aristotle 27, 29, 38, 45 axiology 6–8, 27, 29, 49, 50, 52, 53, 56, 58, 59, 98, 99, 117 bearers of value 8, 35–8, 69, 129, 130–3 pluralism about 37, 133 beauty 31n. 2, 32, 52, 86 Bentham, J. 28 Bradley, B. 87n. 1, 130–3 Brentano, F. 10, 150 Broad, C. D. 19 buck-passing account of value 10–15, 19, 20, 21n. 15, 138–9, 141 arguments against 13–15, 139, 141–3 arguments for 14, 20–1, 139–40 formulations of 10, 13, 14, 141 Chang, R. 102, 109–13 comparability, complete 107

conditionalism about value 93–100 adjusted attitude objection to 96–7 arbitrariness objection to 95–6 consequentialism 6, 50–2, 55, 59, 73–6, 127 agent-relative 74 contextual value 93, 98–100 correctness 10, 32, 58n. 15, 121, 126n. 10, 150, 153–5 D’Arms, J. and Jacobson, D. 144n. 12, 151–3 Dancy, J. 33n. 5, 84–5, 87n. 2, 97n. 13, 98–100, 108, 138n. 5 default value 71, 99 deontology 6, 52, 54, 59n. 17, 76, 139n. 8 duty 8, 74, 136, 137n. 3, 139 Early Moorean Account 136–7, 141, 142 enabling conditions 30–1, 33n. 5, 39, 84–5, 94, 98, 119–20 error theory 18–19, 22n. 18 eudaimonia 27, 29 evaluative inadequacy 90–1 Ewing, A. C. 11, 19n. 12, 118 favouring attitudes 9, 60n. 18, 76, 90, 117–18, 145–9

170 Index

ability to favour 13, 68 comparative attitudes 105–6, 108, 113–14 degree/strength of 122–6 direction of 60, 119–22 kinds of 118–19, 126 fitting attitude account of value (FA) 10–15, 58, 83, 117, 125–6, 135, 157 and absolute value 57–61, 66n. 5 and agent–relative/agent– neutral value 73, 76–8 and cognitivism/ non-cognitivism 17–8 and conditional value 120–1 and conditionalism 9 and error theory 18–19 and final value 16, 25–39, 88, 119–22 and incomparability 104–6, 114 and naturalism/non-naturalism 19–22 and organic unities 88–90 and personal value 65–8, 72–3, 119–22 and thin/thick values 11n. 7, 152–3, 156 and virtual value 90–1 arguments for and against 13–15, 139, 141–3 as leading idea of the book 16, 132, 135 rival accounts to 136–41, 142, 156 fittingness and unfittingness 5, 10–11, 13–17, 19, 21, 28, 36–8, 58n. 15, 60–1, 83, 88–9, 119–22, 127, 133, 150–3, 156–7 compared to normative reasons 13–15

degrees of fittingness 122–6 fittingness and agents 13–15, 19, 22, 52, 60–1, 67–8, 72–3, 73–8, 107–8, 124 and morality 8–9, 10, 57–60, 76, 139n. 8, 144, 150, 151 and value of an attitude 66n. 5, 126 and universalizability 75, 127–8 Foot, P. 50–2, 57–9 for its own sake 27–9, 31–3, 36, 38, 40, 43, 58, 60, 72, 120–2 for someone’s sake 66–9n. 10, 72, 121–2, 143n. 11 friendship 30, 33–5, 38, 40, 41n. 9, 43, 70–1, 103n. 3, 105–6, 119–21, 129 Geach, P. 46–50, 53, 64 Gert, J. 108, 110–12 good as 45, 49, 53–4, 61n. 20 good as an end see value, final good for someone see value, personal good will 27–30, 32, 40–1 good-making properties 4, 8, 11n. 7, 12, 21n. 15, 33, 84, 94, 103–4, 137–8, 140, 142 happiness 27, 28, 29–30, 52 being worthy of 29–30, 40–1 hedonism 6–7, 28–9, 32, 98–100 holism vs. atomism 98–100 Hurka, T. 40, 41n. 9, 95–6, 126n. 9 incommensurability, overall 102–3, 104 incommensurability, pro tanto 103, 104 incomparability/ incommensurability 101–15

Index arguments against 107–8, 109 arguments for 102, 104–6 indifference (as a fitting attitude) 38, 90, 126 Kant, I. 27–30, 32, 38, 40, 131n. 16 Kantian conception of intrinsic value 130–3 Korsgaard, C. 19–20, 33n. 4 Kraut, R. 61n. 19 love 37, 105–6, 153–4 maximization of value 73–6, 127–30, 131–2 argument for 128–9 McDowell, J. 19–20, 156–7 meta-ethics 7, 17–22, 33n. 5, 50–1, 61 Mill, J. S. 28, 55, 98 monism and pluralism about value 6, 29, 30, 98, 99, 100, 103–4, 117, 126–7, 134, 140 Moore, G. E 20, 55 on good for 63–6, 69 on intrinsic value 31–2, 34 on organic unities 85–9 on value and fitting attitudes 58–9, 132n. 17, 136–7 Moorean conception of intrinsic value 32, 35, 85–8, 130–3 moral theory see normative theory Napoleon’s hat 34–9, 42, 119–20 naturalism 19–22, 58n. 15, 61, 83–4 Necessary Equivalence Accounts 137 with Redundancy 138, 141, 142 without Redundancy 138–41, 142

171

no good reasons principle 104–5 non-cognitivism 17–18 non-naturalism 19–22, 61 no-priority view 21, 156–7 normative reasons 8–9, 13–15, 60, 124, 137–40 agent-relative 75–6 compared to fittingness 13–15 holding-reasons vs. contentreasons 150 internalism about 21–2, 61 monism and pluralism about 140 object-given vs. state-given 143–4 peremptory vs. enticing 108 practical, evaluative, epistemic 145–7, 157 primary vs. secondary 147–50 reasons and ability 13 reasons and agents 14–15 reasons to believe, to judge 146–50, 157 shared 153–5 special or personal 67 see also reasons of fittingness; value and fitting attitudes/ normative reasons: general connection between, wrong kind of reasons Normative Redundancy Claim 12, 136, 137–41 normative theory 6–7, 52, 54–5, 58–9 normativity 6–7, 60–1, 83, 136–8, 157 value and 8–15, 20, 136–41, 157 Olson, J. 96, 98, 150 open question argument 20–1 organic unities 16, 85–90 attitude mismatch objection to 88–9

172 Index

normatively idle value objection to 89–90 ought 8, 10, 13, 76, 108, 136, 144 parity 109–14 and vagueness 109–10 argument for 109 objections to 110–12 particularism vs. generalism 99–100 Persson, I. 147–9 Pettit, P. 127–30 Plato 27n. 1 pleasure 6–7, 28–9, 31–2, 84–5, 89–92, 98–100, 103–4 sadistic 86, 89–92, 94–6 pluralism about value see monism and pluralism about value pluralism, methodological 17, 37, 133 Progress and Decline 92–3, 95n. 10 punishment 85–9, 94–7 Rabinowicz, W. 114, 143n. 11, 149 Raz, J. 102–3, 108, 126n. 10, realism 18–20 reasons of fittingness 14, 124, 141–5, 148, 150–2 Reduction Claim 12, 19, 136, 136–41, 156 Regan, D. 107–8 respect 37, 38, 69n. 10, 105, 126, 127–9 resultance 84, 91, 119, 121, 138, 141 rightness 8, 50, 59, 66n. 5, 136, 137n. 3 Rønnow-Rasmussen, T. 67n. 7, 69n. 11, 121–2, 143n. 11, 149

Ross, W. D. 29, 35n. 6, 41, 59n. 16, 139, 141 Scanlon, T. M. 9, 37 on buck-passing 10–11, 20–1, 138–40 on maximization 129 on well-being 69–70 Schroeder, M. 76n. 23, 153–5 sensibility theory see no-priority view Sidgwick, H. 31n. 2, 65–6 Skorupski, J. 146–7 Smith, M. 21, 58n. 15, 77n. 24 Stratton-Lake, P. 12, 143n. 11 supervenience 21n. 15, 81–5, 91, 102, 119, 122 thin/thick values and evaluations 2–6, 11n. 7, 152 Thomson, J. J. 53–8, 60n. 18 trichotomy thesis 101–2, 109 utilitarianism see consequentalism value absolute 45–61, 64–6, 72, 79, 136 arguments against 46–55, 61n. 19 defence of 56–61 attributive and predicative 3–4, 46–50, 56, 60n. 18, 64 essential 39, 44 extrinsic 31–2, 39–43, 44, 91, 119–21 argument for elimination of 41–3 final 25–44, 45, 50, 57n. 12, 58–61, 71–2, 75, 85–6, 90–4, 97–100, 119–22, 127, 129–33

Index arguments against final extrinsic value 35–9 exclusively final value 29, 41n. 9, 43 and extrinsic value 32–5, 119–21, 131n. 16 and intrinsic value 16–17, 26, 31–2, 35–9, 42, 44, 57n. 12, 71n. 17, 85–6, 89, 92–3, 119, 130, 133 and personal value 69–72, 97, 119–22 impersonal 51, 72–3, 77–9, 121 instrumental 3, 26, 28, 33, 41–3, 120–1 intrinsic 26, 31–2, 33nn. 4, 5, 35–7, 44, 56–7, 119, 130–3 and conditional value 39–41 isolation test for 31–2, 34, 36, 86 and organic unities 85–8 personal 3, 6, 45, 55, 56, 61n. 19, 63–79, 97, 121–2, 154 and agent–relative value 73–9 definitions of ‘good for’ 65–8 in relation to impersonal value 69–72 Moore’s argument against 63–4 unconditional and conditional 29–31, 33–4, 38–41, 43, 120–1 value and fitting attitudes/ normative reasons: general

173

connection between 8–10, 16, 71, 76, 89, 106, 107, 110–11, 117, 131, 140–1, 157 value for its own sake see for its own sake; value, final value judgements 1–6, 8, 18, 20, 99, 156–157 value pumps 113–14 value theory debates in 130–3 questions of 5, 7, 16, 22–3 three senses of 6–8 virtual value 90–3 virtue and vice 5, 6–7, 29, 40, 41n. 9, 52, 55, 58, 98, 126n. 9 ways of being good 53–6 welfare see well-being well-being 69–70, 72, 79, 122 see also value, personal Wiggins, D. 20, 156–7 Williams, B. 2, 21–2 wrong kind of reasons (WKR) 13, 14n. 11, 105n. 6, 135, 141–57 less of a problem for rivals to FA 142 WKR scenarios cruel joke 144 cheating in a game of chess 145 evil demon 13, 141, 143 father and daughter 144–5 Zimmerman, M. J. 36–8, 56–8, 67n. 6, 91–2, 119–20